encyclopedia of physical science and technology - energy

P1: GRB/LSX P2: ZCK Final Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology EN003H-867 June 13, 2001 20:13

Coal PreparationRobert A. MeyersRamtech Limited

Janusz S. LaskowskiUniversity of British Columbia

Anthony D. WaltersKilborn Engineering Limited

I. Coal Characteristics Related toCoal Preparation

II. Breaking, Crushing, and GrindingIII. Coal ScreeningIV. Float-and-Sink AnalysisV. Dense-Medium Separation

VI. Separation in a Water MediumVII. Flotation

VIII. Separation EfficiencyIX. Ancillary OperationsX. Coal Preparation FlowsheetsXI. On-Line AnalysisXII. Research into New Beneficiation Processes

GLOSSARY

Dense media (or heavy liquids) Fluids used in dense-medium separation of coal and gangue particles bytheir relative densities. The medium can be any suitablefluid, but in commercial coal preparation operations, itis usually a suspension of fine magnetite in water.

Density (gravity) separation Separation methods basedon differences in density of separated minerals, such asdense-medium separation and jigging.

Float-and-sink tests Tests carried out to determine coal

washability, by which the coal is separated into variousdensity fractions.

Floatability Description of the behavior of coal particlesin froth flotation.

Hardgrove grindability index Measure of the ease bywhich the size of a coal can be reduced. Values decreasewith increasing resistance to grinding.

Metallurgical coal Coal used in the manufacture of cokefor the steel industry.

Near-density material Percentage of material in the feedwithin ±0.1 density range from the separation density.

. 79

html

P1: GRB/LSX P2: ZCK Final


80 Coal Preparation

Organic efficiency (recovery efficiency) Measure ofseparating efficiency, calculated as

actual clean coal yield

theoretical clean coal yield× 100

with both yields at the same ash content.Probable error Ep (Ecart probable moyen) Measure

of separating efficiency of a separator (jig, cyclone,etc.). Calculated from a Tromp curve.

RRB particle size distribution Particle size distribu-tion developed by Rossin and Rammler and found byBennett to be useful in describing the particle size dis-tribution of run-of-mine coal.

Separation cut point δ50 Density of particles report-ing equally to floating and sinking fractions (heavymedia overflow and underflow in the dense mediacyclone). The δ50 is also referred to as partitiondensity.

Separation density δs Actual density of the densemedium.

Thermal coals Coals used as a fuel, mainly for powergeneration. These are lower-rank coals (high volatilebituminous, subbituminous, and liginites).

Tromp curve (partition curve, distribution curve)Most widely used method of determining graphi-cally the value of Ep as a measure of separationefficiency. Synonyms: partition curve, distributioncurve.

Washability curves Graphical presentation of the float-sink test results. Two types of plots are in use: Henry–Reinhard washability curves and M-curves.

Washing of coal (cleaning, beneficiation) Term denot-ing the most important coal preparation unit opera-tion in which coal particles are separated from inor-ganic gangue in the processes based on differences indensity (gravity methods) or surface properties (flota-tion). Cleaning then increases the heating value of rawcoal.

COAL PREPARATION is the stage in coal production—preceding its end use as a fuel, reductant, or conver-sion plant feed—at which the run-of-mine (ROM) coal,consisting of particles, different in size and mineralogi-cal composition, is made into a clean, graded, and con-sistent product suitable for the market; coal prepara-tion includes physical processes that upgrade the qualityof coal by regulating its size and reducing the contentof mineral matter (expressed as ash, sulfur, etc.). Themajor unit operations are screening, cleaning (wash-ing, beneficiation), crushing, and mechanical and thermaldewatering.

I. COAL CHARACTERISTICS RELATEDTO COAL PREPARATION

Coal, an organic sedimentary rock, contains combustibleorganic matter in the form of macerals and inorganic mat-ter mostly in the form of minerals.

Coal preparation upgrades raw coal by reducing its con-tent of impurities (the inorganic matter). The most com-mon criterion of processing quality is that of ash, which isnot removed as such from coal during beneficiation pro-cesses, but particles with a lower inorganic matter con-tent are separated from those with a higher inorganic mat-ter content. The constituents of ash do not occur as suchin coal but are formed as a result of chemical changesthat take place in mineral matter during the combustionprocess. The ash is sometimes defined as all elementsin coal except carbon, hydrogen, nitrogen, oxygen, andsulphur.

Coal is heterogeneous at a number of levels. At thesimplest level it is a mixture of organic and inorganicphases, but because the mineral matter of coal originatedin the inorganic constituents of the precursor plant, fromother organic materials, and from the inorganic compo-nents transported to the coal bed, its textures and libera-tion characteristics differ. The levels of heterogeneity canthen be set out as follows (Table I):

1. At the seam level, a large portion of mineral matter incoal arises from the inclusion during mining of roofor floor rock.

2. At the ply and lithotype level, the mineral matter mayoccur as deposits in cracks and cleats or as veins.

3. At the macerals level, the mineral matter may bepresent in the form of very finely disseminateddiscrete mineral matter particles.

4. At the submicroscopic level, the mineral matter maybe present as strongly, chemically bonded elements.

Even in the ROM coal, a large portion of both coal andshale is already liberated to permit immediate concentra-tion. This is so with heterogeneity level 1 and to someextent with level 2; at heterogeneity level 3 only crushingand very fine grinding can liberate mineral matter, while atlevel 4, which includes chemically bonded elements andprobably syngenetic mineral matter, separation is possibleonly by chemical methods.

Recent findings indicate that most of the mineral matterin coal down to the micron particle-size range is indeeda distinct separable phase that can be liberated by finecrushing and grinding.

The terms extraneous mineral matter and inherentmineral matter were usually used to describe an ash-forming material, separable and nonseparable from coal



Coal Preparation 81

TABLE I Coal Inorganic Impuritiesa

Type Origin Examples Physical separation

Strongly chemically From coal-forming organic Organic sulphur, nitrogen Nobonded elements tissue material

Adsorbed and weakly Ash-forming components in pure water, Various salts Very limitedbonded groups adsorbed on the coal surface

Mineral matter Minerals washed or blown into the peat Clays, quartz Partly separable by physical methodsa. Epiclastic during its formation

b. Syngenetic Incorporated into coal from the very Pyrite, siderite, some clay Intimately intergrown with coal maceralsearliest peat-accumulation stage minerals

c. Epigenetic Stage subsequent to syngenetic; migration Carbonates, pyrite, kaolinite Vein type mineralization; epigeneticof the minerals-forming solutions minerals concentrated along cleats,through coal fractures preferentially exposed during breakage;

separable by physical methods

a Adapted from Cook, A. C. (1981). Sep. Sci. Technol. 16(10), 1545.

by physical methods. Traditionally in coal preparation pro-cesses, only the mineral matter at the first and, to someextent, the second levels of heterogeneity was liberated,the rest remained unliberated and, left with the cleanedcoal, contributed to the inherent mineral matter. Recentvery fine grinding, which also liberates the mineral mat-ter at the third level of heterogeneity, has changed the oldmeanings of the terms inherent and extraneous. The con-tent of the “true” inherent part of ash-forming material(i.e., the part left in coal after liberating and removingthe mineral matter at the first, second, and third levels ofheterogeneity) is usually less than 1%.

In recent years, the emphasis in coal preparation hasbeen placed on reducing sulfur content of coal and on re-covering the combustible material. Sulfur in coal is presentin both organic and inorganic forms. The dominant formof inorganic sulfur is pyrite, but marcasite has also beenreported in many coals. Pyrite occurs as discrete particles,often of microscopic size. It comprises 30–70% of totalsulfur in most coals. Other forms of inorganic sulfur thatmay be present are gypsum and iron sulfates. The sulfatelevel in fresh unoxidized coals is generally less than 0.2%.

Organic sulfur in coal is believed to be contained ingroups such as

S

COAL

THIOPHENE

R S R′

ORGANIC SULFIDE

R SH

MERCAPTAN

R SS

ORGANICDISULFIDE

R′

COAL

The organic sulfur content in coals range from 0.5 to2.5% w/w.

Physical cleaning methods can remove inorganic sul-fates (gypsum) and most of the coarse pyrite; the finelydisseminated microcrystalline pyrite and organic sulfurare usually not separable by such processes. This meansthat in the case of coal containing 70% of sulfur in pyriticform and 30% as organic sulfur, the physical cleaning canreduce the sulfur content by about 50%.

II. BREAKING, CRUSHING,AND GRINDING

The primary objectives of crushing coal are

1. Reduction of the top size of ROM coal to make itsuitable for the treatment process

2. liberation of coal from middlings, and3. size reduction of clean coal to meet market

specification.

Size reduction of coal plays a major role in enablingROM coal to be used to the fullest possible extent forpower generation, production of coke, and production ofsynthetic fuels. ROM coal is the “as-received” coal fromthe mining process. Because the types of mining processesare varied, and size reduction actually begins at the facein the mining operation, it is quite understandable thatthe characteristics of the products from these various pro-cesses differ widely. The type of mining process directlyaffects the top size and particle size distribution of themine product.

During the beneficiation of coal, the problem of treatingmiddlings sometimes arises. This material is not of suf-ficient quality to be included with the high-quality cleancoal product, yet it contains potentially recoverable coal.If this material is simply recirculated through the cleaning



82 Coal Preparation

circuit, little or no quality upgrading can be achieved. Lib-eration can be accomplished if the nominal top size of thematerial is reduced, which permits recovery of the coalin the cleaning unit. Hammer mills, ring mills, impactors,and roll crushers in open or closed circuits can be used toreduce the top size of the middlings. Normally, reducingthe nominal top size to the range 20–6 mm is practiced.

Breaking is the term applied to size operations on largematerial (say +75 mm) and crushing to particle size re-duction below 75 mm; the term grinding covers the sizereduction of material to below about 6 mm. However, theseterms are loosely employed. A general term for all equip-ment is size reduction equipment, and because the termcomminution means size reduction, another general termfor the equipment is comminution equipment.

A. Breaking and Crushing

Primary breakers that treat ROM coal are usually rotarybreakers, jaw crushers, or roll crushers; some of these arelisted below.

Rotary Breakers (Fig. 1). The rotary breaker serves twofunctions—namely, reduction in top size of ROM and re-jection of oversize rock. It is an autogenous size-reductiondevice in which the feed material acts as crushing media.

Jaw Crusher. This type of primary crusher is usuallyused for crushing shale to reduce it to a size suitable forhandling.

Roll Crusher. For a given reduction ratio, single-rollcrushers are capable of reducing ROM material to a prod-uct with a top size in the range of 200–18 mm in a singlepass, depending upon the top size of the feed coal. Double-roll crushers consist of two rolls that rotate in opposite di-

FIGURE 1 Cutaway view of rotary breaker.

rections. Normally, one roll is fixed while the other roll ismovable against spring pressure. This permits the passageof tramp material without damage to the unit. The driveunits are normally equipped with shear pins for overloadprotection.

Hammer Mills. The swinging hammer mill (insteadof having teeth as on a single-roll crusher) has hammersthat are mounted on a rotating shaft so that they have aswinging movement.

B. Grinding

The most common grindability index used in conjunctionwith coal size reduction, the Hardgrove Grindability In-dex, is determined by grinding 50 g of 16 × 30 mesh driedcoal in a standardized ball and race mill for 60 revolutionsat an upper grinding ring speed of 20 rpm. Then the sampleis removed and sieved at 200 mesh to determine W , theamount of material passing through the 200 mesh sieve.The index is calculated from the following formula:

HGI = 13.6 + 6.93W. (1)

From the above formula, it can be deduced that asthe resistance of the coal to grinding increases, the HGIdecreases. The HGI can be used to predict the particle sizedistribution, that is, the Rossin–Rammler–Bennett curve[see Eq. (4)].

The distribution modulus (m) and the size modulus d63.2

must be known to determine the size distribution of aparticular coal.

It has been shown that for Australian coals, the distri-bution modulus can be calculated from the HGI by thefollowing equation:



Coal Preparation 83

HGI = 35.5m−1.54. (2)

The value of d63.2 is a function of the degree of break-age that the coal has undergone during and after mining.Having selected d63.2 and the value of m from the HGI,one can predict the size distribution.

For the use of coal at power stations and for treatmentby some of the newer beneficiation techniques (e.g., forcoal/water slurries), grinding is employed to further re-duce the top size and produce material with a given par-ticle size distribution. The principal equipment used forcoal grinding is the following:

1. air-swept ball mills,2. roll or ball-and-race type mills,3. air-swept hammer mills, and4. wet overflow ball mills.

III. COAL SCREENING

Sizing is one of the most important unit operations in coalpreparation and is defined as the separation of a heteroge-neous mixture of particle sizes into fractions in which allparticles range between a certain maximum and minimumsize. Screening operations are performed for the followingpurposes:

1. Scalping off large run-of-mine coal for initial sizereduction

2. Sizing of raw coal for cleaning in different processes3. Removal of magnetic (dense medium) from clean

coal and refuse4. Dewatering5. Separation of product coal into commercial sizes

The size of an irregular particle is defined as the smallestaperture through which that particle will pass.

In a practical screening operation the feed materialforms a bed on the screen deck (Fig. 2) and is subjectedto mechanical agitation so that the particles repeatedlyapproach the deck and are given the opportunity of passing

FIGURE 2 Four-mesh wire screen.

FIGURE 3 Passage of a spherical particle through a squareaperature.

through. In the simplest case (Fig. 3), a spherical particleof diameter d will only pass through if it does not touchthe sides of the aperture. The condition for passing is thatthe center of the sphere falls within the inner square, sidea − d. The probability P of passing is thus given by the ra-tio between the areas of the inner and outer squares, that is,

P = (a − d)2/a2 = (1 − d/a)2. (3)

The assumption that passage will be achieved onlyif there is no contact with the aperture sides is too re-strictive. They can and do collide with the screen deckwhile passing through. In particular, the following factorscontribute to the probability of passing:

1. Percentage open area2. Particle shape3. Angle of approach4. Screen deck area5. Bed motion6. Size distribution of feed.

To select the correct screen for an application in a coalpreparation plant, a detailed knowledge of the size dis-tribution of the feed is necessary. Size distributions areusually presented graphically, and it is useful to use astraight-line plot, because curve fitting and subsequentinterpolation and extrapolation can be carried out withgreater confidence. In addition, if a function can be foundthat gives an acceptable straight-line graph, the functionitself or the parameter derived from it can be used to de-scribe the size distribution. This facilitates data compari-son, transfer, or storage, and also offers major advantagesin computer modeling or control. The particle size dis-tribution used most commonly in coal preparation is theRossin–Rammler–Bennett,

F(d) = 100(1 − exp[(−d/d63.2)m], (4)

where F(d) is the cumulative percent passing on size d,d63.2 is the size modulus (d63.2 is that aperture throughwhich 63.2% of the sample would pass), and m is thedistribution modulus (the slope of the curve on the RRBgraph paper). A sample of the RRB plot is shown in Fig. 4.

In Fig. 4 one can read off the slope m; the scale forS · d63.2 is also provided. If d63.2 is the particle size (in



84 Coal Preparation

FIGURE 4 Rossin–Rammler–Bennett net with additional scales. Specific surface S is in square meters per cubicdecimeter, d63.2 in millimeters.

millimeters) belonging to F(d) = 63.2%, then the scalesfurnish the specific surface S in m2/dm3.

The types of screens used in coal preparation plantsgenerally fall into the following categories:

1. Fixed screens and grizzlies (for coarse coal).2. Fixed sieves (for fine coal). These are used for

dewatering and/or size separation of slurried coal orrejects. The sieve bend is the most commonly used incoal preparation plants for this application (Fig. 5).

3. Shaking screens. These are normally operated bycamshafts with eccentric bearings. They can bemounted horizontally or inclined, and operate at lowspeeds with fairly long strokes (speeds up to 300 rpmwith strokes of 1–3 in).

4. Vibrating screens. These are the most commonly usedscreens in coal preparation and are to be found invirtually all aspects of operations. A summary of theirapplication is shown in Table II. A recentdevelopment in vibrating screens has been the

introduction of very large screens (5.5 m wide, 6.4 mlong). Combined with large tonnage [1000 metrictons per hour (tph) per screen], screens have beendeveloped (banana screens) with varying slope: firstsection 34◦, second section 24◦, and third section 12◦.

5. Resonance screens. These have been designed to saveenergy consumption. The screen deck is mounted onflexible hanger strips and attached to a balance frame,which is three or four times heavier than the screenitself.

6. Electromechanical screens. This type of screenoperates with a high-frequency motion of very smallthrow. The motion is usually caused by a movingmagnet which strikes a stop.

A. Recent Developments

A new process for the screening of raw coal of highmoisture content at fine sizes has been developed by theNational Coal Board. Their Rotating Probability Screen



Coal Preparation 85

FIGURE 5 Sieve bend.

(Fig. 6) has the advantage that the effective screening aper-ture can be changed while the screen is running. The screen“deck” is made up of small diameter stainless steel rods ra-diating from a central hub. The hub is rotated, and the coalfalls onto the rotating spokes. The undersize coal passesthrough the spokes; the oversize coal is deflected overthem. The speed of rotation dictates the screening aper-ture. The ability to screen coal with high levels of moisturehas taken precedence over the accuracy of size separation.A screening operation by which the proportion of under-flow product can be controlled while the machine is inoperation represents an important advance in the prepara-tion of blended coals in treatment plants where the finesare not cleaned.

IV. FLOAT-AND-SINK ANALYSIS

Most coal cleaning processes that are used to removeinorganic impurities from coal are based on the gravityseparation of the coal from its associated gangue. ROM

coal consists of some proportion of both coal and shaleparticles, already sufficiently liberated, together with coalparticles with inclusions of gangue (i.e., bands of shale).Commercially cleaned coal contains only very dissemi-nated impurities and has a density ranging from 1.2 to1.6. Carbonaceous shale density ranges from 2.0 to 2.6,and pure shale, clay, and sandstone have a density ofabout 2.6. The density of pyrite is about 5.0. The dif-ference in density between pure coal and these impurities,in a liberated state, is sufficient to enable an almost com-plete separation to be achieved fairly easily. However, ithas been shown that the inorganic impurity content, andhence, the ash content, ranges from pure coal containingonly microscopic impurities to shale, which is entirelyfree from carbonaceous matter. Generally speaking, themineral matter content of any coal particle is proportionalto its density and inversely proportional to its calorificvalue.

For the prediction of concentration results (design offlowsheet) and for the control of preparation plant opera-tions, a measure of concentrating operations is needed.



86 Coal Preparation

TABLE II Vibrating-Screen Applications in Coal Preparation Plants

Number InstallationType of decks angle Aperturea Screen deck type Accessories

Run-of-mine scalper Single 17◦–25◦ 6 in. Manganese skid bars, Feed box with liners, extra highAR perforated plate side plates, drive guardwith skid bars enclosures

Raw-coal sizing screen Double 17◦–25◦ 1 in. AR steel perforated Dust enclosures,drive plate,polyurethane, rubber guard enclosures

516 in. Polyurethane, wire, 304 Feed box with liners

stainless steel profiledeck, rubber

Pre-wet screen Double Horizontal 1 in. Wire, polyurethane, rubber Water spray bar, side platedrip angles, drive guard

1 mm Stainless steel profile deck, enclosures, feed box linerspolyurethane

Dense-medium drain and Double Horizontal 1 in. Wire, polyurethane, rubber Side plate drip angles, spray bars,rinse screen (coarse coal) shower box cross flow screen

1 mm 304 stainless steel profile or sieve bend, drip lip angles,deck, polyurethane drive guard enclosures

Dewatering screen Single Horizontal 1 mm 304 stainless steel profile Sieve bend or cross flow screen,(coarse coal) deck, polyurethane dam, discharge drip lip angles,

drive guard enclosures

Desliming screen Single Horizontal 0.5 mm 304 stainless steel profile Sieve bend or cross flow screen,deck, polyurethane spray bars, shower box, drive

guard enclosures

Classifying screen Single 28◦ 100 mesh Stainless steel woven wire Three-way slurry distributor and(fine coal) sandwich screens feel system

Dense-medium drain and Single Horizontal 0.5 mm 304 stainless steel profile Sieve bend or cross flow screen,rinse screen (fine coal) deck, polyurethane spray bars, shower box, drip lip

angles, drive guard enclosures

Dewatering screen (fine coal) Single Horizontal 0.5 mm 304 stainless steel profile Sieve bend or cross flow screen,or 27◦–29◦ deck or woven wire, dam, drip lip angles, drive

rubber, polyurethane guard enclosures

a Typical application.

The best known means of investigating and predictingtheoretical beneficiation results are the so-called washabil-ity curves, which represent graphically the experimentalseparation data obtained under ideal conditions inso-called float–sink tests. Float-and-sink analysis is alsoused to determine the Tromp curve, which measures thepractical results of a density separation. The practical re-sults of separation can then be compared with the idealand a measure of efficiency calculated.

The principle of float–sink testing procedure is asfollows: A weighed amount of a given size fraction isgradually introduced into the heavy liquid of the lowestdensity. The floating fraction is separated from the fractionthat sinks. The procedure is repeated successively withliquids extending over the desired range of densities. Thefraction that sinks in the liquid of highest density is alsoobtained. The weight and ash contents of each densityfraction are determined.

In the example shown in Fig. 7, five heavy liquids withdensities from 1.3 to 1.8 are used. The weight yields of sixdensity fractions are calculated (γ1, γ2, . . . , γ6), and theirash contents are determined (λ1, λ2, . . . , λ6). The resultsare set out graphically in a series of curves referred to aswashability curves (Henry–Reinhard washability curvesor mean-value curve, M-curve, introduced by Mayer).

The construction of the primary washability curve(Henry–Reinhard plot) is shown in Fig. 8. It is noteworthythat the area below the primary curve (shaded) representsash in the sample. The shaded area, changed into the rect-angle of the same surface area, gives the mean-ash contentin the sample (α = 16.44% in our example).

The shape of the primary curve reveals the proportionsof raw coal within various limits of ash content and soshows the proportions of free impurities and middlingspresent. Because the relative ease or difficulty of cleaninga raw coal to give the theoretical yield and ash content



Coal Preparation 87

FIGURE 6 Rotating probability screen.

depends on the proportion of middlings, the shape of theprimary curve also indicates whether the coal is easy ordifficult to clean.

The construction of the mean-value curve (M-curve),also referred to as Mayer’s curve, is shown in Fig. 9. Thepoint where the curve intersects the abscissa gives theaverage ash content of the raw coal (α).

The shape of the primary washability curve and theM-curve is an indication of the ease or difficulty of clean-ing the coal. The more the shape approximates the letter

FIGURE 7 Float–sink analysis procedure.

L the easier the cleaning process will be. (This is furtherillustrated in Fig. 10.)

Figure 10 shows four different cases: (a) ideal separa-tion, (b) easy cleanability, (c) difficult cleanability, and(d) separation impossible.

The primary washability curve for the difficult-to-cleancoal (c) exhibits only a gradual change in slope revealinga large proportion of middlings.

It can be seen from the example that the further theM-curve is from the line connecting the zero yield point



88 Coal Preparation

FIGURE 8 Construction of the primary washability curve (Henry–Reinhard washability diagram).

FIGURE 9 Construction of the M-value curve.



Coal Preparation 89

FIGURE 10 Primary washability and M-value curves for (a) ideal separation, (b) easy-to-clean coal, (c) difficult-to-clean coal, and (d) impossible separation.

with α on the λ axis, the greater the range of variationof ash content in the coal. The more gradual change inslope of an M-curve indicates more difficult washabilitycharacteristics.

The washability curve can be used in conjunction withthe specific-gravity-yield curve (curve D) and the ±0.1near-density curve (curve E) as demonstrated in the ex-ample given in Table III and plotted in Fig. 11.

Curve A is the primary washability curve. Curve B isthe clean-coal curve and shows the theoretical percent ashof the clean-coal product at any given yield. Curve C (thecumulative sink ash) shows the theoretical ash content ofthe refuse at any yield. Curve D, plotted directly from thecumulative percent yield of floats versus density, givesthe yield of floats at any separation density. Curve E, thecurve of near-density material, gives the amount of ma-

TABLE III Set of Float–Sink Test Results

±0.1 Specific gravitydistributionCumulative

Direct floats Cumulative sinksWeight Cumulative Sink

Specific of ash weight weight Specific Yieldgravity Yield γ Ash λ of total of ash Yield Ash of ash Yield Ash gravity (%)

(g/cm3) (1) (wt. %) (2) (%) (3) (%) (4) (%) (5) (%) (6) (%) (7) (%) (8) (%) (9) (%) (10) (g/cm3) (11) (12)

<1.3 46.0 2.7 1.24 1.24 46.0 2.7 15.2 54.0 16.44 — —

1.3–1.4 17.3 7.3 1.26 2.50 63.3 3.9 13.94 36.7 28.22 1.4 63.3

1.4–1.5 11.7 16.3 1.87 4.37 75.0 5.8 12.07 25.0 38.00 1.5 29.0

1.5–1.6 10.0 26.7 2.67 7.04 85.0 8.3 9.4 15.0 48.28 1.6 14.15

1.6–1.8 8.3 46.0 3.82 10.86 93.3 11.6 5.58 6.7 62.66 1.7 8.3

>1.8 6.7 83.3 5.58 16.44 100.0 16.44 — — 83.29 1.8 10.85

100.0 16.44

terial within ±0.1 specific-gravity unit, and is used as anindication of the difficulty of separation. The greater theyield of the±0.1 fraction, the more difficult will be the sep-aration at this separation density. For the example given,the separation process, as indicated by the ±0.1 densitycurve, will be easier at higher densities than at lower ones.

V. DENSE-MEDIUM SEPARATION

The use of liquids of varying density in a simple float–sinktest leads to separation of the raw coal into different den-sity fractions. In a heavy liquid (dense medium), particleshaving a density lower than the liquid will float, andthose having a higher density will sink. The commercialdense-media separation process is based on the same prin-ciple. Since the specific gravity of coal particles varies



90 Coal Preparation

FIGURE 11 Complete set of washability curves (for data given in Table III).

from about 1.2 g/cm3 for low-ash particles to about 2.0g/cm3 for the high-ash particles, the liquids within thisrange of density provide conditions sufficient for theheavy-medium separation of coals.

Of the four types of dense media that can be consideredin such a process—namely, organic liquids (carbon tetra-chloride, bromoform), aqueous solutions (ZnCl2, CaCl2),aqueous suspensions of high-density solids (ferrosilicon,magnetite, barite, quartz sand), and air fluidized bed sus-pension (sand)—the first two are used only in laboratorywashability studies, and only the third has found wide in-dustrial applications.

The main difference between the former two and thelatter two is stability: The first two are homogeneous liq-uids, while the latter are composed of fine solid particlessuspended in water (or air), and as such are highly un-

TABLE IV Classification of Dense-Medium Separatorsa

Static

Shallow Deep

Separation Separation Separation Hydraulic Separation Mechanicalproducts products products removal of products removal of

Rotating-drum removal with removal by removal with Separation removal by Separationseparates paddles belt conveyor scraper chains products airlift products

Examples

Wemco Link belt; SKB-Teska; Ridley–Sholes bath; Tromp shallow; McNally static Wemco cone; Chance; Barvoys;drum Neldco (Nelson– Daniels bath Dutch State bath; Potasse Humboldt Tromp

Davis); Disa; Mines bath; d’AlsaceNorwalt; Drewboy McNally Tromp

a After Laskowski, T. (1958). “Dense Medium Separation of Minerals,” Wyd. Gorn. Hutnicze, Katowice. Polish text.

stable. Magnetite has now become the standard industrialdense medium.

To achieve densities in the range 1.5–1.9, the concen-tration of magnetite in water must be relatively high.At this level of concentration, the suspension exhibitsthe characteristics of a non-Newtonian liquid, resultingin a formidable task in characterizing its rheologicalproperties.

The principles of separation in dense media are depictedin Fig. 12, which shows a rotating drum dense-mediumseparator. The classification of static dense-mediumseparators offered 40 years ago by T. Laskowski is re-produced in Table IV.

Theoretically, particles of any size can be treated in thestatic dense-medium separators; but practically, the treat-able sizes range from a few millimeters to about 150 mm.



Coal Preparation 91

FIGURE 12 Principles of separation in dense media.

Since this process is unable to treat the full size range ofraw coal, it is most important to remove the fines from thefeed, because their presence in the circuit can increase themedium viscosity and increase the loss of media per tonof treated coal. The loss of magnetite in the coarse-coalcleaning is usually much below 1 kg/metric ton of washedproduct.

Dense-media static separators have a capability of han-dling high throughputs of up to 1000 metric ton/h.

Dynamic separators, which are the most efficientdevices for cleaning intermediate size coal (50 downto 0.5 mm), include the dense-medium cyclone, two-and three-product Dynawhirlpool separators, the Vorsyldense-medium separator, and the Swirl cyclone.

The most common are dense-medium cyclones (DMS)(Fig. 13a). Because the raw coal and medium are intro-duced tangentially into the cyclone, and the forces actingon the particles are proportional to V 2/r , where V isthe tangential velocity and r the radius of the cylindri-cal section, the centrifugal acceleration is about 20 timesgreater than the gravity acceleration acting upon particlesin the static dense-medium separator (this acceleration ap-proaches 200 times greater than the gravity acceleration atthe cyclone apex). These large forces account for the highthroughputs of the cyclone and its ability to treat fine coal.

A DMS cyclone cut point δ50 is greater than the densityof the medium and is closely related to the density of thecyclone underflow. The size distribution of the magnetiteparticles is very important for dense-medium cyclones,and magnetite that is 90% below 325 mesh (45 µm) was

FIGURE 13 Cyclones in coal preparation (a) dense-mediumcyclone; (b) water-only cyclone; (c) water-compound cyclone(Visman tricone).

found to provide the best results. Even finer magnetite(50% below 10 µm) was found to be essential for treat-ing −0.5 + 0.075 mm coal. In treating such fine coal ina DMS, which technically is quite possible, the mainproblem that still awaits solving is medium recovery atreasonable cost.

Innovative dynamic separators are now being intro-duced that are able to treat the full size range (100–0.5 mm) in a single separating vessel. The Larcodems(LArge, COal, DEnse, MEdium, SEparator) is basically acylindrical chamber that is mounted at 30◦ to the hori-zontal. Feed medium is introduced under pressure by aninvolute inlet at the bottom, and the raw coal enters sepa-rately through an axial inlet at the top. Clean coal is dis-charged through an axial inlet at the bottom end and rejectexpelled from the top via an involute outlet connected toa vortextractor.

The Tri-flo separator is a large-diameter three-productseparator similar in operation to a Dyna Whirlpool.

A. Dense Media

Any suitable fluid can be used as a dense medium, but onlyfine solid particles in water suspensions have found wideindustrial applications. A good medium must be chemi-cally inert, and must resist degradation by abrasion andcorrosion, have high inherent density, be easily recoveredfrom the separation products for reuse, and be cheap.

There is a close relationship between the medium den-sity and viscosity: the lower the medium density (i.e., thesolids content in suspension), the lower its viscosity. This,in turn, is related to medium stability (usually defined asthe reciprocal of the settling rate) through the size andshape of solid particles: the finer the medium particles,the lower the settling rate (hence, the greater stability),but the higher the viscosity. Medium recovery is easier forcoarser medium particles. Thus the medium density can-not be changed without the viscosity and stability beingaffected. For constant medium density, a higher specificgravity solid used as the medium reduces solid concen-tration in the dense medium and then reduces viscosityand stability; both are also affected by the selection of aspherical medium type (i.e., granulated ferrosilicon).

The term viscosity, as used above, is not accurate. Densemedia at higher solid concentrations exhibit characteris-tics of non-Newtonian liquids, namely, Bingham plasticor pseudo-plastic behavior, as shown in Fig. 14.1

1The use of the modified rheoviscometer, which enables handlingof unstable mineral suspensions, has recently revealed that the CassonEquation fits the flow curve for the magnetite suspension better thanthe typically used Bingham plastic model [see the special issue of CoalPreparation entirely devoted to magnetite dense media: [Coal Prepara-tion (1990). 8(3–4).]



92 Coal Preparation

FIGURE 14 Rheological curves for Newtonian and pseudo-plastic liquids.

As the rheological curves for magnetite dense-mediumshow, for such systems the shear stress is not proportionalto the shear rate; therefore, such systems are not charac-terized by one simple value of viscosity, as in the case ofNewtonian liquids.

The plastic and pseudo-plastic systems are describedby the Bingham equation,

τ = τB + ηpl D, (5)

and so both values τ0 (τB) and ηpl, which can be obtainedonly from the full rheological curves, are needed to char-acterize dense-medium viscosity.

Figure 15 shows the effect of magnetite particle size andconcentration on the medium rheology. As seen, the finer

FIGURE 15 Rheological curves for magnetite dense media. Solidline, magnetite particle size 100% below 200 µm; broken line,magnetite size 100% below 10 µm. [Adapted from Berghofer, W.(1959). Bergbauwissenschaften, 6(20), 493.]

magnetite suspension exhibits much higher viscosity. Itis also seen that while magnetite dense-medium behavesas a Newtonian liquid at magnetite concentration lowerthan 15% by volume, it is clearly pseudoplastic at higherconcentrations.

In a heavy-medium separation process, coal particleswhose density is higher than the medium sink while thelower-density coal particles float; the separation efficiencyand the through put of the separation device depend on thevelocity of coal particles in a dense medium. The viscosityof a medium has little effect on the low-density coal par-ticles or the high-density gangue particles, but becomescritical in the separation of material of a density equal toor near that of the medium; hence, a low viscosity mustbe maintained to separate near-density material at a highrate of feed.

The efficiency of separation (Ep) in dense-mediumbaths was found to depend on the plastic viscosity ofthe media, and a high yield stress (τ0) of the mediumis claimed to cause elutriation of the finer particles intothe float as they (and near-density particles) are unable toovercome the threshold shear stress required before themovement takes place. When a particle is held in a sus-pension, the yield stress is responsible for it; but when itis moving, its velocity is a function of plastic viscosity.It is understandable then that dispersing agents used todecrease medium viscosity may improve separation effi-ciency quite significantly. They may also decrease mag-nitite loses.

As the research on coal/water slurries shows, the sus-pension viscosity can also be reduced by close control ofthe particle size distribution; bimode particle distributionsseem to be the most beneficial.

VI. SEPARATION IN A WATER MEDIUM

A. Jigs

Jigging is a process of particle stratification in which theparticle rearrangement results from an alternate expansionand contraction of a bed of particles by pulsating fluid flow.The vertical direction of fluid flow is reversed periodically.Jigging results in layers of particles arranged by increasingdensity from the top to the bottom of the bed.

Pulsating water currents (i.e., both upward and down-ward currents) lead to stratification of the particles. Thefactors that affect the stratification are the following:

1. Differential Acceleration. When the water is at itshighest position, the discard particles will initially fallfaster than the coal.

2. Hindered Settling. Because the particles are crowdedtogether, they mutually interfere with one another’s



Coal Preparation 93

settling rate. Some of the lighter particles will thus beslowed by hindrance from other particles. This hinderedsettling effect is an essential feature, in that it helps to makeseparation faster; once the separation has been achieved,it helps to ensure that the material remains stratified.

3. Consolidation Trickling. Toward the end of thedownward stroke, the particles are so crowded togetherthat their free movement ceases, and the bed begins tocompact. The further settling of the particles that leadsto this compaction is referred to as consolidation. Thelarger particles lock together first, then the smaller par-ticles consolidate in the interstices. Therefore, the largerparticles cease to settle first, while the smaller particles cantrickle through the interstices of the bed as consolidationproceeds.

1. Discard Extraction Mechanism

The method of discard extraction varies in different typesof jig. Discard is discharged into the boot of the bucketelevator, and the perforated buckets carry it farther out.Some fine discard passes through the screen deck and istransported to the boot by screw conveyor, but the majorportion is extracted over the screen plate. There are manymethods of automatic control of discard.

2. Baum Jig

The most common jig used in coal preparation is the Baumjig, which consists of U-shaped cells (Fig. 16a). One limb

FIGURE 16 Comparison of (a) Baum and (b) Batac jigs. [After Williams, D. G. (1981). In “Coal Handbook” (Meyers,R. A. ed.), p. 265, Marcel Dekker, New York.]

Type Baum jig Batac jig

Year introduced 1892 1965

Present width 2.1 m 7.0 m

Present area 8.5 m2 42.0 m2

Present capacity 450 metric tons/hr 730 metric ton/hr.

of the cell is open and contains the perforated deck; theother terminates in a closed chamber, to and from which airis repeatedly admitted and extracted by rotary valves cou-pled to an air compressor. Particle separation is effectedon the perforated deck.

3. Batac Jigs

For many years, jig technology was largely unchanged(i.e., Baum type), except for advances in the design of airvalves and discard extraction mechanisms. A new concept,the Batac jig, in which the air chamber is placed beneaththe screen plate, was developed initially in Japan and laterin Germany. The principal feature of the design (Fig. 16b)is that instead of the box being U-shaped, the air cham-bers are placed beneath the washbox screen plates. In thisway, the width of the jig can be greatly increased and itsperformance improved by more equal jigging action overthe width of the bed.

Jigs traditionally treat coals in the size range 100–0.6 mm; however, for optimum results, coarse and finecoals should be treated in separate jigs. Fine-coal jigs havetheir own “artificial” permanent bed, usually feldspar.

B. Concentrating Tables

A concentrating table (Fig. 17) consists of a riffled rubberdeck carried on a supporting mechanism, connected to ahead mechanism that imparts a rapid reciprocating motion



94 Coal Preparation

FIGURE 17 Distribution of concentrating table products by par-ticle size and specific gravity. •, coal; �, middling; ©, refuse.

in a direction parallel to the riffler. The side slope of thetable can be adjusted. A cross flow of water (dressingwater) is provided by means of a launder mounted alongthe upper side of the deck. The feed enters just ahead ofthe water supply and is fanned out over the table deck bydifferential motion and gravitational flow. The particlesare stratified in layers by three forces:

1. Friction or adhesion between deck and particle2. Pressure of the moving water3. Differential acceleration due to the table action.

The clean coal overflows the lower side of the table, andthe discard is removed at the far end. The device cleansefficiently over the size range 5–0.5 mm.

C. Water-Only Cyclones

The water-only cyclone (WOC) (first developed by theDutch State Mines) is a cylindro-conical unit with an in-cluded apex angle up to 120◦ (Fig. 13b). Unlike the dense-medium cyclone, it uses no medium other than water; forthis reason it is sometimes referred to as the autogenouscyclone. Geometrically, the principal differences from thedense-medium cyclone are the greater apex angle and themuch longer vortex finder of the water-only cyclone.

The compound water cyclone (Visman Tricon)(Fig. 13c) is a variation of the water-only cyclone. Its dis-tinguishing features are the three sections of the cone partwith different apex angles, the first being 135◦, the second75◦, and the third 20◦.

Water-only cyclones do not make a sharp separation,but are commonly used as a preconcentrator or “rougher.”Their efficiency can be increased by the addition of a sec-ond stage. There is also a sizing effect with water-onlycyclones, and particles finer than 100 µm tend to report tothe overflow regardless of ash content or relative density.

D. Spirals

Spiral separators have been used for many years, sincethe introduction of the Humphrey unit. The early spiralshad simple profiles and were used in easy separations.Later development of Reichert spirals, suitable for a widevariety of applications, increased the use of spirals.

The effective size range for coal spiral operation(−3 mm +0.075 µm) coincides largely with the sizes thatlie between those most effectively treated by heavy-mediacyclones and those best treated by froth flotation. Thus,the principal areas of application might be substitution forwater-only cyclones, for fine heavy-media cyclone sepa-ration, and for coarse froth flotation.

The capacity of Reichert coal spirals is about 2 tph.Ep values for Reichert Mark 9 and Mark 10 spirals haverecently been reported in the range of 0.14–0.18, but theywere found to be much lower for the combined productplus middlings than for the product alone. Reichert spiralsoffer simple construction requiring little maintenance pluslow capital cost and low operating costs. Reichert-typeLD Mark 10 and Vickers spirals have recently been shownto be able to achieve lower Ep values for the finer fractions(0.1 mm) but higher Ep for the coarse fractions (1 mm)than water-only cyclone. In general, the spiral is lessaffected by a change in particle size than WOC, but mostimportantly, the spirals tolerate increasing amounts of clayin the feed with little or no change in separation efficiency,in contrast to the WOC performance, which deteriorateswith increasing amounts of clays. This indicates that thespiral is less influenced by increase in viscosity and isbetter suited for raw coals with significant proportions ofclays.

VII. FLOTATION

Typically, coal fines below 28 mesh (0.6 mm) are cleanedby flotation. In some plants the coarser part of such a feed[+100 mesh (150 µm)] is treated in water-only cyclones,with only the very fine material going to flotation.

The coal flotation process is based on differences insurface properties between hydrophobic low-ash coalparticles and hydrophilic high-ash gangue. Coals of dif-ferent rank have various chemical composition and phys-ical structure, and therefore their surface properties andfloatability change with coalification. While metallurgicalcoals float easily and may require only a frother, flotationof lower-rank subbituminous coal and lignites (as well ashigh-rank anthracites) may be very difficult. In such a pro-cess one may need not only large amounts of oily collectorbut also a third reagent, the so-called promoter (Table V).



Coal Preparation 95

TABLE V Coal Flotation Reagentsa

Type Examples Remarks

Collectors Insoluble in water, oily hydrocarbons, Used in so-called emulsion flotation of coal, in which collectorkerosene, fuel oil droplets must attach to coal particles

Frothers Water-soluble surfactants; aliphatic To stabilize froth; adsorb at oil/water interface and onto coal;alcohols; MIBC have some collecting abilities

Promoters Emulsifiers Facilitate emulsification of oily collector and the attachment of thecollector droplets to oxidized and/or low-rank coal particles

Depressants/dispersants Organic colloids: dextrin, Adsorb onto coal and make it hydrophiliccarboxymethyl cellulose, etc.

Inorganic salts NaCl, CaCl2, Na2SO4, etc. Improve floatability; in so-called salt flotation process may beused to float metallurgical coals, even without any organicreagents; are coagulants in dewatering.

a Adapted from Klassen, V. I. (1963). “Coal Flotation,” Gogortiekhizdat, Moscow. (In Russian.)

The surface of coal is a hydrophobic matrix that con-tains some polar groups as well as hydrophobic inorganicimpurities. Coal surface properties are determined by

1. The coal hydrocarbon skeleton (related to the rank)2. The active oxygen content (carboxylic and phenolic

groups)3. Inorganic matter impurities

Only the third term, the content of inorganic matter, isrelated to coal density. Therefore, in general there is no re-lationship between coal washability and coal floatability;such a relationship can be observed only in some particularcases.

Results obtained by various researchers agree that themost hydrophobic are metallurgical, bituminous coals.The contact angle versus volatile matter content curve,which shows wettability as a function of the rank (Fig. 18),is reproduced here after Klassen. As seen, lower-rankcoals are more hydrophilic, which correlates quite wellwith oxygen content in coal, shown here after Ihnatowicz(Fig. 19). Comparison of Fig. 18 with Fig. 19 also indicatesthat the more hydrophilic character of anthracites cannotbe explained on the same basis.

FIGURE 18 Effect of the rank on coal wettability. [From Klassen,V. I. (1963). “Coal Flotation,” Gosgortiekhizdat, Moscow. (InRussian).]

Two types of reagents are traditionally used in the flota-tion of coal:

Water-insoluble, oily hydrocarbons, as collectorsWater-soluble surfactants, as frothers

Insolubility of the collector in coal flotation requiresprior emulsification or long conditioning time. On theother hand, the time of contact with the frother shouldbe as short as possible to avoid unnecessary adsorption offrother by highly porous solids, such as coal.

In the flotation of coal, the frother (a surface-activeagent soluble in water) adsorbs not only at the water/gasinterface but also at the coal/water and the oil/water inter-faces; it may facilitate, to some extent, the attachment of

FIGURE 19 Oxygen functional groups in coals. (Large O standfor oxygen.) [From Ihnatowicz, A. (1952). Bull. Central ResearchMining Institute, Katowice, No. 125. (In Polish).]



96 Coal Preparation

FIGURE 20 Effect of surface-active agent (frother) on the flota-tion of coal with water-insoluble oily collector. Curve 1 is for flota-tion with kerosene; curve 2 for kerosene and n-octyl alcohol. [FromMelik-Gaykazian, V. I., Plaksin, I. N., and Voronchikhina, V. V.(1967). Dokl. Akad. Nauk SSSR, 173, 883.]

oily droplets to coal particles. This can lead to a substantialimprovement in flotation, as shown in Fig. 20.

In order to float oxidized and/or low-rank coals, newreagents called promotors have recently been tested andintroduced into the industry. These are emulsifying agentsthat facilitate emulsification of the oily collector and theattachment of oily droplets to coal particles.

The reduction of sulfur is one of the principal benefitsof the froth flotation process. Of the two types of sulfur incoal (i.e., inorganic and organic), only the inorganic sul-fur (mainly represented by pyrite) can be separated fromcoal by physical methods. The floatability of coal-pyrite is,however, somewhat different from that of ore-pyrite, andits separation from coal presents a difficult problem. Theresearch based on the behavior of ore-pyrite in the flota-tion process—namely, its poor flotability under alkalineconditions—did not result in a development of a coal-pyrite selective flotation process. The two-stage reverseflotation process proved to be much more successful. In

FIGURE 21 Fine-coal cleaning circuit.

this process, the first stage is conventional flotation withthe froth product comprising both coal and pyrite. Thisproduct is reconditioned with dextrin, which depressescoal in the second stage, and is followed by flotation ofpyrite with xanthates.

The fine-coal cleaning circuits in the newest plantsfrequently comprise water-only cyclones and flotation(Fig. 21). Such an arrangement is especially desirable incleaning high-sulfur coals, because the difference in spe-cific gravity between coal particles (1.25–1.6) and pyrite(5) is extremely large, and flotation does not discrimi-nate well between coal and pyrite. Advantages of suchan arrangement are clearly seen in Table VI, quoted afterMiller, Podgursky, and Aikman.

VIII. SEPARATION EFFICIENCY

The yield and quality of the clean-coal product from an in-dustrial coal preparation plant and the theoretical yield andquality determined from washability curves are known tobe different. In the ideal cleaning process, all coal parti-cles lower in density than the density of separation wouldbe recovered in the clean product, while all material ofgreater density would be rejected as refuse. Under theseconditions the product yield and quality from the actualconcentration process and the yield and quality expectedfrom the washability curves would be identical.

The performance of separators is, however, never ideal.As a result, some coal particles of lower than the separationdensity report to rejects, and some high-ash particles ofhigher than the separation density report to clean coal.These are referred to as misplaced material.

Coal particles of density well below the density of sep-aration and mineral particles of density well above thedensity of separation report to their proper products:clean coal and refuse. But as the density of separationis approached, the proportion of the misplaced materialreporting to an improper product increases rapidly.

Tromp, in a study of jig washing, observed that thedisplacement of migrating particles was a normal or



Coal Preparation 97

TABLE VI Effectiveness of Hydrocycloning and Flotation in Ash andSulfur Removala

Ash removal (%) Sulfur removal (%)

Particle size, Flotation Flotationmesh Hydrocyclone (single stage) Hydrocyclone (single stage)

28 × 100 60–65 50–60 70–80 50–60

100 × 200 40–45 40–45 60–70 30–40

200 × 325 15–18 40–45 35–40 25–30

325 × 0 0–5 50–55 8–10 20–25

a After Miller, F. G., Podgursky, J. M., and Aikman, R. P. (1967). Trans. AIME 238,276.

near-normal frequency (gaussian curve), and from this ob-servation the partition curve (distribution, Tromp curve)in the form of an ogive was evolved.

The partition curve, the solid line in Fig. 22a, illustratesthe ideal separation case (E p = 0), and the broken-linecurve represents the performance of a true separating de-vice. The shaded areas represent the misplaced material.

FIGURE 22 (a) Partition curve plotted on linear graph paper and(b) its anamorphosis plotted on a probability net.

The curves are plotted according to European conventionand represent the percent of feed reporting to reject. InAmerican practice, Tromp curves usually give the percentof feed reporting to washed coal.

The Tromp curve from the mathematical point of viewis a cumulative distribution curve and as such can belinearized on probability graph paper. Such anamorpho-sis is produced by plotting the partition coefficients on aprobability scale versus specific gravity on a linear scale(Fig. 22b) for dense-media separation, and versus log(δ − 1) for jigs.

To determine the partition curve for a cleaning opera-tion, one needs the yield of clean coal from this opera-tion and the results of float–sink tests for both products—that is, for the clean coal and the refuse. Such data al-low the reconstituted feed to be calculated, and from thiscan be found the partition coefficients, which give thepercentage of each density fraction reporting to reject.As seen in Fig. 22b, the particles with densities belowδ50 − 4Ep and the particles of density above δ50 + 4Ep

report entirely to their proper products. The density frac-tions within δ50 ± 4Ep are misplaced. Material of densityvery close to δ50 (near-density material) is misplaced themost. As postulated by Tromp, 37.5% of fractions withinδ50–δ25 and δ75–δ50 are misplaced (this corresponds toδ50 ± Ep), and this percentage falls off drastically withthe distance of the actual density fraction from the δ50

density.Figure 23 shows partition curves for the major U.S.

coal-cleaning devices. As seen, the sharpness of separa-tion in dense-media separators is much better than in jigsor water-only hydrocyclones. Figure 24 shows Ep val-ues plotted versus the size of treated particles for variouscleaning devices. Ep values for the dense-medium bathand dense-medium cyclone are in the range of 0.02–0.04for particles larger than 5 mm; Ep values of jigs and con-centrating tables are in the range of 0.08–0.15; and forwater-only cyclones, Ep values exceed 0.2. As seen, in allcases the efficiency of separation as given by Ep valuesdecreases sharply for finer particles.



98 Coal Preparation

FIGURE 23 Performance of gravity separators. �, hydrocy-clones, 1

4 in. × 200 mesh; , air tables, 2 in. × 200 mesh; •, jigs:Baum, 6 in. × 1

4 in.; Batac 34 in. × 28 mesh; �, concentrating ta-

bles, 34 in. × 200 mesh; ©, dense-medium separators: cyclone,

34 in. × 28 mesh; vessel, 6 in. × 1

4 in. [From Killmeyer, R. P.“Performance characteristics of coal-washing equipment: Baumand Batac jigs,” U.S. Department of Energy, RI-PMTC-9(80).]

Another index developed to characterize the sharpnessof separation is the imperfection I .

I = Ep/δ50 (6)

for dense-medium separation, and

I = Ep/δ50 − 1 (7)

for jigs.According to some authors, I values vary from 0.07 for

dense-medium cyclones and 0.175 for jigs to above 0.2for flotation machines.

Conventional float and sink methods for the derivationof the partition curve are expensive and time-consuming.In the diamond and iron ore industries, the density tracertechnique has been developed to evaluate the separationefficiency. This technique has also been adopted for stud-ies of coal separation.

The tracers, usually plastic cubes prepared to knownspecific gravities rendered identifiable through color cod-ing, are introduced into a separating device, and on re-covery from the product and reject streams, are sortedinto the appropriate specific gravity fractions and counted.This allows the points for the partition curve to be calcu-lated. The technique, as described above, was adopted atthe Julius Kruttschnitt Mineral Research Centre, Brisbane,while tracers made from plastic–metal composites that can

be detected with metal detectors mounted over conveyerbelts, known as the Sentrex system, were developed in theUnited States.

IX. ANCILLARY OPERATIONS

There are many ancillary operations in coal preparationthat are similar to operations in the mineral processingindustry—namely

1. Mechanical dewateringa. Vibrating basket centrifugesb. Screen bowl centrifugesc. Solid bowl centrifugesd. Disk filterse. Drum filtersf. Belt filtersg. Plate and frame filter pressesh. Thickeners, conventional and high capacity

2. Thermal dewateringa. Fluidized bed dryingb. Rotary dryers

3. Stocking and blending systems4. Automatic product loadout systems5. Computer startup and process monitoring

X. COAL PREPARATION FLOWSHEETS

In ROM coal, a large portion of both coal and shale isalready liberated sufficiently to permit immediate con-centration. Because dewatering of coarse coal is muchmore efficient and cheaper than that of fines, and becausemany users prefer larger size fractions for their particu-lar processes, the preparation of feed before cleaning hastraditionally consisted of as little crushing as possible.Therefore, coal preparation consists mostly of gravity con-centration and to some extent flotation, complemented byscreening and solid/liquid separation auxiliary processes.

The precrushing needed to reduce the top size of the ma-terial is widely applied by means of rotary breakers. Thisprocess, as already pointed out, is based on the selectivebreakage of fragile coal and hard inorganic rock, and com-bines two operations: size reduction and preconcentration(because it rejects larger pieces of hard rock). A furtherdevelopment in coal preparation has been the applicationof the three-product separators to treat coarse and inter-mediate size fractions. The recrushed middlings producedin such a cleaning are reprocessed, together with the finepart of the feed, increasing the total coal recovery.

This complete concentration sequence is typical for pro-cessing a metallurgical coal.



Coal Preparation 99

FIGURE 24 Probable error Ep vs mean size of the treated coal for dense-media baths (DMB), dense-media cyclones(DMC), jigs, concentrating tables, and water-only cyclones (WOC). [After Mikhail, M. W., Picard, J. L., and Humeniuk,O. E. (1982). “Performance evaluation of gravity separators in Canadian washeries.” Paper presented at 2nd TechnicalConference on Western Canadian Coals, Edmonton, June 3–5.].

Coal preparation plants are classified according to theextent to which coal is cleaned, as follows (see Fig. 25).

Type I. Only the large size coal is reduced and screenedinto various size fractions.

Type II. The plant cleans coarse coal by gravity methods.The gangue content in coal is reduced in the form of coarsereject from the Bradford breaker and high-gravity rejects.The yield of saleable product is 75–85%.

Type III. The plant cleans coarse and intermediate sizecoal, leaving the fines untreated.

Type IV. These modern coal preparation plants clean thefull size-range of raw coal.

Type V. The plant produces different saleable productsthat vary in quality, including deep-cleaned coal that islow in sulfur and ash.

Type VI. The plant incorporates fine grinding to achievea high degree of liberation, using froth flotation (oil ag-glomeration) and water-only cyclones (or spirals).

Type VI represents a very important stage in the de-velopment of coal preparation strategy. Types I–V are allpredominantly based on coarse coal cleaning, with the

amount of fines reduced to a minimum; but in the type VIplant all coarse coal cleaning operations are aimed at elim-inating coarse hard gangue and constitute a pretreatmentstage, with the final cleaning following after fine crushingand grinding. This future development in coal preparationtechnology is aimed at cleaning and utilization of thermalcoal, and is mainly a result of the new environmental-protection restrictions.

In the future, type VI coal preparation plants will becombined with coal/water slurry technology, and suchslurries will to some extent replace liquid fuels.

XI. ON-LINE ANALYSIS

On-line measurements of parameters such as ash, mois-ture, and sulfur use an inferential technique in which someproperty can be measured that in turn relates to the param-eter in the material.

Most on-line techniques use some form of elec-tromagnetic radiation. When radiation passes throughmatter, there is a loss of intensity by absorption andscattering.



100 Coal Preparation

FIGURE 25 Classification of coal preparation circuits.

A. Ash Analysis

Several devices irradiate the coal with γ - or X-rays, andthe backscatter is measured. At relatively low energies(<100 keV) the change in scattering coefficient withatomic number more than offsets the change in mass ab-sorption coefficient; thus the backscatter can be correlated

to mean atomic number and to the ash for a given coal.The backscattered radiation intensity is a function of bothbulk density and mean atomic number. Thus, instrumentsusing back scatter must be designed to ensure that a con-sistently packed sample of coal is presented for analysis.Some of the gauges using this principle are the GunstonSortex, the Berthold–Humbolt–Wedag, and the Wultex.



Coal Preparation 101

When γ radiation above energy levels of 1022 keV in-teracts with matter, positron–electron pairs are generated.The positron is rapidly annihilated by collision with anelectron and releases gamma rays, the intensity of whichis strongly dependent upon atomic number. The gaugethat uses the pair production principle is the COALSCANModel 4500, which uses Radium (226) emitting γ radia-tion at both the 1760 and 2200 keV levels. The devices thatuse absorption use a low energy level source that is sensi-tive to the mean atomic number (2) of the material. Thereis an eightfold difference in the mass absorption coeffi-cient between the organic portion of coal (Z = 6) and themineral matter (Z = 12). By comparing the differences inabsorption between low and high energy beams, the meanatomic number can be determined, and from this the ashcomposition can be inferred. This is the underlying prin-ciple behind such ash gauges as the COALSCAN Model3500, the SAI 400, and the ASHSCAN slurry gauge.

Another form of radiation used for coal analysis is neu-tron activation. This method, known as Prompt GammaNeutron Activation Analysis (PGNAA), has the distinctadvantage of being able to conduct a complete elementalanalysis rather than just total ash. This method permitsthe determination of the sulfur, carbon, and hydrogencomposition, which also permits the calculation of spe-cific energy. Several commercial PGNAA gauges areavailable—SAI Models 100 and 200, MDH Motherwell,and the Gamma Metrics.

The determination of ash in coal slurries uses the ba-sic principle of measurement of those for solid coal. Theextra problem is that the entrained air could seriously af-fect the accuracy of these devices. Various methods areused to compensate for this, such as pressurizing the sys-tem to collapse the air bubbles (ASHSCAN) or using aneutron attenuation measurement to correct for possibleair entrainment (AMDEL).

B. Moisture Analysis

There are several methods of on-line moisture analysis:

1. Microwave attenuation/phase shift: Involves trans-mitting a microwave beam through a given layer of coal(conveyor belt). The maximum absorption of the energyoccurs at 20 GHz and the energy absorbed by the materialis calibrated against moisture content. However, simple at-tenuation is found to be sensitive to other properties suchas particle size, bulk density, and salt content. To over-come these problems, the phase shift of the microwaveenergy is measured after passage through the sample orthe attenuation is measured at more than one frequency.

2. Capacitance measurements: Based on the princi-ple that electrical conductivity of coal increases with the

quantity of moisture. However, it is only applicable for themeasurement of surface moisture.

3. Nuclear magnetic resonance: This method measuresthe hydrogen in the coal, and is able to distinguish betweenhydrogen in water and in hydrocarbons.

4. Neutron moisture gauges: Neutron transmission andscatter methods can be used to measure total hydrogencontent, and assumes a constant hydrocarbon hydrogen.

C. Sulfur Analysis

The PGNAA method is generally used to measure elemen-tal sulfur. Commercial gauges available are the MDH-Motherwell ELAN, Science Application International-Nucoalyzer, and the Gamma-Metrics.

XII. RESEARCH INTO NEWBENEFICIATION PROCESSES

A. Gravity

1. Otisca Process

The Otisca process is a waterless separation process em-ploying CCl3F (with certain additives) as a heavy liquid.In this process, raw coal with surface moisture up to 10%is subjected to separation in a static bath. The mediumis recovered by evaporation from the separation productsand is recycled. The Otisca process plant is completelyenclosed, and the chemicals used are nonflammable, odorfree, and noncorrosive.

The Ep calculated from the 8-hr plant tests for differ-ent size feeds gave the results (shown in Table VII). Asseen, Ep values increase for very fine material (28 × 100mesh and 100 × 325 mesh) but are still better than formany other methods. This indicates very favorable low-viscosity conditions in the process, which were confirmedby rheological measurements. The favorable rheology ofthe true organic heavy-liquid slurries has further beenconfirmed in experiments on the use of Freon-113(1,1,2-trichloro-1,2,2-trifluoro ethane) in dense-medium

TABLE VII Ep Values for Otisca Process Calcu-lated for Different Size Feedsa

Organic Probable SpecificComposite feed efficiency % error, Ep gravity

38 × 1

4 in. 100 0.008 1.4814 in. × 28 mesh 99 0.015 1.49

28 × 100 mesh 98 0.175 1.57

100 × 325 mesh 96 0.260 1.8038 in. × 325 mesh 98 0.023 1.49

a After Keller, D. V. (1982). In “Physical Cleaning ofCoal” (Liu, Y. A., ed.), p. 75, Marcel Dekker, New York.




cyclones. The separation in cyclones, with use ofFreon-113 as a heavy medium, gave Ep = 0.009 for28 mesh × 0 coal, Ep = 0.045 for 200 mesh × 0, andEp = 0.100 for 400 mesh × 0 coal.

B. Magnetic Separation

This process is being investigated in the separation of coaland pyrite. The difference in magnetic susceptibility ofcoal and pyrite is very small; therefore, the most promis-ing application is for high-gradient magnetic separators(HGMS) or the pretreatment of coal, which could increasethe susceptibility of coal mineral matter.

Because pyrite liberation usually requires very finegrinding (e.g., below 400 mesh), high-gradient magneticseparation, the process developed to upgrade kaolin claysby removing iron and titanium oxide impurities, seemsto be particularly suitable for coal desulfurization. Wetmagnetic separation processes offer better selectivity forthe fine dry coal that tends to form agglomerates. Exper-iments on high-gradient wet magnetic separation have sofar indicated better sulfur rejection at fine grinding.

Perhaps the most interesting studies are on the effect ofcoal pretreatment, which enhances the magnetic propertiesof the mineral matter, on the subsequent magnetic sepa-ration. It has, for example, been shown that microwavetreatment heats selectively pyrite particles in coal withoutloss of coal volatiles, and converts pyrite to pyrrhotite,and as a result, facilitates the subsequent magneticseparation.

In the Magnex process, crushed coal is treated with va-pors of iron carbonyl [Fe(CO)5], followed by the removalof pyrite and other high-ash impurities by dry magneticseparation. Iron carbonyl in this process decomposes onthe surfaces of ash-forming minerals and forms stronglymagnetic iron coatings; the reaction with pyrite leads tothe formation of pyrrhotite-like material.

In 1970, Krukiewicz and Laskowski described a mag-netizing alkali leaching process, which was applied to con-vert siderite particles into magnetic γ -Fe2O3 and Fe3O4.A similar process has recently been tested for high-sulfurcoal, and it was reported that, after pretreatment of thecoal in 0.5-M NaOH at 85◦C and 930 kPa (135 psi) airpressure for 25 min, more than 50% of the coal sulfurwas rejected under the same conditions in which the samehigh-gradient magnetic separation had removed only 5%sulfur without the pretreatment.

C. Processes Based on Surface Properties

1. Oil Agglomeration

As in coal flotation, oil agglomeration takes advantage ofthe difference between the surface properties of low-ash

coal and high-ash gangue particles, and can cope with evenfiner particles than flotation. In this process, coal particlesare agglomerated under conditions of intense agitation.The following separation of the agglomerates from thesuspension of the hydrophilic gangue is carried out byscreening.

The most important operating parameters in this processare the amount of oil, intensity of agitation (agitation timeand agitation speed), and oil characteristics.

The amount of oil that is required is in the range of5–10% by weight of solids. Published data indicate thatthe importance of agitation time increases as oil densityand viscosity increase, and that the conditioning time re-quired to form satisfactory coal agglomerates decreasesas the agitation is intensified. Because the agitation ini-tially serves to disperse the bridging oil to contact theoil droplets and coal particles, higher shear mixing with alower viscosity bridging liquid is desirable in the first stage(microagglomeration), and less intense agitation with theaddition of higher viscosity oil (macroagglomeration) isdesirable in the second stage. Viscous oil may producelarger agglomerates that retain less moisture. With largeroil additions (20% by weight of solids), the moisture con-tent of the agglomerated product can be well below 20%and may be reduced even further if tumbling is used in thesecond stage instead of agitation.

The National Research Council of Canada developedthe spherical agglomeration process in the 1960s. Thisprocess takes place in two stages: First, the coal slurry isagitated with light oil in high shear blenders where mi-croagglomerates are formed; then the microagglomeratesare subjected to dewatering on screen and additional pel-letizing with heavy oil.

Shell developed a novel mixing device to conditionoil with suspension. Application of the Shell PelletizingSeparator to coal cleaning yielded very hard, uniform insize, and simple to dewater pellets at high coal recover-ies. The German Oilfloc Process was developed to treatthe high-clay, −400-mesh fraction of coal, which is theproduct of flotation feed desliming. In the process devel-oped by the Central Fuel Research Institute of India, coalslurry is treated with diesel oil (2% additions) in millsand then agglomerated with 8–12% additions of heavyoil.

It is known that low rank and/or oxidized coals are nota suitable feedstock for beneficiation by the oil agglomer-ation method. The research carried out at the Alberta Re-search Council has shown, however, that bridging liquids,comprising mainly bitumen and heavy refinery residuesare very efficient in agglomeration of thermal bituminouscoals. Similar results had earlier been reported in the flota-tion of low rank coals; the process was much improvedwhen 20% of no. 6 heavy oil was added to 2 fuel oil.




2. Otisca T-Process

This is another oil agglomeration process that can copewith extremely fine particles. In this process, fine raw coal,crushed below 10 cm, is comminuted in hammer crushersto below 250 µm and mixed with water to make a 50%by weight suspension; this is further ground below 15 µmand then diluted with water to 15% solids by weight. Sucha feed is agglomerated with the use of Freon-113, andthe coal agglomerates and dispersed mineral matter areseparated over screen. The separated coal-agglomeratedproduct retains 10–40% water and is subjected to thermaldrying; Freon-113, with its boiling point at 47◦C, evap-orates, and after condensing is returned liquified to thecircuit. The product coal may retain 50 ppm of Freon and30–40% water.

Various coals cleaned in the Otisca T-Process containedin most cases below 1% ash, with the carbonaceous ma-terial recovery claimed to be almost 100%. Such a lowash content in the product indicates that very fine grind-ing liberates even micromineral matter (the third level ofheterogeneity); it also shows Freon-113 to be an excep-tionally selective agglomerant.

3. Selective Agglomeration During Pipeliningof Slurries

In some countries, for example in Western Canada, themajor obstacles to the development of a coal mining in-dustry are transportation and the beneficiation/utilizationof fines. Selective agglomeration during pipelining offersan interesting solution in such cases. Since, according tosome assessments, pipelining is the least expensive meansfor coal transportation over long distances, this ingeniousinvention combines cheap transportation with very effi-cient beneficiation and dewatering. The Alberta ResearchCouncil experiments showed that selective agglomerationof coal can be accomplished in a pipeline operated un-der certain conditions. Compared with conventional oilagglomeration in stirred tanks, the long-distance pipelineagglomeration yields a superior product in terms of wa-ter and oil content as well as the mechanical propertiesof the agglomerates. The agglomerated coal can be sep-arated over a 0.7-mm screen from the slurry. The watercontent in agglomerates was found to be 2–8% for met-allurgical coals, 6–15% for thermal coals (high-volatilebituminous Alberta), and 7–23% for subbituminous coal.The ash content of the raw metallurgical coal was 18.9–39.8%, and the ash content of agglomerates was 8–15.4%.For thermal coals the agglomeration reduced the ash con-tent from 19.8–48.0 to 5–12.8%, which, of course, is ac-companied by a drastic increase in coal calorific value. Be-sides transportation and beneficiation, the agglomerationalso facilitates material handling; the experiments showed

that the agglomerates can be pipelined over distances of1000–2000 km.

4. Flotation

Flotation has progressed and developed over the years;recent trends to achieve better liberation by fine grindinghave intensified the search for more advanced means of im-proving selectivity. This involves not only more selectiveflotation agents but also better flotation equipment. Sincethe froth product in conventional flotation machines con-tains entrained fine gangue, which is carried into the frothwith feed water, the use of froth spraying was suggestedin the late 1950s to eliminate this type of froth “contam-ination.” The flotation column patented in Canada in theearly 1960s and marketed by the Column Flotation Com-pany of Canada, Ltd., combines these ideas in the form ofwash water supplied to the froth. The countercurrent washwater introduced at the top of a long column prevents thefeed water and the slimes that it carries from entering anupper layer of the froth, thus enhancing selectivity.

Of many variations of the second generation columns,perhaps the best known is the Flotaire column marketedby the Deister Concentrator Company.

The microbubble flotation column (Microcel) devel-oped at Virginia Tech is based on the basic premise thatthe rate (k) at which fine particles collide with bubblesincreases as the inverse cube of the bubble size (Db), i.e.,k ∼ 1/D3

b. In the Microcel, small bubbles in the range of100–500 µm are generated by pumping a slurry throughan in-line mixer while introducing air into the slurry at thefront end of the mixer. The microbubbles generated as suchare injected into the bottom of the column slightly abovethe section from which the slurry is with drawn for bubblegeneration. The microbubbles rise along the height of thecolumn, pick up the coal particles along the way, and forma layer of froth at the top section of the column. Like mostother columns, it utilizes wash water added to the frothphase to remove the entrained ash-forming minerals. Ad-vantages of the MicrocelTM are that the bubble generatorsare external to the column, allowing for easy maintenance,and that the bubble generators are nonplugging. An 8-ftdiameter column uses four 4-in. in-line mixers to produce5–6 tons of clean coal from a cyclone overflow containing50% finer than 500 mesh.

Another interesting and quite different column was de-veloped at Michigan Tech. It is referred to as a static tubeflotation machine, and it incorporates a packed-bed col-umn filled with a stack of corrugated plates. The packingelements arranged in blocks positioned at right angles toeach other break bubbles into small sizes and obviate theneed for a sparger. Wash water descends through the sameflow passages as air (but countercurrently) and removes




entrained particles from the froth product. It was shown inboth the laboratory and the process demonstration unit thatthis device handles extremely well fine below 500-meshmaterial.

The Jameson Column developed at the University ofNewcastle is considerably shorter, and its height may beonly slightly greater than its diameter.

Another novel concept is the Air-Sparged Hydrocy-clone developed at the University of Utah. In this device,the slurry fed tangentially through the cyclone header intothe porous cylinder to develop a swirl flow pattern inter-sects with air sparged through the jacketed porous cylin-der. The froth product is discharged through the overflowstream.

D. Chemical Desulfurization

Chemical desulfurization has been investigated as amethod to selectively remove inorganic sulfur (which con-

FIGURE 26 Molten–Caustic–Leaching (Gravimelt Process) flow schematic.

sists of pyritic sulfur with trace amounts of sulfate) andorganic sulfur without need to discard part of the coal richin sulfur and ash as with conventional coal preparation. Ifthis could be accomplished without loss of coal productheat content, a clean burning coal alternative to flue gasdesulfurization would be at hand. Several methods havebeen tested for the chemical removal of pyritic or organicsulfur from coal. Treatment with aqueous sodium hydrox-ide or aqueous sodium hydroxide with added lime at tem-peratures of 200–325◦C and pressures up to 12500 psiremoves only 45–95% of the inorganic sulfur and noneof the organic sulfur, while significantly reducing the heatcontent of the product coal due to hydrothermal oxida-tion. Sulfur dioxide, potassium nitrate with complexingagents, nitrogen dioxide, nitric acid, air oxidation withand without the bacterium Thiobacillus ferrooxidans, hy-drogen peroxide, chlorine, and organic solvents have allbeen tested. In each case, there was only minimal sulfurremoval (based on sulfur content per unit heat content of




product coal) and considerable loss of coal heat contentand/or dilution of the organic coal matrix by bonding withreactant chlorine, nitrate, etc.

However, the use of aqueous ferric sulfate has provedto be effective in selectively removing virtually all of theinorganic sulfur while the heat content increases in di-rect proportion to the amount of iron pyrite removed.The ferric sulfate sulfur removal technology, known asthe Meyers process, was developed through pilot plantdemonstration under the auspicies of the Environmen-tal Protection Agency. Typically, 90–95% of the pyriticsulfur was removed and the ferric salt was simultane-ously regenerated with air oxidation. Elemental sulfur isproduced as a byproduct, and must be extracted fromthe coal by use of an organic solvent or steam distil-lation. The process has not yet been utilized commer-cially as removal of pyritic sulfur alone (usually only halfof the total sulfur, the remainder being mainly organic

FIGURE 27 Rotary kiln reactor in operation at Gravimelt Test Plant Facility. The reactor consists of a 12-ft longfurnace section, a 3-ft cooling zone, and a discharge area.

sulfur) does not meet present air pollution controlstandards.

Subsequently, simultaneous removal of organic sulfurand pyritic sulfur was demonstrated in the laboratory andthen in a test plant utilizing molten sodium and potas-sium hydroxide followed by a countercurrent water washto recover sodium, potassium, and about half of the origi-nal coal mineral matter. A dilute sulfuric acid wash re-moves the remainder of the coal mineral matter (ash).This process, termed Molten–Caustic–Leaching, or theGravimelt Process, results in removal of 90–99% of thecoal inorganic and organic sulfur as well as 95–99% ofthe coal mineral matter. The product coal meets all cur-rent air pollution control standards for sulfur and ashemissions, and has an increased heat content, usually inthe range of 14000 btu/lb. This coal is also very effi-cient as a powdered activated carbon for water purifi-cation. This is due to the high internal pore structure




introduced by removal of sulfur and ash. The coal alsofunctions as a cation exchange medium for water cleanupdue attachment of carboxy groups during the caustictreatment.

A schematic of the Gravimelt Process is shown inFig. 26. Feed coal is premixed with anhydrous sodiumhydroxide or mixed sodium and potassium hydroxides,and then fed to a rotary kiln reactor where the mixtureis heated to reaction temperatures of 325–415◦C, caus-ing the caustic to (1) melt and become sorbed in the coalmatrix, (2) react with the coal sulfur and mineral matter,and (3) dissolve the reaction products containing sulfurand inorganic components. The kiln reactor can be seenin operation in Figure 27.

The reaction mixture is cooled in the cooling section ofthe kiln such that the mixture exits the rotary kiln in theform of pellets that are then washed with water in a seven-stage countercurrent separation system consisting of twofilters and five centrifuges. As an example, four weightsof wash water per weight of coal may be used, resultingin a 50% aqueous reacted caustic solution exiting the firststage of filtration. A large part of the coal-derived ironand a small amount of sodium and/or potassium remainwith the coal cake exiting this countercurrent washingsystem.

The coal cake is next washed to remove the residualiron and caustic in a countercurrent three-stage centrifugesystem. Sulfuric acid is added to the first stage to producea pH of about 2 in order to dissolve the residual alkaliand iron hydroxide. As an example, three weights of washwater per weight of coal may be used, resulting in anacid extract leaving the first stage containing 5900 ppmof mixed sodium and potassium sulfates and 12,500 ppmof iron sulfate. The product coal exiting this acidwashingsystem contains about one weight of sorbed water perweight of coal.

All of the iron sulfate and a major portion of the alkalisulfates can be removed from the acid extract water by asequence of heating and lime treatment to form insolubleminerals such as gypsum. This water can then be recy-cled to the water wash train. Under some (very carefullycontrolled) conditions, the precipitate formed can be ajarosite-like double salt, which can be easily disposed ofor land-filled.

A newer process configuration, with potentially largecost savings to the process, was tested in which the effluentfrom the acid wash train (without acid addition) was usedas the makeup wash water for the water wash train. Thisconfiguration resulted in a product coal with low sulfur

and moderate ash content suitable for those commercialapplications that do not require low ash coal.

The aqueous caustic recovered from the first coun-tercurrent washing circuit is limed to remove the coal-derived mineral matter, sulfur compounds, and carbon-ates. The mixture is provided with sufficient residencetime to permit precipitation of the impurities before beingcentrifuged. The purified liquid is preheated and sent toa caustic evaporator where the water is recovered for re-cycle to the first wash train, while producing anhydrouscaustic (as either a molten liquid or as flakes) for reuse inthe initial leaching of the coal.

SEE ALSO THE FOLLOWING ARTICLES

COAL GEOLOGY • COAL STRUCTURE AND REACTIVITY

• FOSSIL FUEL POWER STATIONS, COAL UTILIZATION •FROTH FLOTATION • FUELS • MINING ENGINEERING

• PETROLEUM REFINING

BIBLIOGRAPHY

Deurbrouck, A. W., and Hucko, R. E. (1981). In “Chemistry of CoalUtilization” (M. A. Elliot, ed.), 2nd Suppl. vol., Chap. 10, Wiley, NewYork.

Gochin, R. J., and Smith, M. R. (1983). Min. Mag., Dec., p. 453–467.Horsfall, D. W., ed. (1980). “Coal Preparation for Plant Operators,” South

African Coal Processing Society, Cape Town.Klassen, V. I. (1963). “Coal Flotation,” Gosgorttiekhizdat, Moscow (Rus-

sian text).Laskowski, T., Blaszczynski, S., and Slusarek, M. (1979). “Dense

Medium Separation of Minerals,” 2nd ed. Wyd. Slask, Katowice(Polish text).

Leonard, J. W., ed. (1979). “Coal Preparation,” 4th ed., American Insti-tute of Mechanical Engineering, New York.

Liu, Y. A., ed. (1982). “Physical Cleaning of Coal,” Dekker, New York.Meyers, R. A. (1977). “Coal Desulfurization,” Marcel Dekker, New York.Misra, S. K., and Klimpel, R. R., eds. (1987). “Fine Coal Processing,”

Noyes Publications, Park Ridge.Osborne, D. G. (1988). “Coal Preparation Technology,” Volumes 1 and

2, Graham and Trotman, London.Speight, J. G. (1994). “The Chemistry and Technology of Coal,” 2nd.,

revised and Expanded, Marcel Dekker, New York.TRW Space and Electronics Group, Redondo Beach, CA Applied Tech-

nology Division. (1993). “Molten–Caustic–Leaching (Gravimelt) sys-tem integration project, final report,” Department of Energy, Washing-ton, DC, NTIS Order Number: DE930412591NZ, Performing Orga-nization’s Report Number: DOE/PC/91257-T18.

U.S. Department of Energy (1985). “Coal Slurry Fuels Preparation andUtilization,” U.S. Department of Energy, Pittsburgh.

Wheelock, T. D., ed. (1977). “Coal Desulfurization,” ACS Symp. Ser.No. 64, American Chemical Society, Washington, DC.

P1: GNB Final Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology EN003B-112 June 13, 2001 20:17

Coal Structure and ReactivityJohn W. LarsenLehigh University

Martin L. GorbatyExxon Research and Engineering Company

I. IntroductionII. Geology: Origin and DepositionIII. Coal ClassificationsIV. Physical PropertiesV. Physical Structure

VI. Chemical StructureVII. Coal Chemistry

GLOSSARY

Anthracite Coals with less than ∼10% volatile matter;the highest coal rank.

Bituminous Coals that have between about 10 and30% volatile matter and coals that agglomerate onheating.

Diagenesis Initial biological decomposition of coal-forming plant material.

Grade Suitability of a coal for a specific use.Lignite Stage of coalification following peat; high-

oxygen, low-heating-value, young coals.Lithotypes Macroscopic rock types occurring in coals,

usually in banded structures.Macerals Homogeneous microscopic constituents of

coals, analogous to minerals in inorganic rocks.Metamorphism Chemical changes responsible for the

conversion of fossil plant materials to coal.

Rank Degree of maturation of a coal.Type Origin and composition of the components of a coal.

IN THE BROADEST TERMS, coal may be defined as anaturally occurring, heterogeneous, sedimentary organicrock formed by geological processes (temperature, pres-sure, time) from partially decomposed plant matter underfirst aerobic and then anaerobic conditions in the presenceof water and consisting largely of carbonaceous materialand minor amounts of inorganic substances. The main usefor coal worlwide is combustion as a source of heat forgenerating steam to make electricity. Coals used for thispurpose are called steam coals. Certain other coals are usedto produce the metallurgical coke required for making ironfrom iron ore. Coal can also serve as the raw material formaking clean, usable liquid fuels, such as gasoline, andchemicals.

107

P1: GNB Final


108 Coal Structure and Reactivity

TABLE I Worldwide Coal Depositsa

Estimated Provenresources resources

Country (billions of tons) (billions of tons)

Australia 600 32.8

Canada 323 4.2

Federal Republic of Germany 247 34.4

India 81 12.4

People’s Republic of China 1,438 98.9

Poland 139 59.6

South Africa 72 43

United Kingdom 190 45

United States of America 2,570 167

Russia 4,860 110

Other 229 55.7

10,749 663

aData from Wilson, C. L. (1980). “Coal—Bridge to the Future, Reportof the World Coal Study,” p. 161, copyright Ballinger Publishing Co.,Cambridge, MA.

I. INTRODUCTION

Coal is perhaps the most abundant fossil fuel, and largedeposits are distributed worldwide. As with other natu-ral resources, the amounts are estimated on two bases:estimated resources and proven reserves. Estimated re-sources are based on geological studies (e.g., seismic sur-veys and exploratory drilling) and are the best estimateof the amount of coal believed to be present in a givenarea. Proven reserves are those amounts of coal known tobe economically recoverable by current technology in agiven area. Proven reserves represent a small fraction ofestimated resources. Table I contains resource and reservedata for some of the major worldwide coal deposits ex-pressed in billions of tons. It must be recognized that theseestimates vary from year to year, as do the methods forobtaining the data. Therefore, the data in the table shouldnot be viewed as being absolute. According to these data,∼663 billion tons, or 6% of worldwide resources of over10 trillion tons, are recoverable by current technology.The worldwide reserves of coal have an energy contentequivalent to ∼2 trillion barrels of petroleum.

II. GEOLOGY: ORIGIN AND DEPOSITION

Coals may be viewed as fossilized plant matter. The pro-cesses by which plant matter was converted to coal arecomplex and involve at least three steps: accumulation,diagenesis, and metamorphism. The starting material for

coal was a diversity of plant types and parts. For the plantmatter to accumulate, a specific set of conditions had toprevail. The deposition rate of plant matter had to be ap-proximately equal to the subsidence of the land on whichit grew. Furthermore, there had to be water present tocover and partially preserve the plant matter. Initially, aer-obic organisms began to decompose the cellulosic partsof the plant, as well as proteinaceous and porphyrin ma-terials. This biological decomposition is called diagen-esis. As the accumulated, partly decayed biomass wasburied deeper, aerobic activity ceased, and under the in-creasing temperature and pressure conditions of burial itwas chemically transformed to coal. This chemical trans-formation is called metamorphism. From the least serveto the most severe conditions, the coals that resulted arearranged as follows: peat → lignite → subbituminous →bituminous → anthracite. Chemically, the transforma-tions can be viewed as combinations of deoxygenation, de-hydrogenation, aromatization, and oligomerization. Thesetransformations are inferred from the elemental analysesshown in Table II. The data in this table are typical; eachclass includes a range of values.

All coals contain minor amounts of inorganic material,called mineral matter, ranging from about 2 to 30% andaveraging ∼10% by weight. Inorganics were incorporatedinto coal as the inorganic matter present in the originalvegetation, mineral detritus deposited by water or windflowing through the decaying biomass (detrital), and byprecipitation or ion exchange of soluble ions from waterpercolating through the coal bed during metamorphism(authigenic). The former two are called syngenetic, thelatter epigenetic to differentiate the processes occurringduring diagenesis from those occurring during and aftermetamorphism.

The major components of mineral matter include alu-minosilicate clays, silica (quartz), carbonates (usuallyof calcium, magnesium, or iron), and sulfides (usuallyas pyrite and/or marcasite). Many other inorganics arepresent, but only in trace quantities of the order of parts permillion.

TABLE II Elemental Analysis CharacterizingMetamorphism as a Deoxygenation/Aromatization Process

Coal orprecursor Carbon (%) Hydrogen (%) Oxygen (%)

Wood 49.3 6.7 44.4

Peat 60.5 5.6 33.8

Lignite 69.8 4.7 25.5

Subbituminous 73.3 5.1 18.4

Bituminous 82.9 5.7 9.9

Anthracite 93.7 2.0 2.2

P1: GNB Final


Coal Structure and Reactivity 109

It is important to differentiate between mineral mat-ter and ash. The former consists of inorganics presentin the coal as mined; the latter is an oxide residue fromcombustion.

III. COAL CLASSIFICATIONS

The diverse nature of deposited plant and mineral matterand the burial conditions make it clear that coal is not onesubstance but a wide range of heterogeneous materials,each with considerably different chemical and physicalproperties from the others. The heterogeneity of coals canbe seen on the macroscopic and microscopic levels, and thescience of classifying coals on the basis of these differenceis called coal petrography.

On the macroscopic level, coals, particularly bitumi-nous coals, appear to be layered or banded. These layers,called lithotypes, fall into four main classes: vitrain (co-herent, black, and high gloss), clarain (layered, glossy),durain (granular, dull sheen), and fusain (soft, friable,charcoal-like).

On the microscopic level, the lithotypes just describedare seen to be composites made up of discrete entitiescalled macerals. Optically differentiated by microscopicexamination of coals under reflected light or by the studyof thin sections of coal using transmitted light (Fig. 1),macerals are classified into three major groups: vitrinite,

FIGURE 1 Transmission photomicrograph of bituminous coal. (Courtesy of D. Brenner, Exxon Research and Engi-neering Company.)

liptinite (often termed exinite), and inertinite. The vitriniteclass appears orange-red in transmitted light and is derivedfrom fossilized lignin. Liptinites appear yellow and arethe remains of spores, waxy exines of leaves, resin bodies,and waxes. Inertinites appear black and are believed to bethe remnants of carbonized wood and unspecified detritalmatter. Chemically, the liptinites are richest in hydrogen,followed by the vitrinites. Both the liptinites and vitrinitesare reactive and give off a significant amount of volatileswhen pyrolyzed, while the inertinites are relatively unre-active. It is not possible to generalize with accuracy theproportional maceral composition of all coals; however,it is fair to say that most bituminous coals contain about70% vitrinite, 20% liptinite, and 10% inertinite.

It is generally believed that maceral composition ul-timately determines a coal’s reactivity in a particularprocess. For example, coals are blended together on a com-mercial scale to make optimum feedstocks for metallur-gical coke based largely on their maceral contents. In thepast, macerals were separated tediously by hand under amicroscope. Today, new techniques are available to sepa-rate a coal into its component macerals much more easily,and this has renewed interest in maceral characterizationand reactivity. It may even be possible one day to tailorcoal feedstocks for specific processes such as pyrolysis,gasification, or liquefaction by separation and blending ofmacerals.

P1: GNB Final



Another method of classifying coals is by the degree ofmetamorphism they have undergone. This classification iscalled rank—the higher the rank, the more mature the coal.Since metamorphism is a deoxygenation–aromatizationprocess, rank correlates generally with carbon content andheat content of the coal, although there are many other,more specific methods of determining rank. Lignites andsubbituminous coals are generally considered to be lowrank, while bituminous coals and anthracite are consideredto be high rank. The term “rank” must be distinguishedfrom type and grade. Type refers to a coal’s composition,the origin of the organic portion, and its organic compo-nents. Grade refers to the quality or suitability of a coalfor a particular purpose. These distinctions are importantbecause coal properties are not solely attributable to meta-morphic alteration; the nature of the source material andits diagenetic alterations are also important in determiningcoal properties.

In general, carbon content, aromaticity (i.e., the per-centage of carbon that is aromatic), number and size ofcondensed rings, and calorific content increase with in-creasing rank, while volatile matter, oxygen content, oxi-dizability, and solubility in aqueous caustic decrease withincreasing rank. Hardness and plastic properties increaseto a maximum, then decrease with increasing rank, whileporosity (i.e., moisture-holding capacity) and density de-crease to a minimum, then increase with increasing rank.Typical values are shown in Table III. Within each rankclassification, a range of values is found.

IV. PHYSICAL PROPERTIES

Physical properties are an important consideration inall uses of coals. Most vary with rank and coal type. Abrief discussion of the most important physical propertiesfollows.

TABLE III Properties of Coals of Different Rank

Propertya Lignite Subbituminous Bituminous Anthracite

Moisture capacity (wt%) 40 25 10 <5

Carbon (wt%), DMMF 69 75 83 94

Hydrogen (wt%), DMMF 5.0 5.1 5.5 3.0

Oxygen (wt%), DMMF 24 19 10 2.5

Volatile matter (wt%), DMMF 53 48 38 6

Aromatic carbon (mol fraction) 0.7 0.78 0.84 0.98

Density (helium) (g/cm3) 1.43 1.39 1.30 1.5

Grindability (Hardgrove) 48 51 61 40

Heat content (btu/lb), DMMF 11.600 12.700 14.700 15.200

aDMMF: dry, mineral-matter-free.

A. Mechanical Properties

1. Elasticity

Because of the heterogeneous nature and the natural oc-currence of cracks and pores, determining elastic mod-uli is experimentally difficult. Under normal conditions,coals are brittle solids. For bituminous coals, the elasticproperties are independent of rank and similar to those ofphenol–formaldehyde resins. At ∼92% carbon, the mod-uli increase rapidly and become anisotropic.

2. Hardness (Grindability)

Grindability is normally measured by determining thenumber of revolutions in a pulverizer that are requiredto achieve a given size reduction. This is an indirect mea-sure of the amount of work required for that size reduction.Grindability increases slowly to a maximum at ∼90% car-bon and decreases rapidly beyond this point. Hardness isascertained by measuring the size of the indentation leftby a penetrator of specified shape pressed onto the coalwith a specified force for a specified time. It changes onlyslightly between 70 and 90% carbon and increases rapidlybeyond 90% carbon.

3. Plastic Deformation

Coals heated to above 400◦C plastically deform, as dothose that have been swollen with pyridine and other sol-vents that are good hydrogen bond acceptors. Very smallcoal particles flow and deform at room temperature undervery high stresses.

B. Thermal Properties

1. Heat of Combustion

Heat of combustion is normally measured calorimetri-cally, but it can also be calculated with good accuracy from

P1: GNB Final



the elemental composition of the coal. It is the heat givenoff when a given amount of coal is completely burned.

2. Specific Heat

The specific heat records the heat necessary to cause agiven temperature rise in a coal. It is rank dependent, de-creasing from ∼0.35 cal/g at 50% carbon to ∼0.23 cal/g at90% carbon. Above 90% carbon, it decreases quite rapidly.

3. Thermal Conductivity

The capacity of a coal to conduct heat is strongly depen-dent on the pore structure and moisture content. It has beenwidely studied. When coals soften thermally, between 400and 600◦C, the thermal conductivity increases as the porescollapse.

C. Electrical Properties

1. Dielectric Constant

The dielectric constant is a sensitive function of theamount of water present in a coal. Bituminous coals areinsulators with static dielectric constants between 3 and 5.Above 88% carbon, the static dielectric constant increasesvery rapidly.

2. Electric Conductivity

The most interesting feature here is that anthracites arenatural semiconductors. Conductivity increases sharplyabove 88% carbon, presumably due to increasing graphiticcharacter.

3. Magnetic Susceptibility

Coals contain numerous unshared electrons, of the orderof 1018 spins per gram. This gives rise to the bulk propertycalled paramagnetic susceptibility, the magnitude of theforce exerted on a sample in a strong magnetic field.

4. Reflectance

Coal samples are often geologically characterized by veryaccurate measurements of the reflectance of light fromtheir components. There is a vast literature dealing withthis empirical technique.

D. Density

The measurement of coal densities is complicated by thepresence of pores in coals. A number of techniques for

dealing with this problem have been developed, and accu-rate data are available for whole coals and macerals. Den-sity is rank dependent in a regular way. Vitrinite densitydecreases slowly from ∼1.5 g/cm3 at 50% carbon to a min-imum of 1.27 at 88% carbon, then increases rapidly, reach-ing values as high as 1.6–1.8. Fusinites and micrinites aremore dense than vitrinites, while liptinites are less dense.The density differences provide a good way of separatingmacerals.

V. PHYSICAL STRUCTURE

Coals are very complex mixtures of several macromolecu-lar components and a number of inorganic materials. Thisfact makes the discussion of physical properties difficult.Each individual component can be isolated and its prop-erties determined, but will the property of the whole coalbe the sum or some weighted average of the individualcomponent properties? Perhaps the easiest approach isthe phenomenological one: simply to report the proper-ties measured for the whole coal while realizing that theseare the result of some complex addition of the proper-ties of many different components. This approach is oftenperfectly satisfactory. If the heat of combustion of a coalis desired, it makes little difference what the individualcomponent contributions to that heat are. The approachused here will be phenomenological, but the reader shouldknow that this obscures the many differences among indi-vidual components.

A. Pore Structure

Coals are highly porous materials, many having void vol-umes as high as 20% of thier total volume. To describe thepore structure we need to know the size distribution of thepores, the size ranges of the pores, and the population ofeach range. The shape of the pores is important, as is thetotal surface area of the coal.

A reagent coming in contact with a coal first encountersits surface. Often, it reacts there. To understand reactivitypatterns, we need to know the surface area. Even if thereagent must diffuse into the coal, the rate at which itdoes this is strongly influenced by the amount of surfaceit contacts. The size and shape of the pores control whethera molecule can enter and contact the surface and so are animportant part of any description of the pore network.

Most information about pores comes from adsorptionmeasurements. A sample of coal that has been evacuated toclean its surface is exposed to vapors of a gas (e.g., CO2),which adsorb and cover the surface. By means of a verysensitive balance, the amount of adsorption is measuredat several pressures. Thus, the amount of gas required to

P1: GNB Final



FIGURE 2 Dependence of coal surface area for dry, ash-free (DAF) coal on coal rank and technique. [Reproducedwith permission from Gan, H., Nandi, S. P., and Walker, P. L. (1972). Fuel 51, 272.]

cover the surface is determined. The cross-sectional areaof the gas is known. This is the amount of surface thateach gas molecule will cover. From these two quantities,the surface area can be calculated.

Figure 2 shows how the surface areas of coals varywith carbon content. Carbon dioxide measures the entiresurface area, pore surfaces as well as external surfaces. Ni-trogen cannot penetrate the pores (it does not have enoughenergy at these low temperatures), and so only the externalsurface area is measured. The surface areas are enormous,demonstrating that the coals must be very porous.

The pore volume can be estimated from some elaborategas adsorption studies or measured directly. If the densityof coal particles is measured in a vacuum and in a gas thatfills the pores, different values are obtained. Density ismass over volume, and for both measurements the volumeis the same. In one case (vaccum), the mass of the coal isrecorded. In the other case, the mass of the coal plus themass of the gas in the pores is recorded. If the density ofthe gas is known, the pore volumes are easily obtained.Some typical values are given in Table IV.

An analysis of the weight increase due to gas adsorptionas the pressure of the gas increases can give the size distri-bution of the pores. At low pressures, pores of very smallsize fill by capillary condensation, while larger pores arestill only surface coated. The results of such an analysisfor a series of coals are given in Table IV. For bituminouscoals and anthracites, most of the surface area is due to mi-cropores, which also make the largest contribution to thepore volume. Lignites have a much more open pore struc-ture, with the largest pores predominating. Penetration of

the pores by liquid mercury at high pressures can also pro-vide information about the size distribution of macroporesand some of the transitional pores. We shall return brieflyto pore structure later to discuss micropores as packingimperfections between macromolecules.

There remains only an analysis of pore shape. This is-sue has been probed mainly by looking at the dependenceof molecular absorption on the shape of the absorbed

TABLE IV Illustrative Gross Pore Distributions in Coala

PoresPore volume

(total), Macro, Transitional, Micro ,4–30,000 A 300–30,000 A 12–300 A 4–12 A

Rankb (cm3/g) (cm3/g) (cm3/g) (cm3/g)

Anthracite 0.076 0.009 0.010 0.057

LV bituminous 0.052 0.014 0.000 0.038

MV bituminous 0.042 0.016 0.000 0.026

HVA bituminous 0.033 0.017 0.000 0.016

HVB bituminous 0.144 0.036 0.065 0.043

HVC bituminous 0.083 0.017 0.027 0.039

HVC bituminous 0.158 0.031 0.061 0.066

HVB bituminous 0.105 0.022 0.013 0.070

HVC bituminous 0.232 0.043 0.132 0.070

Lignite 0.114 0.088 0.004 0.022

0.105 0.062 0.000 0.043

0.073 0.064 0.000 0.009

aReproduced with permission from Gan, H., Nandi, S. P., and Walker,P. L. (1972). Fuel 51, 272.

b LV, Low volatile; MV, medium volatile; HVA, high volatile A; HVB,high volatile B; HVC, high volatile C.

P1: GNB Final



molecules. There are fewer data here than there are onpore volume or surface area. Nevertheless, there is generalagreement that the pores are often slits, oblate elipsoids,that often have constricted necks.

Coals are penetrated by an extensive network of verytiny pores and, because of this, have enormous surface ar-eas. The smaller pores are about the same size as smallmolecules, so coals are molecular sieves, capable of trap-ping small molecules in their pores while denying accessto larger molecules.

B. Macroscopic Structure

The best analogy for the macroscopic structure of coal is afruitcake. The dough corresponds to the principal maceral,vitrinite, while the various goodies correspond to the dif-ferent macerals spread throughout. The ground nuts playthe role of the mineral matter. The composition of the fruit-cake is not rank dependent, but rather is due to the originalcoal deposition and depositional environment. The mac-erals become more like one another as rank increases, buttheir proportions are not rank dependent. In fact, their rel-ative proportions vary with both vertical and horizontalpositions in a coal seam, with the vertical variation occur-ing over a much shorter distance scale.

To make our fruitcake a more accurate model, we mustalter it somewhat. Coals tend to have banded structures,with macerals grouped in horizontal bands. A fruitcakemade by a lazy cook who did not stir the batter throughly isan improved model. The nuts must also be ground to coveran enormous size range, from large chunks to micrometer-size particles. Further more, they must stick to the cake.Even extremely fine grinding in a fluid energy mill will notentirely separate the mineral matter from the organics. Toseparate the macerals on the basis of their different densi-ties, the mineral matter must first be dissolved in acid. Weknow almost nothing about the binding forces or bindingmechanism between the mineral matter and the coal.

C. Extractable Material

As much as 25% of many coals consists of small moleculesthat will dissolve in a favorable solvent and can thereby beremoved from the insoluble portion. The amount of coalthat dissolves is a function of the nature of the coal, thesolvent used, and the extraction conditions. First, let usconsider the solubility of coals in various solvents at theirboiling points at atmospheric pressure. Typical data arecontained in Table V.

Certain solvents stand out as being unusually effective,pyridine and ethylenediamine being the most noteworthy.They are not unique, however, and any solvent having aHildebrand solubility parameter of ∼11 and the capacityto serve as an effective hydrogen bond acceptor will be a

TABLE V Yields of Coal Extracts in Solvents atTheir Boiling Pointsa

Yield of coal extract(wt% of dry,

Solvent ash-free coal)

n-Hexane 0.0

Water 0.0

Formamide 0.0

Acetonitrile 0.0

Nitromethane 0.0

Isopropanol 0.0

Acetic acid 0.9

Methanol 0.1

Benzene 0.1

Ethanol 0.2

Chloroform 0.35

Dioxane 1.3

Acetone 1.7

Tetrahydrofuran 8.0

Pyridine 12.5

Dimethyl sulfoxide 12.8

Dimethylformamide 15.2

Ethylenediamine 22.4

1-Methyl-2-pyrrolidone 35.0

a Data from Marzec, A., Juzwa, M., Betlej, K., andSobkowiah, M. (1979). Fuel Proc. Technol. 2, 35, andrefer to a specific coal; they are illustrative but not repre-sentative.

good solvent. The dependence on coal rank of the amountthat can be extracted by pyridine is shown in Fig. 3. Thepyridine extract seems to resemble the parent coal quiteclosely, and the structural characteristics of the insolublecoal can be taken as present in the soluble portion, whichis much easier to analyze.

Other solvents generally dissolve less of the coal, withsolvents like toluene dissolving only a small percentageand hexane being ineffective. Less material can be ex-tracted from coals by chloroform than the portion of thepyridine extract which is soluble in chloroform. If a coaland the pyridine extract of a coal are extracted with chloro-form, different amounts of material go into solution. Thegreater percentage of the starting coal is dissolved by firstextracting with pyridine. This will be explained in the nextsection.

The solvent power of dense, supercritical fluids is oftengreater than that of the corresponding liquid. Supercriticalorganic fluids at temperatures above 350◦C are often quitegood solvents for coals, and, indeed, a conversion processbased on supercritical fluid extraction has received exten-sive development in Britain. It seems probable that somechemical decomposition is occurring that leads to the high

P1: GNB Final



FIGURE 3 Extraction yield in pyridine at 115◦C vs rank.[Reprinted with permission from Van Krevelen, D. W. (1981).“Coal,” Elsevier, New York.]

extractabilities observed, often approaching 40% of thecoal when toluene is the solvent.

D. Macromolecular Nature of Coals

Coals are believed to be three-dimensionally cross-linkedmacromolecular networks containing dissolved organicmaterial that can be removed by extraction. This modeloffers the most detailed and complete explanation of thechemical and mechanical behavior of coals. It is a rela-tively recent model and is somewhat controversial at thiswriting. The insoluble portion of the coal comprises thecross-linked network, one extraordinarily large moleculelinked in a three-dimensional array. This network is heldtogether by covalent bonds and hydrogen bonds, the weakinteractions that play such a large role in the associationof biological molecules. The extractable portion of thecoal is simply dissolved in this solid, insoluble frame-work. A solvent like pyridine, which forms strong hydro-gen bonds, breaks the weaker hydrogen bonds within thecoal structure. In the presence of pyridine, the networkis less tightly held together. Molecules not tied into thenetwork by covalent bonds can be removed. A solventlike chloroform, which does not break hydrogen bonds incoals, cannot extract all of the material that is soluble init, because some of the soluble molecules are too large tomove through the network and be removed. They do comeout of the coal in pyridine, which expands the network.Once removed from the network, they readily dissolve inchloroform.

The macromolecular structure controls many importantcoal properties but has been relatively little studied. Thenative coal is a hard, brittle, viscoelastic material. This isbecause it is a highly cross-linked solid. The network isheld together by both covalent bonds and a large number ofhydrogen bonds, which tie otherwise independent parts ofthe network together. The network is frozen in place andis probably glassy. Diffusion rates are low because thenetwork cannot move to allow passage of the diffusingmolecule.

If the internal hydrogen bonds are destroyed, either bychemical derivatization or by complexation with a solventthat is a strong hydrogen bond acceptor, the properties ofcoals change enormously. The network is now quite flex-ible, and the coal is a rubbery solid, which is no longerbrittle. Diffusion rates are enormously enhanced. All me-chanical properties have been altered. All of this is dueto the freeing of those portions of the network that wereheld to one another by the hydrogen bonds. This greatly in-creased freedom of motion changes the fundamental char-acter of the solid.

VI. CHEMICAL STRUCTURE

It is impossible to represent accurately and usefully a ma-terial as complex as coal using classical chemical struc-tures. A model large enough to show the distribution of thevery many structures present in a coal would be too largeand unwiedly to be useful. We shall discuss the occur-rence of the various coal functional groups while ignoringtheir immediate environment. A phenolic hydroxyl will belisted as a phenolic hydroxyl irrespective of whether it isthe only functional group on a benzene ring or whether itshares an anthracene nucleus with several other hydroxylgroups. Since the distribution of hydroxyl environmentsis unknown, there is no alternative.

The elements of interest are, in increasing order of struc-tural importance, nitrogen, sulfur, oxygen, and carbon.The chemistry of hydrogen, the remaining major compo-nent, is governed by its position on one of these elementsand need not be discussed separately. Trace elements willbe ignored, and general mineral mater composition is cov-ered in Section II.

A. Nitrogen

Nitrogen is a minor constituent of most coals, being lessthan 2% by weight. Its principal importance is that it maylead to the formation of nitrogen oxides on combustion.Its origin is believed to be amino acids and porphyrinsin the plant materials from which coals were derived. Re-cently, X-ray methods have been developed which allowed

P1: GNB Final



FIGURE 4 Organic nitrogen forms as a function of rank for Ar-gonne Premium Coals. [Reprinted with permission from Kelemen,S. R., Gorbaty, M. L., and Kwiatek, P. J. (1994). Energy Fuels 8,896. Copyright American Chemical Society.]

the accurate speciation and quantification of the forms oforganically bound nitrogen, including the basic pyridinesand quinolines and the weakly acidic or neutral indoles andcarbazoles. X-ray photoelectron spectroscopy (XPS) datashowing the quantitative distributions of nitrogen typesin a suite of well-preserved coals representative of manyranks (Argonne Premium Coals) are shown in Fig. 4 as afunction of weight percent carbon. Pyrrolic nitrogen wasfound to be the most abundant form of organically boundnitrogen, followed by pyridinic, and then quaternary types.It is clear from Fig. 4 that the distributions are rank depen-dent, with the highest abundance of quaternary nitrogenin the lowest rank coals. The quaternary species are at-tributed to protonated pyridinic or basic nitrogen speciesassociated with hydroxyl groups from carboxylic acids orphenols. The concentrations of quaternary nitrogen de-crease, while pyridinic or basic nitrogen forms appearto increase correspondingly as a function of increasingrank.

B. Sulfur

The presence of sulfur in coals has great economic conse-quences. When coal is burned, the resulting SO2 and SO3

must be removed from the stack gasses at great expense ordischarged to the atmosphere at great cost to the environ-ment. The presence of sulfur in coals has been the drivingforce for a great deal of research directed toward its re-moval. Numerous processes for removing mineral matter(including inorganic sulfur compounds) from coal are inuse, as are processes for removing sulfur oxides from stackgasses. Much work has been done on chemical methodsfor removing sulfur from coals before burning, but noneare in large-scale use.

The origin of sulfur in most coals is believed to be sul-fate ion, derived from seawater. During the earliest stagesof coal formation, bacterial decomposition of the coal-forming plant deposits occurs. Some of these bacteria re-duce sulfate to sulfide. This immediately reacts with ironto form pyrite, the principal inorganic form of sulfur incoals. It is also incorporated into the organic portion ofthe coal. The amount and form of sulfur in coals dependmuch more on the coal’s depositional environment thanon its age or rank. In this sense, it is largely a coal-typeparameter, not a rank parameter.

Coals contain a mixture of organic and inorganic sulfur.The inorganic sulfur is chiefly pyrite (fool’s gold). Expo-sure of coal seams to air and water results in the oxidationof the pyrite to sulfate and sulfuric acid, causing acid minedrainage from open, wet coal seams. Much of the pyritecan often be removed by grinding the coal and carryingout a physical separation, usually based on the differencein density between the organic portion of the coal and themore dense mineral matter. Coal cleaning is quite com-mon, is often economically beneficial, and is increasing.The organic sulfur in coals, that is, the sulfur bonded tocarbon, is very difficult to remove. Although many pro-cesses have been developed for this purpose, none hasbeen commercially successful.

In the last decade, major advances have been made inspeciating and quantifying forms of organically boundsulfur in coals into aliphatic (i.e., dialkylsulfides) and aro-matic (i.e., thiophenes and diarylsulfides) forms. Repre-sentative results on a well-preserved suite of coals ob-tained from XPS and two different sulfur K-edge X-rayabsorption near edge structure spectroscopy (S-XANES)analysis methods are plotted together in Fig. 5. The ac-curacy of the latter methods is reported to be ±10 mol%.In Fig. 5, aromatic sulfur forms are plotted against weightpercent carbon. It is seen that in low-rank coals there arerelatively low levels of aromatic sulfur (i.e., significantamounts of aliphatic sulfur) and that levels of aromaticsulfur increase directly as a function of increasing rank.The X-ray methods have been used to follow the chem-istry of organically bound sulfur under mild oxidation inair, where it was shown that the aliphatic sulfur oxidizesin air much more rapidly than the aromatic sulfur, ex situ

P1: GNB Final



FIGURE 5 Aromatic sulfur forms as a function of rank for Argonne Premium Coals: �, � XANES data; ◦◦ XPS data.[Reprinted with permission from Gorbaty, M. L. (1994). Fuel 73, 1819.]

and in situ pyrolysis, single electron transfer, and strongbase conditions.

C. Oxygen

Oxygen is a major component of the plant materials thatform coals. The principal precursor of coals is proba-bly lignin, the structural material of plants. Most of thecellulose and starch is probably destroyed in the earlieststages of deposition and diagenesis, although a contro-versy over this continues. The oxygen content of coalsdecreases steadily as the rank of the coals increases. Goodanalytical techniques exist for the important oxygen func-tional groups, so their populations as a function of rankare known (Fig. 6). The detailed chemistry of the loss andtransformations of these functional groups remains largelyunknown.

The carboxylate (—COOH) group is the major oxy-gen functionality in lignites but is absent in many bitu-minous coals. The mechanism for its rapid loss is un-known. Its presence contributes much to the chemistryof lignites and low-rank coals. It is a weak acid and inmany coals is present as the anion (—COO−) complexedwith a metal ion. Low-rank coals are naturally occurringion-exchange resins. Certain lignites are respectable ura-nium ores, having removed the uranium from ground waterflowing through the seam by ion exchange.

Lignites contain water and can be considered to behydrous gels. Removing the water causes irreversiblechanges in their structure. The water is an integral part

of the native structure of the lignites, not just somethingcomplexed to the surface or present in pores. The natureof the water in the structure is unknown. Its presence isundoubtedly due to the presence of a large number of polaroxygen functional groups in the coal.

Hydroxyl (—OH) is present in phenols or carboxylicacids. It is present in all coals except anthracites. It isslowly lost during the coalification process. The phenolsare much less acidic than the carboxylic acids and remainun-ionized in native coals.

FIGURE 6 Distribution of oxygen functionality in coals: DMMF,dry, mineral-matter-free. [Reprinted with permission from White-hurst, D. D., Mitchell, T. O., and Farcasiu, M. (1980). “Coal Liq-uefaction: The Chemistry and Technology of thermal Processes,”Academic Press, New York.]

P1: GNB Final



The methoxyl group (—OCH3) is a minor componentwhose loss parallels that of carboxylate. The variety ofgroups containing the carbonyl function (✘�C 0) com-prise a small portion of the oxygen.

In some ways, the ether functionality (—O—) is themost interesting. Its loss during coalification is not con-tinuous. The pathway by which it is created is unknown.It is also uncertain how much of the ether oxygen is boundto aliphatic carbon and how much to aromatic carbon, buttechniques capable of providing an answer to this questionhave just arrived on the scene.

D. Carbon

There are two principal types of carbon in coals: aliphatic,the same type of carbon that is found in kerosene and thatpredominates in petroleum, and aromatic, the kind of car-bon found in graphite. Many techniques have been used toestimate the relative amounts of the two types of carbon incoals, but only recently has the direct measurement of thetwo been possible. Their variation with rank is illustratedin Fig. 7.

The carbon in coals is predominantly aromatic and isfound in aromatic ring structures, examples of which are

FIGURE 7 Plot of fraction of aromatic carbon ( fa) vs percentcarbon for 63 coals and coal macerals. [Reprinted with permissionfrom Wilson, M. A., Pugmire. R. J., Karas, J., Alemany, L. B.,Wolfenden, W. R., Grant, D. M., and Given, P. H. (1984). Anal.Chem. 56, 933.]

given below. It is very desirable to know the frequencyof the occurrence of these ring systems in coals, andmany techniques have been applied to this problem. Mostof the techniques involve rather vigorous degradationchemistry, advanced instrumental techniques, and manyassumptions.

The distribution of aromatic ring systems in coals hasbeen estimated for the same Argonne Premium Coal sam-ples discussed previously, based on 13C NMR experi-ments, and these data, expressed as number of aromaticcarbons per cluster (error ≈ ±3), are shown in Fig. 8. Forsamples in this suite varying from lignite to low volatilebituminous, the results show that the average number ofaromatic carbons per ring cluster ranged from between 9(lignite) and 20 carbons (lvb), with most subbituminousand bituminous coals of this suite containing 14. This in-dicates an average size of 2–3 rings per cluster for mostof these coals, with the lignite having perhaps an averageof 1–2 and the highest rank coal having 3–4.

Our knowledge of the aliphatic structures in coals iscurrently inadequate. In many low-rank coals, there is asignificant amount of long-chain aliphatic material. Thereis a debate as to whether this material is bound to the coalmacromolecular structure or whether it is simply tangledup with it and thus trapped.

Much of the aliphatic carbon is present in ring systemsadjacent to aromatic rings. Some is bonded to oxygen andin short chains of —CH2— groups linking together aro-matic ring systems. No accurate distribution of aliphaticcarbon for any coal is known.

E. Maceral Chemical Structure

There are four principal maceral groups. The vitrinite mac-erals are most important in North American coals, com-prising 50–90% of the organic material. They are derivedprimarily from woody plant tissue. Since they are the prin-cipal component of most coals, vitrinite maceral chemistryusually dominates the chemistry of the whole coal. Iner-tinite maceral family members comprise between 5 and40% of North American coals. These materials are not atall chemically inert, as the name implies. They are usuallyless reactive than the other macerals, but their chemicalreactivity can still be quite high. They are thought to bederived from degraded woody tissue. One member of thisfamily, fusinite, looks like charcoal and may be the resultof ancient fires at the time of the original deposition ofthe plant material. The liptinite maceral group makes up5–20% of the coals in question. Its origin is plant resins,spores, and pollens, resinous and waxy materials that giverise to macerals rich in hydrogen and aliphatic structures.Terpenes and plant lipid resins give rise to the variedgroup of resinite macerals. From some unusual western

P1: GNB Final



FIGURE 8 Aromatic cluster sizes in Argonne Premium Coals. [Reprinted with permission from Solum, M. S., Pugmire,R. J., Grant, D. M. (1989). Energy Fuels 3, 187. Copyright American Chemical Society.]

U.S. coals, resinites have been isolated and commerciallymarketed.

The rank dependence of the elemental composition ofthe maceral groups is shown in Fig. 9. The inertinites arethe most aromatic, followed by the vitrinites and then theliptinites. Their oxygen contents decrease in the order vit-

FIGURE 9 Elemental composition of the three main maceralgroups. [Reprinted with permission from Winans, R. E., andCrelling, J. C. (1984). “Chemistry and characterization of coalmacerals.” ACS Symp. Ser. 252, 1.]

rinite > liptinite > inertinite. It is probably impossible tocome up with a universally accurate reactivity scale forthese materials. The conditions and structures vary toowidely. Reactivity refers to the capacity of a substanceto undergo a chemical reaction and usually refers to thespeed with which the reaction occurs. The faster the reac-tion, the greater is the reactivity. The reactivity of differentcompounds changes with the chemical reaction occurringand thus must be defined with respect to a given reactionor a related set of reactions. The reactivity of maceralsas substrates for direct liquefaction and their reactivity ashydrogen donors are parallel and in the order liptinite >

vitrinite > inertinite > resinite. The solubility of theresinites is generally high, so their low reactivity makeslittle difference for most coal processing.

F. Structure Models

A number of individuals have used the structural infor-mation available to create representative coal structures.All these representations contain the groups thought to oc-cur most frequently in coals of the rank and compositionportrayed. Formulating any such structure necessarily in-volves making many guesses. Taken in the proper spirit,they are useful for predicting some types of chemistry andare a gauge of current knowledge. One such structure isshown in Fig. 10. It is typical of recent structure proposals

P1: GNB Final



FIGURE 10 Molecular structure for a bituminous coal. [Reprintedwith permission from Shinn, J. (1984). Fuel 63, 1284.]

for bituminous coals. These structures must be used cau-tiously. They are meant to be representative; they are notcomplete structures containing all groups actually present.

VII. COAL CHEMISTRY

The basic chemical problem in coal conversion to gaseousor liquid products can be viewed as management of thehydrogen-to-carbon atomic ratio (H/C). As Fig. 11 shows,conventional liquid and gaseous fuels have atomic H/Cratios higher than that of coal. Thus, in very general terms,to convert coal to liquid or gaseous fuels one must eitheradd hydrogen or remove carbon.

A. Gasification

Coal can be converted to a combustible gas by reactionwith steam,

C + H2O −→ CO + H2,

reports of which can be traced back to at least 1780.The reaction is endothermic by 32 kcal/mol and requirestemperatures greater than 800◦C for reasonable reactionrates.

The major carbon gasification reactions are shownin Table VI, and selected relative rates are shown inTable VII. The products of the carbon–steam reaction,carbon monoxide and hydrogen, called synthesis gas orsyngas, can be burned or converted to methane:

ReactionName Reaction ∆H298 temp. (◦◦C)

Carbon–steam 2C + 2H2O −−−−−−→ 2CO + 2H2 62.76 950

Water–gas shift CO + H2Ocatalyst−−−−−−→ CO2 + H2 −9.83 325

Methanation CO + 3H2

catalyst−−−−−−→ CH4 + H2O −49.27 3752C + 2H2O −−−−−−→ CH4 + CO2 3.66

In this scheme, hydrogen from water is added to carbon,and the oxygen of the water is rejected as carbon dioxide.

To make this chemistry a viable technology, many othersteps are involved, a key one being acid gas and ammo-nia removal from raw syngas. Coal is not made only ofcarbon, hydrogen, and oxygen, but also contains sulfurand nitrogen. During gasification the sulfur is convertedto hydrogen sulfide and the nitrogen to ammonia. Carbondioxide is also produced. These gases must be removedfrom the raw syngas because they adversely affect the cat-alysts used for subsequent reactions.

Another issue in coal gasification is thermal efficiency.The thermodynamics of the reactions written above indi-cate that the overall C → CH4 conversion should be ther-moneutral. However, in practice, the exothermic heat ofthe water–gas shift and methanation reactions is gener-ated at a lower temperature than the carbon–steam reac-tion and therefore cannot be utilized efficiently. Recentcatalytic gasification approaches are aimed at overcomingthese thermal inefficiencies.

Coal gasification plants are in operation today and areused mainly for the production of hydrogen for ammo-nia synthesis and the production of carbon monoxide forchemical processes and for syngas feed for hydrocarbonsynthesis.

B. Liquefaction

As already discussed, coal is a solid, three-dimensional,cross-linked network. Converting it to liquids requires thebreaking of covalent bonds and the removal of carbon orthe addition of hydrogen. The former method of producing

TABLE VI Carbon Gasification Reactions

∆H298

Name Reaction (kcal/mol)

Combustion C + O2 → CO2 −94.03

C + 12 O2 → CO −26.62

CO + 12 O2 → CO2 −67.41

Carbon–steam C + H2O → CO + H2 31.38

Shift CO + H2O → CO2 + H2 −9.83

Boudouard C + CO2 → 2CO 41.21

Hydrogenation C + 2H2 → CH4 −17.87

Methanation CO + 3H2 → CH4 + H2O −49.27

P1: GNB Final



FIGURE 11 Relationship of resource to end products on the basis of the ratio of atomic hydrogen to carbon.

liquid is called pyrolysis, the latter hydroliquefaction. Coalpyrolysis, or destructive distillation, is an old technologythat started on a commercial scale during the industrialrevolution. Hydroliquefaction, the reaction of hydrogenwith coal to make liquids, was first reported in 1869 andwas developed into a commercial process in the periodfrom 1910 to 1920.

In general, both pyrolysis and hydroliquefaction reac-tions begin with the same step, thermal homolytic bondcleavage to produce free radicals, effecting a molecularweight reduction of the parent macromolecule (Fig. 12).If the radicals are healed, smaller neutral molecules re-sult, leading to liquid and gaseous products. Those rad-icals that are not healed recombine to form products ofthe same or higher molecular weight than the parent,leading eventually to a highly cross-linked carbonaceousnetwork called coke (if the material passed through aplastic phase) or char. In pyrolysis, radicals are healedby whatever hydrogen is present in the starting coal. Inhydroliquefaction, excess hydrogen is usually added asmolecular hydrogen and/or as molecules (such as 1,2,3,4-tetrahydronaphthalene) that are able to donate hydrogento the system. Thus, hydroliquefaction produces largeramounts of liquid and gaseous products than pyrolysisat the expense of additional hydrogen consumption. Asshown in Table VIII, conventional pyrolysis takes place attemperatures higher than those of hydroliquefaction, buthydroliquefaction requires much higher pressures.

Many techniques have been used in hydroliquefaction.They all share the same thermal initiation step but differin how hydrogen is provided: from molecular hydrogen,either catalytically or noncatalytically, or from organic

TABLE VII Approximate Relative Rates ofCarbon Gasification Reactions (800◦C, 0.1 atm)

Reaction Relative rate

C + O2 −→ CO2 1 × 105

C + H2O −→ CO + H2 3

C + CO2 −→ 2 CO 1

C + 2 H2 −→ CH4 3 × 10−3

donor molecules. Obviously, rank and type determine aparticular coal’s response to pyrolysis and hydroliquefac-tion (Table IX), and the severity of the processing con-ditions determines the extent of conversion and productselectivity and quality.

A large number of pyrolysis and hydroliquefaction pro-cesses have been and continue to be developed. Pyroly-sis is commercial today in the sense that metallurgicalcoke production is a pyrolysis process. Hydroliquefac-tion is not commercial, because the economics today areunfavorable.

C. Hydrocarbon Synthesis

Liquids can be made from coal indirectly by first gasifyingthe coal to make carbon monoxide and hydrogen, followedby hydrocarbon synthesis using chemistry discovered byFischer and Tropsch in the early 1920s. Today, commer-cialsize plants in South Africa use this approach for mak-ing liquids and chemicals. Variations of hydrocarbon syn-thesis have been developed for the production of alcohols,specifically methanol. In this context, methanation andmethanol synthesis are special cases of a more general re-action. Catalyst, temperature, pressure, and reactor design(i.e., fixed or fluid bed) are all critical variables in deter-minig conversions and product selectivities (Table X).

n CO + (2n + 1)H2Ocatalyst−−−−→ CnH2n+2 + n H2O

2n CO + (n + 1)H2

catalyst−−−−→ CnH2n+2 + n CO2

n CO + 2n H2

catalyst−−−−→ CnH2n + n H2O

n CO + 2n H2

catalyst−−−−→ CnH2n+1OH + (n − 1)H2O

TABLE VIII Typical Coal Liquefaction Conductions

Reaction Temperature (◦◦C) Pressure (psig)

Pyrolysis 500–650 50

Hydroliquefaction 400–480 1500–2500

P1: GNB Final



FIGURE 12 Schematic representation of coal liquefaction.

Clearly, a broad spectrum of products can be obtained,and research today is aimed at defining catalysts and con-ditions to improve the product selectivity. One recent ad-vance in selectivity improvement involves gasoline syn-thesis over zeolite catalyst from methanol, which is madeselectively from syngas. The good selectivity to gasolineis ascribed to the shape-selective zeolite ZSM-5, whichcontrols the size of the molecules synthesized. CalledMTG (methanol to gasoline), the process operates at about400◦C and 25 psig. Large-scale demonstration plants arecurrently under construction in New Zealand and the Fed-eral Republic of Germany.

D. Coke Formation

The most important nonfuel use of coal is the formationof metallurgical coke. Only a few coals form commer-cially useful cokes, while many more agglomerate whenheated. Good coking coals sell at a premium price. Thesupply of coals suitable for commercial coke making issufficiently small so that blends of good and less goodcoking coals are used. Desirable coke properties are alsoobtained by blending. The coking ability of the blends canbe estimated from the maceral composition and rank of thecoals used in the blends. The best coking coals are bitu-minous coals having ∼86% carbon. Other requirementsfor a commercial coking coal or coal blend are low sulfur

TABLE IX Variation of Hydroliquefaction Conversion withRank

Batch hydroliquefaction

Conversion to liquids Barrelsand gas (%% of coal, oil/ton

Coal rank dry ash-free basis) coal

Lignite 86 3.54

High-volatile bituminous-1 87 4.68

High-volatile bituminous-2 92 4.90

Medium-volatile bituminous 44 2.10

Anthracite 9 0.44

and mineral matter contents and the absence of certain el-ements that would have a deleterious effect on the iron orsteel produced.

In coke making, a long, tall, narrow (18-in.) oven isfilled with the selected coal or coal blend, and it is heatedvery slowly from the sides. At ∼350◦C, the coal startsto soften and contract. When the temperature increasesto between 50 and 75◦C above that, the coal begins tosoften and swell. Gases and small organic molecules (coaltar) are emitted as the coal decomposes to form an eas-ily deformable plastic mass. It actually forms a viscous,structured plastic phase containing highly organized liquidcrystalline regions called mesophase. With further heat-ing, the coal polymerizes to form a solid, which on furtherheating forms metallurgical coke. The chemical processis extraordinarily complex.

TABLE X Hydrocarbon Synthesis Product Distributions(Reduced Iron Catalyst)

Variable Fixed bed Fluid bed

Temperature 220–240◦C 320–340◦C

H2/CO feed ratio 1.7 3.5

Product selectivity (%)Gases (C1–C4) 22 52.7Liquids 75.7 39Nonacid chemicals 2.3 7.3Acids — 1.0

Gases (% total product)C1 7.8 13.1C2 3.2 10.2C3 6.1 16.2C4 4.9 13.2

Liquids (as % of liquid products)C3–C4 (liquified petroleum gas) 5.6 7.7C5–C11 (gasoline) 33.4 72.3C12–C20 (diesel) 16.6 3.4Waxy oils 10.3 3.0Medium wax (mp 60–65◦C) 11.8 —Hard wax (mp 95–100◦C) 18.0 —Alcohols and ketones 4.3 12.6Organic acids — 1.0

Paraffin/olefin ratio ∼1:1 ∼0.25:1

P1: GNB Final



The initial softening is probably due to a mix of physi-cal and chemical changes in the coal. At the temperaturewhere swelling begins, decomposition is occurring and thecoal structure is breaking down. Heat-induced decompo-sition continues and the three-dimensionally linked struc-ture is destroyed, producing a variety of mobile smallermolecules that are oriented by the large planar aromaticsystems they contain. They tend to form parallel stacks.Eventually, these repolymerize, forming coke.


COAL GEOLOGY • COAL PREPARATION • GEOCHEMISTRY,ORGANIC • FOSSIL FUEL POWER STATIONS—COAL UTI-LIZATION • MINING ENGINEERING • POLLUTION, AIR

BIBLIOGRAPHY

Berkowitz, N. (1979). “As Introduction to Coal Technology,” AcademicPress, New York.

Elliott, M. A., ed. (1981). “Chemistry of Coal Utilization,” 2nd suppl.vol., Wiley (Interscience), New York.

Francis, W. (1961). “Coal: Its Formation and Composition,” Arnold,London.

George, G. N., Gorbaty, M. L., Kelemen, S. R., and Sansone, M.(1991). “Direct determination and quantification of sulfur forms incoals from the Argonne Premium Sample Program,” Energy Fuels 5,93–97.

Schobert, H. H., Bartle, K. D., and Lynch, L. J. (1991). “Coal Sci-ence II,” ACS Symposium Series 461, American Chemical Society,Washington, DC.

Gorbaty, M. L. (1994). “Prominent frontiers of coal science: Past, presentand future,” Fuel 73, 1819.

Van Krevelen, D. W. (1961). “Coal,” Elsevier, Amsterdam.

P1: GPJ Final Pages Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology EN005K-222 June 15, 2001 20:35

Energy Efficiency ComparisonsAmong Countries

Dian PhylipsenEcofys

I. Measuring Energy ConsumptionII. Measuring Economic Activity

III. Energy Efficiency IndicatorsIV. Accounting for Sector StructureV. Time Series

VI. Reaggregation of Energy Efficiency Indicators

GLOSSARY

Energy efficiency A measure of the amount of energyconsumed per unit of a clearly defined human activity.

Energy efficiency indicator An approximation of energyefficiency such as the energy intensity or the specificenergy consumption. Most efficiency indicators mea-sure the inverse efficiency; if the efficiency increases,the value of the indicator decreases.

Energy intensity An energy efficiency indicator, usuallybased on economic measures of activity, such as valueadded. Energy intensity generally measures a combi-nation of energy efficiency and structure. The energyintensity is often used at a higher level of aggregationthan the specific energy consumption.

Specific energy consumption (SEC) An energy effi-ciency indicator, usually based on a physical measurefor activity, such as the production volume. The SEC

is also called energy intensity when it measures a com-bination of energy efficiency and sector structure.

Structure The mix of activities or products within a coun-try or sector.

ENERGY EFFICIENCY comparison in the manufactur-ing industry is a tool that can be used by industry to as-sess its performance relative to that of its competitors.For this purpose it has been used for many years by thepetrochemical industry and refineries, and interest on thepart of other industries is growing. The tool can also beused by policy makers when designing policies to re-duce greenhouse gas emissions and prioritizing energysavings options. International comparisons of energy effi-ciency can provide a benchmark against which a country’sperformance can be measured and policies can be evalu-ated. This is done both by national governments (such as

433

P1: GPJ Final Pages


434 Energy Efficiency Comparisons Among Countries

the Netherlands and South Korea) and by internationalorganizations such as the IEA (the International EnergyAgency) and APERC (the Asian Pacific Energy ResearchCenter).

I. MEASURING ENERGY CONSUMPTION

A. System Boundaries

The numerator of the energy efficiency indicator measuresenergy consumption, which can be defined in many ways:

End-use energy measures the energy used for each finalapplication. With surveys of energy consumption and theequipment used (as well as information on frequency ofuse, home size, etc.), many of the key end uses in homes,buildings, vehicles, and even in complex industrial pro-cesses can be determined.

Final energy is the energy purchased by final energyusers, excluding on-site conversion losses, measured aspurchased energy − the consumption by in-plant conver-sion processes + the production by in-plant conversionprocesses. The difference between end-use energy and fi-nal energy is actually that more conversions are taken intoaccount for the latter.

Purchased energy the amount of energy bought fromsuppliers. In cases where extraction of energy carrierstakes place within the industry itself (e.g., wood in thepaper industry or hydropower), purchased energy will un-derestimate the actual energy consumption. Purchased en-ergy overestimates actual energy consumption in caseswhere part of the purchased energy is exported to othersectors after conversion (e.g., coke oven gas in steel pro-duction, excess electricity from combined heat and powergeneration).

Net available energy is the amount of energy that ac-tually becomes available to the user, including both netpurchased energy, stock changes, and the user’s extrac-tion of energy forms.

Demand for primary energy carriers cannot be mea-sured but must be calculated on the basis of, for example,final energy consumption and knowledge of flows in theenergy sector itself. The simplest approach used is that ofadding sectoral fuel consumption to the electricity con-sumption using a conversion efficiency for electric powerplants. More sophisticated methods also take into accountthe losses in production, conversion, and transportation ofthe fuels. Whether to use uniform conversion, factors forall countries and for different points in time or country-dependent and time-dependent factors depends on the goalof the analysis.

B. Units

Energy is measured in many different units around theworld. In addition to the SI (Systeme Internationale) unit,the joule, energy can measured in calories, British thermalunits (Btu), tonnes of oil equivalent (toe), tonnes of coalequivalent (tce), kilowatt-hours (kWh), and in physicalunits (e.g., cubic meters of gas, barrels of oil). The use ofthese physical units poses a problem, because the energydensity of energy carriers can vary substantially betweencountries. For the other units, it suffices to list the SI unit,if necessary, in combination with a local unit.

C. Heating Values

Energy may be measured in terms of the lower heatingvalue (LHV) or the higher heating value (HHV). Higherheating values measure the heat that is freed by the com-bustion of fuels when the chemically formed water is con-densed. Lower heating values measure the heat of com-bustlon when the water formed remains gaseous. Althoughthe energy difference between HHV and LHV may not al-ways be technically recoverable, HHV is a better measureof the energy inefficiency of processes.

II. MEASURING ECONOMIC ACTIVITY

Economic activity within the industry sector can be mea-sured in either physical or economic terms.

A. Economic Measures

Economic indicators that can be used to measure the ac-tivity in the industry sector are as follows:

Gross output is the most comprehensive measure ofproduction, including sales or receipts and other operatingincome plus inventory changes.

The value of shipments is a measure of the gross eco-nomic activity of a particular sector, including total re-ceipts for manufactured products, products bought andsold without further manufacturing, and rendered services.As such, it includes the costs of inputs used in that sector.

The value of production measures the value of theamount of goods produced, instead of the amount of goodssold and is calculated by correcting the value of shipmentsfor inventory changes.

Value added is defined as the incremental value addedby sectors or industrial processes, usually measured bythe price differential between total cost of the input andprice of the output (including services). This is also calledgross value added. A net value added can also be definedby subtracting expenditures for depreciation from gross

P1: GPJ Final Pages


Energy Efficiency Comparisons Among Countries 435

value added. Another distinction can be made into (netor gross) value added at market prices or at factor costs.The latter is calculated by subtracting indirect taxes andsubsidies from the former. Usually gross value added atmarket prices is reported.

Value added represents the net economic output for agiven sector or process, and as such the value added forsectors and processes can be added without double count-ing. The total value added of all sectors more or less equalsGDP.

A production index reports the ratio of production ina given sector in a given year to the production in thesame sector in a base year. Often the index is partly phys-ical, partly economic: Physical production is weighed onthe basis of the economic value of the different products.This measure is mainly used for studying changes in onecountry over time.

To be able to compare activity measured in economicterms between countries, national currencies have to beconverted to one single currency. This can be done by us-ing market exchange rates or by using purchasing powerparities. The first approach simply converts currencies intothe reference currency (e.g., U.S. dollars) by applying av-erage annual market exchange rates. Purchasing powerparities (PPP) are the rates of currency conversion usedto equalize the purchasing power of different currencies.For developing countries, PPP-based, estimates of GDPmay be much (2 to 3 times) higher than market exchange-based rates. This means that the same amount of valueadded would comprise more goods and activities in de-veloping countries than it would in OECD countries. Acertain amount of money, converted at PPP rates to na-tional currencies, will buy the same amount of goods andservices in all countries.

If comparisons are made for various years, the devalua-tion of currencies also has to be taken into account. Timeseries are therefore usually expressed in fixed-year prices;actual expenditures are converted to, for example, 1990dollars. Because inflation rates vary from country to coun-try and between industries, national currencies should becorrected for domestic inflation rates before they are con-verted into the reference currency.

B. Physical Measures

Physical production represents the amount of salable prod-uct. The most widely used physical measure of activity isvolume based (i.e., metric tonnes). Physical measures ofactivity are the same for different countries or are easilyconvertible (e.g., short tonnes to metric tonnes). The mea-sure also varies less over time. Therefore, physical mea-

sures of activity are more suitable for energy-intensive in-dustries, such as the iron and steel industry and the cementindustry. For light industry sectors, an economic measureof activity may be more appropriate because the mix ofproducts is very diverse.

III. ENERGY EFFICIENCY INDICATORS

An energy efficiency indicator gives the amount of en-ergy that is required per unit of human activity or pro-duction. Based on the type of demand indicators used,either economic or physical energy efficiency indicatorsare distinguished.

A. Economic Efficiency Indicators

The most aggregate economic energy efficiency indicatoris generally called the energy intensity. A commonly usedeconomic indicator in industry is based on value added.Most national statistics produce figures for value added.Because activities in various sectors are expressed in onemeasure, it is possible to add them, which is an importantadvantage of this type of indicator.

The most important shortcoming of economic indica-tors is the lack of ability to reflect the structural differencesamong countries, or within a country in time. To some ex-tent this can be corrected for by using decomposition anal-ysis (see Section IV.C). However, usually only intersec-toral differences in structure are accounted for in this typeof analysis. In addition, economic efficiency indicatorscan be strongly influenced by changes in product prices,feedstock prices, etc. Furthermore, parts of the economydo not generate value added (as measured in the officialstatistics), such as trade in informal economies (black mar-kets) or barter economies, intermediate products that areused within the same company, etc. For historical analy-ses, we also have the a problem that planned economiesdid not use concepts like GDP and value added.

B. Physical Efficiency Indicators

The physical energy efficiency indicator is called specificenergy consumption (SEC). It is defined as the ratio ofenergy consumption to a measure of human activity inphysical terms, i.e., per weight unit of product.

The advantage of physical indicators is that there is adirect relationship with energy efficiency technology: Theeffect of energy saving measures, in general, is expressedin terms of a decrease of the SEC. Since physical energyefficiency indicators are necessarily developed for a loweraggregation level than economic efficiency indicators, theoccurrence of structural effects is by definition less of a

P1: GPJ Final Pages



problem than for economic indicators. However, at thelevel of product mix structural differences may also playa role for physical indicators. In cases where the SECmeasures a combination of energy efficiency and structure,it can also be referred to as energy intensity.

A problem with physical efficiency indicators is thatit is not easy to count together different types of activ-ity. Furthermore, there is no one-to-one relation betweenphysical indicators and the monetary quantities that formthe basis of economic models.

C. Comparing Economic and Physical EnergyEfficiency Indicators

Studies have shown that for the heavy industries the cor-relation between different economic indicators is poor.Differences can be so large that one type of economicindicator may suggest an increase in energy efficiency,while another type of economic indicator suggests a de-crease in energy efficiency. Comparisons between physi-cal and economic energy efficiency indicators also showlarge differences. No one single economic efficiency in-dicator correlates best with the physical. Rather, correla-tion varies by country and by sector. Generally, physicalenergy efficiency indicators are considered to provide abetter estimate of actual energy efficiency developmentsin the heavy industry.

IV. ACCOUNTING FOR SECTORSTRUCTURE

At the national level, structure is defined by the share thedifferent economic sectors have in gross domestic prod-uct, for exampale, the share of manufacturing versus theshare of the service sector. At a lower level of aggregation,structure is defined by the mix of products and those as-pects that influence product mix and product quality (suchas feedstock).

Because of the influence of sector structure on energyintensity, cross-country or cross-time comparisons cannotbe made based solely on trends in the absolute value ofthe energy efficiency indicator (SEC) for each country.To be able to compare developments in energy efficiencybetween countries and over time, differences and changesin economic structure have to be taken into account. Twoapproaches are used for this, the reference plant approachand decomposition analysis.

A. Reference Plant Approach

In the reference plant approach the actual SEC is comparedto a reference SEC that is based on a country’s sector struc-

ture. A commonly used definition of the reference SECis the lowest SEC of any commercially operating plantworldwide. Such a best plant is defined for each processand a country’s sectoral reference SEC is established bycalculating the weighted average of the reference SECs ofindividual processes and/or products present in that coun-try. This means that both the actual SEC and the referenceSEC are similarly affected by changes in sector structure.

The difference between the actual SEC and the refer-ence SEC is used as a measure of energy efficiency, be-cause it shows which level of energy efficiency is achiev-able in a country with a particular sector structure. Notethat this is a somewhat conservative estimate, since the ref-erence SEC is usually not an absolute minimum require-ment, such as a thermodynamic minimum. Usually, theSEC of the best plant observed worldwide can, technicallyspeaking, still be reduced, but the necessary measures maynot be implemented for a variety of reasons, such as thecontinuity of production, economic considerations, etc. Itmust also be noted that technological and economic lim-itations change over time, resulting in a lower referenceSEC.

The relative differences between actual and referenceSEC can be compared between countries. Usually this isdone by calculating an energy efficiency index (EEI): theratio between actual SEC and reference SEC. The closerthe EEI approaches 1, the higher the efficiency is.

Figure 1 explains the methodology with an examplefor the pulp and paper industry. The SEC is depicted as afunction of the ratio between the amount of pulp producedversus the amount of paper produced. This accounts fordifferences between countries in the use of waste paperas a feedstock for paper production and in import/exportpatterns of the energy-intensive intermediate pulp. In thegraph, called the structure/efficiency plot, both the actual(national average) SEC and the reference SEC are shown(respectively, the upper end and the lower end of the verti-cal lines in Fig. 1). The diagonal line represents the refer-ence SEC at various pulp/paper ratios for a given productmix (such as high-quality printing paper, wrapping paperor newsprint). The reference level of individual countriesmay differ from this line if the manufactured product mixwithin a country is different (such as for country C).

If the countries in Fig. 1 (A, B, C, and D) were to becompared on the basis of the SEC alone, one would tend toconclude that countries A and B are comparably efficient,country C is less efficient, and country D is, by far, the leastefficient of the four countries shown. However, the s/e plotshows that country D’s high specific energy consumptionis largely caused by a more energy-intensive product mix.

A comparison of the relative differences between ac-tual and reference SEC shows that countries A and D areequally efficient. Country B is the most efficient, while

P1: GPJ Final Pages



FIGURE 1 A structure/efficiency (s/e) plot for the paper industry.

country C is the least efficient country. The EEIs of A,B, C, and D are 112, 105, 113, and 121, respectively. AnEEI of 105 means that the SEC is 5% higher than the ref-erence level, so that 5% of energy could be saved at thegiven sector structure by implementing the reference leveltechnology.

B. Decomposition Analysis

Methods have been developed in which changes in over-all energy consumption are decomposed into changes inactivity level, sector structure and energy efficiency. Suchmethods are called decomposition analysis. These meth-ods are based on the following relation:

Et = A∑

Si,t • Ii,t ,

where

Et = sectoral energy consumption in year tA = sectoral output

Si,t = structure of subsector i in year t , measured byAi/A

Ai,t = output of subsector i in year tIi,t = energy intensity of activity i in year t

In decomposition analysis, indices are constructed thatshow what sectoral energy consumption would have beenin year t if only one of the above parameters (A, S, orI ) had changed, while the others remained the same asthe level in the base year (year 0). Depending on the ex-

act methodology used, these indices are called Divisia orLaspeyres indices. The main differences between method-ologies occur in which base year is chosen (e.g., first year,last year, or the middle year of the analysis) and whetherthe base year is fixed or not. Fixed year Laspeyres indicescan be expressed as follows:

L At = At∑

Si,0 • Ii,0

E0,

L St = A0∑

Si,t • Ii,0

E0,

L It = A0∑

Si,0 • Ii,t

E0.

Here, L A measures the change in energy use that wouldhave occurred given actual changes in output, but at fixedstructure and intensity. L S measures the change that wouldhave occurred due to structural change if output and in-tensity had both remained constant; L I shows the devel-opment energy consumption that would have existed ifoutput and structure did not change.

As shown by the use of the term energy intensity inthe equation, decomposition analysis is usually carriedout on the basis of economic energy efficiency indicators.This also brings with it the problems associated with thosetype of indicators as mentioned in Section III. However,recently a method has been developed to carry out a de-composition analysis on the basis of physical indicators.

In this method, activity is measured in physical units.Also the effects of a change in energy efficiency are not

P1: GPJ Final Pages



estimated using the energy intensity, but on the basis of aphysical production index:

PPI =∑

Px • wx

where

PPI = physical production indexPx = production of product x (kg)wx = weighting factor

The weighting factors are chosen in such a way that theyrepresent the amount of energy needed to produce each ofthe products. An appropriate set of weighting factors maybe based on the best plant SECs as used in the referenceplant.

When the weighting factors are the same for all the yearsin the analysis, the development of the PPI represents afrozen efficiency development of the energy consumption.The energy consumption of a sector can then also be writ-ten according to:

∑E =

∑P × PPI∑

P×

∑E

PPI.

Here,∑

P is the parameter for activity (total amount pro-duced in kilograms), PPI/

∑P is related to the composi-

tion of the product mix (structure) and∑

E/PPI is an in-dicator of the energy efficiency (on a physical basis). Notethat this ratio is actually the same as the energy efficiencyindex, EEI, as defined in Section IV.A. Figure 2 shows oneexample of a decomposition analysis on a physical basis.

Note that the countries are both indexed to their own ref-erence situation (in 1973), not to one common reference

FIGURE 2 A decomposition analysis based on physical indicators for the pulp and paper industry. The relativechange (compared to 1973) is shown for total energy consumption in the industry, and for the underlying changes inactivity level, sector structure (product mix), and energy efficiency.

situation. This means that an index of 100% for one coun-try represents a different situation than it does for the othercountry. In the case of the two countries listed in Fig. 2, theJapanese paper industry was more efficient in 1973 thanthe U.S. paper industry. As a result, the 1993 differencesin energy efficiency are larger than Fig. 2 suggests.

V. TIME SERIES

The reference plant approach described in Section IV.Acannot only be used for a cross-country comparison ofenergy efficiency, but also for a cross-time analysis ofchanges in energy efficiency in one or more countries.

To construct a time series, both actual SEC and ref-erence SEC are determined for different points in time.Here too, actual comparison is not based on the absolutelevel of the actual SEC alone, but on the relative differencebetween actual and reference SEC, the EEI.

Figure 3 shows a cross-time and cross-country compar-ison based on this approach. As stated before, a countrybecomes more efficient as EEI approaches 1. Note thatan EEI of 1 can represent a different specific energy con-sumption for different points in time (and for differentcountries).

VI. REAGGREGATION OF ENERGYEFFICIENCY INDICATORS

After energy efficiency indicators have been determinedat a low level of aggregation, often further reduction of

P1: GPJ Final Pages



FIGURE 3 The development of the EEI in the pulp and paper industry over time for various countries. Countries’ actualprimary energy consumption is indexed to the primary energy consumption of the best plant observed worldwide.

the number of indicators into more aggregate figures isdesirable (e.g., one per sector).

One approach for this reaggregation could be to calcu-late an aggregate SEC, based on the actual SEC and thereference SEC of all individual products and processesaccording to the following equation. In doing this, each ofthe products is weighed by their relevance for energy con-sumption (which is, for this aim, more appropriate thanweighing by economic value or product quantity):

secag =∑

mi • SECi∑mi • SECref,i

=∑

Ei∑mi • SECref,i

,

FIGURE 4 The aggregated energy efficiency index for a number of countries. The sectors included are the ironand steel industry, the aluminium industry, the pulp and paper industry, the cement industry, and a major part of thechemical industry (the production of petrochemicals, ammonia, and chlorine).

where

secag = aggregate specific energy consumption (dimen-sionless)

Ei = energy consumption for product imi = production quantity of product i

SECi = specific energy consumption of product iSECref,i = a reference specific energy consumption of

product i

The aggregated SEC equals 1 if all products have a SECthat is equal to the reference SEC. The higher the SECsof each individual products are, the higher the aggregatevalue (and the higher the room for improvement). Figure 4

P1: GPJ Final Pages



shows how such a reaggregation might look for a numberof countries.

Another possibility is to define a physical indicator ac-cording to a “basket of goods” such as 1 tonne of sev-eral primary industrial products. Aggregate specific en-ergy consumption would then be defined as:

SECag =∑

(Wi • SECi ),

where Wi is a weight factor equivalent to the energy con-sumption of industry i , divided by the total energy con-sumption of the basket. One problem with this approacharises when a country does not produce all the products inthe basket.


CARBON CYCLE • ENERGY FLOWS IN ECOLOGY AND IN

THE ECONOMY • ENERGY RESOURCES AND RESERVES

• GREENHOUSE EFFECT AND CLIMATE DATA • GREEN-HOUSE WARMING RESEARCH

BIBLIOGRAPHY

Phylipsen, G. J. M., Blok, K., and Worrell, E. (1998). “Handbook onInternational Comparisons of Energy Efficiency in the ManufacturingIndustry,” Department of Science, Technology and Society, UtrechtUniversity, The Netherlands.

Phylipsen, G. J. M., Blok, K., and Worrell, E. (1998). “Benchmarkingthe Energy Efficiency of the Dutch Energy-Intensive Industry; A Pre-liminary Assessment of the Effects on Energy Consumption and CO2

Emissions,” Department of Science, Technology and Society, UtrechtUniversity, The Netherlands.

Schipper, L., Meyers, S., Howard, R. B., and Steiner, R. (1992). “En-ergy Efficiency and Human Activity: Past Trends, Future Prospects,”Cambridge University Press, Cambridge.

Schipper, L. (1997). “Indicators of Energy Use and Efficiency: Under-standing the Link Between Energy and Human Activity,” InternationalEnergy Agency, Paris.

Schipper, L., and Haas, R. (1997). “Energy Policy Special Issue on Cross-Country Comparisons of Indicators of Energy Use, Energy Efficiencyand CO2 Emissions,” June/July.

P1: GNH/GLT P2: FQP/MAG QC: FYD Final Pages Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology EN005I-223 June 15, 2001 20:45

Energy Flows in Ecologyand in the Economy

Sergio UlgiatiUniversity of Siena

I. IntroductionII. Paradoxes about Energy and Resource UseIII. Is Energy a Scarce Resource?IV. How Do We Evaluate Resource Quality? The

Search for an Integrated Assessment inBiophysical Analyses

V. Toward an Integrated Evaluation ApproachVI. Putting Problems in a Larger PerspectiveVII. Case Study: Industrial Production of Solid

Oxide Fuel Cells and HydrogenVIII. A Basis for Value

IX. Conclusions

GLOSSARY

Cumulative exergy consumption The “total consump-tion of exergy of natural resources connected with thefabrication of the considered product and appearing inall the links of the network of production processes,”according to J. Szargut.

Ecological footprint A concept introduced to portray theenvironmental impact of a process or an economy, bymeasuring the amount of ecologically productive landthat is needed to support it.

The concept has been refined by suggesting that theenvironmental support to a process (i.e., the resources

and environmental services) is measured in terms ofsolar emergy.

Emergy The amount of exergy of one form that is directlyor indirectly required to provide a given flow or storageof exergy or matter. In so doing, inputs to a process aremeasured in common units. Emergy is measured inemjoules (eJ). When inputs are measured in units ofsolar exergy, the solar Emergy is calculated, measuredin solar emergy joules (seJ).

Emergy is a memory of the available energy (ex-ergy) previously used up in a process and is used whenthe global support from biosphere to the process isinvestigated.

441

P1: GNH/GLT P2: FQP/MAG QC: FYD Final Pages


442 Energy Flows in Ecology and in the Economy

Converting flows of different nature into flows ofone kind requires conversion coefficients called trans-formities (seJ/J). Emergy accounting acknowledgesthat different forms of exergy may require a differ-ent environmental support and may show differentproperties.

Exergy The amount of work obtainable when somematter is brought to a state of thermodynamicequilibrium with the common components of thenatural surroundings by means of reversible processes,involving interaction only with the above-mentionedcomponents of nature. Exergy is measured in joules orkcal.

Real processes are irreversible. Therefore, mea-sures of actual work compared with theoreticalvalue provide indicators of exergy losses due to theirreversibility of a process. Exergy is therefore athermodynamic assessment of potential improvementsat the scale of the investigated process.

Life-cycle analysis A technique for assessing the en-vironmental aspects and potential impacts associatedwith a product, by (1) compiling an inventory of rele-vant inputs and outputs of a process, from the extrac-tion of raw resources to the final disposal of the productwhen it is no longer usable (so called “from cradle tograve” analysis); (2) evaluating the potential environ-mental impacts associated with those inputs and out-puts; and (3) interpreting the results of the inventory andimpact phases in relation to the objectives of the study.

Reversible/irreversible A reversible change in a processis a change that can be reversed by an infinitesimalmodification of any driving force acting on the system.The key word “infinitesimal” sharpens the everydaymeaning of the word “reversible” as something thatcan change direction. In other words, the state of thesystem should always be infinitesimally close to theequilibrium state, and the process should be describedas a succession of quasi-equilibrium states.

A system is in equilibrium with its surroundings ifan infinitesimal change in the conditions in oppositedirections results in opposite changes in its state.

It can be proved that a system does maximum workwhen it is working reversibly. Real processes are irre-versible, therefore there is always a loss of availableenergy due to chemical and physical irreversibilities.

Sustainable/sustainability Sustainable development is aform of development that integrates the production pro-cess with resources conservation and environmentalenhancement. It should meet the needs of the presentwithout compromising our ability to meet those of thefuture.

Some of the aspects involved are the relationshipbetween renewable and nonrenewable resources

supporting society, environmental impact in relationto economic actions, intergenerational equity, andindirect impacts of trade.

PATTERNS of energy conversion and use, including en-ergy quality evaluation and sustainable use of the energythat is available, are not easy issues to deal with. A generalconsensus is still far from being reached among energyexperts and economists about evaluation procedures andenergy policy making. The rationale underlying energyevaluation methodologies and the need for integration ofdifferent approaches for a comprehensive assessment arepresented in this paper. How the evaluation approachesand space-time scales of the investigations are linked isstressed. Finally, a case study (industrial production ofhydrogen and fuel cells) is presented and discussed byusing an integrated biophysical approach, to show whateach approach is able to contribute. The impossibility ofachieving a universal theory of value is emphasized, butsynergies achieved by integration of different evaluationprocedures are pointed out.

I. INTRODUCTION

According to Howard Odum; “A tank of gasoline drives acar the same distance regardless of what people are will-ing to pay for it.” It is as a result of Odum’s pioneer-ing work in the sixties and subsequent work that the im-portance of energy as the simultaneous driving force ofour economies accompanied by changing ecosystems hasbeen increasingly recognized. Until recently, neoclassi-cal economics has been unsuitable for understanding therole of the biophysical support provided at no cost by theenvironment to the development of economic processes.Whatever we may be willing to pay for our gallon of gaso-line, if we cannot afford the cost, we are unable to fill thetank. Money flows parallel to biophysical flows support-ing them, so that they must both be accounted for. Theproblem is that while money flows are easily dealt withby using market mechanisms, biophysical flows are not.Economists and policy makers must recognize the role ofenergy and other resource flows in support of human soci-eties. Integration between economic and biophysical toolsfor a better management of available resources is urgentlyneeded and may lead to a very different way of dealingwith production processes and their relationship to theenvironment.

What do we pay for when we trade dollars for gasoline?Dollars are a cumulative measure of labor and servicesinvolved through the whole chain of processes from re-source extraction to refining, storage, and transport to the



Energy Flows in Ecology and in the Economy 443

end user. We pay people for what they do to supply uswith gasoline or food. In none of the steps of any of theseprocesses do we pay nature for making the raw materialor for concentrating it from dispersed states. “We do notpay nature for each acre of land taken out of biologicalproduction, nor do we pay nature for the millions of yearsof work it did in making coal or oil,” M. Gilliland wrotein 1979.

The price of gasoline is highly variable, depending onoil-supply constraints at the sources, the cost of labor andrefining processes, taxation in a given country (these taxesare used for services provided by other sectors of the econ-omy), etc. In May 2000, the U.S. price was about $1.50 pergallon, and the price was about U.S. $4 per gallon in Italy,where heavy taxes (including a carbon tax) are levied onoil products. Refining oil may be much more expensive(cost more money) in Italy than in Saudi Arabia, but theprocess of refining takes about the same amount of energyper liter of gasoline in both countries, since essentially thesame technology is used. Energy costs to make things aremore constraining than current dollar costs. In principle,we may always find a country where the manufacturingprocess is cheaper (i.e., where labor is less expensive), butwe are unlikely to find a country where the energy cost isgreatly lowered, because the best technology is universallypreferred.

Why does money appear to be so important and energynot? It is because money appears to be a scarce resourceand depends on the amount of economic activity a givencountry or a person is able to control. Governments andpeople are not universally convinced that this scarcity alsoapplies to energy and other biophysical resources.

II. PARADOXES ABOUT ENERGYAND RESOURCE USE

T. R. Malthus recognized in 1798 that when a resourceis available, it is used with a boosting effect on systemgrowth. Systems grow using a resource until the resourcebecomes scarce due to the increasing size of the system.In other words, resource availability increases demandwhich, in turn, reduces the resource availability becauseof increasing resource use.

Let us assume now that we are able to save some en-ergy by means of a more efficient domestic heater. In sodoing, we save a corresponding amount of money. A partof this money (say the money saved in the first two tothree years) will cover the cost of the new heater and thesafe disposal of the old one. After that, we have somemoney available for other uses. We may therefore decide topurchase new books, to go more frequently to the movies,to buy new clothes, or to go to an expensive restaurant.

We might invest money in traveling for leisure or in re-modeling our apartment. All of these activities are drivenby labor, material, and energy inputs, resulting in makingnew goods and services available to us. Labor and mate-rial are ultimately supported by an additional energy input.The end of the story is that the energy saved in the heat-ing is expended on the energy invested in making the newheater and the energy which supports the new activities andgoods.

Depending on the kinds of commodities and servicesand their market prices (i.e., the costs of labor and servicessupporting them), we may cause the expenditure of moreenergy than the amount previously saved by using a moreefficient heater.

A different strategy is also possible. The higher effi-ciency of the new heater makes it possible to supply moreheat for the same amount of fuel input. We may thereforedecide to heat our apartment more than in the past to feelmore comfortable and eventually wear very light clothesat home. We will get used to a higher temperature andwill continue this lifestyle even as the heater gets olderand has reduced efficiency. This trend requires increasedenergy input over time. The final result may be the sameor even a higher energy expenditure, in addition to the en-ergy cost of making the new heater. The increased use ofa resource as a consequence of increased efficiency is thewell-known Jevons paradox.

Assume, finally, that increased costs of energy (due totaxes purposefully aimed at decreasing energy use) forceus to run the heater at a lower temperature or for a shortertime (this happens in large Italian cities as the result ofefforts to lower the environmental impacts of combus-tion). We are now unable to save money and thereforewe are unable to invest this money in different kindsof energy-expensive goods. Tax revenues may, however,be reinvested by the government in development plansand services, thereby using energy to support these plans.We save energy, but the government does not. Also inthis last case, no real energy savings occur, nor do weknow if energy consumption remains stable or is likely toincrease.

Only when a decrease of energy use at the individ-ual level is coupled to strategies aimed at lowering thetotal energy expenditures at the country level will therebe decreasing imports or decreasing depletion of localstorages. This scenario happens, for instance, when thedecreased buying power of the local currency of an oil-importing country or the increased oil costs in the inter-national market force a country to buy less oil and applyenergy-conservation strategies, as many European coun-tries did after the 1973 and 1980 oil crises. When sucha case occurs, depletion of oil storages in oil-exportingcountries is slowed down and some oil remains unused




for a while. The oil price may drop down to levels thatother countries are able to afford and thus the use rate willrise again.

What do we learn from these facts? We learn that whena resource is abundant or cheap, it is very likely to be de-pleted quickly by systems which are well organized for thispurpose (Darwinian survival of the fittest). Increased effi-ciency calls for additional consumption. When a resourceis scarce (high cost is a temporary kind of scarcity), usebecomes more efficient to delay (not to prevent) final de-pletion. This trend was clearly recognized in 1922 by A. J.Lotka as a Maximum Power Principle, suggesting that sys-tems self-organize for survival and that under competitiveconditions systems prevail when they develop designs thatallow for maximum flow of available energy. As environ-mental conditions change, the response of the system willadapt so that maximum power output can be maintained. Inthis way, systems tune their thermodynamic performanceaccording to the changing environment.

III. IS ENERGY A SCARCE RESOURCE?

In 1956, M. K. Hubbert pointed out that all finite resourcesfollow the same trend of exploitation. Exploitation is firstvery fast due to abundance and availability of easily ac-cessible and high-quality resources. In the progress of theexploitation process, availability declines and productionis slowed down. Finally, production declines, due to bothlower abundance and lower quality.1 This trend allowsthe investigator to forecast a scenario for each resource,based upon estimates of available reserves and exploita-tion trends. New discoveries of unknown storages willshift the scenario some years into the future, but it cannotultimately be avoided.

Hubbert also predicted the year 1970 for oil productionto peak in the U.S. He likewise predicted the year 2000for oil production to peak on a planetary scale. He wasright in the first case and it seems he was not very far fromright in the second prediction. Based upon this rationaleand energy analysis data, C. A. S. Hall and C. J. Clevelandand Cleveland and co-workers have predicted that U.S. oilwould no longer be a resource soon after the year 2000.Around this time, extraction energy costs would becomehigher than the actual energy yield, due to increased energycosts for research, deep drilling as well as to the lowerquality and accessibility of still available oil storages.

Finally, in 1998 C. Campbell and J. Laherrere predictedthe end of cheap oil within the next decade, based upon

1In principle, if an unlimited energy supply were available, therewould be no material shortages because every substance would becomereusable. This would shift the focus to energy resources, further enhanc-ing their importance.

TABLE I World Proved Reserves and World Consumption ofFossil Fuelsa

World proved Annual world Reserve-to-reservesb consumptionc consumption

(109 tons of oil (109 tons of oil ratio, R/C,Fossil fuel equivalent) equivalent) yearsd

Oil 140 3.46 40.5

Coal 490 2.13 229.9

Natural gas 134 2.06 65.2

Totale 764 8.53 99.8

a BP Amoco, 2000; based on 1999 estimates.b Proved reserves are defined as those quantities which geological

and engineering information indicates with reasonable certainty can berecovered in the future from known reservoirs under existing economicand operating conditions.

c Indicates the amount of the total reserve that is used up in a givenyear.

d Ratio of known reserves at the end of 1999 to energy consumptionin the same year. Indicates how long the reserve will last at the presentconsumption rate. A reserve-to-annual production ratio, R/P, can also beused, with negligible differences due to annual oscillations of consump-tion and production rates.

e Estimated, assuming that oil, coal, and natural gas can be substitutedfor each other. Of course, while this is true for some uses (e.g., use inthermal power plants followed by conversion to electricity), it is not easyfor others unless conversion technologies are applied and some energy islost in conversion (e.g., converting coal to a liquid fuel for transportationuse or to hydrogen for making electricity by using fuel cells).

geological and statistical data. It is important to empha-size here that they did not predict the end of oil produc-tion but only the end of oil production at low cost, withforeseeable consequences on the overall productive struc-ture of our economies running on fossil fuels. The presentfossil resources (known energy storages that can be ex-ploited by means of present technology) and resource-to-consumption ratios are listed in Table I. We are notincluding nuclear energy in Table I, due to the fact that theincreasing environmental concerns and public opinion inmost industrialized countries are reducing the use of thistechnology and the so-called “intrinsically safe nuclearreactor” remains at a research stage of development.

Other authors believe that energy availability is not yetgoing to be a limiting factor for the stability of developedeconomies as well as to the growth of still-developingcountries due to the possible exploitation of lower qualitystorages (e.g., oil from tar sands and oil shales, coal sup-plies) at affordable energy costs. However, they admit thateconomic costs are very likely to increase. Still availablefossil fuels, coupled with energy-conservation strategies,as well as new conversion technologies (among which arecoal gasification and gas reforming for hydrogen produc-tion, and use of fuel cells) might help delay the end of thefossil-fuel era, allowing for a transition period at higher




costs. The transition time may be long enough to allow fora redesigning of economic, social, and population-growthstrategies to come out of the fossil-fuel era in an acceptablemanner. It is still under debate whether the new pattern willbe characterized by additional growth supported by newenergy sources or will be a global and controlled down-sizing of our economies, which was anticipated by H. T.Odum as a “prosperous way down.”

Other kinds of scarcity, especially economic and envi-ronmental factors, are more likely to be strong constraintson energy use. We have already pointed out that an in-crease of the energy price is a kind of scarcity factor. In-deed, it affects availability of energy for those economiesand individuals who are not able to afford higher costs.The economic cost of energy is closely linked to develop-ment, creation of jobs, and possible use of other energysources that could not compete in times of cheap oil andmight provide alternatives for future development in timesof reduced fossil-fuel use. On the other hand, environmen-tal concerns (related to nuclear waste, the greenhouse ef-fect and global warming potential, need for clean urbanair, etc.) affect societal assets in many ways and may al-ter production patterns when they are properly taken intoaccount. All of these issues call for correct and agreedupon procedures of energy-sources evaluation, energy-conversion processes, energy uses, and energy costs.

IV. HOW DO WE EVALUATE RESOURCEQUALITY? THE SEARCH FOR ANINTEGRATED ASSESSMENT INBIOPHYSICAL ANALYSES

The relation between energy and the economy has beendeeply investigated by J. Martinez-Alier who often em-phasized that thermodynamic analyses cannot be the onlybases for policy, nor will they lead to a new theory ofeconomic value. They can, instead, do something that ismuch more important. In addition to providing a biophys-ical basis for economic descriptions, they may help toexpand the scope of economics, away from a single nu-meraire or standard of evaluation toward systems thinkingand a multicriteria framework. For this to be done, it mustbe accepted that solutions of different problems may re-quire different methodologies to be dealt with properly.Each approach may only answer a given set of questionsand may require the support of other methodologies fora more complete view of system behavior. The correctuse and integration of complementary approaches mightease descriptions and enhance understanding of complexsystems dynamics, including human economies.

The demand for integration of methodologies has beengreat during the past decade. Previous studies in the first

two decades of energy analyses (the 1970s and 1980s)were very often devoted to assessing and demonstratingthe superiority of one approach over others. After intro-duction of systems thinking, hierarchy theory, nonlineardynamics, fractal geometry, and complex systems analy-sis, it is now increasingly clear that different approachesare very often required by the very nature of the problemswe deal with to build a set of complementary descriptionsthat may be used to provide different assessments or dif-ferent views of changes defined on different space-timescales. Biophysical analyses make it possible to bridgethese different perspectives. The values of variables de-scribed at a certain level will affect and be affected bythe values of variables describing the system at a differentlevel. A clear assessment and call for integration was madeat the beginning of the 1990s by the economist C. F. Kiker.Comparing available energy (exergy) analysis and emergyaccounting methodologies, he came to the conclusion that“while the backgrounds of available energy and emergyanalysts are substantially different, the underlying con-ceptual bases are sufficiently the same for the two meth-ods to be used jointly. If clear analytical boundaries arespecified for human-controlled systems, for natural sys-tems and for systems derived from the interaction of thesetwo types, available energy analysis and emergy analy-sis can be used together to evaluate the fund of servicefrom these systems.” He concluded that “what is needednow are real world examples that actually do this.” M. T.Brown and R. Herendeen, in 1998, have compared in thesame way the embodied energy approach and the emergyapproach, looking for similarities and trying to make aclear assessment of differences and still unresolved ques-tions. M. Giampietro and co-workers, also in 1998, apply-ing concepts of complex systems theory to socioeconomicsystems, described as adaptive dissipative holarchies,2 em-phasize that “improvements in efficiency can only be de-fined at a particular point in time and space (by adopt-ing a quasi–steady state view); whereas improvementsin adaptability can only be defined from an evolutionaryperspective (by losing track of details). Short-term andlong-term perspectives cannot be collapsed into a singledescription.”

P. Frankl and M. Gamberale, in 1998, presented, theLife-Cycle Assessment (LCA) approach as a tool for en-ergy policy “to identify (future) optimal solutions whichmaximize environmental benefits, . . . to identify priorities

2“ . . . Supersystems (e.g., ecosystems or organisms) consist of sub-systems (organisms and cells). They are nested within each other, andfrom this view are inseparable, since they, though clearly individual,consist of each other,” wrote R. V. O’Neill and co-workers in 1986.Open subsystems of higher order systems are called holons, hierarchiesof holons are called holarchies. They form a continuum from the cell tothe ecosphere.




for orienting research and development activities and iden-tifying best fiscal and market incentive measures.” Theseauthors recall that life-cycle analysis is a technique forassessing the environmental aspects and potential impactsassociated with a product, by compiling an inventory ofrelevant inputs and outputs of a system, evaluating thepotential environmental impacts associated with those in-puts and outputs, and finally interpreting the results of theinventory and impact phases in relation to the goals ofthe study. They emphasize that “LCA results should notbe reduced to a simple overall conclusion, since trade-offs and complexities exist for the systems analyzed atdifferent stages of their life cycle,” and that “there is nosingle method to conduct LCAs . . . LCA is a valuable ana-lytical tool in the decision-making process in conjunctionwith other tools.” It clearly appears that LCA could greatlybenefit from integration procedures with other approachessuch as exergy and emergy evaluations.

A call for the importance of using integrated method-ologies has come from scientists attending the Interna-tional Workshop” Advances in Energy Studies. EnergyFlows in Ecology and Economy” at Porto Venere, Italy,in 1998, to improve definitions of scales and boundaries,methods of analysis, policy initiatives, and future researchneeds required to generate a balanced evaluation of hu-manity in the environment. In a final Workshop document,S. Ulgiati and colleagues challenge the scientific commu-nity to reframe previous insights and research activity intoa systems perspective, enhancing complementarity of ap-proaches, systems thinking, and multi scale, multi criteriaevaluation tools. Finally, understanding the mechanism ofthe environmental support to human wealth and its em-bodiment in the goods that human societies make and usehas been suggested as a research and policy priority todesign sustainable interactions of humans and nature bythe participants in the Second International Workshop on“Advances in Energy Studies. Exploring Supplies, Con-straints, and Strategies,” Porto Venere, Italy, held in 2000.

V. TOWARD AN INTEGRATEDEVALUATION APPROACH

Evaluation procedures so far proposed by many authorshave been applied to different space-time windows of in-terest and were aimed at different investigations and policygoals. Many of these have offered valuable insights towardthe understanding and description of aspects of resourceconversions and use. However, without integration of dif-ferent points of view, it is very unlikely that biophysicaltools can be coupled to economic tools to support envi-ronmentally sound development policies.

We must also be aware that nature and economies areself-organizing systems, where processes are linked andaffect each other at multiple scales. Investigating only thebehavior of a single process and seeking maximization ofone parameter (efficiency, production cost, jobs, etc.) isunlikely to provide sufficient insight for policy making. Itis of primary importance to develop approaches that focuson the process scale and then to expand our view to thelarger scale of the global system in which the process isembedded, to account for the interaction with other up-stream, downstream, and parallel processes.

Finally, systems are dynamic and not static. They mayappear static when a short time scale is chosen as thewindow of interest, but their behavior very often followsoscillating patterns. This also applies to societies. Manyscientists and policy makers suggest that human societiescan grow to a defined state, where resource supply and useare balanced by limiting their rate of depletion to the rateof creation of renewable substitutes. Since the planet as awhole is a self-organizing system, where storages of re-sources are continuously depleted and replaced at differentrates and matter is recycled and organized within a self-organization activity, “sustainability concerns managingand adapting to the frequencies of oscillation of naturalcapital that perform best. Sustainability may not be thelevel ‘steady-state’ of the classical sigmoid curve but theprocess of adapting to oscillation. The human economicsociety may be constrained by the thermodynamics thatis appropriate for each stage of the global oscillation,” ac-cording to Odum3. Evaluation approaches must be able toaccount for this aspect too.

A. Describing Matter Flows

No process description can be provided, at any level,without a clear assessment of matter flows. Either amechanical transformation or a chemical reaction may beinvolved and quantifying input and output mass flows isonly a preliminary step for a careful evaluation. Withoutnuclear transformations, matter is conserved accordingto a fundamental law of nature. Mass cannot be createdor destroyed. Whenever we burn 40 kg of gasoline in ourengine, we are not only supplying energy to run the car,we are also pushing far more than 40 kg of combustioncompounds formed from gasoline into the air. There is a

3When resources are abundant, systems that are able to maximizeresource throughput prevail, no matter their efficiency. Maximum effi-ciency is not always the best strategy for survival. Only when resourcesare scarce, systems self-organize to achieve maximum output by increas-ing efficiency. They therefore choose different thermodynamic strategiesdepending on external constraints and available resources.




complex set of chemical reactions involving carbon andatmospheric oxygen and nitrogen to yield carbon dioxideand nitrogen oxides. The mass of each individual atomicspecies is conserved in the process and the total massof reactants must equal the mass of reaction products.If these changes are carefully accounted for, the processcan be described quite well while we make sure that weare not neglecting any output chemical species that couldbe profitably used or that should be safely disposed of.In addition, when we expand our scale of investigation,we realize that each flow of matter supplied to a processhas been extracted and processed elsewhere. In so doing,additional matter is moved from place to place, processedand then disposed of. For instance, nickel is used inthe ceramic components of some kinds of fuel cells.The fraction of nickel in pentlandite (nickel ore) is only5.5 × 10−3 g Ni/g ore. To supply 1 g of pure nickel toan industrial process requires that at least 181 g ofpentlandite be dug. If the nickel mine is underground,the upper layers of topsoil and clay are also excavated,perhaps a forest is also cut and biotic material degraded.Finally, some process energy in the form of fossil fuelsis used up and a huge amount of process water is veryoften needed as input and formed as output. In addition,combustion also requires oxygen and releases carbondioxide. Accounting for the amount of biotic and abioticmaterial involved in the whole chain of steps supportingthe investigated process has been suggested as a measureof environmental disturbance by the process itself. Wetherefore find two main aspects of the material balance.When focusing on the product side, we must make surethat economically and environmentally significant matterflows have not been neglected. When addressing the inputside, we must allow for the total mass transfer supportinga process and thereby measure indirectly how the processaffects the environment. Both of these two points ofview are based on the law of conservation of mass innonrelativistic systems. Mass demand per unit of product(gin/gout) and total input mass supporting a process aresuitable indicators. Output/input mass ratios (equal toone in the case of careful accounting4) and mass of by-products released per unit of main product (gbyprod/gout)are also needed. After mass flows have been carefullyaccounted for, economic, energy, and environmentalevaluations are made much easier, as with all of them agood process description is clearly the starting point.

4When chemical reactions are involved, all reactants and productsmust be accounted for. Atmospheric oxygen and nitrogen, for instance,are involved in the combustion reaction of fossil fuels via a large numberof elementary reaction steps, with known or estimated reaction rates.Their masses must also be accounted for.

B. Describing Heat Flows

First-law heat accounting should always be coupled tomass accounting as a required step of any biophysical in-vestigation. The same item may be identified both as a ma-terial and heat source. For instance, oil can be measured inkilograms of hydrocarbons or in joules of combustion heatthat it can generate. Both measures are useful, dependingupon the goal of the evaluation. In plastic polymer produc-tion, oil is the substrate to be converted to the final product.Measuring oil in joules when it is more a substrate thana fuel may not be useful. In electricity production, oil isthe fuel. Its combustion releases heat to power the turbineand generate electricity. In both cases, some waste heat isreleased to the environment while the main products aregenerated. In times of cheap oil, waste-heat flows may beneglected, since the focus is on the main product. In timesof decreasing oil availability or increasing costs, there maybe uses for waste heat, depending upon its temperature,and we may then redefine waste heat as a by-product fordomestic heating or some industrial application. However,total input heat flow must always be equal to total outputheat flow for isothermal systems. When heat is producedby combustion to make a product from raw mineral, wehave a mass output in the form of the product and an addedwaste-heat output equal to the input energy used. When theproduct is a form of energy, such as electricity, the outputwill be the sum of waste heat and produced electricity.

Environmental as well as economic concerns may mo-tivate us to investigate the consequences of releasing to theenvironment a resource characterized by a higher tempera-ture than the environmental temperature. To address theseopportunities, a careful description and quantification ofinput and output heat flows is needed.

The result of a first-law energy accounting is two-fold.As in the mass-accounting procedure, we remain with anintensive indicator of efficiency (energy output/energy in-put or energy input/unit of output) and an extensive ac-counting of total heat input and total heat output. The ratioof total energy output to total energy input must be equalto one, unless some output flows have been neglected.

C. Accounting for (User-Side) ResourceQuality: The Exergy Approach

Not all forms of energy are equivalent with respect to theirability to produce useful work. While heat is conserved, itsability to support a transformation process must decreaseaccording to the second law of thermodynamics (increas-ing entropy). This is very often neglected when calculatingefficiency based only on input and output heat flows (first-law efficiency) and leads to an avoidable waste of still




usable energy and to erroneous efficiency estimates. Thesame need for quality assessment applies to every kindof resource supplied or released in a process. The abilityof resources to supply useful work or to support a furthertransformation process must be taken into account and isthe basis for inside-the-process optimization procedures,recycling of still usable flows, and downstream allocationof usable resource flows to another process.

The ability of driving a transformation process and, as aspecial case, producing mechanical work, may be quanti-fied by means of the exergy concept. According to Szargutand co-workers in 1988, exergy is “the amount of workobtainable when some matter is brought to a state of ther-modynamic equilibrium with the common components ofthe natural surroundings by means of reversible processes,involving interaction only with the above-mentioned com-ponents of nature.” Except for cases such as electricityproduction by means of nuclear reactors, the gravitationalpotential of water, or the geothermal potential of deep heat,chemical exergy is the only significant free energy sourcein most processes. In the procedure of Szargut and co-workers, chemical exergy is calculated as the Gibbs freeenergy relative to average physical and chemical param-eters of the environment. Szargut and co-workers calcu-lated the chemical exergy of a resource, bch,res, relative tocommon components of nature, by means of the followingprocedure:

1. The most stable end products of reactions involv-ing the resource under consideration are chosen as refer-ence chemical species “rs” in the environment (air, ocean,or earth crust). These reference species are not in ther-modynamic equilibrium, as the biosphere is not a ther-modynamic equilibrium system. Reference species areassigned a specific free energy (exergy) brs due to theirmolar fraction zrs in the environment by means of theformula brs = −RT ln zrs, where the reference species areassumed to form an ideal mixture or solution.

2. In accord with the classical procedure of Gibbs, achemical reaction linking a resource “res” to its referencespecies is chosen. The chemical exergy difference �bch

between the two states is calculated as �bch = �bf +∑bch,i, where �bf is the standard chemical exergy of

formation of the resource from elements, and bch,i is thestandard chemical exergy of the i th element relative to itsreference species. Tabulated values of free energy can beused for this purpose.

3. The total chemical exergy of the resource is finallycalculated as bch,res = �bf +

∑bch,i +

∑brs. In so do-

ing, tables of exergies per unit mass or per mole (specificexergy) can be constructed for each resource.

By definition, the exergy (ability of doing reversible work)is not conserved in a process: the total exergy of inputs

equals the total exergy of outputs (including waste prod-ucts) plus the exergy losses due to irreversibility. Quantify-ing the exergy losses due to irreversibility (which dependson deviations from an ideal, reversible case) for a pro-cess offers a way to calculate how much of the resourceand economic cost of a product can be ascribed to theirreversibility affecting the specific technological devicethat is used (as in exergoeconomics as described by M. A.Lozano and A. Valero in 1993 and A. Lazzaretto and col-leagues in 1998), as well as to figure out possible processimprovements and optimization procedures aimed at de-creasing exergy losses in the form of waste materials andheat. Exergy losses due to irreversibilities in a process arevery often referred to as “destruction of exergy.”

Szargut, in 1998, suggested that the “cumulative exergyconsumption” required to make a given product could be ameasure of the ecological cost of the product itself. Finally,R. U. Ayres and A. Masini,l in 1998, have suggested that“there is a class of environmental impacts arising simplyfrom uncontrolled chemical reactions in the environmentinitiated by the presence of ‘reactive’ chemicals. The for-mation of photochemical smog in the atmosphere is oneexample . . . .” Exergy may therefore be a useful tool forpotential harm evaluation, but requires further investiga-tions for quantitative analyses of this type.

After exergy accounting has been performed, we re-main with exergy efficiency indicators (bout/bin, bwaste/bin,bin/unit of product) that can be used for optimization andcost-allocation purposes.

VI. PUTTING PROBLEMS INA LARGER PERSPECTIVE

Optimizing the performance of a given process requiresthat many different aspects be taken into account. Someof them, mostly of a technical nature, relate to the localscale at which the process occurs. Other technological,economic, and environmental aspects are likely to affectthe dynamics of the larger space and time scales in whichthe process in embedded. These scale effects require thata careful evaluation be performed of the relation betweenthe process and its “surroundings” so that hidden conse-quences and possible sources of inefficiencies are clearlyidentified.

A. Factors of Scale and System Boundaries

It is apparent that each evaluation (mass, energy, andexergy-flow accounting) can be performed at differentspace scales. In Fig. 1 we assume that we are evaluatinga power plant, with natural gas used to produce elec-tricity. The immediate local scale includes only plantcomponents and buildings. An input mass balance at this




Power

FIGURE 1 Schematic showing the hierarchy of space and time scales that should be considered when evaluating aproduction process. W1 is the flow of usable waste materials that can be recycled within the economic process; W2is the flow of waste materials released to the environment in a form that is not yet completely degraded. Flows to theheat sink represent waste heat and materials that cannot be further degraded. Systems symbols after Odum, 1996.

scale (over the actual plant area for a time of one year)should only include the actual masses of componentsand building materials discounted over the plant lifetimeplus the fuel supplied. The same fraction of buildingmaterials from decommissioning plus airborne emissionsshould be accounted for as mass output. Similarly, anenergy balance should only account for input naturalgas and output electricity and waste heat. As the scale isexpanded to the regional level [whatever the regional areais, it should include the production process for machinerycomponents (boiler, turbine, insulating materials, etc.)and plant building materials like concrete and steel],new mass and energy inputs must be taken into account.For instance, the huge amount of electricity needed forcomponent manufacture from metals contributes heavilyto the energy cost, while the amount of fossil fuels neededto produce the electricity remains assigned to the materialflow at the input side. If the scale is further expanded, themass of raw minerals that must be excavated to extract thepure metals for making plant components may contributeheavily to all of the calculated indicators [considerthe above examples of nickel ore or of iron, which is25–30% raw iron ore]. At this larger scale, raw oil usedup in the extraction and refining of minerals and the oil

itself must also be accounted for. It is a common andsomehow standardized practice to account for embodiedlarge-scale input flows while evaluating the process atthe local scale. In so doing the evaluation is more easilydone and results are summarized shortly. We do notsuggest this procedure, as large-scale processes mayhide local-scale performance and optimization needs. Weprefer to perform the evaluation at three different scales(local, regional, and global), each one characterized bywell-specified processes (respectively, resource finaluse, components manufacturing and transport, resourceextraction and refining), so that inefficiencies at eachscale may be easily identified and dealt with. Thelarger the space scale, the larger the cost in terms ofmaterial and energy flows (i.e., the lower the conversionefficiency). If a process evaluation is performed at a smallscale, its performance may not be well understood andmay be overestimated. Depending upon the goal of theinvestigation, a small-scale analysis may be sufficientto shed light on the process performance for technicaloptimization purposes, while a large-scale overview isneeded to investigate how the process interacts withother upstream and downstream processes as well asthe biosphere as a whole. Defining the system boundary




and making it clear at what time scale an assessment isperformed is therefore of paramount importance, evenif the scale of the assessment is sometimes implicit inthe context of the investigation. It is very importantto be aware that a “true” value of net return or otherperformance indicators does not exist. Each value ofa given indicator is “true” at the scale at which it iscalculated. When the same value is used at a differentscale, it does not necessarily become false. It is, however,out of the right context and inapplicable.

Finally, how is the time scale accounted for in a processevaluation? The simplest case is when we have inputs,whose lifetime exceeds the time window of the analysis.For instance, assets in a farm have a lifetime of about20–25 years, agricultural machinery never lasts more thanabout 10 years, while fertilizers and fuels are used up soonafter they are supplied to the process. It is therefore easyto transform long-life inputs into annual flows by dividingby their lifetime (in years).

Another and perhaps more important time scale is hid-den in the resources used, that is, the time it takes to makea given resource. When we use oil, it is not the extractiontime or the combustion time that makes the differencebut rather the very long time it took for crude oil for-mation in the deep storages of the earth crust. It is thistime that makes oil a nonrenewable resource. Similar ar-guments apply to other resources, so that we may distin-guish between renewable, slowly renewable, and nonre-newable resources. Turnover time (i.e., the time it takes toreplace a resource) is often a good measure of resource re-newability. An effort to go beyond the concept of turnovertime in resource evaluations is the introduction of emergy-accounting procedures. This approach is dealt with in thefollowing section. It is an attempt to place a value on re-sources by calculating the environmental work and time ittook to supply a given resource to a user process. Emergyhas been suggested as an aggregated measure of environ-mental support from present (space scale) and past (timescale) activities of the ecosystem. This important conceptis discussed in succeeding sections.

B. Life-Cycle Assessment

As already pointed out, careful material and energy as-sessments at different space-time scales are prerequisitesfor a good description of process dynamics. Assessingmaterial and energy input and output flows is the basisfor evaluations of an environmental disturbance generatedby withdrawal of resources and release of waste, as wellas energy costs associated with resource processing anddisposal. Life-cycle assessment (LCA) methodologies areused to assess environmental impacts of a product (or aservice) from “cradle to grave” or better “from cradle to

cradle,” including recycle and reclamation of degraded en-vironmental resources. More than a specific methodology,LCA is a cooperative effort performed by many investiga-tors throughout the world (many working in the industrialsectors) to follow the fate of resources from initial ex-traction and processing of raw materials to final disposal.Different approaches have been suggested, thousands ofpapers published, and meetings organized. This effort isconverging toward developing standard procedures or atleast a common framework that will make procedures andresults comparable. SETAC (the International Society forEnvironmental Toxicology and Chemistry) has created asteering committee to “identify, resolve, and communicateissues regarding LCAs (and to) facilitate, coordinate, andprovide guidance for the development and implementa-tion of LCAs.” In a series of workshops between 1990 and1993, SETAC developed a “code of practice” to be adoptedas a commonly agreed procedure for reliable LCAs. Ac-cording to this code, LCA investigations should be dividedinto four phases: Goal Definition and Scope, Inventory,Impact Assessment, and Improvement Assessment. TheSETAC standardization trial has been followed by a ro-bust effort of the International Standard Office in 1997and 1998 to develop a very detailed investigation proce-dure for environmental management, based on LCA, andfor a comparable quality assessment. The InternationalStandard clearly “recognizes that LCA is still at an earlystage of development. Some phases of the LCA technique,such as impact assessment, are still in relative infancy.Considerable work remains to be done and practical expe-rience gained in order to further develop the level of LCApractice.” The ISO documents suggest clear and standardprocedures for descriptions of data categories, definitionsof goals and scope, statements of functions and functionalunits, assessments of system boundaries, criteria for inclu-sions of inputs and outputs, data quality requirements, datacollections, calculation procedures, validations of data, al-locations of flows and releases, reuse and recycling, andreporting of results. Nowhere in the ISO documents is apreference given to a particular approach. LCA is sug-gested as a standardized (and still to be improved) frame-work, where most of the methodologies already devel-oped for technical and environmental investigations maybe included and usefully contribute. This last statementis of fundamental importance, as it gives LCA a largerrole than was recognized in earlier steps toward develop-ment. LCA may become a preferred pattern for replacingthousands of evaluations of case studies that are not com-parable, since these may have had different boundaries (ordid not describe a boundary at all), different approaches,different foci, different calculation procedures, differentfunctional units, different measure units. It is no wonderthat environmental and energy assessments have not yet




been accepted as a required practice in resource policymaking. Standardization of procedures is a key feature inmaking significant progress toward policy and economicrelevance of environmental assessments.

C. Accounting for (Donor-Side) ResourceQuality: The Emergy Approach

It takes roughly5 1.25 g of crude oil to make 1 g of refinedoil in a typical refinery process; it takes 3 J of refined oil tosupply 1 J of electricity in a typical power plant. Therefore,it takes about 3.8 J of crude oil to supply 1 J of electricityfor Carnot-efficiency-limited conversions in current prac-tice. Some additional oil is also invested in machinery andother goods that are components of the refinery and powerplant. Oil is also needed to transport components. The oilenergy that has been invested in the overall production pro-cess is no longer available. It has been used up and is notcontained in the final product. The actual energy contentor availability (measured as combustion enthalpy, H.H.V.,L.H.V., exergy, etc.) of the product differs from the totalinput oil energy (or exergy) because of losses in manyprocesses leading to the final product. This required totalenergy in the form of crude oil equivalent is sometimesreferred to as “embodied energy” by energy analysts, al-though it may also be simply referred to as the sum ofusable and lost energy in product manufacture. Embodiedexergy has been suggested as another useful term.

1. From Embodied Exergy to Emergy

While a conservation law (first law) always applies to thetotal energy content of a resource, such a law is obviouslynot applicable to the embodied energy (exergy) concept.As defined, the embodied energy (exergy) of a usable re-source is a measure of the sum of the product energy (ex-ergy) and the energy (exergy) previously used up to makethe product available. The same product may be producedvia different production pathways and with different en-ergy (exergy) expenditures depending on the technologyused and other factors. The embodied energy (exergy) as-signed to an item does not depend only on the initial andfinal states, but also on boundary conditions that may varyfrom case to case and are strongly affected by the levelof process irreversibility. If we were able to use reversibleprocesses from the raw resources to the final products, wewould, of course, obtain the optimal value of the embod-ied energy (exergy). Unfortunately, this is not the case andwe therefore remain with sets of values that may be aver-aged or given as a range of values. Embodied energy (ex-ergy) algebra is therefore more a “memory algebra” than a

5All numbers are approximate.

“conservation algebra.” It is a process-dependent estimatewhich may vary significantly for a given product.

However, resources are always available in a limitedsupply. When a chain of processes is investigated to assessthe actual resource investment that is required, we mayneed to quantify the relation of resources with the environ-mental work which supports them. In other words, a givenresource might require a larger environmental work thanothers. The use of such a resource means a larger appropri-ation of environmental support and services, and may beassumed as a measure of sustainability and/or pressure onthe environment by the system. As a further developmentof these ideas, Odum has introduced the concept of formemergy 6, that is, “the total amount of exergy of one kindthat is directly or indirectly required to make a given prod-uct or to support a given flow.” We may therefore have anoil emergy, a coal emergy, etc., according to the specificgoal and scale of the process. In some way, this concept ofembodiment supports the idea that something has a valueaccording to what was invested into making it. This wayof accounting for required inputs over a hierarchy of levelsmight be called a “donor system of value,” while exergyanalysis and economic evaluation are “receiver systems ofvalue” (i.e., something has a value according to its useful-ness to the end user).

The focus on crude oil and other fossil fuels (solar en-ergy stored by ecosystem work) as prime energy sourcesis very recent in human history. Because we need andpay for these energy sources, we worry about possibleshortages. Before fossil fuels were used, the focus wason free, direct solar radiation as well as indirect solar en-ergy in the form of rain, wind, and other environmentalsources. Everybody was well aware that the availabilityof direct and indirect solar energy was the main driv-ing force of natural ecosystems and human societies aswell. H. T. Odum therefore switched his focus from theinterface human societies–fossil storages to the interfacehuman societies–environment, identifying the free envi-ronmental work as the source of each kind of resourcesupporting human activities.

Expanding the scale of the investigation requires theconcept of oil emergy to be expanded to solar emergy,a measure of the total environmental support to all kindsof processes in the biosphere by means of a new unit,the solar emergy joule. This new point of view also ac-counts for energy concentration processes through a hier-archy of processes that are not under human control andmay therefore follow a different optimization pattern thanhumans do. Human societies (deterministic and productoriented in their strategies) usually maximize efficiency,

6The name emergy and a valuable effort at developing the emergynomenclature were provided by D. Scienceman in 1987.




short time-scale return on investment, employment, profit,and one-product output. At the other end, natural processesare stochastic and system-oriented and are assumed to tryto maximize the utility of the total flow of resources pro-cessed through optimization of efficiencies and feedbackreinforcement.

2. Solar Emergy—Concepts and Definitions

The solar emergy is defined as the sum of all inputs of so-lar exergy (or equivalent amounts of geothermal or grav-itational exergy) directly or indirectly required in a pro-cess. Usually these inputs are produced by another processor sequence of processes, by which solar emergy is con-centrated and upgraded. In mathematical terms, the solaremergy Em assigned to a product is defined as

Em =∫ ∫ ∫

ε(λ, τ, σ ) dλ dτ dσ,

where ε(λ, τ, σ ) is the total exergy density of any source atthe soil level, depending on the time τ and land area σ overthe entire space and time scales involved in the investigatedprocess. In principle, it also depends on the wavelength λ

of the incoming solar radiation. Sources that are not fromsolar source (like deep heat and gravitational potential) areexpressed as solar equivalent energy by means of suitabletransformation coefficients.

The amount of input emergy dissipated (availabilityused up) per unit output exergy is called solar transfor-mity. For investigators who are used to thinking in termsof entropy, the transformity may also be considered asan indicator of entropy production, from which the totalentropy produced over the whole process can be in prin-ciple calculated. The solar transformity (emergy per unitproduct) is a measure of the conversion of solar exergythrough a hierarchy of processes or levels; it may there-fore be considered a “quality” factor which functions as ameasure of the intensity of biosphere support for the prod-uct under study. The total solar emergy of a product maybe expressed as

solar emergy (seJ) = exergy of the product (J)× conversion factor (solar transformity in seJ/J).

Solar emergy is usually measured in solar emergy joules(seJ), while the unit for solar transformity is solar emergyjoules per joule of product (seJ/J). Sometimes emergy perunit mass of product or emergy per unit of currency is alsoused (seJ/g, seJ/$, etc.). In so doing, all kinds of flows to asystem are expressed in the same unit (seJ of solar emergy)and have a built-in quality factor to account for the con-version of input flows through the biosphere hierarchy. Itis useful to recall that emergy is not energy and therefore

it is not conserved in the way that energy is. The emergyof a given flow or product is, by definition, a measure ofthe work of self-organization of the planet in making it.Nature supplies resources by cycling and concentratingmatter through interacting and converging patterns. Sincethere is energy in everything including information andsince there are energy transformations in all processes onearth and possibly in the universe as well, all these pro-cesses can be regarded as part of an energy hierarchy. Allthese energy transformations are observed to be connectedto others so as to form a network of energy transforma-tions, where many lower components support a few unitsat higher hierarchical position. A part of this work hasbeen performed in the past (viz., production of fossil fu-els) over millions of years. A part is the present work ofself-organization of the geobiosphere. The greater the flowthat is required, the greater the present and past environ-mental costs that are exploited to support a given processor product.

According to the process efficiencies along a given path-way, more or less emergy might have been required toreach the same result. The second law of thermodynamicsdictates that there is a lower limit below which the prod-uct cannot be made. There is also some upper limit abovewhich the process would not be feasible in practice al-though, in principle, one could invest an infinite amount offuel in a process and thus have an infinitely high transfor-mity. Average transformities are used whenever the exactorigin of a resource or commodity is not known or when itis not calculated separately. It follows that: (1) Transfor-mities are not constants nor are they the same for the sameproduct everywhere, since many different pathways maybe chosen to reach the same end state. (2) Emergy is not apoint function in the way energy and other thermodynamicstate functions are. Its value depends upon space and timeconvergence, as more emergy is used up (and assigned tothe product) over a pathway requiring a higher level of pro-cessing. The emergy value has a “memory” of resourcesinvested over all processes leading to a product. While theexergy content of a given resource indicates somethingthat is still available, the emergy assigned to a given itemmeans something that has already been used up and de-pends upon the processes followed in making the product.This last characteristic is shared with the embodied exergyand cumulative exergy consumption concepts, as it comesout of the different irreversibilities that characterize eachpathway. (3) Optimum performance for specified externalconstraints may be exhibited by systems that have under-gone natural selection during a long “trial and error” pe-riod and that have therefore self-organized their feedbackfor maximum power output. Their performance may resultin optimum (not necessarily minimum) transformity.




Transformities are a very central concept in emergyaccounting and are calculated by means of preliminarystudies for the production process of a given good or flow.Values of transformities are available in the scientific lit-erature on emergy. When a large set of transformities isavailable, other natural and economic processes can beevaluated by calculating input flows, throughput flows,storages within the system, and final products in emergyunits. After emergy flows and storages in a process or sys-tem have been evaluated, it is possible to calculate a set ofindices and ratios that may be suitable for policymaking.

Emergy as defined here provides emergy indicators thatexpand the evaluation process to the larger space and timescales of the biosphere. While the emergy approach is un-likely to be of practical use in making decisions aboutthe price of food at the grocery store or the way a pro-cess should be improved to maximize exergy efficiency atthe local scale, its ability of linking local processes to theglobal dynamics of the biosphere may provide a valuabletool for adapting human driven processes to the oscilla-tions and rates of natural processes. As pointed out inSection V, this may be a useful step toward developingsustainable patterns of human economies.

VII. CASE STUDY: INDUSTRIALPRODUCTION OF SOLID OXIDEFUEL CELLS AND HYDROGEN

Mass, energy, exergy and emergy accounting, when usedtogether, provide a valuable tool for multicriteria life-cycleassessment. Data from Tables II to VII may help clarifyhow system performance may be investigated. The tablesrefer to electricity production by using solid oxide fuel

TABLE II Mass Balance for a Solid Oxide Fuel Cell. Production and Use of a Single Cella

Material inputOil requirement per unit of

Amount per unit product CO2 NO2

Product produced of product (including oil) emissions emissions

Anode 0.55 g 369 g/g 3320 g/g 1170 g/g 4.8 g/g

Cathode 0.55 g 451 g/g 3850 g/g 1400 g/g 5.7 g/g

Electrolyte 1.60 g 1087 g/g 9550 g/g 3460 g/g 14.2 g/g

Interconnector 25.00 g 25 g/g 192 g/g 79.1 g/g 8.1 g/g

Total cell 27.70 g 102 g/g 16900 g/g 322 g/g 1.3 g/g

Electricityb 192 kWh 214 g/kWh 4250 g/kWh 578 g/kWh 1.1 g/kWh

a Ulgiati et al., 2000.b Including the fractions of energy and matter expenditures for the whole plant. The data account for all input

flows, from extraction of raw materials to processing and final production of electricity (see the text, Section VII.A).A fraction of inputs for stack and plant manufacturing is also accounted for and assigned to the cell when evaluatingthe production of electricity.

cells (SOFCs). Fuel cells (FCs) run on hydrogen that canbe produced within the cell (internal reforming) if natu-ral gas or other fuels are supplied. Hydrogen can also besupplied directly (from external reforming or water elec-trolysis). We have investigated both FC production andoperating phases and hydrogen production from naturalgas via steam reforming and from water via electrolysisto calculate a set of performance parameters that could beused to support decision making about this innovative anddeveloping technology. Figure 2 shows diagrams of thetwo investigated pathways for hydrogen production.

The questions to be answered relate to energy efficiency,irreversibility of process steps, material demand as a mea-sure of environmental disturbance, and ecological foot-print as a measure of environmental support. We need aprocedure that differentiates processes to support planningand decision making.

A. Solid Oxide Fuel Cells

A fuel cell is an electrochemical device for converting thechemical energy of a fuel directly into electricity, withoutthe efficiency losses of conventional combustion processeswhich are constrained by the Carnot limit. In principle, theconversion efficiency is 100% for FCs, while it is around35–40% for the Carnot conversion at the temperaturesused. In real cases, efficiencies of FCs are lowered bythe requirements of other plant components, such as re-forming and preheating phases as well as irreversibilitiesaffecting the process.

A fuel and an oxidizing gas are involved. The fuel ishydrogen which is oxidized at the anode according to thereaction H2 → 2H+ + 2e−. Positive ions migrate from theanode to the cathode through the electrolyte conductor




FIGURE 2 Energy systems diagrams of hydrogen production processes: (a) steam reforming of natural gas; (b)water electrolysis.

of the cell, while electrons migrate through the outsidecircuit to the cathode where the oxidizing gas is reducedaccording to the process O2 + 4e− → 2O2−. The overallreaction is 4H+ + 2O2− → 2H2O, so that emissions con-sist mainly of water vapor. Fuel cells also operate on fuelsother than hydrogen (CH4, methanol, gasoline, etc.). Inthis case, preliminary reforming (oxidation) is needed toconvert the fuel primarily to hydrogen and carbon diox-ide. The electric output of a fuel cell is in the form of cur-rent (DC) that is usually converted into alternating current(AC) before use. Our investigation deals with the wholeprocess, from extraction of raw resources to production ofelectricity in the FC power plant. Reforming of natural gasand preheating of input air and fuel are also accounted for.Local-scale evaluation includes preheating of input gases,reforming of CH4, and conversion of H2 to electricity.Downstream combustion of CH4, CO, and H2 in exhaustgases is also included, as it provides heat to the preheatingstep and other uses. Regional-scale evaluation includescomponents manufacturing and transport. Finally, globalbiosphere-scale assessment includes additional expensesfor extraction and refining of raw resources. Performancevalues calculated at the local scale are always higher thanthose calculated at larger scales. Performance indicatorsare also affected by the assumptions made for natural gasproduction costs. Available energy-cost estimates for nat-ural gas are in the range 1.10–1.40 g of oil equivalentper gram of natural gas and depend on the location and

the way natural gas is extracted, refined, and transported.These values affect the energy ratio and the material bal-ance of our case study (the higher the energy cost the lowerthe performance). Our calculated data refer to an energycost equal to 1.1 g of oil equivalent per gram of natural gas.

Table II shows an overview of material and energy per-formance of the cell components as well as of the cell as awhole, including the electricity produced. Electrolyte pro-duction appears to be the most oil-intensive component, aswell as the most mass-intensive and CO2 releasing compo-nent. The opposite is true for the interconnector. Globally,it takes 214 g of crude oil equivalent to produce one kWhof electricity. In addition to this direct and indirect fuelinput, it takes about 4 × 103 g of other material inputs ofdifferent nature (minerals, oxygen, etc.) per kWh of elec-tricity delivered. Energy data from Table III show a globalefficiency of 40.25%, which is a relatively good perfor-mance compared with other fossil-fueled power plants in-vestigated (Table IV). The energy efficiency of the singlecell alone is the ratio of delivered electricity (192 kWh)to the energy of the hydrogen input from the reformer(7.5 × 108 J), yielding a value of 92% which is very closeto the theoretical maximum. This clearly indicates that ifhydrogen could be produced in ways different than steamreforming (for instance, electrolysis from excess hydro-electricity or high-temperature catalysis), FCs may be-come a very effective way of producing electricity. Thetransformity, that is, the environmental support for the




TABLE III Energy and Emergy Accounting for a Solid Oxide Fuel Cell. Production and Use Referto a Single Cella

Energy Emergy assigned Emergy assignedAmount requirement per to a unit of product, to a unit of product,of item unit of product with services without services

Product produced (MJ/unit) ( ××1012) ( ××1012)

Anode 0.55 g 15.45 1.37 seJ/g 1.08 seJ/g

Cathode 0.55 g 18.91 1.20 seJ/g 1.03 seJ/g

Electrolyte 1.60 g 45.62 3.06 seJ/g 2.77 seJ/g

Interconnector 25.00 g 1.04 0.23 seJ/g 0.06 seJ/g

Total cell 27.70 g 4.26 0.91 seJ/g 0.26 seJ/g

Electricityb 192 kWh 8.94 0.57 seJ/kWh 0.48 seJ/kWhTransformity, Transformity,

Output/input with services without serviceselectric efficiency ( ××105) ( ××105)

η = 40.25% 1.59 seJ/J 1.32 seJ/J

a Ulgiati et al., 2000.b Including the fractions of energy and matter expenditures for the whole plant. The data account for all input

flows, from extraction of raw materials to processing and final production of electricity (see the text, Section VII.A).A fraction of inputs for stack and plant manufacturing is also accounted for and assigned to the cell when evaluatingthe production of electricity.

process, calculated without including labor and services,is 1.32 × 105 seJ/J, while inclusion of labor and servicesyields a higher value of 1.59 × 105 seJ/J (Table III). Thedifference (0.27 × 105 seJ/J, i.e., 17% of the total value)is a measure of the fraction depending upon the labor andservices cost in the country, i.e., a measure of the “eco-nomic part” of the process. This fraction might be lower ina different economic context with lower labor costs. The

TABLE IV Performance of Solid Oxide Fuel Cell System,Compared with Selected Electricity Power Plantsa

CO2 emissions Transformityper unit of without including

Source of Output/input product serviceselectricity energy ratio (g/kWh) ( ××104 seJ/J)

Geothermal 20.76 655 14.0energy

Hydroelectric 14.02 19 6.10power

Wind 7.67 35 5.89

MCFCb,c 0.43 561 14.1

SOFCc 0.40 578 13.2

Natural Gas 0.35 766 14.5

Oil 0.30 902 14.9

Coal 0.25 1084 13.0

a Ulgiati et al., 2000.b Molten Carbonate Fuel Cells.c Including reforming of natural gas. In principle, fuel cells are ex-

pected to show the highest overall efficiencies. These are lowered in realplants by the up-stream reforming process, due to losses in the conversionto H2 and the need for preheating of input gases.

emergy input from natural gas accounts for about 76% ofthe total, while the emergy of technological inputs (thestack of fuel cells and other devices plus process energy)only accounts for about 7%. The technological fractionmight change over time, depending on technological im-provements. However, the contribution of the fuel (76%)is not likely to change. This means that the value of thecalculated transformity can be assumed to be very sta-ble and only a small percentage may change dependingon variations in the economy and technology. The smallfraction of “technological emergy” might appear negli-gible when compared with the high input of emergy inthe fuel. However, it is not negligible at the regional scalewhere cell components are manufactured and cells assem-bled. Improvements at this level would affect the perfor-mance and balance of the cell-manufacturing plant, andtherefore translate into an advantage to the producer andthe regional environment. The results derived from thedata in Tables II to IV are the following: (1) Additionaleffort is needed to decrease the energy and material in-tensity of electrolyte production. (2) SOFC emissions ofCO2 are lower than emissions of other fossil-based powerplants. (3) Indirect resource flows supporting labor andservices do not appear to be important (emergy) inputs tothis technology. (4) At the larger biosphere scale, FC oper-ation affects overall performance more than construction.At the regional scale, construction inputs are important fora given design. At this level, an exergy evaluation may beneeded to obtain optimized production patterns. (5) Theelectric energy efficiency of SOFCs is greater than the ef-ficiencies of other fossil-fuel power plants investigated.




TABLE V Performance of a Single SOFC Evaluated at Differ-ent Scales, from Local to Globala,b

Local Regional GlobalIndex scale scale scale

Material input per unit of 3250 3390 4250product (g/kWh)

CO2 released per unit of 379 409 578product (g/kWh)

NO2 released per unit of 0.11 0.23 1.07product (g/kWh)

Energy efficiency, whole FC 54.67 51.54 40.25plant (%)

Transformity of electricity 1.32 × 105 1.32 × 105 1.32 × 105

produced, without includingservices (seJ/J)

a Ulgiati et al., 2000.b Including reforming of natural gas. In principle, SOFC cells are

expected to show the highest overall efficiencies. These are loweredin real plants by the up-stream reforming process, due to losses in theconversion to H2 and the need for preheating of input gases.

(6) The transformity of electricity from SOFCs (i.e., theenvironmental support per unit of product energy) is in thesame order of magnitude than that for electricity produc-tion from conventional fossil-fuel power plants (Table IV)and higher than that from nonfossil sources. This is mostlydue to energy inputs for upstream steps of reforming andpreheating of input gases. These steps should be carefullystudied to improve their efficiencies. However, it should beemphasized that fossil-fuel power plants are optimized byuse, while FCs still are at a demonstration or precommer-cial stage, with large potential for improvement. Table Vshows that performance indicators depend on the scale ofthe investigation as we emphasized in Section VI.A. Forthe sake of simplicity, we refer only to data at the globalscale in Tables II to IV.

Preliminary results from a similar study on molten car-bonate fuel cells (MCFCs) compared with SOFCs show ahigher output/input energy ratio (42.7%), a higher trans-formity (1.41 × 105 seJ/J) and lower CO2 emission (561 gCO2/kWh) for the whole plant system (Table IV).

B. Industrial Production of Hydrogen Fuel

Tables VI and VII present some results from an LCAfor hydrogen production by means of natural-gas steam-reforming and alkaline electrolysis. The two technologi-cal pathways to hydrogen have been investigated by usingmass, energy, exergy, and emergy accounting.

Steam-reforming shows quite good energy and exergyconversion efficiencies, 46 and 42%, and a solar transfor-mity of the expected order of magnitude (1.15 × 105 seJ/Jof hydrogen, labor and services included). Steam-

reforming supplies hydrogen to fuel cells. The previouslydiscussed evaluation of SOFCs has yielded a transformityof about 1.59 × 105 seJ/J of electricity delivered. If weconsider that an SOFC module operates in a two-stepprocess (natural-gas reforming and electrochemical con-version in a cell) a lower transformity for the reformingprocess alone is perfectly consistent with SOFC data.Alkaline electrolysis shows lower performance due tothe fact that the production of electricity is performed byconventional combustion of natural gas at 39% efficiency.This is the bottleneck of the process, whose globalefficiency is quite good at 40% due to the better perfor-mance of the water-splitting process. The oil-equivalentinput per unit of product as well as CO2 emission (otherchemicals released are not shown) do not show significantdifferences between the two processes. Instead, it mayappear surprising that there is a huge difference in thetotal material demand per unit of output, which is fourtimes higher in the steam-reforming process. The reasonis the huge water consumption, for both natural gasoxidation and cooling. Water is a cheap resource that canstill be obtained at low energy cost. There is no evidentreason to design a process that saves water, becauseno one perceives this problem as crucial. However, ifhydrogen is to become the favorite energy carrier formany uses, the shortages of water foreseen for the nextdecades may become limiting factors in some countries.Water supplies may become a research priority that ismore important than the need for raising the alreadygood energy and exergy efficiencies. Finally, alkalineelectrolysis requires environmental support that is greaterthan for steam-reforming or electricity production fromSOFCs, as is clearly shown by the higher transformity.This makes alkaline electrolysis a less decidable steptoward the production of hydrogen as a fuel.

Our case studies provide both performance data andidentification of bottlenecks for the investigated processes.Improvements can be made at different scales by look-ing at the values of the calculated indicators and design-ing technological changes where appropriate. It appearsfrom our case studies that none of the methodologies usedprovides a comprehensive, multiscale assessment. Theircombined use leads to synergistic results that are useful indecision making about these nonconventional processes.

VIII. A BASIS FOR VALUE

Investigating a system performance is by itself a very dif-ficult task, due to the complexity of the problems thatare always involved. When a simplified model is adopted,this is certainly a way to address part of the problem at




TABLE VI Energy Evaluation of Hydrogen Production from Natural-Gas Steam-Reformingand from Alkaline Electrolysisa

Amount of Solarhydrogen Output/input Exergy transformity

Production process produced (g/yr) energy ratio efficiency ( ××105 seJ/J)

Natural-gas steam-reforming 7.18 × 108 0.46 0.42 1.15

Alkaline electrolysisb 1.94 × 1011 0.40 0.38 1.74

a Ulgiati et al., 2000.b The electricity is produced with in the plant using natural-gas combustion.

the cost of leaving unsolved another part of it. Dependingupon the goal of the investigation, this might be a veryuseful procedure.

Investigators very often run the risk of neglecting thecomplexity of the problem and taking their model for thereality. As a consequence, they assign a value to a pro-cess product according to the results of their simplifiedinvestigation. The outcome of this evaluation process isthen used in other following evaluations and translatedinto economic and policy actions. In so doing, the com-plexity is lost: the reality does not fit the model and theplanned policy fails or is inadequate.

Complexity and the consequent difficult task of reach-ing agreement on a decision can be shown by means ofa simple example that recalls the paradoxes already citedin Section II. Assume that I commute every day due tomy job and that a friend of mine does the same. The bustakes one and a half hours. We spend half an hour readingour newspapers, then we talk for a while, until we reachour destination. To save money and the environment, wedecide to purchase only one newspaper and take turnsreading it, one reading the paper, while the other one talkswith other passengers or is involved in other reading. Wesave therefore about 20 newspapers per month, that trans-lates into a savings of about U.S.$7 each. We feel thatwe are generating a lower pressure on the environment(fewer trees are cut for papermaking, less ink is used, less

TABLE VII Mass-Flow Evaluations for Hydrogen Production from Natural-Gas Steam-Reforming and from Alkaline Electrolysisa

Amount of Oil requirement Material input per CO2 emissionshydrogen per unit of unit of product per unit ofproduced product (including oil) product

Production process (g/yr) (g/g H2) (g/g H2) (g/g H2)

Natural-gas steam-reforming 7.18 × 108 7.44 196c 24.18

Alkaline electrolysisb 1.94 × 1011 8.60 47d 39.84

a Ulgiati et al., 2000.b The electricity is produced within the plant using natural-gas combustion.c Mostly due to the very high requirement of process and cooling water.d Mostly due to the requirement for process water.

garbage is to be disposed of). At first glance, it seemsthat our approach aimed at saving money and resourcesworks well. We might also decide to purchase two dif-ferent newspapers and read both of them while traveling,thereby gaining insight into different political opinions.The decision about these two alternatives has no roots inany biophysical analysis; it is just a different approach fora different return on investment. Thus we may get twocopies of the same newspaper, two different newspapers,one shared newspaper plus additional goods purchasedwith the saved money.

At the local scale of our personal life, we find the follow-ing consequences: (1) we will have more money availableand may decide to purchase a magazine each week; (2)we may invest this money into drinking more coffee at thecafeteria of the train station, which will possibly affect ourhealth and therefore the health care bill at the personal andnational levels (the process of coffee-making may have alarger environmental impact than paper-making); (3) as weread at different times, we talk less during travel and wetherefore have a reduced exchange of ideas and personalfeelings. When we expand the scale to take into accountother possible advantages (or disadvantages) generated byour action, we find: (4) the decrease in newspaper salesmay cause a decrease of jobs in the sectors of paper andnewspaper making and the government will be required toinvest money to support this activity; (5) the government




may get less revenue from taxing the newspaper businessas well as from taxing a decreased number of workers; (6) areduced environmental pressure is generated by the down-sizing of the newspaper sector, but an increased pressureresults from our investment of money in other sectors. Wedo not know which one of the two alternatives is worseunless a careful evaluation is performed. It is very hardto reach a balance among these different gains and losses.Maybe some of these interests are irreducible to a commonoptimization pattern. In any case, whatever we do affectsthe environment in a measure that is partially predictableby means of a careful biophysical analysis.

Let us now assume that we have developed a well-standardized multicriteria, multiscale procedure for thebiophysical and environmental evaluation of a processor for the development of the whole economy. We re-tain many unsolved problems that call for new aspectsand values affecting the process of decision making. Letus consider the case of the Maheshwar hydroelectric dam(Narmada Valley, State of Madhya Pradesh, Central India),as described by the Indian writer Arunhati Roy in her 1997best-seller “The God of Small Things.” One of the largesthydroelectric projects in the world will make the villageof Sulgaon and 60 other villages disappear, while 40,000people will have to move somewhere else. A fertile area,filled with cultural and social traditions and a good level ofeconomic development based on agriculture (sugar cane,cotton, cash crops), will become an artificial lake to sup-port the dam turbines. A political and legal controversyis now in progress. What is the problem here? There isno doubt that the people have the right to live in the areawhere they were born, where they have a home and a job,where they have traditions and social structures, etc. Thereis no doubt that forcing them to move will cause social dis-ruptions and the collapse of that part of the economy thatis supported in this area. On the other hand, there is nodoubt that the additional electricity available to the eco-nomic system will support new activities, will create jobsand wealth, and will eventually give rise to new villagesand social structures. It appears that the diverse projectionsare not reducible to an agreed upon strategy. No biophys-ical theories of value appear to be able to determine theprocess of decision making. The problem is not only con-version efficiency or environmental degradation, nor is ita purely economic cost-benefit evaluation.

There is no way that an absolute value can be put on anitem or a strategy for a given approach because the conceptof value is scale- and goal-dependent. Of course, valuesassigned to an item may fit the goals and scale of validityof each approach very well. There is no consensus amongscientists on the whole set of approaches that is neededfor a complete evaluation. This is not necessarily bad, as itleads to cultural diversity. As for natural systems, diversityis a fundamental prerequisite for stability and progress.

The commonly accepted economic belief is that capitaland labor form the real basis of wealth. It does not assignany special value to resources and services that are sup-plied for free by nature and therefore are not accountedfor by market mechanisms. This belief has been a sourceof huge environmental degradation and is being graduallyabandoned by concerned environmental economists. Aspreviously pointed out, most evaluation systems are basedon utility of what is received from an energy transforma-tion process. Thus, fossil fuels are evaluated on the basisof the heat that will be received when they are burned.Economic evaluation is very often based on the willing-ness to pay for perceived utility. An opposite view of valuein the biosphere could be based on what is put into some-thing rather than what is received. In other words, the moreenergy, time, and materials that are “invested” in some-thing, the greater its value. As a consequence of this laststatement, the source of resources and the way they areused should always be carefully assessed and form thebasis of a new economic view that assigns an additionalvalue, complementary to utility values, to something suchas natural services that have no market by definition.

We started this paper observing that both physical andeconomic aspects are involved in the process of runninga car. We have stressed many aspects involved in the trialfor developing an acceptable unique theory of value thatis able to account for all aspects of a technological oreconomic process. Should we keep trying, by again gettinginvolved in this never-ending debate? In our opinion, morefruitful research would deal with the nonreducibility ofdifferent theories of value while searching for integrationand synergistic patterns.

On the basis of this discussion, it can be concluded that:(1) any evaluation process that is not supported by a pre-liminary integrated biophysical assessment is not reliableor usable. Biophysical theories provide the scientific basisfor choices by including productivity, climate, efficiency,pollution, and environmental degradation scenarios. (2)Biophysical assessments based on a single approach ornumeraire may have research value but are unlikely tocontribute to effective policy action. (3) After an inte-grated biophysical assessment has been performed (i.e.,the source of natural value is recognized), other values(ethical, social, monetary, political, etc.) should be inte-grated. The process of decision making will have to takeall of these into account, each fitting a different scale andwith different goals.

IX. CONCLUSIONS

A consensus is being developed on the urgent need for in-tegrated and dynamic assessment tools. These are neededboth at the larger scale of the economy and at the local




scale of technological processes. Among the latter, energy-conversion processes may be dealt with by using inte-grated approaches, as is shown for the case studies inves-tigated in this article. The results show the feasibilities andadvantages of each energy process and provide compre-hensive assessments for governments, industrial corpora-tions, and the public, thereby leading to economically andenvironmentally sound policies.

It may appear surprising that we have emphasized in thispaper the need for biophysical evaluation procedures as afundamental and useful support tool in decision makingand, at the same time, we have not supported the possi-bility of policy making when relying only on biophysicaltheories of value. In our opinion, the two statements arenot contradictory. As repeatedly emphasized in this article,the applicability of any evaluation depends on the space-time scale of the investigation and its utility depends on the(scale-related) goal of the study. When dealing with policyactions involving human societies and short time scales,values other than biophysical assessments are very likelyto affect governmental choices and trends, even if there isno doubt that in the long run human societies, as well asany other self-organizing systems in the biosphere, willbe constrained by the availability of energy and resources.Limits to growth have very often been claimed since thepublication of the Meadows’ report in 1972, which dealtwith the consequences of finite resource bases. Integratedbiophysical tools are needed to identify which constraintsaffect the development of our societies as a whole. Theyare also helpful in understanding how the uses of new tech-nological tools or developments are affected by these con-straints. Biophysical evaluation procedures may provideanswers to local and short-time scale problems, while pro-viding estimates of planetary and long-run consequencesof our actions.


BIOENERGETICS • CARBON CYCLE • ENERGY EFFICIENCY

COMPARISONS AMONG COUNTRIES • ENERGY RE-SOURCES AND RESERVES • FOSSIL FUEL POWER STA-TIONS • FUEL CELLS, APPLICATIONS IN STATIONARY

POWER SYSTEMS • GREENHOUSE WARMING RESEARCH

• HYDROELECTRIC POWER STATIONS • PETROLEUM

GEOLOGY • THERMODYNAMICS • TRANSPORTATION

APPLICATIONS FOR FUEL CELLS • WASTE-TO-ENERGY

SYSTEMS

BIBLIOGRAPHY

Atkins, P. W. (1994). “Physical Chemistry; 5th ed.,” Oxford UniversityPress, Oxford, U.K.

Ayres, R. U., and Masini, A. (1998). Waste exergy as a measure ofpotential harm. In “Advances in Energy Studies. Energy Flows in

Ecology and Economy” (S. Ulgiati, M. T. Brown, M. Giampietro,R. A. Herendeen, and K. Mayumi, eds.), pp. 113–128. Musis, Rome,Italy.

BP Amoco (2000). Statistical review of world energy. June 2000.Brown, M. T., and Herendeen, R. (1996). Embodied energy analysis and

emergy analysis: A comparative view. Ecol. Econ. 19, 219–235.Brown, M. T., and Ulgiati, S. (1999). Emergy evaluation of the biosphere

and natural capital. Ambio 28(6), 486–493.Boustead, I., and Hancock, G. F. (1979). “Handbook of Industrial Energy

Analysis,” Wiley, New York.Campbell, C., and Laherrere, J. (1998). The end of cheap oil. Sci. Am.

March 1998, 60–65.Cleveland, C. J., Hall, C. A. S., and Kaufman, R. (1984). Energy and the

U.S. economy: A biophysical perspective. Science 225, 890–897.Foran, B., and Crane, D. (1998). The OzECCO embodied energy model

of Australia’s physical economy. In “Advances in Energy Studies.Energy Flows in Ecology and Economy” (S. Ulgiati, M. T. Brown,M. Giampietro, R. A. Herendeen, and K. Mayumi, eds.), pp. 579–596. Musis, Rome, Italy.

Frankl, P., and Gamberale, M. (1998). The methodology of life-cycleassessment (LCA) and its application to the energy sector. In “Ad-vances in Energy Studies. Energy Flows in Ecology and Economy”(S. Ulgiati, M. T. Brown, M. Giampietro, R. A. Herendeen, andK. Mayumi, eds.), pp. 241–256. Musis, Rome, Italy.

Giampietro, M., Mayumi, K., and Pastore, G. (1998). A dynamic modelof socio-economic systems described as adaptive dissipative hol-archies. In “Advances in Energy Studies. Energy Flows in Ecology andEconomy” (S. Ulgiati, M. T. Brown, M. Giampietro, R. A. Herendeen,and K. Mayumi, eds.), pp. 167–190. Musis, Rome, Italy.

Hall, C. A. S., and Cleveland, C. J. (1981). Petroleum drilling and pro-duction in the United States: Yield per effort and net energy analysis.Science 211, 576–579.

Herendeen, R. A. (1998). Embodied energy, embodied everything. . . now what? In “Advances in Energy Studies. Energy Flows in Ecol-ogy and Economy” (S. Ulgiati, M. T. Brown, M. Giampietro, R. A.Herendeen, and K. Mayumi, eds.), pp. 13–48. Musis, Rome, Italy.

Hinterberger, F., and Stiller, H. (1998). Energy and material flows. In“Advances in Energy Studies. Energy Flows in Ecology and Econ-omy” (S. Ulgiati, M. T. Brown, M. Giampietro, R. A. Herendeen, andK. Mayumi, eds.), pp. 275–286. Musis, Rome, Italy.

Hubbert, M. K. (1956). Nuclear energy and the fossil fuels. In “Drillingand Production Practices,” pp. 7–25. American Petroleum Institute,New York.

Hubbert, M. K. (1968). Energy resources. In “Resources and Man,”pp. 157–242. National Academy of Sciences. W. H. Freeman, SanFrancisco.

ISO. (1997). Environmental management—Life-cycle assessment—Principles and framework (ISO 14040:1997). International StandardOrganization, June 1997, pp. 16. Brussels.

ISO. (1998). Environmental management—Life-cycle assessment—Goal and scope definition and inventory analysis (ISO 14041:1998).International Standard Organization, October 1998, pp. 27. Brussels.

Jarach, M. (1985). Sui valori di equivalenza per l’analisi e il bilan-cio energetici in agricoltura. Rivista di Ingegneria Agraria 2, 102–114.

Jørgensen, S. E. (1992). “Integration of ecosystem theories: A pattern,”Kluwer, Dordrecht and Boston.

Kiker, C. F. (1992). Emergy and available energy analyses: Are theseconsistent means for delineating natural and artifactal capital? Paperpresented at the Second Meeting of the International Society for Eco-logical Economics, 3–6 August 1992, Stockholm, Sweden.

Lazzaretto, A., Macor, A., Mirandola, A., and Stoppato, A. (1998). Po-tentialities and limits of exergoeconomic methods in the design, anal-ysis and diagnosis of exergy conversion plants. In “Advances in En-ergy Studies. Energy Flows in Ecology and Economy” (S. Ulgiati,




M. T. Brown, M. Giampietro, R. A. Herendeen, and K. Mayumi, eds.),pp. 515–530. Musis, Rome, Italy.

Lozano, M. A., and Valero, A. (1993). Theory of the exergetic cost.Energy 18(9), 939–960.

Martinez-Alier, J. (1987). “Ecological Economic,” Basil Blackwell, NewYork.

Odum, H. T. (1983). Maximum power and efficiency: A rebuttal. Ecol.Model. 20, 71–82.

Odum, H. T. (1988). Self organization, transformity and information.Science 242, 1132–1139.

Odum, H. T. (1994). The emergy of natural capital. In “Investing in Nat-ural Capital” (A. M. Jansson, M. Hammer, C. Folke, and R. Costanza,eds.), pp. 200–212. Island Press, Covelo, CA.

Odum, H. T. (1996). “Environmental Accounting. Emergy and Environ-mental Decision Making,” Wiley, New York.

Odum, H. T., and Pinkerton, R. C. (1955). Time’s speed regulator: Theoptimum efficiency for maximum power output in physical and bio-logical systems. Am. Sci. 43, 331–343.

O’Neill, R. V., DeAngelis, D. L., Waide, J. B., and Allen, T. F. H. (1986).“A Hierarchical Concept of Ecosystems,” Princeton University Press,Princeton.

Penner, S. S. (2000). Policy issues in providing energy supplies for the21st century and beyond. Manuscript presented at the Second Interna-tional Workshop “Advances in Energy Studies. Exploring Supplies,Constraints, and Strategies,” Porto Venere, Italy, May 23–27, 2000.Book of Proceedings, 2001, pp. 367–377.

Scienceman, D. (1987). Energy and Emergy. In “Environmental Eco-nomics” (G. Pillet and T. Murota, eds.), pp. 257–276. RolandLeimgruber, Geneva, Switzerland.

Smil, V. (1991). “General Energetics. Energy in the Biosphere and Civ-ilization,” Wiley, New York.

Szargut, J. (1998). Exergy analysis of thermal processes. Ecological cost.In “Advances in Energy Studies. Energy Flows in Ecology and Econ-omy” (S. Ulgiati, M. T. Brown, M. Giampietro, R. A. Herendeen, andK. Mayumi, eds.), Musis, Rome, Italy.

Szargut, J., Morris, D. R., and Steward, F. R. (1988). “Exergy Analy-sis of Thermal, Chemical, and Metallurgical Processes,” Hemisphere,London.

Ulgiati, S., and Brown, M. T. (1998). Monitoring patterns of sustainabil-ity in natural and man-made ecosystems. Ecol. Model. 108, 23–36.

Ulgiati, S., Brown, M. T., Bastianoni, S., and Marchettini, N. (1995).Emergy-based indices and ratios to evaluate the sustainable use ofresources. Ecol. Eng. 5, 519–531.

Ulgiati, S., Brown, M. T., Giampietro, M., Herendeen, R. A., andMayumi, K. (eds.) (1998). “Advances in Energy Studies. Energy Flowsin Ecology and Economy,” Musis, Rome, Italy.

Ulgiati, S., Bargigli, S., Raugei, M., and Tabacco, A. M. (2001). Life-cycle assessment and environmental impact of advanced energy sys-tems. In “Advances in Energy Studies. Exploring Supplies, Con-straints, and Strategies” (S. Ulgiati, M. T. Brown, M. Giampietro,R. A. Herendeen, and K. Mayumi, eds.), Palombo, Rome, Italy.pp. 29–42.

P1: GLM Final Pages Qu: 00, 00, 00, 00


Energy Resources and ReservesS. S. PennerUniversity of California, San Diego

I. IntroductionII. Fossil-Fuel ResourcesIII. Resources for Nuclear Fission, Breeder,

and Fusion ReactorsIV. Renewable ResourcesV. Recent Estimates of Fossil-Fuel Reserves

and Resources Likely to Become Usableover the Near Term

VI. The Kyoto ProtocolVII. Policy Considerations

GLOSSARY

Fossil fuels Coals, petroleums, natural gases, oils fromshales and tar sands, methane hydrates, and any othersupplies from which hydrocarbons for energy applica-tions may be extracted.

Fuels for nuclear breeder reactors These include U-238and Th-232, which may be converted to fissile isotopes(e.g., U-233, U-235, Pu-239, and Pu-241) as the resultof neutron capture.

Fuels for (nuclear) fission reactors The naturally occur-ring fissile isotope U-235, as well as Pu-239 and Pu-241produced by neutron capture in U-238 and U-233 fromTh-232.

Fuels for (nuclear) fusion reactors Deuterium-2 (whichoccurs in water as HDO or D2O) and Li-6, which con-stitutes about 7.4% of naturally occurring lithium.

Nonrenewable resources (nonrenewables) Resourceslocated on the planet with estimable times for exhaus-tion at allowable costs and use rates.

Renewable resources (renewables) Usually defined asextraterrestrial energy supplies such as solar resources,but some authors include energy supplies of suchtypes and magnitudes that they will be available forthe estimated duration of human habitation on theplanet.

Reserves Energy supplies which are immediately usableat or very close to current prices.

Resources The totality of energy supplies of specifiedtypes that include reserves and may become usable intime at competitive prices with improved technologies.

AN ASSESSMENT of energy reserves and resources isproperly made in the context of needs for an acceptablestandard of living for all humans. This approach may bequantified by beginning with a discussion of current en-ergy use in a developed country such as the United States,noting ascertainable benefits from high energy-use levels,estimating the possible impacts of a vigorously pursued,

461

P1: GLM Final Pages


462 Energy Resources and Reserves

stringent energy-conservation program, and then defininghuman needs for the estimated world population by 2050on the assumption that population growth will level offafter this time. This is an egalitarian approach based ondocumented benefits to humans of high levels of energyuse while asserting the right of all people everywhere toenjoy the good life which technology has brought to thebeneficiaries of large energy supplies at acceptable costsin a small number of developed countries. Of special in-terest are reserves and resources with a time to exhaustionthat is commensurate with the very long time span whichwe optimistically foresee for human existence on ourplanet.

I. INTRODUCTION

A. Historical Perspective of CurrentEnergy-Use Levels in the United States

Per capita energy consumption in the United States cor-responds to about 12 kW of thermal power used contin-uously. This high level of continuous power use may beappreciated by examining two examples of rapid powerconsumption and the associated energy use.

Top power consumption for an elite athlete exceeds2 kW (Smil, 1999, p. 90), whereas running a 100-m dashin 10 sec requires a power level of 1.3 kW for a typicalworld-class sprinter (Smil, 1999, p. XV, Table 5). Thus,the power expenditure of a world-class sprinter, if main-tained continuously, would be about a factor of 10 smallerthan that enjoyed by an average American.

Another striking example of how rapidly we consumeenergy to attain the good life we take for granted may begiven by asking how many Frenchmen the disciplinarianFinance Minister Jean Baptiste Colbert of King LouisXIV (1643–1715) would have had to motivate in expend-ing all of their usable energy for the King’s benefit toallow Louis XIV to luxuriate at the energy-expenditurelevel we have reached in the United States on the average.A worker consuming 3000 cal/day and converting all ofthe food energy input to useful work would have a poweroutput1 of

(3000 cal/p-day) × (3.73 × 10−7 hp-hr/0.24 cal)

× (1 day/24 hr) = 2 × 10−4 hp/p,

where p stands for person and hp for horsepower. Atthis power-output level per person, Colhert’s requirementbecomes

1The numerator and denominator in the fractions contained withinthe middle set of parentheses both equal 1 J.

(12 kW) × (3.73 × 10−7 hp-hr/2.78

× 10−7 kW-hr)/(2 × 10−4 hp/p) = 80,000 p.

Thus, Colbert needed a veritable army of 80,000Frenchmen who would convert all of their 3000-cal/dayfood consumption into useful work for the king with100% efficiency in order to provide him with the energy-use opportunities enjoyed on the average by the peopleof the United States.

B. Is a High Level of Energy ConsumptionNecessary or Desirable for Human Welfare?

There are many definitions for the good life or desirablestandard of living. Demands for a good life include low in-fant mortality at birth, good health, long life expectancy,high per capita income levels to allow purchase of themany goods and services we take for granted in the devel-oped part of the world, a clean and attractive environment,good educational opportunities, the needed time for andaccess to major cultural events, etc. There has been a stronghistorical correlation everywhere between per capita in-come and energy use on the one hand and between allof the desired amenities and income on the other, i.e. thelevel of energy consumption is a direct determinant ofthe standard of living defined in terms of the specifiedbenefits. Two notable examples are reductions in infantmortality from about 150 per 1000 for undeveloped coun-tries with low energy-use-rate levels to fewer than 15 per1000 in the United States and a corresponding increase inlife expectancy from about 40 years to nearly double thisnumber. The gradual introduction of energy efficiency im-provements may briefly disrupt historical trends, but hasalways been overwhelmed over the longer term by newdiscoveries providing new amenities and hence escalatingenergy consumption.

We shall ignore the historical record and assume that wecan achieve significant energy conservation while main-taining our high standard of living. This desirable goal willbe achieved through accelerated inventions and continuingemphasis on the desirability of minimizing energy use inorder to reduce fossil-fuel consumption because we mustreduce the rate of growth of atmospheric greenhouse-gasconcentrations and also decelerate the rate of accumula-tion of long-lived radioactive isotopes produced in nuclearfission reactors. Whereas the more ambitious projectionsof energy-use reductions top out at about 30% for con-stant income levels, we shall use a reduction of 40% asthe optimistic result of concentrated worldwide energy-conservation efforts. In summary, we assume that success-ful energy-conservation measures will allow us to attainthe desired standard of living for 10 billion people by theyear 2050 at 60% of the late-20th-century per capita uselevels in the United States.

P1: GLM Final Pages


Energy Resources and Reserves 463

During 1999, total U.S. energy use was approximately1 × 1017 Btu/yr, or

(1 × 1017 Btu/yr) × (103 J/Btu) × (1 EJ/1018 J)

= 102 EJ/yr,

where 1 EJ = 1 exajoule = 1018 J and Btu and J stand forthe British thermal unit and the joule, respectively. By theyear 2050, there will be around 10 billion people on theearth. If all of these people were to enjoy 60% of the U.S.per capita energy consumption used during 1999, thenannual worldwide energy use would be

0.6 × 1010 p × (102 EJ/yr)/3 × 108 p = 2 × 103 EJ/yr.

In summary, 10 billion people will hopefully enjoy thegood life enjoyed by U.S. citizens at the end of the 20thcentury provided they each consume 60% of the U.S. percapita average during the year 1999. At that time, world-wide energy use will reach 2 × 103 EJ/y.

C. Choice of a Common Unit to Compare theMagnitudes of Resources and Reserves

In order to gauge the relative importance of nonrenewableand renewable energy supplies over the long term, it is de-sirable to list all of their magnitudes in terms of a commonunit. We choose here the cgs unit of energy or erg (1 erg =10−25 EJ). There are large uncertainties in resourceestimates, as well as variabilities in the heats of com-bustion of fossil fuels bearing the same name (such ascoals, petroleum, tar oils, and shale oils). As an appro-ximation, the following conversion factors are used herefor energy deposits: one standard cubic foot of natural gas(1 SCF or 1 ft3 of NG) = (103 Btu/ft3) × (1010 erg/Btu) =1013 erg/ft3; one barrel (bbl or B) of oil = 5.8 × 106 Btu =5.8 × 1016 erg; one metric ton of coal (1 mt of coal; ≈1.1short ton, 1 t, of coal) = 28 × 106 Btu = 2.8 × 1017 erg;1 mt of oil = 4.3 × 107 Btu ≈ 7.3 bbl ≈ 4.3 × 1017 erg.

For nonrenewable resources,2 which are discussed inSections II and III, the listed entries for the total availableenergies represent estimates of values that have been iden-tified in some manner as probably available and, over thelong term and after sufficient research and developmenthas been carried out, actually usable as reserves providedresource recovery can be implemented at acceptable costs.The estimated size of the nonrenewable resource base may

2The distinction between nonrenewable and renewable resources istenuous, with the latter generally referring to energy resources that arethe result of extraterrestial inputs (e.g. solar energy in its many manifes-tations) but often including also the utilization of geophysical featuresof the earth such as its rotation about an axis or heat flow from thecore to the surface. Some authors mean by renewables those resourceswhich will be usable as long as the earth remains suitable for humanhabitation.

increase with time as improved exploration and assess-ments are performed. Renewable resources are discussedin Section IV.

II. FOSSIL-FUEL RESOURCES

As used here and by most other authors, the terms re-sources and reserves have greatly different meanings. Re-sources are potential supplies if no attention is paid tocommercial readiness at currently paid market prices andgenerally include also supplies for which recovery andutilization techniques remain to be developed. Reserves,on the other hand, are supplies that may be utilized withknown technologies at or very near current market prices.In the normal course of events, resource recoveries pre-cede reserve classification, and reclassification as reservesmay require new technologies or at least improvements ofexisting, tested, or developing technologies.

Estimations of resources and reserves tend to be grosslyoversimplified. For this reason, the use of a single signif-icant figure is often appropriate. An entire paragraph ofassumptions and constraints may be needed to justify anumerical estimate. To illustrate this point, we begin withthe potentially largest of the fossil-fuel supplies, namely,shale oil. A classical worldwide assessment was publishedby Duncan and Swanson (1965). The supplies were la-beled as falling into the following categories: known re-sources, possible extensions of known resources, undis-covered and unappraised resources, and total resources.The shales were further subdivided according to shalegrade in terms of the number of gallons (ga) of oil whichcould be recovered from one short ton (t) of the raw mate-rial with subdivisions according to shale grades of 5–10,10–25 and 25–100 ga/t. Nothing is said about shale gradesbelow 5 ga/t, of which there are enormous deposits. Need-less to say, the greater the shale grade, the easier and morecost-effective is the recovery. As an example, a total re-source estimate of potentially usable oil shale may thenbe made by including all categories of deposits and shalegrades down to 10–25 ga/t. The estimate derived from thedata of Duncan and Swanson (1965) becomes 344,900 ×109 bbl of oil or, using 5.8 × 106 Btu/bbl,

3.45 × 1014 bbl × (5.8 × 106 Btu/bbl)

× (107 erg/10−3 Btu) = 2 × 1031 erg.

If we decided to broaden the resource base by includingshale grades down to 5–10 ga/t, the preceding resourceestimate would be increased by

1.75 × 1015 bbl × (5.8 × 106 Btu/bbl)

× (107 erg/10−3 Btu) = 10 × 1031 erg.

P1: GLM Final Pages



Problems similar to those for shale oil arise for coalresources (for which estimates customarily depend on thethickness of the coal seams and on the depth below thesurface where the coal is recovered), as well as for oilsfrom tar sands, petroleum, and natural gas.

Averitt (1969) published estimates of the total originalcoal resources of the world and of regions of the world onthe basis of (a) resources determined by mapping and ex-ploration and (b) probable additional resources in un-mapped and unexplored areas for 12-foot-thick or thickercoal beds and (in nearly all cases) less than 4000 feetbelow the surface of the earth. The total estimatewas 16,830 × 109 t × (0.909 mt/t) × (2.8 × 1017 erg/mt) =4 × 1030 erg. Of this total, the North American contri-bution was about one-fourth or 1 × 1030 erg. The coalresources are predominantly bituminous and subbitumi-nous coals with perhaps 20% representing low-rank (i.e.low carbon content) lignite and peat and less than 1%high-rank (nearly 100% C content) anthracite. Since theheat of combustion we use for coal applies to bitumi-nous coals and generally decreases with rank (or car-bon content), we may be overestimating the resourcebase by as much as 30%. Worldwide resource estimateswere discussed in 1972 (Cornell, 1972) with L. Rymancontributing values for the ultimately recoverable re-sources of NG and oil. For NG, the estimate was about1.5 × 1016 SCF = 1.5 × 1029 erg and for oil it was about4 × 1012 bbl = 2.3 × 1029 erg. All of these estimates donot refer to resources regardless of cost, but rather to re-sources that are not unlikely to become reserves overtime. The resource base for natural gas may be greatlyexpanded in the future if deep-lying hydrates are included(see Section V.B).

The worldwide base for oil from tar sands has not beenwell determined. For the Athabasca tar sands in Alberta,Canada, where commercial recovery has been pursued formore than 25 years, a resource value of 625 × 109 bblwas estimated by Spragins (1967) and corresponds to3.6 ×1028 erg; a 1981 resource estimate for Athabascawas 8 × 1028 erg.

We may summarize this qualitative overview by thestatement that, over a long period of time, shale oil couldcontribute substantially more than 2 ×× 1031 erg, coal morethan 4 ×× 1030 erg, NG and oil each more than 2 ×× 1029 erg,and worldwide oil from tar sands probably considerablymore than 2 × 1029 erg. Thus, the combined fossil-fuel re-source base certainly exceeds 2 × 1031 erg by a substantialmargin.

It is interesting to compare the fossil-fuel resource esti-mates with past fossil-fuel extraction. According to Smil(1999, p. 135), total global extraction of fossil fuels (inthe form of coal, oil, and natural gas) from 1850 to 1990

was 3 × 102 EJ or (3 × 102 × 1018 J) × (107 erg/J) = 3 ×1027 erg, which equals about 1.5% of the resource es-timates for the least abundant fossil fuels. Total globalelectricity generation from 1900 to 1990 produced (Smil,1999, p. 135) 1.3 × 104 terrawatt-hours (TWhe) = 1.3 ×1016 Whe, which corresponds approximately to 1.3 ×1013 kWhe × 4 × (107 erg/2.78 × 10−7 kWhe), or a fossil-fuel input of about 1.9 × 1027 erg, where the subscript eidentifies electrical energy and we have assumed that fourunits of thermal energy were required per unit of electri-cal energy produced during the first century of electricitygeneration, i.e. the historical average efficiency for elec-tricity generation has been taken to be 25%. Worldwidecoal production from 1800 to 1990 (Smil, 1999, p. 137)was 4.8 × 103 Mt (1 Mt = 106 t) = 4.8 × 109 t × (2.78 ×1016 erg/t) = 1.3 × 1027 erg. The conclusion to be reachedfrom this historical overview is that more than 98% of es-timated fossil-fuel resources of all types remained unusedas of the end of the 20th century.

III. RESOURCES FOR NUCLEAR FISSION,BREEDER, AND FUSION REACTORS

We shall estimate the resources for nuclear fission,breeder, and fusion reactors by using geological data. Inbreeder reactors, the “fertile” isotopes U-238 and Th-232are converted to the “fissile” isotopes U-233, U-235, Pu-239, and Pu-241 as the result of neutron capture. Thoriumis a very widely distributed element3 and does not rep-resent a limiting supply when used in breeder reactorswith uranium. For this reason, the following discussionis restricted to uranium resources for fission and breederreactors.

A. Fuels for Nuclear Fission Reactors

Fission reactors consume the only naturally occurring fis-sile isotope, namely, U-235. Fission occurs as the resultof absorption of slow (thermal) neutrons. Uranium orescontain the following isotope distribution: 6 × 10−5 ofU-234, 7.11 × 10−3 of U-235, and 0.99283 of U-238.Chattanooga shale is a typical deposit; it was formed 33–29 million years ago and is widely distributed in Illinois,Indiana, Kentucky, Ohio, and Tennessee (Swanson, 1960).The Gassaway member of this shale deposit is about16 feet thick.

It is appropriate to estimate the ultimate resources forfission reactors by following the procedure that is custom-

3A 1966 assessment (IAEA, 1966) placed U.S. resources at2.26 × 107 t for costs of $50–100 and at 2.72 × 109 t for costs of $100–500.

P1: GLM Final Pages



arily applied to fusion reactors, i.e. to estimate the totalresource on the planet without regard to costs or near-termavailability. Seawater (Best and Driscoll, 1980, 1981) con-tains 3.3 × 10−9 part by weight of uranium (U). Hence,the 1.5 × 1018 m3 of water on the earth contains 1.5 ×1018 m3 of water × (103 kg of water/m3 of water) ×(3.3 × 10−9 kg of U/kg of water) × (7.11 × 10−3 kg ofU-235/kg of U) × (107 erg/1.22 × 10−14 kg of U-235) =4 × 1031 erg, where we use the estimate that the fission-energy equivalent of 107 erg is derived from 1.22 ×10−14 kg of U-235.

The uranium content of the earth’s crust has been es-timated to be 6.5 × 1013 mt (Koczy, 1954). For use in afission reactor, the uranium content of the crust has anenergy content of (6.5 × 1013 mt of U) × (7.11 × 10−3 mtof U-235/mt of U) × (103 kg of U-235/mt of U-235) ×(107 erg/1.22 × 10−14 kg of U-235) = 4 × 1035 erg.

The current availability of uranium is generally speci-fied in terms of availability at a specified price. A 1983DOE estimate for the United States was 433,400 t of Uat $80/kgU, corresponding to the sum of the probablepotential (251,500 t), possible potential (98,800 t), andspeculative potential (83,100 t) resources; at $260/kg U,the corresponding values were 725,700 + 323,800 +250,700 = 1,300,200 t of U. Uranium prices declinedfrom nearly $110/kg in 1980 to about $40/kg by mid-1984 and remained at about this level during the next4 years. Needless to say, the current availability in theUnited States and elsewhere of uranium resources even atthe highest prices is smaller by many orders of magnitudethan the ultimately available resources. A similar remarkapplies to resources outside of the United States. Thus, thetotal Western world resources (not including the centrallyplanned economies) were estimated in 1987 to exceed theU.S. resources by factors of about two to four. In 1981,the United States was the world’s dominant uranium pro-ducer, with 33.5% of total Western production; by 1986,the U.S. share had declined to 14.3%, while Canada hadbecome the dominant producer with 31.6% of total pro-duction. In 1986, South African (at 12.5%) and Australian(at 11.2%) production followed U.S. production closely.

B. Fuels for Nuclear Breeder Reactors

When burned in a breeder reactor, the energy content ofnaturally occurring uranium (Cohen, 1983) is approxima-tely (1 MW-day/g of U) × (24 hr/day) × [103 kW/MW] ×(107 erg/2.78 × 10−7 kWh) = 8.6 × 1017 erg/g of U. It fol-lows that for use in breeder reactors, the uranium contentsof the oceans have an energy value of 1.5 × 1018 m3 ofwater × (1 mt of water/m3 of water) × (3.3 × 10−9 mt ofU/mt of water) × (8.6 × 1017 erg/g of U) × (106 g of U/mt

of U) = 4 × 1033 erg. The uranium content of the earth’scrust has been estimated as 6.5 × 1013 mt. The energyequivalent for use in a breeder reactor of this resourceis (6.5 × 1013 mt of U) × (106 g of U/mt of U) × (8.6 ×1017 erg/g of U) = 5.6 × 1037 erg; at the estimated year2050 consumption rate of 2 × 1028 erg/year (see Table I),this energy source would last for 2.8 × 109 years or a largefraction of the remaining life on the planet provided bothworld populations and per capita energy use become sta-bilized at the estimated level for the year 2050.

C. Fuels for Nuclear Fusion Reactors

The sequence of nuclear fusion reactions of deuterium toproduce helium with atomic weights of 3 and 4 is

D + D → He-3 + n + 3.2 MeV,

D + D → T + H + 4.0 MeV,

D + T → He-4 + n + 17.6 MeV.

Thus the fusion of five deuterium atoms (D), includingfusion with the reaction intermediate tritium (T), releases24.8 MeV of energy or 4.96 MeV per D atom, where 1 J =6.24 × 1012 MeV = 0.949 × 10−3 Btu. Since 1 m3 of liq-uid water contains 34.4 g of D, fusion of the deuteriumcontained in 1 m3 of liquid water releases (4.96 MeV/Datom) × [(6.023 × 1023/2) D atoms/g of D] × (34.4 g ofD/m3 of H2O) = (5.14 × 1025 MeV/m3) × (0.949 × 10−3

Btu/6.24 × 1012 MeV) = 7.8 × 109 Btu/m3 of H2O. Sincethe total amount of water on the surface of the earth is about1.5 × 1018 m3, the resource estimate for planetary fusionenergy from deuterium becomes (7.8 × 109 Btu/m3) ×(1.5 × 1018 m3) × (107 erg/0.949 × 10−3 Btu) = 1.2 ×1038 erg.

Lithium occurs in water at 1 part in 107 with 7.42% ofLi being Li-6; mineable deposits of Li2O are 5 × 106−10 × 106 mt. At a sufficiently high neutron flux, Li-6 un-dergoes the fission reaction

Li-6 + n → He + T + 4.8 MeV,

which may be followed by tritium fusion with deu-terium to release 17.6 MeV, thus leading to the overallreaction

Li-6 + D + n → 2He-4 + n + 22.4 MeV.

This energy release must be reduced by 4.96 MeV/D atomsince the fusion of deuterium and lithium decreases thesupply of deuterium that can fuse with itself, i.e. the netincremental fusion energy becomes

P1: GLM Final Pages



TABLE I Worldwide Energy Resources without Consideration of Utilization Efficiencies or Technical Readinessa

Energy (EJ) Years of availability“Nonrenewable” energy sources (1 EJ = 1018 J = 1025 erg) at 2 × 103 EJ/yr

Fusion energy from D in 1.5 × 1018 m3 of water 1 × 1013 5 × 109

Uranium in the earth’s crust used in breeder reactors 6 × 1012 3 × 109

Uranium in the earth’s crust used in fission reactors 4 × 1010 2 × 107

Fusion of lithium contained in seawater 3 × 109 1.5 × 106

Uranium in seawater used in breeder reactors 4 × 108 2 × 105

Uranium in seawater used in fission reactors 4 × 106 2 × 103

Worldwide fossil-fuel resources (shale oils, coals, oils from tar sands, petroleum, >2 × 106 >1 × 103

natural gas, natural gas liquids)

U.S. fossil-fuel resources ∼1 × 105 ∼5 × 10

Fusion of lithium in terrestrial Li2O deposits 8 × 104 4 × 10

Hydrothermal energy stored to a depth of 10 km worldwide 4 × 104 2 × 10

U.S. uranium used in water-moderated fission reactors 2 × 104 1 × 10

Hydrothermal energy stored to a depth of 3 km worldwide 8 × 103 4

Power Multiple of need at“Renewable” energy sources (EJ/yr) 2 × 103 EJ/yr

Solar energy at the “outer boundary” of the atmosphereb 5 × 106 2.5 × 103

Worldwide wind energy 2 × 106 1 × 103

Ocean thermal energy conversion (OTEC)c 3.78 × 105 190

Geothermal energy (outward flow of heat from the earth’s core) 8 × 102 4 × 10−1

Tidal energy 1 × 102 5 × 10−2

Hydroelectric energy worldwide 9 × 101 4.5 × 10−2

Commercial, worldwide energy use in 1995 4 × 102 —

Stabilized annual worldwide energy demand as of 2050 with conservation for 10 × 109 people 2 × 103 —and allowance of energy use to establish an acceptable standard of living for alld

a See the text for sources of these values.b This is the total solar input which may be recovered in part as biomass, wind energy, OTEC, direct water photolysis, photovoltaic energy, etc.c A 1964 assessment is 2 × 103 EJ for the Gulf Stream alone. If 0.1% of the solar input power is used for tropical waters with accessible temperature

differences >22◦C (area ≈ 60 × 106 km2), the continuous power production is 12 × 106 GWe year = 378 × 103 EJ/year.d This long-range, steady-state value refers to 10 × 109 people (now expected by the year 2050) with each person using 60% of the U.S. per capita

consumption of 1998, i.e. each person enjoying the average 1998 U.S. standard of living if conservation efforts lead to a 40% energy-use reduction overthe long term.

17.44 MeV

Li-6 atom× 107 erg

6.24 × 1012 MeV× 1 g of Li

107 g of H2O

× 7.42 × 10−2 g of Li-6

g of Li× 106 g of H2O

m3 of water

× 6.023 × 1023 atoms of Li-6

6.94 g of Li-6× 1.5

× 1018 m3of water = 2.7 × 1034 erg.

Li-7 may also breed with T, but with lower efficiency thanLi-6.

Mineable deposits of Li2O have been estimated (Penner,1978) to be as much as 11 × 107 mt of which the lithiumweight is about 4.64 × 106 mt, while that of Li-6 is equalto 3.44 × 105 mt. The incremental fusion energy (after

subtracting that of the deuterium used in the process) thenbecomes about

(3.44 × 105 mt of Li-6) × 17.44 MeV

atom of Li-6

× 6.023 × 1023 atoms of Li-6

6.94 g of Li-6× 103 kg of Li-6

mt of Li-6

× 103 g of Li-6

kg of Li-6× 107 erg

6.24 × 1012 MeV= 8.3 × 1029 erg.

D. Concluding Observations on Resourcesfor Breeder and Fusion Reactors

It is apparent that the resource bases for breeder and fusionreactors are so vast that the ascertainable amounts of both

P1: GLM Final Pages



of these materials, when used in currently contemplatedsystems, could last until the white dwarf that is our suncools and becomes a red giant of vastly increased radius(increased by about a factor of 10), at which time the riversand oceans of our planet will boil away, thereby causinga fiery demise for all of the earth’s inhabitants.

While the resource estimates are not greatly different,the status of technological development (Zebroski et al.,1998) is not comparable for fusion and breeder reactors.For fusion reactors, large-scale demonstrations have nowbeen deferred by three or more decades because of thelack of adequate funding for R&D on the scale needed forrapid commercialization. Although large-scale breeder re-actors represent a demonstrated technology, both technol-ogy improvement and commercialization have been sus-pended in nearly all countries where breeders were beingdeveloped because they are not currently cost-competitivewith fossil-fuel electricity generators or with conventionalfission-nuclear reactors.

The procedure used for resource estimations of the“nonrenewable” breeder and fusion fuels makes these es-timates, in common with those of renewable resource es-timates considered in the following section, unachievablelimiting values in practice because costs for the recoveryof resources must escalate as the source concentrations arereduced through recovery.

IV. RENEWABLE RESOURCES

For renewable resources, the listed values represent limitsassociated with defined physical parameters such as thesolar constant and area of the earth’s surface (for solarenergy), estimates of the total energy that could possi-bly be extracted (Palmen and Newton, 1969) from atmo-spheric winds (for wind energy), the known geothermalheat flow for the entire earth (for geothermal energy re-covery from this source), known water flows in all of therivers of the planet (for hydroelectric energy utilization),and total tidal friction as determined by the gravitationalpulls of the moon and sun on the earth’s oceans, whichlead to differences in tidal range (for tidal energy). Theselimits are absolute values that cannot increase with timeor be augmented in magnitude through improved technol-ogy. What can increase with time is the fraction of thetotal resource base that may be usable at acceptable costs.The renewable resource base may also increase throughidentification and utilization of physical effects that havenot been included in the listed resource base (e.g., col-lection of solar energy in geostationary orbits, followedby conversion to microwave energy of the electric energygenerated in photovoltaic cells used for collection, beforetransmission of the remaining collected energy to the sur-

face of the earth; or using the differences in salinities atthe mouths of rivers between waters in rivers and estuar-ies where they are discharged into the oceans; or an as yetunidentified scheme to utilize the rotation of the earth ina manner not involving planetary winds; or cost-effectiveutilization of wave energy, etc.).

The magnitudes of entries for renewable resources willnow be justified by utilizing known physical phenomena.

A. Solar Energy Input to theSurface of the Earth

The solar constant Q0 is defined as the radiation powerdensity at the top of the earth’s atmosphere which facesthe sun directly. Its value is close to the easily rememberedestimate 2 cal/min-cm2; it is about 3% greater than the av-erage value during the summer in the northern hemisphereand about 3% smaller during the winter in the northernhemisphere because the earth’s orbit around the sun is notexactly circular. Using 2 cal/min-cm2, we may estimate areasonable average radiant power density for the earth asa whole by proceeding as follows.

We first multiply the approximate value 2 cal/min-cm2

by factors of unity in order to convert the solar constant tothe power density in watts per square meter (W/m2),viz., Q0 = 2.0 (cal/min-cm2) × (104 cm2/m2) × (1 J/0.24 cal) × (1 min/60 sec) × (1 W-sec/J) or Q0 =1389 W/m2. The approximate average solar radiationpower density at the top of the earth’s atmosphere is ob-tained from the solar constant by dividing by 4 to allow forvariations of seasons and times of day around the globe.The resulting global average radiant solar power densityis then about 345 W/m2. Since the mean radius of theearth is re = 6.371 × 106 m and its surface area is approxi-mately 4πr2

e = 5.10 × 1014 m2, the radiant power incidenton the entire surface of the earth becomes (345 W/m2) ×(5.10 × 1014 m)2 = 1.77 × 1017 W × (107 erg/1 W-sec) ×(3.154 × 107 sec/year) ≈ 5 × 1031 erg/year, where wehave slightly reduced the calculated value of 5.6 × 1031

erg/year and neglected the small difference between theradius of the earth and the radial distance to the top ofthe atmosphere from the center of earth. This widelyused numerical value is itself only a fair approximationto the actual value because of the many idealizationsmade in the calculations (e.g., the earth is spherical, Q0

is exactly 2 cal/min-cm2 rather than a slightly smallervalue, etc.).

It should be noted that local solar radiant power in-puts are variable not only because of seasonal and di-urnal changes but also because of variability of opticalproperties of the atmosphere (changes in water vapor con-centration, clouds, and particulate concentrations) and ofsurface reflectivities of the earth. A careful evaluation of

P1: GLM Final Pages



these effects is very complex and requires use of the com-plete equations for radiant energy transfer with absorption,scattering, and reflection from variable atmospheres andhighly variable surfaces.

The number 5 × 1031 erg/year provides a reasonable re-source estimate for a discussion of solar energy availabilityover the entire planetary surface.

B. Wind, Geothermal, Hydroelectric,and Tidal Energy

The numerical value listed for the worldwide wind-energy potential (Palmen and Newton, 1969) resultsfrom the estimate that 2 × (10 × 1010 MW) are availableif the average winds are the same in the northern andsouthern hemispheres, which corresponds to 2 × 10 ×1010 MW × (106 W/MW) × (107 erg/2.78 ×10−4 W-hr)× (8.76 × 103 hr/year) = 6 × 1031 W/yr. The listed valuefor the outward flow of heat from the core of theearth equals 8 × 1017 Btu/yr = 8 × 1027erg/year and in-cludes heat flow to the large areas covered by theearth’s oceans; at locations where this geothermalheat flow is usable, the conversion efficiency to electricityis not likely to exceed about 25%. The limiting world-wide potential for hydroelectric energy is 2.86 × 106 MW× [106 W/MW] × (107 erg/2.78 × 10−4 W-hr) × (8.76 ×103 hr/yr) = 9 × 1026 erg/year; the fossil-fuel equivalentfor this resource base as a source of electric energyis about three times larger or about 3×1027 erg/year.Based on the increase in the length of the day during thelast 100 years, the total tidal energy has been estimated tobe about 3 × 106 MW = 3 × 1012 W × (107 erg/2.78 ×10−4 W-hr) × (8.76 × 103 hr/year) = 1 × 1027 erg/year.The achieved conversion efficiencies for tidal to electricalenergy are in the range of 10–20%.

C. Ocean Thermal Energy Conversion [OTEC]

A total assessment of the worldwide potential for energyrecovery using OTEC does not appear to have been made.That the magnitude of this resource is very large is certainin view of a 1964 estimate of the potential for electricitygeneration using only the existing thermal gradients inthe Gulf Stream (Anderson and Anderson, 1964, cited inAnderson, 1966). The estimate is 1.82 × 1014 kWeh/yr.Assuming 33% conversion efficiency to electricity, thethermal fossil-energy equivalent becomes

(1.82 × 1014 kWeh/yr) × (3 kW/kWe) × (107 erg/2.78

× 10−7 kWh) = 1.96 × 1028 erg/yr

The worldwide potential (see the article in 18.05 G byW. H. Avery) is probably at least 20 times greater, i.e.

>4 × 1029 erg/year, and hence may represent a complete,sustainable solution to the world’s energy needs at theestimated steady-state requirement beyond 2050 for 5%recovery of the worldwide ocean thermal energy conver-sion potential.

V. RECENT ESTIMATES OF FOSSIL-FUELRESERVES AND RESOURCESLIKELY TO BECOME USABLEOVER THE NEAR TERM

Estimations of reserves should be more accurate than thoseof resources since they presumably refer to currently avail-able supplies at acceptable costs. Nevertheless, listed tab-ulations have built-in inaccuracies arising from diversecauses such as the following: (1) different reserve holders(e.g., international and domestic oil companies, govern-mental agencies, coal companies, drillers for NG, etc.)may not use compatible evaluation procedures; (2) na-tional representatives of oil interests may overestimatereserves in order to obtain increased leverage in dealingwith monopolistic consortia such as OPEC and thereby in-fluence approval of increased recovery and sales; and (3)companies may find it expedient to underestimate reservesfor the purpose of augmented drilling-right approvals or toleverage assets and stock prices over time. It is thereforenot surprising to find diverse estimates from “authorita-tive” sources.

A. Oil Reserves

In a Scientific American article addressed to a generalaudience, Campbell and Laherrere (1998) express doubtsabout official reserve reports from six OPEC countries(Saudi Arabia, Iran, UAE, Iraq, Kuwait, and Venezuela),which show an increase in oil reserves from about370 × 109 to 710 × 109 bbl between 1980 and 1997, onthe grounds that the increases were reported without ma-jor discoveries of new fields.4 This type of argument isreinforced by noting a decline in announcements of newlydiscovered oil fields since 1980, which has been accompa-nied by depletions of aging fields. Omitted from this typeof analysis are likely beneficial impacts from enhanced

4Developed countries typically maintain petroleum stocks at levelscorresponding to about 1 month of consumption, with somewhat smallerlevels in Canada and Italy and somewhat larger stocks in Germany andJapan. These low ratios of petroleum stock to annual consumption makethe OECD countries highly sensitive to short-term supply disruptions.Petroleum production in non-OPEC countries was smaller than in OPECcountries during the period 1973–1977, slightly exceeded OPEC produc-tion during 1978, and exceeded it by substantial amounts in later years.Since 1995, OPEC production has grown more rapidly than non-OPECproduction and may well again supply more than 50% of totals by 2010if present trends continue.

P1: GLM Final Pages



oil recovery and the possibility that the accounting usedin 1980 may have been low in order to justify high oilprices. Without alternative oil supplies or new discoveries,the authors expect OPEC’s supplies to reach 30% of worldsupplies (as in the 1970s) during the year 2000 or 2001 andthe 50% level by 2010, thereby suggesting the possibilityof arbitrarily set increased oil prices within a decade.

A recent staff report of The Wall Street Journal (Cooper,1999) refers to a reservoir off Louisiana at EugeneIsland (E.G. 330) and describes output peaking at about15,000 bbl/day following initiation of recovery after 1973.Production slowed to about 4000 bbl/day by 1989 but thenbegan to recover and reached 13,000 bbl/day in 1999. As aresult, reserve estimates have changed from 60 to 400 mil-lion bbl for E.G. 330. The conjecture is that the reservoiris being refilled from a deep source miles below the sur-face of the earth. During the last 20 years, the reserveestimates for the Middle East have doubled, while activeexploitation was in progress and no major new discoverieswere announced. A theory to account for the observations,namely, that superheated methane was carrying crude oilupward from greater depths, was apparently not substanti-ated by a drilling test.5 Although more research is clearlyneeded to explain the augmented oil recovery achievedat E.G. 330 and perhaps also the growing estimates forMiddle East reserves, the suggestion that deep oil reser-voirs of great magnitude remain to be tapped is not un-reasonable, especially in view of recent successes withdeep-sea drilling off the coast of West Africa.

In the same issue of Scientific American, Anderson(1998) discusses the likelihood of augmented recoveryabove the conventional 25–35% of oil in place as the resultof underground imaging, gas flooding, steerable drilling,and discoveries of giant deposits off the coast of WestAfrica.

In its 1998 Annual Report to shareholders, Exxon showsa bar graph for each of the years 1994–1998 labeled“proved reserves replacement,” which exceeds 100% foreach of the 5 years and includes an excess of about 120%for 1997, i.e., the company’s reserve estimates have beensteadily increasing. Major oil and gas resource additionsare shown for Alaska, the Gulf of Mexico, in Europe (forthe North Sea off the coasts of England and Germany andnear northeastern Spain), and offshore West Africa nearAngola and also for Papua New Guinea. These resource

5In this connection, mention should be made of a hypothesis by Gold(1999), who describes a biosphere under pressure with temperaturesaround 100◦C at depths near 7 km that exceeds in contents the surface bio-sphere where we live. There “chemilithotropic” reactions involving pri-mordially stored hydrogen and hydrocarbons lead to coal and petroleumformations that continuously augment the availability of fossil-fuel re-sources. Some supporting evidence has been found in bringing to thesurface downhole materials from 6.7-km depths.

additions are defined as “new discoveries and acquisitionsof proved reserves and other resources that will likely bedeveloped.” Areas with the largest new resource additionsare the Commonwealth of Independent States (formerSoviet Union) and Africa, closely followed by Europe.It is not credible to interpret data of this type as pointingto near-term oil-reserve depletions.

Another company perspective on oil reserves is pro-vided by the British Petroleum Company’s “StatisticalReviews of World Energy,” which show that generally ris-ing consumptions of oil and natural gas since the end ofWorld War II have been accompanied by increasing re-serve estimates: world production (3.3616 × 109 mt) andconsumption (3.3128 × 109 mt) of oil are in balance, as arethose for other fossil fuels. In 1996, the yearly reserve-to-production (R/P) ratio for the world as a whole was givenas 42.2, with 8.2 for Europe, 15.7 for the Asia–Pacific Re-gion, 18.1 for North America, about 25 for both Africa andthe CIS, 36.1 for South and Central America, and 93.1 forthe Middle East. These values do not suggest strong priceescalation over the near term. Price determinants throughmarket changes are discussed by Hannesson (1998), aswell as declines in production rates from reservoirs. Aninteresting associated topic is the intelligent managementof fossil-fuel wealth through partial revenue contributionsto a permanent fund for the purpose of leveling the incomeof a country over time as oil revenues decline (Hannesson,1998, Chapter 7).

Assertions of declining oil stocks accompanied byrapidly rising prices have a substantial history of beingfaulty, in spite of the fact that announcements of huge newoil discoveries often turn out to represent overly optimisticreserve estimates. Echoing statements from Exxon andBritish Petroleum, wells in waters 3300–4600 feet deepoff the coast of West Africa are being hailed as showing“excellent” results (George, 1998). According to Exxon(1998–1999), some of these new fields are believed to con-tain reserves of 1 × 109 bbl or more (which are referred toas “elephants” by geologists).

In 1995, the U.S. Geological Survey (USGS) performeda U.S. assessment of oil and gas resources. Measured(proven) reserves in BBO (billions of barrels of oil) were20.2, augmented by (a) 2.1 BBO of continuous-type accu-mulations in sandstones, shales, and chalks, (b) 60.0 BBOexpected reserve growth in conventional fields, and (c) ex-pected, but as yet undiscovered conventional resources of30.3 BBO, thus yielding total resources of 112.6 BBO.Similarly, in TCFG (trillions of cubic feet of naturalgas), the estimates were (a) measured reserves of 135.1TCFG, (b) continuous-type accumulations in coal beds of49.9 TCFC, (c) continuous-type accumulations in sand-stones, shales, and chalks of 308.1 TCFG, (d) expectedreserve growth in conventional fields of 322.0 TCFG, and

P1: GLM Final Pages



(e) expected, but as yet undiscovered conventional re-sources of 258.7 TCFG, for a total of 1073.8 TCFG. Fin-ally, for natural gas liquids (NGL), the estimates in BBOwere (a) measured reserves of 6.6, (b) continuous-typeaccumulations in sandstones, shales, and chalks of 2.1,(c) expected reserve growth in conventional fields of 13.4,and (d) expected, but as yet undiscovered conventionalresources of 7.2, for a total resource estimate of 29.3 BBO.These fossil-fuel reserves and soon-to-be-usable re-sources have a combined energy content of approximately(112.6 × 109 bbl of oil × 5.8 × 1016 erg/bbl) + (1.074 ×103 × 1012 feet3 of NG × 1013 erg/foot3] + ( 29.3 ×109 bbl of oil equivalent × 5.8 × 1016 erg/bbl of oil equi-valent) = 6.5 × 1027 erg as oil + 1.1 × 1028 erg fromNG + 1.7 × 1027 erg from NGL ≈ 1.9 × 1028 erg total.

A year 2000 estimate by the USGS (2000; for a briefsummary, see Science, 2000) shows a very substantial in-crease in oil resources that should become usable over thenear term. A conclusion is that worldwide peak conven-tional oil production is more than 25 years in the futureas compared with a prior estimate of worldwide peak oilproduction during the first decade of the 21st century. Thelarge supply increases are very unevenly distributed overthe globe with most occurring in the Middle East andthe CIS. The absence of large new discoveries in NorthAmerica may reflect the absence of significant recent ex-ploration in this region.

1. Oil Recovery

Recent U.S. oil production of 6.4 MBO/day involvesmostly primary recovery (which is accomplished by usingonly the energy stored in the reservoir), yielding 17% ofthe oil in place, while secondary recovery may contribute51% and tertiary recovery (or enhanced oil recovery, EOR)an additional 12% (US DOE, 1996). Secondary recoverymay be accomplished with water flooding. Tertiary re-covery or EOR (using thermal, gas-miscible, chemical,or microbiological techniques) could potentially augmentU.S. production by 50 BBO. However, current high costshave limited the use of EOR. The U.S. DOE program onProduction and Reservoir Extension is aimed at achievinglower cost secondary and tertiary oil recovery from knownand previously worked reservoirs.

2. The U.S. Strategic Petroleum Reserve

The Strategic Petroleum Reserve (SPR) was established in1975 as a buffer against possible oil embargoes of the typethat occurred during 1973–1974 (US Congress, 1998).The SPR holds currently about 564 MBO (million bar-rels of oil), well below the initially planned level of 1000MBO. Since its establishment, a sufficient number of tests

have been performed to ensure that filling and emptying(drawdown) procedures for reserve additions and subtrac-tions function properly. As of 1998, the SPR had becomea depository for the sale of oil in order to generate rev-enue for the government, while additions were curtailed toreduce federal spending, on the assumption that the like-lihood of supply interruptions was low because OPEC’sshare of U.S. imports had been reduced to about 30% whilethe unregulated oil markets operated efficiently with re-spect to price allocation. For reasons that are unclear, aconsensus has emerged that an SPR containing around500 MBO is sufficient, even though this level of supply isused up in the United States in little more than 30 days. Areview of management procedures used for the SPR maysuggest that reserve additions have been made preferen-tially when oil prices were high while reserve drawdownsoccurred occasionally when oil prices were low. In viewof the privatization of the U.S. Navy’s Elk Hill petroleumreserve, the SPR may be said to provide now about thesame level of domestic security as was available prior toestablishment of the SPR and sale of the Navy reserve.

The total input into the SPR through 1995 was612.8 MBO, with 41.9% or 256.7 MBO from Mexico,147.3 MBO (24.0%) from the North Sea, 48.1 MBO(7.8%) from U.S. production, 27.1 MBO (4.4%) fromSaudi Arabia, 23.7 MBO (3.9%) from Libya, 20.0 MBO(3.3%) from Iran, and lesser amounts from other countries.Drawdown during the Gulf War reduced SPR holdings to568.5 MBO, whereas subsequent purchases led to refillingto 591.6 MBO.

The drawdown and distribution capability was designedto be 4.3 MBO/day for 90 days. Gas seepage and saleor transfer of 69 MBO from the Week Island site, whichwas decommissioned during 1996, reduced the drawdowncapacity to 3.7 MBO/day by the spring of 1998.

We refer to the U.S. Congress (1998) for some discus-sion of political decisions that have determined drawdown,refilling, and sale of SPR oil.

B. Natural Gas Reserves and MethaneHydrate Resources

The reserve-to-production ratio for natural gas (NG) waslarger than that for oil in 1996.6 The world total was 62.2;it was 11.8 for North America, 18.6 for Europe, 40.1for the Asia-Pacific region, 70.2 for South and CentralAmerica, 80.1 for the CIS, and more than 100 for Africaand the Middle East. NG supply estimates, like those forconventional oil, were greatly increased in the year 2000USGS revisions.

6As cited in the British Petroleum Company’s “Statistical Reviews ofWorld Energy” for that year.

P1: GLM Final Pages



Resource estimates for methane hydrates are highly un-certain but clearly much greater than the known resourcesof NG. Examples are the following (Mielke, 1998): (1)The world’s total gas hydrate accumulations range from1 × 105 to 3 × 108 trillion feet3 of NG, i.e. from 1 × 1030 to3 × 1033 erg. These energy estimates range from severaltimes the known NG reserves to those of uranium fromseawater used in breeder reactors, with the larger valuesufficient to supply 10 × 109 people at the estimated year2050 energy demand for 150,000 years. (2) The USGS1995 appraisal of NG for the United States is 3 × 105 tril-lion feet3, or 3 × 1030 erg, which exceeds the known U.S.reserves of NG by about a factor of 140. Seafloor stabilityand safe recovery of hydrates are open issues.

C. Reserves and Resources of Coals

Based on 1978 technology, the world’s coal reserves(Penner, 1987, p. 627) were estimated to equal about3 × 1012 bbl of oil-equivalent (bble) = 1.7 × 1029 erg, withthe U.S. share of 28% exceeding that of Europe (20%), theCIS (17%), and China (16%). Estimates generally includecoals of all ranks (from peat, lignite, subbituminous andbituminous coals, to anthracite).

1. U.S. Coal Resources and Coal Use

U.S. coal resources are being reassessed by the USGS withinitial emphasis on (a) the Appalachian and Illinois basins,(b) the Gulf Coast, (c) the northern Rocky Mountainsand Great Plains, and (d) the Rocky Mountains–ColoradoPlateau (USGS, 1996). Even before completion of thisstudy, it is apparent that U.S. coal reserves will not soonbe exhausted. About one billion short tons (t) of coal wereproduced for the first time in the United States in 1994after a continuous sequence of production rises extendingfor a period of time longer than a century.

Consumption during 1994 in the United States was930 × 106 t, of which electric utilities used 817 × 106 tand 71 × 106 t were exported. Nonutility use was mostlyconfined to steel production.

2. Environmental Issues in Coal Use

Concerns have often been expressed about coal utilizationas a source of toxic and also of radioactive materials sincecoals contain uranium and thorium at concentration levelsof one to four parts per million (ppm), with uranium atlevels of 20 ppm rare and thorium as high as 20 ppm veryrare (USGS, 1997). Coal combustion releases both U andTh from the coal matrix. As a result, more than 99.5% ofthese elements ends up in the 10% of solid ash which isformed on the average, i.e., the concentrations of radioac-

tive U and Th in the ash are approximately 10 times thosein the original coals. The resulting levels of radioactiveelements are comparable to or lower than those normallyfound in granitic and phosphate rocks and lower than thoseof Chattanooga shales (10-85 ppm of U). The radioactivecombustion products impact the environment when theybecome dispersed in air and water or in flyash from coals,which is incorporated in such commercial products asconcrete building blocks. Upper-bound estimates suggestpossible radioactive dose enhancement up to about 3%in residences, corresponding to roughly 10% of the dosefrom indoor radon that is normally present and contributesabout 65% of the total radioactive load. Of the man-maderadiation sources, X-rays for diagnostic purposes are thelargest contributors at about 11% of the total exposure.About 75% of the annual production of flyash ends up inimpoundments, landfills, and abandoned mines or quar-ries. Standard tests to ensure contamination levels below100 times those of potable water for leachability of toxicflyash constituents such as arsenic, selenium, lead, andmercury show that the actual amounts of dissolved con-stituents justify classification of flyash as a nonhazardoussolid waste. The U.S. Environmental Protection Agencyhas established radioactivity standards for drinking waterat 5 picocuries per liter for dissolved radium; standardsfor U and radon are to be established soon. The leachabil-ity of radioactive elements increases greatly for stronglyalkaline (pH � 8) or strongly acidic (pH � 4) water-basedsolutions. Further studies are required, especially for pastflyash disposals in leaky landfills. Under current regula-tions for flyash disposal, leaching of radioactive elementsor compounds should not pose a significant environmentalhazard.

D. Oil Reserves for Oil Recoveryfrom Tar Sands

Of the very large Athabasca tar-sand resources, about300 × 109 bbl = 1.7 × 1028 erg are recoverable with cur-rent technology (George, 1998) and may therefore be clas-sified as reserves. This reserve assessment is expectedto increase greatly with oil price escalation above about$20/bbl.

E. Shale-Oil Reserves

Although shale-oil resources are enormous and limitedcommercial recovery has been in progress during mostof the 20th century (e.g., in China and Estonia), a ma-jor U.S. development utilizing in situ recovery during the1970s and 1980s (Penner and Icerman, 1984, pp. 10–67)was terminated because of a decline in oil prices. It is un-likely that this program will be reactivated before stable oil

P1: GLM Final Pages



prices rise above about $30/bbl, at which time substantialreserve estimates may be demonstrated for shale oil.

VI. THE KYOTO PROTOCOL

The Kyoto Protocol was signed by the Clinton Adminis-tration in November 1998 but has not yet been ratified bythe U.S. Senate. It obligates the United States to reducecarbon emissions 7% below 1990 levels during the 2008-2012 time period.

Workers at the Energy Information Agency (EIA) of theU.S. Department of Energy have considered six projec-tions to 2012 for U.S. carbon emissions, which range froman upper bound representing the “reference case” or busi-ness as usual to the protocol-mandated reduction levels.Year 2012 U.S. carbon-emission levels range from about1.25 × 109 to 1.87 × 109 mt.

The carbon reduction scenarios have large associatedcosts. The year 2010 U.S. gross domestic product is esti-mated to be reduced by about $400 billion dollars to meetprotocol requirements by 2010 without interdicted emis-sions trading with less developed countries. Emissions re-ductions may be achieved in a number of ways, e.g., byimposing a carbon emission tax, by extensive fuel switch-ing from coal to NG for electricity generation and fromoil to natural gas for transportation, by curtailing gasolineuse through oil-price hikes, etc. According to the EIA,meeting protocol requirements will increase the averageyear 2010 household energy costs by $1740.

It is unlikely that appreciable carbon mitigation can beaccomplished as early as 2010 through implementationof such new energy-efficient technologies as fuel-cell-powered vehicles. Probably the most cost-effective proce-dure for reducing carbon emissions is a large-scale changeto nuclear power generation, perhaps to levels exceeding50%, which were reached some years ago in France, Japan,and some other developed countries.

VII. POLICY CONSIDERATIONS

We have seen that there are abundant low-cost energy sup-plies, especially in the form of fossil fuels and fissionableuranium, which could be used to supply the world’s en-ergy needs for many years to come. Unfortunately, thisis not the end of story. The use of any energy reserve orresource for human well-being is accompanied not onlyby an increase in global entropy but also by a variety ofdetrimental environmental effects. Among these, publicattention has been focused primarily on possibly disas-trous long-range climate changes associated with the verywell documented relentless escalation of the atmosphericCO2 concentration since the beginning of large-scale an-thropogenic conversion of gaseous, liquid, and solid fossil

fuels to combustion products, of which benign water andperhaps not so benign CO2 are the chief constituents.

A. Mitigation of the Greenhouse Effect

The increased gaseous water production associated withfossil fuel use is of no special environmental consequencefor three reasons: (1) The resulting increase in the atmo-spheric water concentration is negligibly small comparedwith the preexisting steady-state atmospheric water con-centrations, (b) consequently, the changes in atmosphericheating produced by any increases in the dominant green-house gas, water, are also negligibly small, and (c) finally,the residence time of the added water in the atmosphereis very short (on the order of days to weeks) and thereforewater cannot accumulate over very long periods of time.Unfortunately, the story is reversed for anthropogenic CO2

addition to the atmosphere because (a) the added amountsof CO2 are not negligibly small compared with preexistinglevels (manmade contributions have already raised preex-isting CO2 levels by upward of 25%), (b) the added CO2

increases the greenhouse effect measurably because long-wavelength absorption by the added CO2 augments atmo-spheric heating, and (c) the residence time of the addedCO2 is of the order of 100–200 years and may thereforebe responsible for long-term atmospheric heating changesas the result of absorption of long-wavelength radiationemitted from the planetary surface. The inevitable conclu-sions derived from these physical effects are that whereashydrogen would be an ideal fuel, fossil fuels may causeunacceptable long-term climate changes because of CO2

emissions to the atmosphere. Hence, we must face pro-posals to eliminate fossil-fuel use without CO2 mitigationover the long term.

B. Elimination of Long-LivedRadioactive Products

The second supply source that is now in large-scale useinvolves nuclear fission reactors. With these, we also facedifficult environmental issues, resulting from (a) inade-quately designed, maintained, or improperly operating re-actors leading to calamitous accidents (e.g., Chernobyl,Three-Mile Island, Tokaimura, and elsewhere) and (b) thecertain production of long-lived radioactive isotopeswhich will require safe storage for very long periods oftime, exceeding hundreds of thousands of years in somecases. Whereas the accident issues will be controllable byconstructing passively safe nuclear reactors (i.e. by con-structing reactors that cannot cause calamitous accidentsas the result of human errors, which are designs that willnot produce runaway heating or radioactive emissions be-cause of the judicious use of physical phenomena thatwill inhibit runaway calamities as the result of human

P1: GLM Final Pages



TABLE II A Global, Large-Scale Energy Supply Trade-off Matrix (1999)a

Overriding consideration in source selection

No long-lived Low risk because ofLow current No net CO2 radioactive established large-scale

Large-scale energy source cost emission products technology

Fossil fuels (all) 5–10 2 10 9–10

Coals 10 2 10 10

Oils 8–10 2 10 10

Natural gases 9–10 2 10 10

Shale oils 5–10 2 10 9–10

Oils from tar sands 7–10 2 10 9–10

Solar technologies (all) 2–10 3–10 10 2–10

Wind energy 4–10 9–10 10 7–9

Photovoltaics 2–10 2–10 10 2–10

Biomass 2–10 7–10 10 2–10

Hydrogen production 2 3–10 10 2

Ocean thermal energy production 4–8 4–10 10 4–8

Hydroelectric power (partial supply) 10 10 10 10

Hydrothermal energy (partial supply) 10 10 10 10

Hot dry-rock geothermal energy (partial supply) 2–8 10 10 2–8

Nuclear fission reactors (all) 10 9–10 2 10

Nuclear breeder reactors (all) 7 9–10 2 9

Fusion reactors 2 9–10 7–9 2

a 10, Very high probability of meeting goals; 1, very low probability of meeting goals.

errors), long-term safe storage of highly radioactive re-actor products remains an inevitable component of the“Faustian bargain” (to quote Alvin Weinberg) which ac-companies our use of nuclear fission reactors. Proponentsof nuclear fission reactors consider long-term, safe stor-age in stable geological strata to provide a proper tech-nical response to this issue. Opponents of nuclear powercontinue to believe that we are incapable of coping withthe required long-term safe storage issues, especially be-cause their use would, in principle, facilitate access tonuclear weapons materials and thereby invite weaponsproliferation.

The preceding arguments lead to the conclusion that wemust change to renewable resources, provided their use isnot so costly as to cause economic collapse for the grow-ing world population, which appears inevitably reachingaround 10 billion people in about 50 years, only a fewyears longer than the historical time scale for replacingan established energy technology by a new energy tech-nology. After more than 25 years of investments in andsubsidies for renewable technologies, the absence of sig-nificant progress makes it appear likely that this programwill not lead to acceptable, cost-effective, large-scale re-placements of fossil and nuclear energy supply systemsover the near term.

The preceding considerations are summarized inTable II in terms of numerical estimates reflecting theauthor’s assessments of the current state of technologicaldevelopment on a scale from 1 to 10, with 10 representinga currently competitive technology, 2 a research agendaaimed at creating a competitive technology, and numbersbetween 2 and 10 representing subjective assessments ofhow close we have come thus far in creating replacementsfor currently competitive technologies on a large scaleand at acceptable costs.

While one may view Table II as a proper basis for ra-tional progress, it is important to remember that experts inother countries may find our approach for defining a newenergy road map to the future either uninteresting or irrel-evant to their needs and preferences. With an inevitableemphasis on low-cost supplies for local well-being,Table III for China, India, France, Norway, and OPECmembers reflects local political preferences and goalsrather than the entries in Table II.

C. Research and Development Agendafor Future Energy Security

If a political decision is made to eliminate man-made,long-lived radioactive products from the environment, no

P1: GLM Final Pages



TABLE III Local Energy Supply Trade-off Matrices for China, India, France, Norway, and OPEC Membersa

Overriding consideration in source selection

Low local Export Established EnvironmentalCountry Energy source cost value technology opposition

China Fossil fuels (all) 8–10 8–10 10 Minimal

Coals 10 10 10 Minimal

Oils 10 10 10 Minimal

Natural gases 10 10 10 Minimal

Shale oils 8 10 10 Minimal

Oils from tar sands 8 10 10 Minimal

Nuclear fission reactors 10 10 10 Minimal

Nuclear breeder reactors 7 7 9 Minimal

Hydroelectric power 10 10 10 Minimal

India Coal 10 10 10 Minimal

Fuelwood 10 1 10 Moderate

Hydroelectric power 8–10 1–10 10 Moderate to strong

Biomass (large scale) 5–10 1–3 7 Moderate

Nuclear fission reactors 10 10 10 Minimal

Nuclear breeder reactors 7 10 9 Minimal

France Nuclear fission reactors 10 10 10 Low

Nuclear breeder ractors 7 10 9 Low

Norway Oils 10 10 10 Minimal

Hydroelectric power 10 10 10 Minimal

Natural gases 10 10 10 Minimal

OPEC members Oils 10 10 10 Minimal

a 10, Very high probability of meeting goals; 1, very low probability of meeting goals.

special supply problems will arise for many years becausewe can continue and augment the use of low-cost fossilfuels. The practical agenda will stress improved energyutilization efficiencies, while R&D emphasis will remainon CO2 sequestration at the source and on a variety of solartechnologies and the development of fusion reactors (seeTable IV). Unfortunately, our present perspective mustassign low probabilities of success to all of the identifiedprograms if the focus is on low-cost, large-scale supplysources by about the year 2025.

If our political choice becomes elimination of CO2

emissions to the atmosphere, or disposal in the deep sea orstable caverns of the enormous amounts of CO2 generatedfrom fossil-fuel burning, then we see again no disastrousnear-term or long-term calamities since the nuclear fissionoptions remain for known, relatively low-cost supply. Inthis environment, near-term construction emphasis mustproperly shift to the use of passively safe nuclear fissionand breeder reactors at somewhat elevated costs in order toensure the availability over the very long term of adequateenergy supplies for an increasingly affluent and growingworld population. While the R&D program on solar and

fusion reactors continues as before, greatly augmented ef-fort should be directed also at the commercialization ofCO2 sequestration with the hope of retaining some of thefossil-fuel supply sources, especially during the transitionperiod to large-scale use of nuclear breeder technologies(see Table IV).

What should be done if both fossil-fuel and nuclear fis-sion reactors are interdicted? As Table IV shows, we arenow left without usable, low-cost, large-scale energy sup-ply sources and only the promise of better things to come ifwe can succeed in the cost-effective commercialization oflarge-scale solar or fusion technologies. During the fore-seeable future, this will be unacceptable to people every-where and will not be followed by their elected officialsbecause the economic consequences would be calamitousand stable political structures could not survive.

D. The Preferred Research and DevelopmentAgenda for the Future

There is little doubt that the best energy technology forthe world at large is the ability to produce low-cost H2 on

P1: GLM Final Pages



TABLE IV Selected Research and Development Agenda

Probabilityof success at

Goal Energy source Agenda acceptable costa

Early elimination of Fossil fuels (all) CO2 capture at the source and utilization for the production 2–4CO2 emissions to of salable chemicals; CO2 capture and separation at thethe atmosphere source prior to long-term storage in the sea or elsewhere 2–4

Solar technologies (all) Direct H2 production from water (e.g., by water splitting inlow-temperature thermal cycles, direct water photolysis, etc.); 2large-scale development of energy-from-the-sea technologiesfollowed by hydrogen or other fuel production; genetically 6–8altered biomass to improve crop yields for energy use; solar 2energy collection in space orbits followed by conversion and 2transmission to earth stations; large-scale use of wind power 2or photovoltaics

Nuclear fission reactors Implement large-scale use of passively safe nuclear fission reactors 10

Nuclear breeder reactors Implement large-scale use of passively safe nuclear breeder reactors 10

Fusion reactors Implement large-scale use of systems 2

Early cessation of the Fossil fuels (all) Improve utilization efficiencies 10

production of long-lived Solar technologies (all) As above

radioactive isotopes Fusion reactors Implement large-scale use of systems 2

No CO2 emissions and no Solar technologies (all) As above

nuclear fission reactors Fusion reactors Implement large-scale use of systems 2

a 10, High probability of success; 1, low probability of success.

the required global scale, i.e. the preferred goal is whatenthusiasts have long labeled “the hydrogen economy.”It matters little how the hydrogen is ultimately produced.A viable fusion energy supply is, of course, completelycompatible with the hydrogen economy since watersplitting by using electrical energy is a long-establishedtechnology. Similarly, the use of solar energy to produceH2 in low-temperature thermal cycles or by the direct pho-tolysis of water constitute priority methods for H2 pro-duction, as is also NH3 synthesis in an ocean thermalenergy conversion system followed by conversion to H2

after transmission to user locations, etc. The emphasis oncombining new energy supply options with large-scale hy-drogen production should become the worldwide priorityR&D program.


ENERGY EFFICIENCY COMPARISONS BETWEEN COUN-TRIES • ENERGY FLOWS IN ECOLOGY AND IN THE ECON-OMY • FOSSIL FUEL POWER STATIONS • HYDROELEC-TRIC POWER STATIONS • NUCLEAR POWER REACTORS •OCEAN THERMAL ENERGY CONVERSION • PETROLEUM

GEOLOGY • RENEWABLE ENERGY FROM BIOMASS •SOLAR THERMAL POWER STATIONS • TIDAL POWER SYS-

TEMS • WASTE-TO-ENERGY SYSTEMS • WIND POWER

SYSTEMS

BIBLIOGRAPHY

Anderson, J. H., and Anderson, J. H., Jr. (1964). “Power from the sun, viathe sea,” Paper presented at the Susquehanna Section of the ASME,November 1964.

Anderson, J. H., and Anderson, J. H., Jr. (April 1966). “Thermal powerfrom sea water: 100,000 kW at 3 mills/kWh,” Mech. Eng. 88, 41–46.

Anderson, R. N. (1998). “Oil production in the 21st century,” Sci. Am.1998(March), 86–91.

Averitt, P. (1969). “Coal Resources of the United States,” U.S. GeologicalSurvey Bulletin 1275, Washington, DC.

Best, F. R., and Driscoll, M. J. (1980). Trans. Am. Nucl. Soc. 34, 380.Best, F. R., and Driscoll, M. J. (1981). Trans. Am. Nucl. Soc. 38, 200.Campbell, C. J., and Laherrere, J. H. (1998). “The end of cheap oil,” Sci.

Am. 1998(March), 78–83.Cohen, B. L. (1983). “Breeder reactors: A renewable energy source,”

Am. J. Phys. 51, 1.Cooper, C. (1999). “It’s no crude joke: This oil field grows even as it’s

tapped,” Wall Street Journal, Friday, April 16, p. A1.Cornell (1972). “Cornell University Workshop on Energy and the Envi-

ronment,” US Government Printing Office, Washington, DC.Duncan, D. C., and Swanson, V. E. (1965). “Organic-Rich Shales of the

United States and World Land Areas,” US Geological Survey Circular523, Washington, DC.

Exxon (1998–1999). “The Lamp,” Exxon Corporation, Shareholder Ser-vices, Irvine, TX.

P1: GLM Final Pages



George, R. L. (1998). “Mining for Oil,” Sci. Am. 1998(March), 84–85.Gold, T. (1999). “The Deep Hot Biosphere,” Springer-Verlag, New

York.Hannesson, R. (1998). “Petroleum Economics: Issues and Strategies of

Oil and Natural Gas Production, II,” Quorum Books, Westport, CT.IAEA. (1966). “Utilization of thorium in power reactors,” In Technical

Report Series No. 52, pp. 10–11, International Atomic Energy Agency,Vienna.

Koczy, F. F. (1954). In “Nuclear Geology” (H. Faul, ed.), Wiley, NewYork.

Mielke, J. E. (1998). “Methane Hydrates: Energy Prospect or Natu-ral Hazard?” Congressional Research Services Report, The Com-mittee for the National Institute for the Environment, Washington,DC.

Palmen, E. H., and Newton, C. W. (1969). “Atmospheric Circulation Sys-tems: Their Structure and Physical Interpretations,” Academic Press,New York.

Penner, S. S. (ed.). (1978). “Lithium Needs and Resources,” PergamonPress, Oxford.

Penner, S. S. (ed.). (1987). “Coal Gasification,” Pergamon Press, Oxford[reprinted from Energy–The Int. J. 12(8/9), 623–903 (1987)].

Penner, S. S., and Icerman, L. (1984). “Energy,” Vol. II, “Non-Nuclear

Energy Technologies,” 2nd ed., Addison-Wesley, Reading, MA.Science. (2000). “USGS Optimistic on World Oil Prospects,” Science

289, 237; review by R. A. Kerr.Smil, V. (1999). “Energies,” MIT Press, Cambridge, MA.Spragins, F. K. (1967). “Mining at Athabasca—A new approach to oil

production,” J. Petroleum Technol. 19, 1337–1343.Swanson, V. E. (1960). “Oil yield and uranium content of beach shales,”

In “Uranium and Carbonaceous Rocks,” US Geological Survey Paper356, Washington, DC.

US Congress. (1998). “Strategic Petroleum Reserve,” Congressional Re-search Service (CRS) Report 8705, December 7, 1998.

US DOE. (1996). US Department of Energy Internet data.USGS. (1996). USGS Fact Sheet FS-157-96, US Geological Survey,

Washington, DC.USGS. (1997). USGS Fact Sheet FS-163-97, US Geological Survey,

Washington, DC.USGS. (2000). Web site http://greenwood.cr.usgs.gov/energy/World/

Energy/DDS-60.Zebroski, E. L., Zaleski, C.-P., Simnad, M., Baker, C. S., and Penner, S. S.

(eds.). (1998). “Advanced Nuclear Reactors—Current Developmentsand Future Prospects,” Elsevier, Oxford [reprinted from Energy–TheInt. J. 23, 521–694].

http://greenwood.cr.usgs.gov/energy/World/Energy/DDS-60

http://greenwood.cr.usgs.gov/energy/World/Energy/DDS-60

P1: GLM Final Pages Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology EN005G-868 June 15, 2001 20:36

Enhanced Oil RecoveryRonald E. TerryBrigham Young University

I. IntroductionII. Fundamentals of Fluid ProductionIII. Miscible FloodingIV. Chemical FloodingV. Thermal Flooding

VI. Microbial FloodingVII. Screening Criteria for EOR Processes

VIII. Summary

GLOSSARY

Capillary pressure Difference in the pressure betweentwo fluids measured at the interface between the twofluids.

Interfacial tension Measure of the ease with which a newinterface between two fluids can be made.

Miscible When two or more phases become singlephased.

Mobility Measure of the ease with which a fluid movesthrough porous media.

Permeability Measure of the capacity of a porous med-ium to conduct a fluid.

Polymer Large-molecular-weightchemicalused to thick-en a solution.

Primary production Production of oil using only the nat-ural energy of the formation.

Reservoir Volume of underground porous media usuallycontaining rock, oil, gas, and water.

Residual oil Amount of oil remaining in a reservoir afterprimary and secondary production.

Secondary production Production of oil when gas, wa-ter, or both are injected into the formation and the in-jected fluid immiscibly displaces the oil.

Surfactant Molecule that is made up of both hydrophilicand hydrophobic entities and can reduce the interfacialtension between two fluids.

Sweep efficiency Measure of how evenly a fluid hasmoved through the available flow volume in a porousmedium.

Viscosity Property of a fluid that is a measure of its resis-tance to flow.

ENHANCED OIL RECOVERY refers to the process ofproducing liquid hydrocarbons by methods other than theconventional use of reservoir energy and reservoir repres-surizing schemes with gas or water. On the average, con-ventional production methods will produce from a reser-voir about 30% of the initial oil in place. The remaining oil,nearly 70% of the initial resource, is a large and attractivetarget for enhanced oil recovery methods.

503

P1: GLM Final Pages


504 Enhanced Oil Recovery

I. INTRODUCTION

A. Classification of Hydrocarbon Production

The initial production of hydrocarbons from an under-ground reservoir is accomplished by the use of naturalreservoir energy. This type of production is termed pri-mary production. Sources of natural reservoir energy thatlead to primary production include the swelling of reser-voir fluids, the release of solution gas as the reservoir pres-sure declines, nearby communicating aquifers, and gravitydrainage. When the natural reservoir energy has been de-pleted, it becomes necessary to augment the natural energywith an external source. This is usually accomplished bythe injection of fluids, either a natural gas or water. Theuse of this injection scheme is called a secondary recoveryoperation. When water injection is the secondary recoveryprocess, the process is referred to as waterflooding. Themain purpose of either a natural gas or a water injectionprocess is to repressurize the reservoir and then to maintainthe reservoir at a high pressure. Hence, the term pressuremaintenance is sometimes used to describe a secondaryrecovery process.

When gas is used as the pressure maintenance agent, itis usually injected into a zone of free gas (i.e., a gas cap)to maximize recovery by gravity drainage. The injectedgas is usually natural gas produced from the reservoir inquestion. This, of course, defers the sale of that gas untilthe secondary operation is completed and the gas can berecovered by depletion. Other gases, such as N2, can beinjected to maintain reservoir pressure. This allows thenatural gas to be sold as it is produced.

Waterflooding recovers oil by the water’s movingthrough the reservoir as a bank of fluid and “pushing”oil ahead of it. The recovery efficiency of a waterflood islargely a function of the sweep efficiency of the flood andthe ratio of the oil and water viscosities.

Sweep efficiency is a measure of how well the water hascome in contact with the available pore space in the oil-bearing zone. Gross heterogeneities in the rock matrix leadto low sweep efficiencies. Fractures, high-permeabilitystreaks, and faults are examples of gross heterogeneities.Homogeneous rock formations provide the optimum set-ting for high sweep efficiencies.

When an injected water is much less viscous than theoil it is meant to displace, the water could begin to finger,or channel, through the reservoir. This is referred to asviscous fingering and leads to significant bypassing ofresidual oil and lower flooding efficiencies. This bypassingof residual oil is an important issue in applying enhancedoil recovery techniques as well as in waterflooding.

Tertiary recovery processes were developed for appli-cation in situations in which secondary processes had be-come ineffective. However, the same tertiary processes

were also considered for reservoir applications for whichsecondary recovery techniques were not used because oflow recovery potential. In the latter case, the name ter-tiary is a misnomer. For most reservoirs, it is advanta-geous to begin a secondary or a tertiary process concur-rent with primary production. For these applications, theterm enhanced oil recovery (EOR) was introduced andhas become popular in referring to, in general, any re-covery process that enhances the recovery of oil beyondwhat primary and secondary production would normallybe expected to yield.

Enhanced oil recovery processes can be classified intofour categories:

1. Miscible flooding processes2. Chemical flooding processes3. Thermal flooding processes4. Microbial flooding processes

The category of miscible displacement includes single-contact and multiple-contact miscible processes. Chem-ical processes are polymer, micellar–polymer, and alka-line flooding. Thermal processes include hot water, steamcycling, steam drive, and in situ combustion. In general,thermal processes are applicable in reservoirs containingheavy crude oils, whereas chemical and miscible displace-ment processes are used in reservoirs containing lightcrude oils. Microbial processes use microorganisms to as-sist in oil recovery.

B. Hydrocarbon Reserves and Potentialof Enhanced Recovery

In the United States, the remaining producible reserve isestimated to be 21 billion barrels. Of this 21 billion, cur-rently implemented EOR projects are expected to recover3 billion barrels. A 1998 report in the Oil and Gas Jour-nal listed a production of 759,653 barrels of oil per day(b/d) from EOR projects in the United States. This amountrepresented about 12% of the total U.S. oil production.

A somewhat dated but highly informative study con-ducted by the U.S. National Petroleum Council (NPC)and published in 1984 determined that, with current EORtechnology, an estimated 14.5 billion barrels of oil couldbe produced in the United States over a 30-yr period. Thisamount includes the 3 billion barrels that are expected tobe produced from current EOR projects. The 14.5-billion-barrel figure was derived from a series of assumptions andsubsequent model predictions. Included in the assump-tions was an oil base price of $30 per barrel in constant1983 U.S. dollars. The ultimate oil recovery was projectedto be very sensitive to oil price, as shown in Table I.

P1: GLM Final Pages


Enhanced Oil Recovery 505

TABLE I Ultimate Oil Recovery from EnhancedOil Recovery Methods as a Function of Oil Pricea

Oil price per bbl Ultimate recovery(1983 U.S. dollars) (billions of bbl)

20 7.4

30 14.5

40 17.5

50 19.0

a bbl, barrel(s).

The NPC study also attempted to predict what the ul-timate recoveries could reach if technological advance-ments were made in EOR processes. A potential 13–15billion barrels of oil could be added to the figures inTable I if research activities led to improvements in EORtechnology.

Interest in EOR activity is increasing outside the UnitedStates. In the same Oil and Gas Journal report that wasreferenced earlier, the worldwide production from EORprojects was listed as 2.3 million b/d. The largest projectin the world at the time of the 1998 Oil and Gas Journalreport was a steam drive in the Duri field in Indonesia. TheDuri steam drive was producing about 310,000 b/d in thefirst quarter of 1998. Total estimated recovery from theDuri field is expected to be nearly 2 billion barrels of oil.Canada continues to report significant EOR production of400,000 b/d, while China and Russia have increased theirproduction to 280,000 and 200,000 b/d, respectively.

II. FUNDAMENTALS OFFLUID PRODUCTION

A. Overall Recovery Efficiency

The overall recovery efficiency E of any fluid displace-ment process is given by the product of the macroscopic,or volumetric, displacement efficiency Ev and the micro-scopic displacement efficiency Ed:

E = Ev Ed.

The macroscopic displacement efficiency is a measure ofhow well the displacing fluid has come in contact with theoil-bearing parts of the reservoir. The microscopic dis-placement efficiency is a measure of how well the dis-placing fluid mobilizes the residual oil once the fluid hascome in contact with the oil.

The macroscopic displacement efficiency is made upof two other terms, the areal, Es, and vertical, Ei, sweepefficiencies:

Ev = Es Ei.

B. Microscopic Displacement Efficiency

The microscopic displacement efficiency is affected by thefollowing factors: interfacial and surface tension forces,wettability, capillary pressure, and relative permeability.

When a drop of one immiscible fluid is immersed inanother fluid and comes to rest on a solid surface, the sur-face area of the drop will take a minimum value due to theforces acting at the fluid–fluid and rock–fluid interfaces.The forces per unit length acting at the fluid–fluid androck–fluid interfaces are referred to as interfacial tensions.The interfacial tension between two fluids represents theamount of work required to create a new unit of surfacearea at the interface. The interfacial tension can also bethought of as a measure of the immiscibility of two fluids.Typical values of oil–brine interfacial tensions are on theorder of 20–30 dyn/cm. When certain chemical agents areadded to an oil–brine system, it is possible to reduce theinterfacial tension by several orders of magnitude.

The tendency for a solid to prefer one fluid over an-other is called wettability. Wettability is a function of thechemical composition of both the fluids and the rock. Sur-faces can be either oil-wet or water-wet, depending on thechemical composition of the fluids. The degree to whicha rock is either oil-wet or water-wet is strongly affectedby the adsorption or desorption of constituents in the oilphase. Large, polar compounds in the oil phase can absorbonto the solid surface; this leaves an oil film that may alterthe wettability of the surface.

The concept of wettability leads to another significantfactor in the recovery of residual oil. This factor is capillarypressure. To illustrate capillary pressure, let us considera capillary tube that contains both oil and brine, the oilhaving a lower density than that of the brine. The pressurein the oil phase immediately above the oil–brine inter-face in the capillary tube will be slightly greater than thepressure in the water phase just below the interface. Thisdifference in pressure is called the capillary pressure Pc

of the system. The greater pressure will always occur inthe nonwetting phase. An expression relating the contactangle θ , the radius r of the capillary, the oil–brine interfa-cial tension γwo, and the capillary pressure Pc is given inEq. (1):

Pc = (2γwo cos θ )/r. (1)

This equation suggests that the capillary pressure in aporous medium is a function of the chemical composi-tion of the rock and fluids, the pore size distribution, andthe saturation of the fluids in the pores. Capillary pressureshave also been found to be a function of the saturation his-tory, although this dependence is not reflected in Eq. (1).Because of this, different values will be obtained duringthe drainage process (i.e., displacing the wetting phase

P1: GLM Final Pages



with the nonwetting phase) than those obtained during theimbibition process (i.e., displacing the nonwetting phasewith the wetting phase). This historesis phenomenon isexhibited in all rock–fluid systems and is an impor-tant consideration in the mathematical modeling of EORprocesses.

It has been shown that the pressure required to force anonwetting phase through a small capillary can be verylarge. For instance, the pressure drop required to force anoil drop through a tapering constriction that has a forwardradius of 0.00062 cm, a rearward radius of 0.0015 cm, acontact angle of 0◦, and an interfacial tension of 25 dyn/cmis 0.68 psi. If the oil drop were 0.01 cm long, a pressuregradient of 2073 psi/ft would be required to move the dropthrough the constriction. This is an enormous pressuregradient, which cannot be achieved in practice. Typicalpressure gradients obtained in reservoir systems are of theorder of 1–2 psi/ft.

Another factor affecting the microscopic displacementefficiency is the fact that two or more fluids are usuallyflowing in an EOR process. When two or more fluid phasesare present, the saturation of one phase affects the perme-ability of the other(s), and relative permeabilities have tobe considered. Figure 1 is an example of a set of rela-tive permeability curves plotted against the wetting phasesaturation (water in this case).

The relative permeability to oil, Kro, is plotted on the leftvertical axis, whereas the relative permeability to water,Krw, is plotted on the right. The curve for Kro goes to zeroat Sor, the residual oil saturation. Once the oil saturationhas been reduced to this point in a pore space by a water-flood, no more oil will flow, since Kro is zero. Similarly,at saturations below the irreducible water saturations, Swr

FIGURE 1 Typical water–oil relative permeability curves for aporous medium.

in the figure, water will not flow, since Krw is zero. Thecurves are strong functions of wettability and do exhibita historesis effect (especially for the nonwetting phasepermeability).

C. Macroscopic Displacement Efficiency

Factors that affect the macroscopic displacement effi-ciency are the following: heterogeneities and anisotropy,the mobility of the displacing fluids compared with themobility of the displaced fluids, the physical arrangementof injection and production wells, and the type of rockmatrix in which the oil exists.

Heterogeneities and anisotropy of an oil-bearing for-mation have a significant effect on the macroscopic dis-placement efficiency. The movement of fluids through thereservoir will not be uniform if there are large variationsin such properties as porosity, permeability, and clay con-tent. Limestone formations generally have wide fluctua-tions in porosity and permeability. Also, many formationshave a system of microfractures or large macrofractures.Any time a fracture occurs in a reservoir, fluids will beinclined to travel through the fracture because of the highpermeability of the fracture. This may lead to substantialbypassing of residual oil. The bypassing of residual oil byinjected fluids is a major reason for the failure of manypilot EOR projects. Much research is being conducted onhow to improve the sweep efficiency of injected fluids.

Mobility is a relative measure of how easily a fluidmoves through porous media. The apparent mobility isdefined as the ratio of effective permeability to fluid vis-cosity. Since the effective permeability is a function offluid saturations, several apparent mobilities can be de-fined. The mobility ratio M is a measure of the relativeapparent mobilities in a displacement process and is givenby Eq. (2):

M = mobility of displacing phase

mobility of displaced phase. (2)

When a fluid is being injected into a porous medium con-taining both the injected fluid and a second fluid, the appar-ent mobility of the displacing phase is usually measuredat the average displacing phase saturation as the displac-ing phase just begins to break through at the productionsite. The apparent mobility of the nondisplacing phase ismeasured at the displacing phase saturation that occursjust before the beginning of the injection of the displacingphase. Sweep efficiencies are a strong function of the mo-bility ratio. The phenomenon called viscous fingering cantake place if the mobility of the displacing phase is muchgreater than the mobility of the displaced phase.

The arrangement of injection and production wells de-pends primarily on the geology of the formation and the

P1: GLM Final Pages



size (areal extent) of the reservoir. When an operator isconsidering an EOR project for a given reservoir, he or shewill have the option of using the existing well arrangementor drilling new wells in different locations. If the operatoropts to use the existing well arrangement, there may be aneed to consider converting production wells to injectionwells or vice versa. This will require analysis of tubing andother factors to determine whether the existing equipmentcan withstand the properties of the chemicals or thermalenergy to be injected. An operator should also recognizethat when a production well is converted to an injectionwell, the production capacity of the reservoir has been re-duced. Often this decision can lead to major cost itemsin the overall project and should involve a great deal ofconsideration. Knowledge of any directional permeabil-ity effects and other heterogeneities can aid in the con-sideration of well arrangements. The presence of faults,fractures, and high-permeability streaks could dictate theshutting in of a well near one of these heterogeneities. Di-rectional permeability trends could lead to a poor sweepefficiency in a developed pattern and might suggest thatthe pattern be altered in one direction or that a differentpattern be used.

Sandstone formations are characterized by a more uni-form pore geometry than that of limestones. Limestoneshave large holes (vugs) and can have significant fractures,which are often connected. Limestone formations are as-sociated with connate water that can have high levels of di-valent ions such as Ca2+ and Mg2+. Vugular porosity andhigh divalent ion content in their connate waters hinderthe application of EOR processes in limestone reservoirs.Conversely, a sandstone formation can be composed ofsuch small sand grain sizes and be so tightly packed thatfluids will not readily flow through the formation.

D. Capillary Number Correlation

In a water-wet system, during the early stages of a water-flood, the brine exists as a film around the sand grains andthe oil fills the remaining pore space. At an intermediatetime during the flood, the oil saturation has been decreasedand exists partly as a continuous phase in some pore chan-nels but as discontinuous droplets in other channels. At theend of the flood, when the oil has been reduced to residualoil saturation Sor, the oil exists primarily as a discontinu-ous phase of droplets or globules that have been isolatedand trapped by the displacing brine.

The waterflooding of oil in an oil-wet system yields adifferent fluid distribution at Sor. Early in the waterflood,the brine forms continuous flow paths through the centerportions of some of the pore channels. The brine entersmore and more of the pore channels as the waterfloodprogresses. At residual oil saturation, the brine has entered

a sufficient number of pore channels to shut off the oil flow.The residual oil exists as a film around the sand grains. Inthe smaller flow channels this film may occupy the entirevoid space.

The mobilization of the residual oil saturation in awater-wet system requires that the discontinuous globulesbe connected to form a flow channel. In an oil-wet porousmedium, the film of oil around the sand grains needs tobe displaced to large pore channels and be connected in acontinuous phase before it can be mobilized. The mobi-lization of oil is governed by the viscous forces (pressuregradients) and the interfacial tension forces that exist inthe sand grain–oil–water system.

There have been several investigations of the effects ofviscous forces and interfacial tension forces on the trap-ping and mobilization of residual oil. From these studiescorrelations between a dimensionless parameter called thecapillary number Nvc and the fraction of oil recovered havebeen developed. The capillary number is the ratio of vis-cous forces to interfacial tension forces and is defined byEq. (3):

Nvc = V µw/γow = K0 �P/φγowL , (3)

where V is the Darcy velocity, µw the viscosity of displac-ing fluid, γow the interfacial tension between displaced anddisplacing fluid, K0 the effective permeability to displacedphase, φ the porosity, and �P/L the pressure drop asso-ciated with Darcy velocity.

Figure 2 is a schematic representation of the capillarynumber correlation. The correlation suggests that a capil-lary number greater than 10−5 is necessary for the mobi-lization of unconnected oil droplets. The capillary number

FIGURE 2 Capillary number correlation.

P1: GLM Final Pages



increases as the viscous forces increase or as the interfacialtension forces decrease.

III. MISCIBLE FLOODING

A. Introduction

In the previous section, it was noted that the microscopicdisplacement efficiency is largely a function of interfacialforces acting between the oil, rock, and displacing fluid. Ifthe interfacial tension between the trapped oil and the dis-placing fluid could be lowered to 10−2 to 10−3 dyn/cm, theoil droplets could be deformed and could squeeze throughthe pore constrictions. A miscible process is one in whichthe interfacial tension is zero; that is, the displacing fluidand the residual oil mix to form one phase. If the interfacialtension is zero, then the capillary number Nvc becomesinfinite and the microscopic displacement efficiency ismaximized.

Figure 3 is a schematic of a miscible process. Fluid A isinjected into the formation and mixes with the crude oil,which forms an oil bank. A mixing zone develops betweenfluid A and the oil bank and will grow due to dispersion.Fluid A is followed by fluid B, which is miscible withfluid A but not generally miscible with the oil and whichis much cheaper than fluid A. A mixing zone will alsobe created at the fluid A–fluid B interface. It is importantthat the amount of fluid A that is injected be large enoughthat the two mixing zones do not come in contact. If thefront of the fluid A–fluid B mixing zone reaches the rearof the fluid A–oil mixing zone, viscous fingering of fluidB through the oil could occur. Nevertheless, the volume offluid A must be kept small to avoid large injected-chemicalcosts.

Consider a miscible process with n-decane as the resid-ual oil, propane as fluid A, and methane as fluid B. The

FIGURE 3 Schematic of an enhanced oil recovery process re-quiring the injection of two fluids.

FIGURE 4 Ternary diagram illustrating typical hydrocarbonphase behavior at constant temperature and pressure.

system pressure and temperature are 2000 pounds persquare inch absolute (psia) and 100◦F, respectively. Atthese conditions both the n-decane and the propane existas liquids and are therefore miscible in all proportions. Thesystem temperature and pressure indicate that any mixtureof methane and propane would be in the gas state; there-fore, the methane and propane would be miscible in allproportions. However, the methane and n-decane wouldnot be miscible for similar reasons. If the pressure were re-duced to 1000 psia and the temperature held constant, thepropane and n-decane would again be miscible. However,mixtures of methane and propane could be located in atwo-phase region and would not lend themselves to a mis-cible displacement. Note that in this example the propaneappears to act as a liquid when it is in the presence of n-decane and as a gas when it is in contact with methane. Itis this unique capacity of propane and other intermediategases that leads to the miscible process.

There are, in general, two types of miscible processes.One is referred to as the single-contact miscible processand involves such injection fluids as liquefied petroleumgases (LPGs) and alcohols. The injected fluids are misciblewith residual oil immediately on contact. The second typeis the multiple-contact, or dynamic, miscible process. Theinjected fluids in this case are usually methane, inert fluids,or an enriched methane gas supplemented with a C2–C6

fraction. The injected fluid and oil are usually not miscibleon first contact but rely on a process of chemical exchangebetween phases to achieve miscibility.

B. Single-Contact Miscible Processes

The phase behavior of hydrocarbon systems can be de-scribed with the use of ternary diagrams such as Fig. 4.Researchers have shown that crude oil phase behavior can

P1: GLM Final Pages



be represented reasonably well by three fractions of thecrude. One fraction is methane (C1). A second fraction is amixture of ethane through hexane (C2–C6). The third frac-tion is the remaining hydrocarbon species lumped togetherand called C7+. Figure 4 illustrates the ternary phase di-agram for a typical hydrocarbon system with these threepseudocomponents at the corners of the triangle. Thereare one-phase and two-phase regions in the diagram. Theone-phase region may be vapor or liquid (to the left ofthe dashed line through the critical point C) or gas (to theright of the dashed line through the critical point). A gascould be mixed with either a liquid or a vapor in appro-priate percentages and yield a miscible system. However,when liquid is mixed with a vapor, often the result is acomposition in the two-phase region. A mixing processis represented on a ternary diagram as a straight line. Forexample, if compositions V and G are mixed in appro-priate proportions, the resulting mixture would fall on theline V G. If compositions V and L are mixed, the resultingoverall composition M would fall on the line V L but themixture would yield two phases. If two phases are formed,their compositions, V1 and L1, would be given by a tie lineextended through the point M to the phase envelope. Thepart of the phase boundary on the phase envelope fromthe critical point C to point V0 is the dew point curve. Thephase boundary from C to L0 is the bubble point curve.The entire bubble point–dew point curve is referred to asthe binodal curve.

The oil–LPG–dry gas system will be used to illustratethe behavior of the first-contact miscible process on aternary diagram. Figure 5 is a ternary diagram with thepoints O , P , and V representing the oil, LPG, and drygas, respectively. The oil and LPG are miscible in all pro-portions. A mixing zone at the oil–LPG interface will grow

FIGURE 5 Ternary diagram illustrating the single-contact misci-ble process.

FIGURE 6 Ternary diagram illustrating the multicontact dry gasmiscible process.

as the front advances through the reservoir. At the rear ofthe LPG slug, the dry gas and LPG are miscible and amixing zone will also be created at this interface. If thedry gas–LPG mixing zone overtakes the LPG–oil mixingzone, miscibility will be maintained unless the contact ofthe two zones yields mixtures inside the two-phase region(see line M0 M1, Fig. 5).

Reservoir pressures sufficient to achieve miscibility arerequired. This limits the application of LPG processes toreservoirs having pressures at least of the order of 1500psia. Reservoirs with pressures less than this might beamenable to alcohol flooding, another first-contact misci-ble process, since alcohols tend to be soluble with bothoil and water (the drive fluid in this case). The two mainproblems with alcohols are that they are expensive andthey become diluted with connate water during a floodingprocess, which reduces the miscibility with the oil. Alco-hols that have been considered are in the C1–C4 range.

C. Multiple-Contact Miscible Processes

Multiple-contact, or dynamic, miscible processes do notrequire the oil and displacing fluid to be miscible imme-diately on contact but rely on chemical exchange betweenthe two phases for miscibility to be achieved. Figure 6 il-lustrates the high-pressure (lean-gas) vaporizing process.The temperature and pressure are constant throughout thediagram at reservoir conditions. A vapor denoted by V inFig. 6, consisting mainly of methane and a small percent-age of intermediates, will serve as the injection fluid. Theoil composition is given by the point O . The following se-quence of steps occurs in the development of miscibility:

1. The injection fluid V comes in contact with crudeoil O . They mix, and the resulting overall composition is

P1: GLM Final Pages



given by M1. Since M1 is located in the two-phase region,a liquid phase L1 and a vapor phase V1 will form withthe compositions given by the intersections of a tie linethrough M1 with the bubble point and dew point curves,respectively.

2. The liquid L1 has been created from the original oilO by vaporization of some of the light components. Sincethe oil O was at its residual saturation and was immo-bile due to Kro’s being zero, when a portion of the oil isextracted, the volume, and hence the saturation, will de-crease and the oil will remain immobile. The vapor phase,since Krg is greater than zero, will move away from theoil and be displaced downstream.

3. The vapor V1 will come in contact with fresh crudeoil O , and again the mixing process will occur. The overallcomposition will yield two phases, V2 and L2. The liquidagain remains immobile and the vapor moves downstream,where it comes in contact with more fresh crude.

4. The process is repeated with the vapor phase compo-sition moving along the dew point curve, V1–V2–V3, andso on, until the critical point, c, is reached. At this point,the process becomes miscible. In the real case, becauseof reservoir and fluid property heterogeneities and disper-sion, there may be a breaking down and a reestablishmentof miscibility.

Behind the miscible front, the vapor phase compositioncontinually changes along the dew point curve. This leadsto partial condensing of the vapor phase with the result-ing condensate being immobile, but the amount of liquidformed will be quite small. The liquid phase, behind themiscible front, continually changes in composition alongthe bubble point. When all the extractable componentshave been removed from the liquid, a small amount of liq-uid will be left, which will also remain immobile. Therewill be these two quantities of liquid that will remain im-mobile and will not be recovered by the miscible process.In practice, operators have reported that the vapor fronttravels anywhere from 20 to 40 ft from the well bore be-fore miscibility is achieved.

The high-pressure vaporizing process requires a crudeoil with significant percentages of intermediate com-pounds. It is these intermediates that are vaporized andadded to the injection fluid to form a vapor that will even-tually be miscible with the crude oil. This requirement ofintermediates means that the oil composition must lie tothe right of a tie line extended through the critical pointon the binodal curve (see Fig. 6). A composition lying tothe left, such as denoted by point A, will not contain suffi-cient intermediates for miscibility to develop. This is dueto the fact that the richest vapor in intermediates that canbe formed will be on a tie line extended through point A.Clearly, this vapor will not be miscible with crude oil A.

As pressure is reduced, the two-phase region increases.It is desirable, of course, to keep the two-phase regionminimal in size. As a rule, pressures of the order of 3000psia or greater and an oil with an American PetroleumInstitute (API) gravity greater than 35 are required formiscibility in the high-pressure vaporizing process.

The enriched-gas-condensing process is a second typeof dynamic miscible process (Fig. 7). As in the high-pressure vaporizing process, where chemical exchange ofintermediates is required for miscibility, miscibility is de-veloped during a process of exchange of intermediateswith the injection fluid and the residual oil. In this case,however, the intermediates are condensed from the injec-tion fluid to yield a “new” oil, which becomes misciblewith the “old” oil and the injected fluid. The followingsteps occur in the process (the sequence of steps is sim-ilar to those described for the high-pressure vaporizingprocess but contain some significant differences):

1. An injection fluid G rich in intermediates mixes withresidual oil O .

2. The mixture, given by the overall composition M1

separates into a vapor phase V1 and a liquid phase L1.3. The vapor moves ahead of the liquid that remains

immobile. The remaining liquid L1 then comes in contactwith fresh injection fluid G. Another equilibrium occurs,and phases having compositions V2 and L2 are formed.

4. The process is repeated until a liquid is formed fromone of the equilibration steps that is miscible with G. Mis-cibility is then said to have been achieved.

Ahead of the miscible front, the oil continually changesin composition along the bubble point curve. In contrastwith the high-pressure vaporizing process, there is the po-tential for no residual oil to be left behind in the reservoir as

FIGURE 7 Ternary diagram illustrating the multicontact enriched-gas-condensing miscible process.

P1: GLM Final Pages



long as there is a sufficient amount of G injected to supplythe condensing intermediates. The enriched gas processmay be applied to reservoirs containing crude oils withonly small quantities of intermediates. Reservoir pressuresare usually in the range 2000–3000 psia.

The intermediates are expensive, so usually a dry gasis injected after a sufficient slug of enriched gas has beeninjected.

D. Inert Gas Injection Processes

The use of inert gases, in particular carbon dioxide, CO2,and nitrogen, N2, as injected fluids in miscible processeshas become extremely popular. The ternary diagram repre-sentation of the process with CO2 or N2 is exactly the sameas that for the high-pressure vaporizing process except thateither CO2 or N2 becomes a component and methane islumped with the intermediates. Typically the one-phase re-gion is largest for CO2, with N2 and dry gas having aboutthe same one-phase size. The larger the one-phase region,the more readily miscibility will be achieved. Miscibilitypressures are lower for CO2, usually in the neighborhoodof 1200–1500 psia, whereas N2 and dry gas yield muchhigher miscibility pressures (i.e., 3000 psia or more).

The capacity of CO2 to vaporize hydrocarbons is muchgreater than that of natural gas. It has been shown that CO2

vaporizes hydrocarbons primarily in the gasoline and gas–oil range. This capacity of CO2 to extract hydrocarbons isthe primary reason for the use of CO2 as an oil recoveryagent. It is also the reason CO2 requires lower miscibilitypressures than those required by natural gas. The presenceof other diluent gases such as N2, methane, or flue gas withthe CO2 will raise the miscibility pressure. The multiple-contact mechanism works nearly the same with a diluentgas added to the CO2 as it does for pure CO2. Frequentlyan application of the CO2 process in the field will toleratehigher miscibility pressures than what pure CO2 wouldrequire. If this is the case, the operator can dilute the CO2

with other available gas, which will raise the miscibilitypressure but also reduce the CO2 requirements.

The pressure at which miscibility is achieved is bestdetermined by conducting a series of displacement exper-iments in a long, slim tube. A plot of oil recovery ver-sus flooding pressure is made, and the minimum misci-bility pressure is determined from the plot. There havebeen several attempts to correlate miscibility pressuresfor CO2–oil systems. An early study showed that thereis a correlation among the API gravity of the crude, thetemperature of the reservoir, and the minimum miscibil-ity pressure. A second study revealed that the molecu-lar weight of the C5+ fraction was a better variable touse than the API gravity. In further investigations, it hasbeen found that the minimum miscibility pressure is the

pressure that will achieve a minimum density in the CO2

phase.

E. Problems in Applying the Miscible Process

Because of differences in density and viscosity betweenthe injected fluid and the reservoir fluid(s), the miscibleprocess often suffers from poor mobility. Viscous finger-ing and gravity override frequently occur. The simultane-ous injection of a miscible agent and a brine was suggestedin order to take advantage of the high microscopic dis-placement efficiency of the miscible process and the highmacroscopic displacement efficiency of a waterflood. Theimprovement was not as good as hoped for since the mis-cible agent and brine tended to separate due to densitydifferences, with the miscible agent flowing along the topof the porous medium and the brine along the bottom.

Several variations of the simultaneous injection schemehave been suggested and researched. They typically in-volve the injection of a miscible agent followed by brineor the alternating of miscible agent–brine injection. Thelatter variation has been named the WAG (water alternategas) process and has become the most popular. A bal-ance between amounts of injected water and gas must beachieved. Too much gas will lead to viscous fingering andgravity override of the gas, whereas too much water couldlead to the trapping of reservoir oil by the water. The addi-tion of foam-generating substances to the brine phase hasbeen suggested as a way to aid in reducing the mobility ofthe gas phase. Research is continuing in this area.

Operational problems involving miscible processes in-clude transportation of the miscible flooding agent, corro-sion of equipment and tubing, and separation and recyclingof the miscible flooding agent.

IV. CHEMICAL FLOODING

A. Introduction

Chemical flooding relies on the addition of one or morechemical compounds to an injected fluid either to reducethe interfacial tension between the reservoir oil and the in-jected fluid or to improve the sweep efficiency of theinjected fluid.

There are three general methods in chemical flood-ing technology. The first is polymer flooding, in whicha large macromolecule is used to increase the displacingfluid viscosity. This leads to improved sweep efficiencyin the reservoir. The second and third methods, micellar–polymer and alkaline flooding, make use of chemicals thatreduce the interfacial tension between an oil and a displac-ing fluid.

P1: GLM Final Pages



B. Polymer Processes

The addition of large-molecular-weight molecules calledpolymers to an injected water can often increase the ef-fectiveness of a conventional waterflood. Polymers areusually added to the water in concentrations ranging from250 to 2000 parts per million (ppm). A polymer solution ismore viscous than a brine without polymer. In a floodingapplication, the increased viscosity will alter the mobilityratio between the injected fluid and the reservoir oil. Theimproved mobility ratio will lead to better vertical andareal sweep efficiencies and thus higher oil recoveries.Polymers have also been used to alter gross permeabilityvariations in some reservoirs. In this application, poly-mers form a gel-like material by cross-linking with otherchemical species. The polymer gel sets up in large perme-ability streaks and diverts the flow of any injected fluid toa different location.

Two general types of polymers have been used. Theseare synthetically produced polyacrylamides and biologi-cally produced polysaccharides. Polyacrylamides are longmolecules with a small effective diameter. Thus, they aresusceptible to mechanical shear. High rates of flow throughvalves will sometimes break the polymer into smaller en-tities and reduce the viscosity of the solution. A reductionin viscosity can also occur as the polymer solution triesto squeeze through the pore openings on the sand face ofthe injection well. A carefully designed injection schemeis necessary. Polyacrylamides are also sensitive to salt.Large salt concentrations (i.e., greater than 1–2 wt %)tend to make the polymer molecules curl up and lose theirviscosity-building effect.

Polysaccharides are less susceptible to both mechanicalshear and salt. Since they are produced biologically, caremust be taken to prevent biological degradation in thereservoir. As a rule, polysaccharides are more expensivethan polyacrylamides.

Polymer flooding has not been successful in high-temperature reservoirs. Neither polymer type has ex-hibited sufficiently long-term stability above 160◦F inmoderate-salinity or heavy-salinity brines.

Polymer flooding has the best application in moderatelyheterogeneous reservoirs and reservoirs containing oilswith viscosities less than 100 centipoise (cP). In the UnitedStates, there has been a significant increase in the num-ber of active polymer projects since 1978. The projectsinvolve reservoirs having widely differing properties, thatis, permeabilities ranging from 20 to 2000 millidarcies(mD), in situ oil viscosities of up to 100 cP, and reservoirtemperatures of up to 200◦F.

Since the use of polymers does not affect the micro-scopic displacement efficiency, the improvement in oil re-covery will be due to an improved sweep efficiency over

what is obtained during a conventional waterflood. Typi-cal oil recoveries from polymer flooding applications arein the range of 1–5% of the initial oil in place. It has beenfound that operators are more likely to have a successfulpolymer flood if they start the process early in the produc-ing life of the reservoir.

C. Micellar–Polymer Processes

The micellar–polymer process uses a surfactant to lowerthe interfacial tension between the injected fluid and thereservoir oil. A surfactant is a surface-active agent thatcontains a hydrophobic (“dislikes” water) part to themolecule and a hydrophilic (“likes” water) part. The sur-factant migrates to the interface between the oil and waterphases and helps make the two phases more miscible. In-terfacial tensions can be reduced from ∼30 dyn/cm, foundin typical waterflooding applications, to 10−4 dyn/cm withthe addition of as little as 0.1–5.0 wt % surfactant to water–oil systems. Soaps and detergents used in the cleaningindustry are surfactants. The same principles involved inwashing soiled linen or greasy hands are used in “washing”residual oil off rock formations. As the interfacial tensionbetween an oil phase and a water phase is reduced, thecapacity of the aqueous phase to displace the trapped oilphase from the pores of the rock matrix increases. Thereduction of interfacial tension results in a shifting of therelative permeability curves such that the oil will flowmore readily at lower oil saturations.

When surfactants are mixed above a critical satura-tion in a water–oil system, the result is a stable mixturecalled a micellar solution. The micellar solution is madeup of structures called microemulsions, which are homo-geneous, transparent, and stable to phase separation. Theycan exist in several shapes, which depend on the concen-trations of surfactant, oil, water, and other constituents.Spherical microemulsions have typical size ranges from10−6 to 10−4 mm. A microemulsion consists of externaland internal phases sandwiched around one or more lay-ers of surfactant molecules. The external phase can be ei-ther aqueous or hydrocarbon in nature, as can the internalphase.

Solutions of microemulsions are known by several othernames in the petroleum literature. These include surfactantsolutions and soluble oils.

Figure 3 can be used to represent the micellar–polymerprocess. A certain volume of the micellar or surfactantsolution fluid A is injected into the reservoir. The sur-factant solution is then followed by a polymer solution,fluid B, to provide a mobility buffer between the surfac-tant solution and a drive water, which is used to push theentire system through the reservoir. The polymer solu-tion is designed to prevent viscous fingering of the drive

P1: GLM Final Pages



water through the surfactant solution as it starts to build upan oil bank ahead of it. As the surfactant solution movesthrough the reservoir, surfactant molecules are retained onthe rock surface due to the process of adsorption. Oftena preflush is injected ahead of the surfactant to precon-dition the reservoir and reduce the loss of surfactants toadsorption. This preflush contains sacrificial agents suchas sodium tripolyphosphate.

There are, in general, two types of micellar–polymerprocesses. The first uses a low-concentration surfactant so-lution (<2.5 wt %) but a large injected volume (up to 50%pore volume). The second involves a high-concentrationsurfactant solution (5–12 wt %) and a small injected vol-ume (5–15% pore volume). Either type of process has thepotential of achieving low interfacial tensions with a widevariety of brine–crude oil systems. Both have been usedin pilot field trials with moderate technical success.

Whether the low-concentration or the high-concentra-tion system is selected, the system is made up of sev-eral components. The multicomponent facet leads to anoptimization problem, since many different combinationscould be chosen. Because of this, a detailed laboratoryscreening procedure is usually undertaken. The screeningprocedure typically involves three types of tests: (1) phasebehavior studies, (2) interfacial tension studies, and (3) oildisplacement studies.

Phase behavior studies are typically conducted in small(up to 100 mL) vials in order to determine what type, ifany, of microemulsion is formed with a given micellar–crude oil system. The salinity of the micellar solution isusually varied around the salt concentration of the fieldbrine where the process will be applied. Besides the mi-croemulsion type, other factors examined could be oil up-take into the microemulsion, ease with which the oil andaqueous phases mix, viscosity of the microemulsion, andphase stability of the microemulsion.

Interfacial tension studies are conducted with variousconcentrations of micellar solution components to de-termine optimal concentration ranges. Measurements areusually made with the spinning drop, pendent drop, orsessile drop techniques.

The oil displacement studies are the final step in thescreening procedure. They are usually conducted in twoor more types of porous media. Often initial screeningexperiments are conducted in unconsolidated sand packsand then in Berea sandstone. The last step in the sequenceis to conduct the oil displacement experiments in actualcored samples of reservoir rock. Frequently, actual coresamples are placed end to end in order to obtain a coreof reasonable length since the individual core samples aretypically only 5–7 in. long.

If the oil recoveries from the oil displacement testswarrant further study of the process, the next step is

usually a small field pilot study involving anywhere from1 to 10 acres.

The micellar–polymer process has been applied in sev-eral pilot projects and one large field-scale project. Theresults have not been very encouraging. The process hasdemonstrated that it can be a technical success, but theeconomics of the process has been either marginal or poorin nearly every application.

D. Alkaline Processes

When an alkaline solution is mixed with certain crude oils,surfactant molecules are formed. When the formation ofsurfactant molecules occurs in situ, the interfacial tensionbetween the brine and oil phases could be reduced. Thereduction of interfacial tension causes the microscopic dis-placement efficiency to increase, which thereby increasesoil recovery.

Alkaline substances that have been used include sodiumhydroxide, sodium orthosilicate, sodium metasilicate,sodium carbonate, ammonia, and ammonium hydroxide.Sodium hydroxide has been the most popular. Sodiumorthosilicate has some advantages in brines with a highdivalent ion content.

There are optimum concentrations of alkaline and saltand optimum pH where the interfacial tension values expe-rience a minimum. Finding these requires a screening pro-cedure similar to the one discussed above for the micellar–polymer process. When the interfacial tension is loweredto a point where the capillary number is greater than 10−5,oil can be mobilized and displaced.

Several mechanisms have been identified that aid oilrecovery in the alkaline process. These include the fol-lowing: lowering of interfacial tension, emulsification ofoil, and wettability changes in the rock formation. All threemechanisms can affect the microscopic displacement ef-ficiency, and emulsification can also affect the macro-scopic displacement efficiency. If a wettability change isdesired, a high (2.0–5.0 wt %) concentration of alkalineshould be used. Otherwise, concentrations of the order of0.5–2.0 wt % of alkaline are used.

The emulsification mechanism has been suggested towork by either of two methods. The first is by formingan emulsion, which becomes mobile and later trappedin downstream pores. The emulsion “blocks” the pores,which thereby diverts flow and increases sweep efficiency.The second mechanism is by again forming an emulsion,which becomes mobile and carries oil droplets that it hasentrained to downstream production sites.

The wettability changes that sometimes occur with theuse of alkaline affect relative permeability characteristics,which in turn affect mobility and sweep efficiencies.

Mobility control is an important consideration in thealkaline process as it is in all EOR processes. Often, it is

P1: GLM Final Pages



necessary to include polymer in the alkaline solution inorder to reduce the tendency of viscous fingering to takeplace.

Not all crude oils are amenable to alkaline flooding. Thesurfactant molecules are formed with the heavier, acidiccomponents of the crude oil. Tests have been designed todetermine the susceptibility of a given crude oil to alkalineflooding. One of these tests involves titrating the oil withpotassium hydroxide (KOH). An acid number is found bydetermining the number of milligrams of KOH required toneutralize 1 g of oil. The higher the acid number, the morereactive the oil will be and the more readily it will formsurfactants. An acid number larger than ∼0.2 mg KOHsuggests a potential for alkaline flooding.

In general, alkaline projects have been inexpensive toconduct. However, recoveries have not been large in thepast field pilots.

E. Problems in Applying Chemical Processes

The main technical problems associated with chemicalprocesses include the following: (1) screening chemicalsto optimize the microscopic displacement efficiency,(2) making contact with the oil in the reservoir, and (3)maintaining good mobility in order to lessen the effects ofviscous fingering. The requirements for screening ofchemicals vary with the type of process. Obviously, as thenumber of components increases, the more complicatedthe screening procedure becomes. The chemicals mustalso be able to tolerate the environment they are placedin. High temperature and salinity may limit the chemicalsthat could be used.

The major problem experienced in the field to date inchemical flooding processes has been the inability to makecontact with residual oil. Laboratory screening proce-dures have developed micellar–polymer systems that havedisplacement efficiencies approaching 100% when sandpacks or uniform consolidated sandstones are used as theporous medium. When the same micellar–polymer sys-tem is applied in an actual reservoir rock sample, however,the efficiencies are usually lowered significantly. This isdue to the heterogeneities in the reservoir samples. Whenthe process is applied to the reservoir, the efficiencies be-come even worse. Research is being conducted on meth-ods to reduce the effect of the rock heterogeneities and toimprove the displacement efficiencies.

Mobility research is also being conducted to improvedisplacement sweep efficiencies. If good mobility is notmaintained, the displacing fluid front will not be effectivein making contact with residual oil.

Operational problems involve treating the water usedto make up the chemical systems, mixing the chemicalsto maintain proper chemical compositions, plugging theformation with particular chemicals such as polymers,

dealing with the consumption of chemicals due to adsorp-tion and mechanical shear and other processing steps, andcreating emulsions in the production facilities.

V. THERMAL FLOODING

A. Introduction

Primary and secondary production from reservoirs con-taining heavy, low-gravity crude oils is usually a smallfraction of the initial oil in place. This is due to the factthat these types of oils are very thick and viscous and as aresult do not migrate readily to producing wells. Figure 8shows a typical relationship between the viscosity of aheavy, viscous crude oil and temperature. As can be seen,for certain crude oils, viscosities decrease by orders ofmagnitude with an increase in temperature of 100–200◦F.This suggests that if the temperature of a crude oil in thereservoir can be raised by 100–200◦F over the normalreservoir temperature, the oil viscosity will be reducedsignificantly and will flow much more easily to a produc-ing well. The temperature of a reservoir can be raised byinjecting a hot fluid or by generating thermal energy insitu by combusting the oil. Hot water or steam can be in-jected as the hot fluid. Three types of processes will bediscussed in this section: steam cycling, steam drive, andin situ combustion. In addition to the lowering of the crudeoil viscosity, there are other mechanisms by which thesethree processes recover oil. These mechanisms will alsobe discussed.

Most of the oil that has been produced by EOR methodsto date has been as a result of thermal processes. There is

FIGURE 8 Typical viscosity–temperature relationships for sev-eral crude oils.

P1: GLM Final Pages



a practical reason for this, as well as several technical rea-sons. In order to produce more than 1–2% of the initial oilin place from a heavy-oil reservoir, operators had to em-ploy thermal methods. As a result, thermal methods wereinvestigated much earlier than either miscible or chemi-cal methods, and the resulting technology was developedmuch more rapidly.

B. Steam Stimulation Processes

The steam stimulation process was discovered by acci-dent in the Mene Grande Tar Sands, Venezuela, in 1959.During a steam injection trial, it was decided to relievethe pressure from the injection well by backflowing thewell. When this was done, a very high oil production ratewas observed. Since this discovery, many fields have beenplaced on steam stimulation.

The steam stimulation process, also known as the steamhuff and puff, steam soak, or cyclic steam injection, be-gins with the injection of 5000–15,000 bbl of high-qualitysteam. This can take a period of days to weeks to accom-plish. The well is then shut in, and the steam is allowed tosoak the area around the injection well. This soak periodis fairly short, usually from 1 to 5 days. The injection wellis then placed on production. The length of the produc-tion period is dictated by the oil production rate but canlast from several months to a year or more. The cycle isrepeated as many times as is economically feasible. Theoil production will decrease with each new cycle.

Mechanisms of oil recovery due to this process include(1) reduction of flow resistance near the well bore by re-ducing the crude oil viscosity and (2) enhancement of thesolution gas drive mechanism by decreasing the gas solu-bility in an oil as temperature increases.

Often, in heavy-oil reservoirs, the steam stimulationprocess is applied to develop injectivity around an injec-tion well. Once injectivity has been established, the steamstimulation process is converted to a continuous steamdrive process.

The oil recoveries obtained from steam stimulation pro-cesses are much smaller than the oil recoveries that couldbe obtained from a steam drive. However, it should beapparent that the steam stimulation process is much lessexpensive to operate. The cyclic steam stimulation processis the most common thermal recovery technique. Recov-eries of additional oil have ranged from 0.21 to 5.0 bbl ofoil per barrel of steam injected.

C. Steam Drive Processes

The steam drive process is much like a conventional wa-terflood. Once a pattern arrangement is established, steamis injected into several injection wells while the oil is pro-duced from other wells. This is different from the steamstimulation process, whereby the oil is produced from the

same well into which the steam is injected. As the steamis injected into the formation, the thermal energy is usedto heat the reservoir oil. Unfortunately, the energy alsoheats the entire environment such as formation rock andwater. Some energy is also lost to the underburden andoverburden. Once the oil viscosity is reduced by the in-creased temperature, the oil can flow more readily to theproducing wells. The steam moves through the reservoirand comes in contact with cold oil, rock, and water. Asthe steam comes in contact with the cold environment, itcondenses and a hot water bank is formed. This hot waterbank acts as a waterflood and pushes additional oil to theproducing wells.

Several mechanisms have been identified that are re-sponsible for the production of oil from a steam drive.These include thermal expansion of the crude oil, viscos-ity reduction of the crude oil, changes in surface forces asthe reservoir temperature increases, and steam distillationof the lighter portions of the crude oil.

Steam applications have been limited to shallow reser-voirs because as the steam is injected it loses heat energyin the well bore. If the well is very deep, all the steam willbe converted to liquid water. Recently, interest has beenshown in downhole steam generation; research to developan economical system is continuing in this area.

Steam drives have been applied in many pilot and field-scale projects with very good success. Oil recoveries haveranged from 0.3 to 0.6 bbl of oil per barrel of steaminjected.

D. In Situ Combustion

Early attempts at in situ combustion involved what isreferred to as the forward dry combustion process. Thecrude oil was ignited downhole, and then a stream of airor oxygen-enriched air was injected in the well wherethe combustion was originated. The flame front was thenpropagated through the reservoir. Large portions of heatenergy were lost to the overburden and underburden withthis process. To reduce the heat losses, researchers deviseda reverse combustion process. In reverse combustion, theoil is ignited as in forward combustion but the airstreamis injected in a different well. The air is then “pushed”through the flame front as the flame front moves in the op-posite direction. Researchers found the process to work inthe laboratory, but when it was tried in the field on a pilotscale, it was never successful. What they found was thatthe flame would be shut off because there was no oxygensupply and that where the oxygen was being injected, theoil would self-ignite. The whole process would then revertto a forward combustion process.

When the reverse combustion process failed, a newtechnique called the forward wet combustion processwas introduced. This process begins as a forward dry

P1: GLM Final Pages



combustion does, but once the flame front is established,the oxygen stream is replaced by water. As the water comesin contact with the hot zone left by the combustion front, itflashes to steam, using energy that otherwise would havebeen wasted. The steam moves through the reservoir andaids the displacement of oil. The wet combustion processhas become the primary method of conducting combustionprojects.

Not all crude oils are amenable to the combustion pro-cess. For the combustion process to function properly, thecrude oil has to have enough heavy components to serve asthe fuel source for the combustion. Usually this requiresan oil of low API gravity. As the heavy components inthe oil are combusted, lighter components as well as fluegases are formed. These gases are produced with the oiland raise the effective API gravity of the produced oil.

The number of in situ combustion projects has de-creased since 1980. Environmental and other operationalproblems have proved to be more than what some opera-tors want to deal with.

E. Problems in Applying Thermal Processes

The main technical problems associated with thermaltechniques are poor sweep efficiencies, loss of heat en-ergy to unproductive zones underground, and poor injec-tivity of steam or air. Poor sweep efficiencies are due tothe density differences between the injected fluids and thereservoir crude oils. The lighter steam or air tends to riseto the top of the formation and bypass large portions ofcrude oil. Data have been reported from field projects inwhich coring operations have revealed significant differ-ences in residual oil saturations in the top and bottom partsof the swept formation. Research is being conducted onmethods of reducing the tendency for the injected fluidsto override the reservoir oil. Techniques involving foamsare being employed.

Large heat losses continue to be associated with thermalprocesses. The wet combustion process has lowered theselosses for the higher-temperature combustion techniques,but the losses are severe enough in many applications toprohibit the combustion process. The losses are not as largewith the steam processes because they involve smallertemperatures. The development of a feasible downholegenerator will significantly reduce the losses associatedwith steam injection processes.

The poor injectivity found in thermal processes islargely a result of the nature of the reservoir crudes. Oper-ators have applied fracture technology in connection withthe injection of fluids in thermal processes. This has helpedin many reservoirs.

Operational problems include the following: the for-mation of emulsions, the corrosion of injection and pro-duction tubing and facilities, and the creation of adverse

effects on the environment. When emulsions are formedwith heavy crude oil, they are very difficult to break. Oper-ators need to be prepared for this. In the high-temperatureenvironments created in the combustion processes andwhen water and stack gases mix in the production wellsand facilities, corrosion becomes a serious problem. Spe-cial well liners are often required. Stack gases also poseenvironmental concerns in both steam and combustion ap-plications. Stack gases are formed when steam is gener-ated by either coal- or oil-fired generators and, of course,during the combustion process as the crude is burned.

VI. MICROBIAL FLOODING

A. Introduction

Microbial enhanced oil recovery (MEOR) flooding in-volves the injection of microorganisms that react withreservoir fluids to assist in the production of residual oil.The U.S. National Institute for Petroleum & Energy Re-search (NIPER) maintains a database of field projects thathave used microbial technology. There has been signifi-cant research conducted on MEOR, but few pilot projectsare currently being conducted. The Oil and Gas Journal’s1998 survey reported just 1 ongoing project in the UnitedStates related to this technology out of 199 projects. How-ever, China reported several pilot projects that indicatedmild success with MEOR.

There are two general types of MEOR processes—thosein which microorganisms react with reservoir fluids to gen-erate surfactants and those in which microorganisms reactwith reservoir fluids to generate polymers. Both processesare discussed below along with a few concluding com-ments regarding the problems in applying them. The suc-cess of MEOR processes will be highly affected by reser-voir characteristics. MEOR systems can be designed forreservoirs that have either a high or a low degree of chan-neling. Therefore, MEOR applications require a thoroughknowledge of the reservoir. Mineral content of the reser-voir brine will also affect the growth of microorganisms.

B. Microbial Processes

Microorganisms can be reacted with reservoir fluids togenerate either surfactants or polymers in the reservoir.Once either the surfactant or the polymer is produced,the processes of mobilizing and recovering residual oilbecome similar to those discussed with regard to chemicalflooding.

Most pilot projects have involved near-well treatmentsof stripper wells in an application of the huff and puffprocess discussed with regard to thermal flooding. A so-lution of microorganisms is injected along with a nutrient,usually molasses. When the solution of microorganismshas been designed to react with the oil to form polymers,

P1: GLM Final Pages



the injected solution will enter high-permeability zonesand react to form the polymers that will then act as apermeability-reducing agent. When oil is produced dur-ing the huff stage, oil from lower permeability zones willbe produced. Conversely, the solution of microorganismscan be designed to react with the residual crude oil to forma surfactant. The surfactant lowers the interfacial tensionof the brine-water system, which thereby mobilizes theresidual oil. The oil is then produced in the huff part ofthe process.

The reaction of the microorganisms with the reservoirfluids may also produce gases, such as CO2, N2, H2, andCH4. The production of these gases will result in an in-crease in reservoir pressure, which will thereby enhancethe reservoir energy.

C. Problems in Applying Microbial Processes

Since microorganisms can be reacted to form either poly-mers or surfactants, a knowledge of the reservoir charac-teristics is critical. If the reservoir is fairly Heterogeneous,then it would be desirable to generate polymers in situthat could be used to divert fluid flow from high-to low-permeability channels. If the reservoir has low injectivity,then using microorganisms that produced polymers couldbe very damaging to the flow of fluids near the well bore.Hence, a thorough knowledge of the reservoir characteris-tics, particularly those immediately around the well bore,is extremely important.

Reservoir brines can inhibit the growth of the microor-ganisms. Therefore, some simple compatibility tests canresult in useful information as to the viability of the pro-cess. These can be simple test-tube experiments in whichreservoir fluids and/or rock are placed in microorganism–nutrient solutions and growth and metabolite productionof the microorganisms are monitored.

MEOR processes have been applied in reservoirs withbrines up to less than 100,000 ppm, rock permeabilitiesgreater than 75 mD, and depths less than 6800 ft. Thisdepth corresponds to a temperature of about 75◦C. MostMEOR projects have been performed with light crude oilshaving API gravities between 30 and 40. These shouldbe considered “rule of thumb” criteria. The most impor-tant consideration in selecting a microorganism–reservoirsystem is to conduct compatibility tests to make sure thatmicroorganism growth can be achieved.

VII. SCREENING CRITERIAFOR EOR PROCESSES

A. Introduction

A large number of variables are associated with a givenoil reservoir, for instance, pressure and temperature, crudeoil type and viscosity, and the nature of the rock matrix

and connate water. Because of these variables, not everytype of EOR process can be applied to every reservoir. Aninitial screening procedure would quickly eliminate someEOR processes from consideration in particular reservoirapplications. This screening procedure involves the analy-sis of both crude oil and reservoir properties. This sectionpresents screening criteria for each of the general types ofprocesses previously discussed in this article except mi-crobial flooding. (A discussion of MEOR screening crite-ria was presented in Section VI). It should be recognizedthat these screening criteria are only guidelines. If a par-ticular reservoir–crude oil application appears to be ona borderline between two different processes, it may benecessary to consider both processes. Once the numberof processes has been reduced to one or two, a detailedeconomic analysis will have to be conducted.

Some general considerations can be discussed beforethe individual process screening criteria are presented.First, detailed geological study is usually desirable, sinceoperators have found that unexpected reservoir hetero-geneities have led to the failure of many EOR fieldprojects. Reservoirs that are found to be highly faultedor fractured typically yield poor recoveries from EORprocesses. Second, some general comments pertainingto economics can be made. When an operator is con-sidering EOR in particular applications, candidate reser-voirs should contain sufficient recoverable oil and be largeenough for the project to be potentially profitable. Also,deep reservoirs could involve large drilling and comple-tion expenses if new wells are to be drilled.

B. Screening Criteria

Table II contains the screening criteria that have been com-piled from the literature for the miscible, chemical, andthermal techniques.

The miscible process requirements are characterizedby a low-viscosity crude oil and a thin reservoir. Alow-viscosity oil will usually contain enough of theintermediate-range components for the multicontactmiscible process to be established. The requirement of athin reservoir reduces the possibility that gravity overridewill occur and yields a more even sweep efficiency.

In general, the chemical processes require reservoirtemperatures of less than 200◦F, a sandstone reservoir,and enough permeability to allow sufficient injectivity.The chemical processes will work on oils that are moreviscous than what the miscible processes require, but theoils cannot be so viscous that adverse mobility ratios areencountered. Limitations are set on temperature and rocktype so that chemical consumption can be controlled toreasonable values. High temperatures will degrade mostof the chemicals that are currently being used in theindustry.

P1: GLM Final Pages



TABLE II Screening Criteria for Enhanced Oil Recovery Processes

Oil Oil Oil Net Averagegravity viscosity saturation Formation thickness permeability Depth Temp

Process (◦API) (cP) (%) type (ft) (mD) (ft) (◦F)

Miscible

Hydrocarbon >35 <10 >30 Sandstone or carbonate 15–25 —a >4500 —a

Carbon dioxide >25 <12 >30 Sandstone or carbonate 15–25 —a >2000 —a

Nitrogen >35 <10 >30 Sandstone or carbonate 15–25 —a >4500 —a

Chemical

Polymer >25 5–125 —b Sandstone preferred —a >20 <9000 <200

Surfactant–polymer >15 20–30 >30 Sandstone preferred >10 >20 <9000 <200

Alkaline 13–35 <200 —b Sandstone preferred —a >20 <9000 <200

Thermal

Steamflooding >10 >20 >40–50 Sand or sandstone >10 >50 500–5000 —a

with high porosity

Combustion 10–40 <1000 >40–50 Sand or sandstone >10 >50 >500 —a

with high porosity

a Not critical but should be compatible.b Ten percent mobile oil above waterflood residual oil.

In applying the thermal methods, it is critical to havea large oil saturation. This is especially pertinent to thesteamflooding process, because much of the produced oilwill be used on the surface as the source of fuel to firethe steam generators. In the combustion process, crude oilis used as fuel to compress the airstream on the surface.The reservoir should be of significant thickness in orderto minimize heat loss to the surroundings.

VIII. SUMMARY

The recovery of nearly two-thirds of all the oil that hasbeen discovered to date is an attractive target for EORprocesses. The application of EOR technology to existingfields could significantly increase the world’s proven re-serves. Several technical improvements will have to bemade, however, before EOR processes are widely im-plemented. The current economic climate will also haveto improve because many of the processes are eithermarginally economical or not economical at all. Steam-flooding and polymer processes are currently economi-cally viable, with produced oil prices per barrel withina range of $20 to $28. In comparison, the CO2 processis more costly, $26–$39. The micellar–polymer processis even more expensive, at $35–$46. (These prices arein 1985 U.S. dollars.) An aggressive research program isneeded to assist in making EOR processes more techno-logically and economically sound.

Enhanced oil recovery technology should be consideredearly in the producing life of a reservoir. Many of theprocesses depend on the establishment of an oil bank inorder for the process to be successful. When oil saturationsare high, the oil bank in easier to form. It is crucial forengineers to understand the potential of EOR and the wayEOR can be applied to a particular reservoir.


FLUID DYNAMICS • MINING ENGINEERING • PETROLEUM

GEOLOGY • PETROLEUM REFINING • POLYMER PROCESS-ING • ROCK MECHANICS

BIBLIOGRAPHY

Bryant, R. S. (1991). “MEOR screening criteria fit 27% of U.S. oilreservoirs,” Oil and Gas J. (Apr. 15).

“EOR oil production up slightly,” (1998). Oil and Gas J. (Apr. 20).Gogerty, W. B. (1983). J. Petrol. Technol. (Sept. and Oct., in two parts).Lake, L. W. (1989). “Enhanced Oil Recovery,” Prentice-Hall, Englewood

Cliffs, N.J.National Petroleum Council (1984). “Enhanced Oil Recovery,” U.S.

Govt. Printing Office, Washington, D.C.Prats, M. (1982). “Thermal Recovery” (Soc. Pet. Eng. Monograph Ser.),

Vol. 7, Soc. Pet. Eng. of AIME, Dallas.Stalkup, F. I. (1983). “Miscible Displacement” (Soc. Pet. Eng. Mono-

graph Ser.), Vol. 8, Soc. Pet. Eng. of AIME, Dallas.van Poolen, H. K., and Associates, Inc. (1980). “Enhanced Oil Recovery,”

PennWell Books, Tulsa, Okla.

P1: GLQ/GLE P2: GLQ Final Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology EN006F-257 June 29, 2001 18:17

Fossil Fuel Power Stations—Coal Utilization

L. Douglas SmootLarry L. BaxterBrigham Young University

I. Coal Use in Power StationsII. Power Station TechnologiesIII. Coal Properties and PreparationIV. Coal Reaction ProcessesV. Coal-Generated Pollutants

VI. Numerical Simulation of Coal Boilersand Furnaces

VII. The Future

GLOSSARY

Ash The oxidized product of inorganic material found incoal.

ASTM American Society for Testing and Materials, oneof several organizations promulgating standards forfuel analysis and classification.

BACT Best available control technology.CAAA Clean Air Act amendments of 1990 in the United

States, which introduced substantial new controls fornitrogen oxides, sulfur-containing compounds, and,potentially, many metals emissions by power plants.

Char Residual solid from coal particle devolatilization.Char oxidation Surface reaction of char with oxygen.Clean Air Act A federal law governing air quality.Coal devolatilization Thermal decomposition of coal to

release volatiles and tars.

Coal rank A systematic method to classify coals basedon measurement of volatile matter, fixed carbon andheating value.

Comprehensive code Computerized numerical modelfor describing flow and combustion processes.

Criteria pollutants Pollutants for which federal lawgoverns emissions; (nitrogen oxides, sulfur-containingcompounds, CO, and particulates are the most signifi-cant criteria pollutants for coal-based power systems).

daf Dry, ash-free, a commonly cited basis for many coalanalyses.

Fouling and slagging Accumulation of residual ash oninner surfaces of power plant boilers, more genericallyreferred to as ash deposition.

Greenhouse gases Gases (e.g., CO2, CH4, N2O, halocar-bons) that contribute to atmospheric warming.

Inorganic material The portion of coal that contributes

Encyclopedia of Physical Science and Technology, Third Edition, Volume 6Copyright C© 2002 by Academic Press. All rights of reproduction in any form reserved. 121

P1: GLQ/GLE P2: GLQ Final


122 Fossil Fuel Power Stations—Coal Utilization

to ash or is not a fundamental portion of the organicmatrix.

ISO International Organization for Standards, an interna-tional association of national/regional standards orga-nizations that publishes standards for fuel characteri-zation and classification.

Mineral matter The portion of the inorganic materialin coal that occurs in the form of identifiable mineralstructures.

NAAQS National Ambient Air Quality Standard.NSPS New source performance standard; regulatory

standards applied to new power stations.PSD Prevention of Significant Deterioration.SCR Selective catalytic reduction of nitrogen oxides

by NH3 over a generally vanadium-based catalyticsurface.

SNCR Selective noncatalytic conversion of nitrogen ox-ides by NH3 via gas-phase reactions.

COAL is the world’s most abundant fossil fuel and isthe energy source most commonly used for generation ofelectrical energy. World use of electrical energy representsabout 35% of the total world consumption of energy andthis share is increasing. Even so, coal presents formidablechallenges to the environment. This article outlines thecommercial technology used for generation of electric-ity and mentions developing technologies. Properties ofcoals and their reaction processes are discussed and ad-vantages and challenges in the use of coal are enumerated.New analysis methods for coal-fired electrical power gen-erators are noted and future directions are indicated.

I. COAL USE IN POWER STATIONS

Most of the world’s coal production of over 5 billion shorttons a year is used to help generate nearly 14 trillionkWhr of electricity, with especially significant use in theUnited States, China, Japan, Russia, Canada, andGermany. Coal generates about 56% of the world’s elec-trical power.

World generation of electricity is growing at a morerapid rate than total world energy consumption. The useof fossil fuels (i.e., coal, natural gas, liquified naturalgas, petroleum) dominates total energy and electric powerproduction, accounting for 85–86% of the total for boththe United States and the world. Petroleum provides thelargest fraction of the U.S. (39%) and the world (37%) en-ergy requirements, while coal is the most commonly usedfuel for electrical power generation in the United States(52% of total kWhr) and the world. Approximately 87%of all coal consumed in the United States is for gener-

ation of electricity. The U.S. production of coal has in-creased by 55% in the past 20 years, during the time thatU.S. production of natural gas and petroleum has heldsteady or declined. During this same period, world coalproduction has increased 39% (with little increase in thepast decade). Proven world recoverable reserves of coalrepresent 72% of all fossil fuel reserves, with petroleumand natural gas each representing about 14% of the to-tal reserve. The world depends increasingly on electricalenergy, while coal represents the most common fuel ofchoice for electric power generation and the world’s mostabundant fossil fuel reserve. With electricity generationfrom nuclear energy currently declining and with hydro-electric power increasing more slowly than electric powerfrom fossil fuels, coal may continue to be the world’s pri-mary energy source for electricity generation. Yet coalis not without substantial technical challenges, which in-clude mining, pollutant emissions, and global warmingconcerns.

II. POWER STATION TECHNOLOGIES

A. Power Station Systemsand Conversion Cycles

Fuel conversion technologies of current commercial sig-nificance include cyclone boilers, pulverized coal boilers,combustion grates, and fluidized beds. Similar technolo-gies, generally operated at less extreme conditions of pres-sure and temperature, provide steam and hot water in re-gions with high demand for district or process heating.The most important advanced technologies include gasi-fiers, high-temperature furnaces and fuel cells. Some of thegeneral characteristics of these technologies are given inTable I, and characteristics of advanced systems are shownin Table II. The advantages and challenges of the varioustechnologies occur in part from the thermodynamic cyclesinvolved in these heat engines.

1. Carnot Cycle

The most efficient means of converting thermal energy intowork or electricity is the Carnot cycle. The Carnot cyclerepresents an idealized heat engine that is unachievablein practical systems but that sets a useful standard forheat engine performance and comparisons. The Carnotefficiency is given by

η = 1 − Tlo

Thi, (1)

where Thi and Tlo are the peak and exhaust temperatures,respectively, of the working fluid (expressed as absolute



Fossil Fuel Power Stations—Coal Utilization 123

TABLE I Figures of Merit for Coal-Based Power Generation Systemsa

Pulverized coal boilers Fluid beds

Total Wall-fired Tangential Cyclones Total Circulating Bubbling Grate-fired

Capital cost ($/kW) 950 950 950 850 1350 1450 1250 1250

Peak furnace temperature (K) 2000 2000 2000 2400 1700 1700 1700 1700

Energy/fuel efficiency (%) 38/100 38/100 38/100 38/100 37/99 38/99 36/98 32/95

Fuel flexibility Moderate Moderate Moderate Excellent Excellent Excellent Excellent Excellent

Average/max size (MWe) 250/1000 250/1000 250/1000 600/1000 30/250 30/250 30/250 10/150

Pollutant emissions Moderate Moderate Moderate High Low Low Low Moderate

Installed capacity (%)

World 82 43 39 2 8 4 5 1

North America 80 43 38 8 1 <1 <1 <1

Western Europe 80 43 37 0 10 4 6 8

Asia 84 43 41 0 8 4 4 8

Australia 88 43 45 0 3 1 2 2

aValues are illustrative of typical trends but vary significantly on case-by-case bases.

temperatures). The strong dependence of efficiency ontemperature is clear from this expression.

2. Rankine Cycle

Nearly all steam-based power boilers use the Rankine cy-cle to generate electricity. A simple Rankine cycle is basedon pressurizing water with a pump, heating the pressur-ized water beyond its boiling point to produce superheatedsteam in the boiler, expanding the steam through a steamturbine, cooling the residual steam to regenerate water, andreturning the water to the pump to start through the cycleagain. Commercial boilers operate at both supercritical(>221.2 bar) and subcritical pressures, with supercriticalboilers exhibiting higher efficiencies and subcritical boil-ers encountering less severe pressure requirements.

The efficiency of the Rankine cycle can be improvedby returning the steam to the boiler after partial steam ex-pansion in a process referred to as reheat. Reheat cyclesadd significant complexity to the turbine, the boiler, andthe controls but, at large scale, the increased complexityand cost can be justified by the increase in efficiency ofa few percent. Other significant factors affecting overall

TABLE II Figures of Merit for Coal-Based Advanced PowerGeneration Systems

High-temperatureGasifiers Fuel cells furnaces

Capital cost ($/kW) 2000 2500 1100

Energy/fuel 50/85 45/100 40/100efficiency (%)

Fuel flexibility Moderate Very poor Poor

Average/max 10/200 <0.2/15 250/1000size (MWe)

efficiency in the Rankine cycle are parasitic losses (soot-blowing, pumps, fans, etc.) and nonidealities in pumps,turbines, etc. In theory, Rankine efficiencies of about 60%are achievable. Actual plant efficiencies for coal-firedsystems range from approximately 40% (on a higher heat-ing value or gross calorific value basis) for large-scalesystems to less than 20% for small-scale systems. Mathe-matical expressions for the efficiency of the Rankine cycleare more cumbersome than those for the Carnot efficiency,but the strong reliance on peak temperature is common toall of them.

3. Brayton (or Joule) Cycle

Essentially all gas turbines are based on the Brayton cy-cle, which is sometimes referred to as a Joule cycle. In thiscycle, fuel and air are pressurized, burned, pass through agas turbine, and exhausted. The exhaust gases are gener-ally used to preheat the fuel or air. Modern gas turbinesoperate at significantly higher temperatures than steamturbines. The turbine blades present the most critical ma-terials issues, and blades are thus cooled by internal circu-lation or transpiration. The Brayton cycle is of little cur-rent commercial significance in coal-fired power plantsbut is a major consideration in the most promising pro-posed schemes for the next generation of coal-driven heatengines. Gas turbine efficiencies can be as high as 60%and gas turbines enjoy a significant turndown ratio withlittle loss in efficiency.

B. Power Station Technologies

The Rankine cycle forms the basis of power generationfrom nearly all current coal-fired power stations. The




Brayton cycle offers efficiency and other advantages andis incorporated into many future coal-based technologies,generally in combination with the Rankine cycle (so-called combined cycle systems). The Brayton cycle is alsothe basis of gas- and oil-fired turbine technologies. Someof the distinguishing characteristics of the major coal con-version technologies and the leading advanced technolo-gies are summarized here.

Most of the energy associated with the efficiency lossescompared with ideal systems is available as heat, but notas work or electricity. Somewhat over half of the totalenergy potentially available from fuel is not converted toelectricity/work because of these losses. In many countries(including the United States) the available heat from theselosses is rarely used for industrial processes or districtheating. Recovering this heat would require collocationof power stations with industrial parks or communitiesand installation of district heating or process heating sys-tems in the communities or collocated industries. Thesechanges are fraught with social and economic challenges.Nevertheless, these energy losses represent major ineffi-ciencies in the overall energy budgets in many countries.

1. Pulverized-Coal Boilers

The most widely implemented coal-based technology forpower generation is the pulverized-coal (pc) boiler. Theseboilers provide 88% of the total coal-based electric ca-pacity in the United States and similar large fractions ofelectric capacity abroad. A typical pulverized coal boileris shown in Fig. 1. Steps in the process are (1) mining ofthe coal from near-surface or deep mines, (2) transportingcrushed coal several centimeters in size to the power sta-

FIGURE 1 Tangentially fired boiler. [Reproduced with permis-sion from Singer, J. G. (ed.), (1981). “Combustion: Fossil FuelSystems,” 3rd Ed., Combustion Engineering, Windsor, CT.

tion, (3) pulverizing the coal to a powder predominantlyless than 100 µm in size, (4) pneumatically transportingthis coal dust with combustion air into a large combustionchamber with steel walls cooled by water flowing in walltubes, (5) ignition and combustion of the coal dust and airto create high temperatures, which heats the pressurizedcooling water to high temperatures, (6) expansion of thehigh-pressure water through large steam turbines to cre-ate electricity, (7) cleaning of the combustion off-gases toremove gaseous pollutants, (8) marketing or disposal ofthe noncombustible ash, and (9) quality control, cooling,and recycling of the water.

Fuel preparation for a pc boiler occurs in a mill, wherecoal is typically reduced to 70% through a 200-mesh(74-µm) screen. Figure 2 shows a particle-size distributionfrom a bowl-mill grinder typically used in power plants.A typical utility specification for particle size distributionof the coal is that 70% of the mass should pass througha 200-mesh screen, which is equivalent to 70% less than74 µm. Sometimes the top size is limited to about 120 µm.The pulverized fuel is pneumatically transported to burnerlevels (commonly one level fed by each mill) with eachlevel made up of a series of burners.

FIGURE 2 Measured pulverized-coal size distribution for a full-scale exhauster-type mill operating at a steady state. [Reproducedwith permission from Beer, J. M., Chomiak, J., Smoot, L. D. (1984).Progress Energy Combustion Sci. 10, 229–272.




Two basic types of burners define the two major designsof pc boilers: wall and corner burners. Wall burners aregenerally configured as a series of annular flows, withthe pulverized coal entering near the center and air, oftenswirled, entering through one or more outer flows. Wallburners are located on either the front or the back wall, orboth. Corner burners comprise a stack of individual ductsin or near the boiler corners that individually contain airor a coal suspended in air. The direction of the air and coalstreams is biased either clockwise or counterclockwise,creating a large vortex that fills the furnace. The burnertilt and yaw (angle made in the horizontal plane) can beadjusted as operational parameters, although in practicethe yaw is generally fixed.

The furnace exit is generally defined as the location atwhich the flue gas enters in-flow heat exchangers. Suchheat exchangers are first encountered either in the form ofplatens hanging in the top of the furnace or in the form oftube bundles over the furnace nose. Beginning at the in-flow heat exchangers, the gas exits the furnace and entersthe convection pass. The highest temperature tube bundlesare called superheaters, which are the last heat exchangersthrough which steam passes before entering the turbine.The steam temperature and pressure entering the turbineare two of the most closely controlled set points in a boiler.The steam temperature range is 500–650◦C, with 540◦Cbeing a common value. Pressures vary from subcriticalto supercritical (>221 bar), with subcritical boilers beingmost common and generally requiring less maintenance,and supercritical boilers having higher efficiencies. Steamturbines in large-scale power plants are split into two orthree sections. Steam exits the high-pressure turbine aftera partial expansion, where it flows through another set ofheat exchangers called reheaters before returning to theintermediate-pressure turbine.

2. Gasifiers

Coal gasification is a process in which coal is convertedto a low-grade gas; it can be regarded in many ways asfuel-rich combustion. Oxygen-blown coal gasifiers oper-ate at overall equivalence ratios (fuel to oxidizer ratio nor-malized by fuel to oxidizer ratio required to fully oxidizefuel) greater than two. Commercial or near-commercialsystems are available that are either air-blown or oxygen-blown. Different designs use suspension firing, slurry fir-ing, or fluid beds. They can be slagging or dry-bottom sys-tems. They operate at either atmospheric or high pressure.The most commercially significant systems for coal areentrained-flow, high-temperature, high-pressure systems.Gasification offers potential environmental and efficiencyadvantages over conventional coal combustion. Efficiencybenefits derive from the potential combustion of the gasi-

fied coal, commonly referred to as producer gas, in a gasturbine. A Brayton cycle for a gas turbine is a more effi-cient heat engine than the Rankine cycle used in a boiler. Ifthe gases are introduced into the turbine at their peak gasi-fication temperatures and if a conventional Rankine cycleis used to burn the residual char and for the low-gradeheat, overall thermal efficiencies as high as 60% (higher-heating-value basis) can be achieved. Gas-phase sulfurforms H2S in fuel-rich systems, while gas-phase nitrogenforms HCN and NH3, all of which are much easier toscrub from combustion gases than are their fully oxidizedcounterparts (SO2 and nitrogen oxides, NOx). Cleaningcoal-derived gases to the standards demanded by gas tur-bines (in particular, residual ash, alkali, and sulfur) is atechnical challenge.

Other commercial systems include cyclone boilers,fluidized-bed boilers, grate-based furnaces, and fuel cells.

C. Fuel Considerations

A boiler converts chemical energy stored in fuels intosteam that is used to produce electricity or for process use.The theoretical and practical limitations of this process de-termine the ultimate technology and systems making upthe overall power plant. Carbon, hydrogen and oxygen areresponsible for the majority of the calorific value of thefuel.

A convenient illustration of the relationships betweenmany fuels is based on the carbon, hydrogen, and oxygencontent and is referred to as a coalification diagram whenrestricted to coal (Fig. 3). Natural gas has a hydrogen-to-carbon ratio of about 3.6 and an oxygen-to-carbon ratio ofnear 0, but is not conveniently illustrated at the scale of thisdiagram. As is seen, major solid fuels fall into consistentregions of the diagram.

The remaining fuel components are typically nitro-gen, sulfur, and inorganic impurities. It is ultimately the

FIGURE 3 Coalification diagram indicating the relationshipsamong a variety of solid fuels in terms of their chemicalcomposition.




impurities in the fuel and process materials that drive en-gineering decisions regarding power station technologiesand operating conditions. The contributions most directlyassociated with carbon, hydrogen, and oxygen such aspeak flame temperature have less influence on perfor-mance than idealized theoretical analyses might suggest.Fuel impurities and limitations of current materials placethe greatest restrictions on boiler design and operation.With low-grade fuels these restrictions can be substantial.

Coal is commonly classified by rank, with the ma-jor classifications being, in order of decreasing rank, an-thracite, bituminous coal, subbituminous coal, and lig-nite. The data in Fig. 3 illustrate how fuel propertiesvary with rank and how they relate to each other. Coalsmost commonly used for steam generation have molecu-lar hydrogen-to-carbon ratios ranging from 0.7 to 0.9 andmolecular oxygen-to-carbon ratios ranging from near 0 toabout 0.2.

While coal rank correlates with oxygen and hydrogencontent, coal rank is determined principally by heatingvalue except for the highest rank coals and anthracites,which are distinguished on the basis of fixed carbon con-tents (according to ASTM standards). Heating value is theamount of energy released from a fuel during combustionand is discussed later in this article.

III. COAL PROPERTIES AND PREPARATION

A. Coal Classification

A common means of characterizing the broad range ofcoals in use is by rank, moisture, inorganic material, andheating value, which ranges from about 9 to 35 MJ/kg(4000 to about 15,000 Btu/lb). Moisture and inorganicmaterial percentages typically range from 1% to 30% andfrom 5% to over 50%, respectively. Table III summarizesseveral properties of various coals. The heating valuesare conveniently cited for scientific purposes on dry, ash-free bases, but for commercial purposes they are typi-cally moist, inorganic-matter-free values. In the UnitedStates, coals are ranked primarily on a heating-value ba-sis up to a value of 33 MJ/kg (14,000 Btu/lb). Coals withheating values higher than 33 MJ/kg are ranked based onvolatile matter/fixed carbon contents. Coals range in rankfrom lignites to low-volatile, bituminous coals and an-thracites, with the bulk of the coals representing either sub-bituminous, low-sulfur coals or high-volatile bituminouscoals.

Figure 3 illustrates the relationship of anthracite, bi-tuminous and subbituminous coals, lignite, fuel oil, andbiomass in terms of the atomic hydrogen-to-carbon (H:C)and oxygen-to-carbon (O:C) ratios. This type of diagram

indicates some chemical structure, combustion, and inor-ganic aspects of coals and other solid fuels. For example,increasing the H:C or O:C ratio implies decreasing aro-maticity of the fuel. Increasing the O:C ratio implies in-creasing hydroxyl, carboxyl, ether, and ketone functionalgroups in the fuel. Both the aromaticity and the oxygen-containing functional groups influence the modes of oc-currence of inorganic material in fuel and inorganic trans-formations during combustion.

B. Inorganic Matter

One type of inorganic material in coal is inherent inor-ganic material that can be volatile at combustion temper-atures and includes some of the alkali and alkaline earthmetals, most notably sodium and potassium. Anthracitestypically lose less than 10% of their mass by pyrolysis. Bi-tuminous and subbituminous coals lose between 5% and65% of their mass by this process. Lignites, peats, andbiomass can lose over 90% of their mass in this first stageof combustion. The large quantities of gases or tars leavingcoals, lignites, and biomass fuels can convectively carryinorganic material out of the fuel, even if the inorganic ma-terial itself is nonvolatile. The combination of high oxygencontent and high organic volatile matter in subbituminouscoals indicates a potential for creating large amounts ofinorganic vapors during combustion.

The second class of inorganic material in solid fuelsincludes material that is added to the fuel from extrane-ous sources. Geologic processes and mining techniquescontribute much of this material to coal. This adventitiousmaterial is often particulate in nature (in contrast to theatomically dispersed material) and is the dominant con-tributor to fly ash particles larger than about 10 µm. Ex-amples include silica, pyrite, calcite, kaolinite, illite, andother silicates.

Typically, a boiler designed for combustion of subbi-tuminous coals is about 40% taller and 20% wider anddeeper than a similarly rated boiler designed for bitumi-nous coal. This increase in size is required even thoughthe low-rank fuel combusts notably faster than the high-rank fuel. The primary reason for the increase in size isto accommodate differences in the behavior of the inor-ganic material associated with these two coals, specificallythe properties of the ash deposits that they form. The in-creased surface area in the boiler for the low-rank fuelallows the operator to manage this ash behavior withoutcompromising boiler lifetime or damaging boiler compo-nents. In the past, boilers largely burned coals supplied bylocal or regional mines under long-term contracts. A com-bination of environmental pressures and deregulation hascreated much larger variation in the coals used by mostcurrent boilers and has led to widespread fuel blending.




TABLE III Major Fuel Properties as Determined by Standardized Analyses for Typical Samples of EachMajor Coal Rank

Beulah Black Thunder Pittsburgh #8 Pocahontas #3

Nominal rank Lignite Subbituminous Bituminous Low Volatile Bituminous

Proximate analysis

Moisture (as received) 26.89 21.3 1.02 0.62

Ash (dry basis) 13.86 6.46 10.68 4.51

Volatile matter (dry basis) 42.78 54.26 40.16 18.49

Fixed carbon (dry basis) 43.36 39.28 49.16 77

Ultimate analysis (daf basis)

Carbon 70.49 74.73 80.06 91.65

Hydrogen 4.75 5.4 5.63 4.46

Nitrogen 1.22 1 1.39 1.31

Sulfur 2.14 0.51 5.35 0.79

Oxygen (by difference) 21.36 18.27 7.57 1.62

Chlorine 0.05 0.08 0 0.17

Ash chemistry/elemental ash (% of ash)

SiO2 21.23 30.67 41.7 37.92

Al2O3 13.97 16.46 20.66 24.16

TiO2 0.42 1.29 0.9 1.14

Fe2O3 12.25 5.1 29.24 17.14

CaO 16.36 21.32 2.08 7.67

MgO 4.46 4.8 0.79 2.4

K2O 0.22 0.35 1.74 1.84

Na2O 6.5 1.43 0.4 0.83

SO3 24.6 17.67 2.35 6.79

P2O5 0 0.92 0.15 0.1

Heating value (MJ/kg, daf basis) 23.4 29.8 32.2 35.0

Form of sulfur (% of daf fuel)

Sulfatic 0.16 0 0.15 0

Pyritic 0.47 0.08 1.81 0.21

Organic 1.18 0.4 2.82 0.54

Ash fusion temperature (◦C)

Reducing atmosphere

Initial deformation 774 1161 1047 1191

Hemispherical 1636 1178 1082 1249

Spherical 1650 1189 1179 1278

Fluid 1659 1210 1222 1331

Oxidizing atmosphere

Initial deformation 1714 1179 1337 1297

Hemispherical 1753 1200 1372 1313

Spherical 1756 1209 1381 1337

Fluid 1769 1243 1389 1364

In the past, long-term experience with a limited numberof coals partially compensated for an inability to quantita-tively anticipate the behavior of the inorganic material inthese coals. Current fuel purchasing trends require muchmore adept boiler operation.

In general, ash behavior depends on fuel properties,boiler design, and boiler operation. Industrial practice hasled to the development of a wide variety of indices, nearly

all of which are based on fuel properties, for predict-ing ash behavior. These include arithmetic combinationsof elemental mass fractions in the ash, fusion tempera-tures, and slag viscosity-related measurements. It is onthe basis of these parameters that the coal-fired electri-cal power industry has developed a large, stable, reliablesuite of boilers. However, there is a critical need for im-proved accuracy, given the combination of the effect of ash




on boiler performance and current market trends leadingto increased variation and generally decreased quality ofcoals used in most boilers.

C. Standardized Analysis Methods

A large number of coal analyses have been standardized byASTM, ISO, and similar organizations around the world.The most commonly cited and generally useful amongthese include energy-content, proximate, ultimate, ash-chemistry, and ash-fusion-temperature analyses. The spe-cific procedures by which these analyses are performeddiffer slightly from organization to organization, but gen-erally they are more similar than they are different. Thefollowing briefly discusses the general objectives of theanalyses, some typical results, and a few easily misinter-pretable features of these analyses.

1. Coal Energy Content and Flame Temperatures

Energy content represents possibly the single most rel-evant property of a fuel. The standard measure of afuel’s energy content is the heating value or calorificvalue, which typically ranges from 23 to 35 MJ/kg(10,000–15,000 Btu/lb) on a dry, ash-free basis. Sincesome low-rank coals contain large quantities of bothash and moisture, as-received values may be as low as13 MJ/kg (6000 Btu/lb) or sometimes lower in excep-tional cases. The practical importance of this parameterfor combustion technology is obvious. The heating valueis more commonly referred to as calorific value in coun-tries that use SI units in commerce. Two different heatingvalues/calorific values are commonly cited. The higherheating value (gross calorific value) assumes that waterforms a liquid product after combustion, whereas the lowerheating value (net calorific value) assumes that steam va-por is the final form of the water, the difference betweenthe two being the heat of vaporization of the water formedduring combustion and contained in the fuel. Most com-putations in the United States and Australia are basedon higher heating value or gross calorific value, whereasmost of the rest of the world bases calculations on netcalorific value. Performance statistics such as efficienciesare consequently not directly comparable among variouscountries.

All standardized heating value analyses measure theheat released from the fuel when burned with enough airto fully oxidize the fuel (carbon forms carbon dioxide,hydrogen forms water, etc.), normalized by the mass ofthe fuel, not the mass of all combustion products. Thisvalue determines the amount of fuel required to releasea given amount of heat during combustion. It does notdirectly indicate the peak temperatures of the resultingflames. These temperatures play an important role in some

FIGURE 4 Variation of heating value and adiabatic flame tem-perature with fuel type. All values are normalized by those forbituminous coal.

theoretical analyses of boiler performance such as the heatengine cycles discussed previously.

A useful figure of merit for most fuels is the adiabaticflame temperature. The adiabatic flame temperature de-fines the temperature of the products of combustion af-ter all chemical reactions have reached equilibrium andwhen no heat is allowed to escape (or enter) the combus-tor. Each fuel has a unique adiabatic flame temperature fora given amount of air. As the ratio of fuel to air is varied,the adiabatic flame temperature varies. The highest valuegenerally occurs at an air-to-fuel ratio with slightly lessair than is required to convert all of the carbon of CO2 andhydrogen to H2O.

The adiabatic flame temperature does not correlate wellwith heating value over a broad range of fuels. Figure 4illustrates both heating values and adiabatic flame temper-atures for a variety of fuels. All values are normalized tothose for a high-volatile bituminous coal and all calcula-tions assume dry fuels. The lack of correlation betweenheating value and flame temperature is clear. A compari-son of hydrogen and carbon monoxide vividly illustratesthis point. The heating values for these two fuels differ bya factor of about 14, whereas the adiabatic flame temper-atures differ by less than 6%, with that of CO exceedingthat of hydrogen. The wide disparity in heating values butsimilar adiabatic flame temperatures arises primarily fromthe varying oxygen content of the fuels. Fuels that containsignificant oxygen require less air to completely burn, sothe heat release is distributed within less mass. The neteffect of less heat generation and less mass to absorb thatheat is a flame temperature that does not significantly varyfrom fuel to fuel, presuming dry fuels.

Despite the dominant role that flame temperature playsin theoretical analysis of cycle efficiencies, peak flame




temperature has little role in practical systems. The effi-ciency of virtually all combustion systems is limited ul-timately by materials, not flame temperatures. Most dryfuels have peak adiabatic flame temperatures that slightlyexceed 2000◦C (assuming air as an oxidizer). However,no commercially viable boiler materials can withstandsuch temperatures as either structural members or pres-sure parts. Therefore, the peak temperatures achieved ingas and steam turbines are determined by materials con-straints, not fuel constraints. Furthermore, simple theo-retical cycles such as the Carnot cycle assume ideal heattransfer and fluid flow that are not achievable in practice,further limiting the efficiency of real systems. A conse-quence of these limitations is that most coal combustionsystems convert less than 40% of the energy in the fuel(as measured by gross calorific value or higher heatingvalue) to electricity. The departure from ideal behaviorarises, in order of significance, from a lack of suitablehigh-temperature materials, nonidealities in fluid flow andheat transfer, and process-related losses of energy for aux-iliary equipment. Steam turbines typically operate withpeak steam temperatures of 500–600◦C (900–1100◦F).Gas turbines typically operate with peak temperatures ashigh as 1425◦C (2600◦F), with the high-temperature limitdepending on sophisticated blade-cooling systems. Thereare many examples of lower performance systems, espe-cially in small systems and those that are optimized forsteam rather than power production.

2. Proximate and Ultimate Analyses

The proximate analysis indicates the moisture, ash, andvolatiles content of a fuel. The volatile yield is sensitive toheating rate and peak temperature. The yield experiencedduring the relatively slow and cool proximate analysis islower than that experienced by coal in many combustorsby about 30%. However, the relative volatile yields ofmany coals are approximately indicated by the proximateanalysis.

The ultimate analysis apportions the composition of theorganic portion of the fuel among the elements C, H, N,and S. The remaining mass is generally assumed to be oxy-gen. Some details of this analysis are commonly misun-derstood with respect to sulfur, chorine, and ash. Ash typi-cally includes some sulfate. The sulfur content reported inthe ultimate analysis is most commonly based on a sampleheated to 1350◦C in an oxidizing environment, well abovethe proximate ashing temperature of about 750–815◦C.Often, 20% or more of the ash from western coals is in theform of sulfates or other forms represented as SO3. Whenproximate analysis ash is included with the ultimate anal-ysis, the sum of ash, carbon, hydrogen, nitrogen, sulfur,and oxygen will exceed 100%. The excess arises from aportion of the sulfur being counted twice, once in the ash

and once in the ultimate analysis for sulfur. Since oxygenis normally determined by difference, this confusion po-tentially biases both the sulfur and oxygen values. To avoidsuch confusion, ash should be included with the proximateanalysis and the ultimate analysis should be reported on anash-free basis. A significant fraction of the sulfur in high-rank U.S. coals derives from pyrite; an accurate estimateof the amount of pyrite in the fuel can be obtained fromrelatively simple, standardized forms of sulfur analyses.

Chlorine analyses generally are performed separatelyfrom ultimate analyses. Essentially all of the chlorine incoals is released during the ultimate analysis process. Cor-rections to the oxygen content should be made to accountfor the presence of chlorine in the evolved gases if oxy-gen is determined by difference. U.S. coals typically haveless than 0.15% chlorine, with 1% chlorine being an ex-treme upper bound. Oxygen content, on the other hand,ranges from 4% to 20% depending strongly on rank. Evensmall amounts of chlorine can have large impacts on in-organic transformation and deposition. Chlorine is muchmore significant in some international coals and in manyfuels fired in conjunction with coal.

3. Ash Chemistry and Ash Fusion Temperatures

The most uniquely suited standardized analyses for ashdeposition include ash chemistry and ash fusion temper-ature. The total ash content from proximate analysis andash composition provide the fuel information that goesinto the majority of common empirical indices of ash be-havior, along with ash fusion temperature. The ash chem-istry analysis typically reports the ash elemental compo-sition on an oxide basis. This does not mean that all of thespecies exist as oxides in the fuel (which they do not). It is aconvenient method of checking the consistency of the data.The sum of the oxides should be about the same as the totalash content. The analysis is fundamentally an elementalanalysis with no distinction of the chemical speciation ofthe inorganic species.

D. Advanced Analysis Methods

A variety of advanced techniques exists for determin-ing the species composition of the fuel, which is vitalto improved predictive methods for coal, and especiallyash behavior, in boilers. X-ray diffraction, performed onlow-temperature ash, is the most widely used techniquefor qualitatively identifying the presence of minerals intheir crystalline form in concentrations of a few weightpercent or greater. Thermal analytical techniques suchas differential thermal analysis (DTA) and thermogravi-metric analysis (TGA) have been used as a signatureanalysis based on changes in physical properties withtemperature.




Microanalytical techniques, including scanning elec-tron microscopy (SEM) equipped with energy dispersivemicroscopy (TEM), resolve particles to submicrometerscales. The techniques may be applied to low-temperatureash as well as raw coal and provide visual images of themorphology of the inorganic structure. Extended X-rayanalysis (XANE) is another signature analytical techniquesometimes applied to coal and petroleum coke. Mossbauerspectroscopy has been widely used for characterizingthe inorganic forms of iron in coal as well as in slagsand deposits. Computer-controlled scanning electron mi-croscopy (CCSEM) is a commonly used technique thatmeasures the chemical composition and size of individ-ual inorganic grains in coal particles. It forms the basis ofseveral models of ash deposition during coal combustion.Chemical fractionation is a nonstandardized but relativelyaccessible technique for characterization of the modes ofoccurrence of both mineral and nonmineral inorganic ma-terial in coal. These analyses, particularly the last two, arebecoming increasingly important for predicting ash be-havior in boilers. Detailed descriptions of inorganic ma-terial in coal can be inferred from chemical fractionation,as illustrated in Fig. 5.

IV. COAL REACTION PROCESSES

A. Reaction Types

Coal particle reactions are an essential aspect of allcoal combustion processes. These reactions have a di-rect impact on the formation of fine particles, nitrogen-

FIGURE 5 Major forms of inorganic material in a suite of U.S. coals, arranged in rank order and by mineral class.Species are presented from bottom to top in the order indicated in the legend.

TABLE IV Typical Process Characteristics for Various Coal-Related Combustion Technologiesa

Flame TypicalParticle temperature residence

Process size (µm) (K) time (sec)

Entrained flow 1–100 1900–2000 <1

Fluidized bed 1500–6000 1000–1450 10–500

Stoker/fixed bed 10,000–50,000 <2000 500–5000

aAdapted from Table 3.1, in Smoot, L. Douglas, and Smith, Philip J.(1985). “Coal Combustion and Gasification,” Plenum Press, New York.

and sulfur-containing species, and other pollutants. Theyalso dominate heat transfer and fluid dynamic processes.Table IV summarizes typical residence times, flame tem-peratures, and particle sizes for three generic processtypes: entrained flow, fluidized bed, and fixed bed.

A schematic diagram of a reacting coal particle is givenin Fig. 6. The diagram suggests that the particle, at anytime in the reaction process, may be composed of mois-ture, (raw) coal, char, inorganic matter, and ash. Ash rep-resents the condensed-phase product of inorganic mattertransformations in a manner similar to char represent-ing the condensed-phase product of coal pyrolysis. Theparticle may be reacting in either oxidizing or reducinggases, depending on boiler design and operation and thelocation of the particle. Commonly, the center of the par-ticle is in a reducing environment while the surface isoxidizing. Reactions of the organic portion of the coal pro-ceed by two primary processes: devolatilization and charoxidation.




FIGURE 6 Schematic diagram of coal particle reactions duringcombustion.

1. Coal Devolatilization

Coal devolatilization or pyrolysis dominates the earlystages of combustion. This process involves the homoge-neous, solid-phase thermal decomposition of coal chemi-cal structure, producing gaseous and solid products. Thegaseous products include light gases (CO, CO2, CH4, H2O,H2, etc.) and heavy hydrocarbon gases that condense atroom temperature. The latter are called tars. The general,but not universal, convention is to call this process de-volatilization when it occurs under unspecified conditionsor combinations of reducing and oxidizing conditions.When it occurs under neutral or reducing conditions, itis sometimes called pyrolysis. Devolatilization generatesthe gases that form visible flames in wood (and coal) fires.

This part of the reaction cycle occurs as the raw coal isheated and thermally decomposes. The particle may softenand undergo internal transformation. Moisture present inthe coal evolves early as the temperature rises. As thetemperature continues to increase, gases and heavy tarrysubstances are emitted as the coal chemical structure ther-mally decomposes. The extent of devolatilization can varyfrom a few percent up to 70–80% of the total particleweight and can take place in a few milliseconds to sev-eral minutes or longer, depending on coal size and typeand temperature. Devolatilization typically occurs in a fewmilliseconds and accounts for about 60% of the organicmass loss in pulverized coal processes. The residual mass,enriched in carbon and inorganic material and depleted inoxygen and hydrogen, is called char. The char particleoften has many cracks or holes made by escaping gases,may have swelled to a larger size than the original coalparticle, and can have high porosity. The nature of thechar is dependent on the original coal type and size andthe conditions of devolatilization.

Five principal devolatilization phases have been iden-tified. The first phase is moisture vaporization near theboiling point of water. The second phase (627–675 K)

FIGURE 7 Comparison of predicted and measured tar and totalvolatiles yields for a wide range of coals. Carbon content is used toillustrate coal rank. Predictions were made with a coal devolatiliza-tion network model. Solid and dashed lines represent correlationof the measured total volatiles and tar yields, respectively. [Repro-duced with permission from Fletcher, T. H., Solum, M. S., Grant,D. M., Pugmire, R. J. (1992). “The Chemical Structure of Chars inTransition from Devolatilization to Combustion,” Energy and Fuels6, 643–650.]

is associated with the initial evolution of carbon diox-ide and carbon monoxide and a small amount of tar.The third phase (800–1000 K) is the evolution of chemi-cally formed water and carbon oxides. The fourth phase(1000–1200 K) involves the final evolution of tar, hy-drogen, and hydrocarbon gases. The fifth phase is theformation of carbon oxides at high temperatures. Struc-tural properties of the coal strongly affect devolatilizationmechanisms and rates. Yield of volatiles depends on ex-ternal factors such as peak temperature, heating rate, pres-sure, and particle size as well as coal type and chemicalstructure. Proportions of gases and tars vary widely, asshown in Fig. 7. Rates and amounts of weight loss duringdevolatilization differ among coal types, with much moreliquid and tar products for the bituminous coals and moregaseous products (including H2O) from lower rank coals.

Weight loss is not significantly size dependent for parti-cles up to about 400 µm in diameter. However, very largeparticles behave differently from finely pulverized coal.Larger particles heat less rapidly and less uniformly, sothat a single temperature cannot be used to characterizethe entire particle. The internal char surface provides siteswhere secondary reactions occur. Devolatilization prod-ucts generated near the center of a particle must migrateto the outside to escape. During this migration, they maycrack, condense, or polymerize with some carbon deposi-tion and smaller yield of volatiles. Devolatilization ratesand yields consequently decrease with increasing particlesize.

2. Char Oxidation

Char oxidation is an inherently heterogeneous, gas–solidreaction with the gas composition dominating the rates




and yields. When this reaction occurs with oxygen it isgenerally called oxidation. When it occurs with CO2,H2O, or other gases it can be called gasification. Gasi-fication is sometimes used to describe the combinationof devolatilization/pyrolysis reactions and heterogeneousreactions with the char under overall reducing conditions.

The residual char particles can be oxidized or burnedaway by direct contact with oxygen at sufficiently hightemperature. This reaction of the char and oxygen is het-erogeneous, with gaseous oxygen diffusing toward andinto the particle, adsorbing and reacting on the particlesurface, and desorbing as either CO or CO2. For smallparticles at high temperature, CO is the dominant sur-face product. This heterogeneous process is often muchslower than the devolatilization process, requiring sec-onds for small particles to several minutes or more forlarger particles. These rates vary with coal type, tempera-ture, pressure, char characteristics (e.g., size, surface area,porosity), presence of inorganic impurities, and oxidizerconcentration. At atmospheric pressure, gasification reac-tions of char with CO2 and H2O are much slower thanoxidation.

These two processes (i.e., devolatilization and char ox-idation) may take place simultaneously, especially at veryhigh heating rates. If devolatilization takes place in an ox-idizing environment (e.g., air), then the fuel-rich gaseousand tar products react further in the gas phase to producehigh temperatures in the vicinity of the coal particles.

B. Rate Processes and Equations

1. Coal Devolatilization

The earliest coal devolatilization models were simplesingle-step, single-equation expressions in which coal wasconceived to form gases and char with a specified yield,as follows:

dv

dt= k(v∞ − v), (2)

k = A exp(−E/RT ), (3)

where v is the mass of volatiles emitted, t is time, k isthe reaction rate coefficient, v∞ is the ultimate volatilesyield, T is temperature, and A and E are respectivelythe preexponential factor and the activation energy thatdescribe the reaction rate coefficient.

This single-step process lacks the flexibility required todescribe much of the experimental data and nonisothermaleffects. More sophisticated but still essentially empiricalmodels describe devolatilization with two parallel, first-order reactions, by a first-order reaction with a statisticaldistribution of activation energy, and by various combina-

FIGURE 8 A schematic illustration of the major steps during coaldevolatilization (pyrolysis).

tions of parallel and sequential first-order reactions. Themost successful of these empirical models describe the de-pendence of volatile yields on temperature, heating rate,and pressure with some amount of differentiation of thespecies produced by devolatilization (tars, light gases, andchars).

The organic properties and structural characteristics ofcoal have a large impact on its devolatilization behavior.Fundamental devolatilization models base predictions onmeasurable organic properties of the coal. For example,devolatilization of higher rank, melting and swelling coalshas been described by the following two-step process, asillustrated in Fig. 8. In the first step, coal undergoes areduction of hydrogen bonding, and the macromolecularnetwork softens to partially form a metaplast containingliquid coal components. Tars are evaporated from the gen-erated metaplast, and the mobile phase is released duringthis stage. Not all of the macromolecular structure is de-polymerized into metaplast. Low-rank coals produce verylittle metaplast. In the second step, further bond break-ing leads to the evolution of tars and gases and repoly-merization of coal fragments from char. Competition be-tween bond scission and cross-linking reactions controlsthe rate of tar evolution and the properties of char. Gasformation is correlated with the composition of the coal.Large amounts of char and primary gases are generateddirectly from the coal macromolecular phase, especiallyfor low-rank coals. Tar formation is viewed as a combineddepolymerization and vaporization process. The structuralevolution of char during pyrolysis suggests that functionalgroups are released from the aromatic clusters, with cross-linking occurring among the aromatic clusters. At the endof devolatilization, the carbon skeletal structures of charsfrom a variety of coals become very similar chemically.However, the physical structure of chars varies signifi-cantly depending on the type and rank of the coal fromwhich they are derived.




Coal network models have more recently been devel-oped to relate the structures of coal to devolatilizationprocesses and rate measurements. Model parameters havebeen obtained from several optical or spectral measure-ments of coals and from rate measurements. Initially,physical melting occurs due to weakening of hydrogenbonds in the network, and the mobile phase is releasedas early tar. Further scission of covalent bonds in the net-work generates fragments corresponding to the main evo-lution of aromatic tars. Depending on coal rank, oxygenstructures in coal such as carboxyl and hydroxyl groupsthermally decompose early in the devolatilization process,generating radical groups that rapidly cross-link the coalnetwork structure. Tar precursors are then not allowed toform and evolve. Thus, early cross-linking is empiricallycorrelated with the evolution of CO2.

Coal structure is typically modeled with four basicgroups: aromatic nuclei, labile bridges, char or stablebridges, and peripheral groups. These structural parame-ters have been determined from many advanced coal char-acterization techniques. Competition occurs in the evolv-ing coal structure between the cleavage of labile bonds andthe formation of stable char links. With increasing tem-perature, the labile bonds are rapidly consumed and thedecomposition of methyl and methoxy groups forms rad-icals that rapidly cross-link the structure into char. Withfurther increases in temperature, methane and hydrogenare evolved from the char structure. It has been shownthat two moieties, aryl–alkyl ethers and hydroaromaticstructures, can be used to correlate weight loss and taryield.

Network models use lattice statistics to quantitativelydescribe the thermal breakup of the coal macromolecu-lar structure and interpret the interrelationships in the lat-tice characteristics of coal, tar, and char. The capabilityof one network model to predict total volatiles and charrelease for a large number of coals of various rank withoutthe use of any fitting parameters was illustrated in Fig. 7.Char is formed from charring reactions within the networkand from cross-linking of nonvolatile metaplast.

2. Coal Volatiles Ignition and Combustion

Many products are produced during devolatilization ofcoal, including tars and hydrocarbon liquids, hydrocarbongases, CO2, CO, H2, H2O, and HCN. These products reactwith oxygen in the vicinity of the char particles, increasingtemperature and depleting oxygen. This complex reactionprocess is important to control soot formation, flame sta-bility, and char ignition. Rate processes include the follow-ing: volatiles release, tar condensation and repolymeriza-tion in char pores, hydrocarbon cloud evolution throughsmall pores from the moving particle, cracking of the hy-

drocarbons to smaller hydrocarbon fragments with localproduction of soot, condensation of gaseous hydrocarbonsand agglomeration of sooty particles, macromixing of thedevolatilizing coal particles with oxygen, micromixing ofthe volatiles cloud and oxygen, oxidation of the gaseousspecies to combustion products, production of nitrogenoxides and sulfur oxides by reaction of devolatilized prod-ucts with oxygen, and heat transfer from the reacting fluidsto the char particles.

At high combustion temperatures, up to 70% or more ofthe reactive coal mass can be consumed through this pro-cess. Volatiles combustion in practical systems is com-plicated by turbulent mixing of fuel and oxidizer, sootformation and radiation, and near-burner fluid dynamics.Such systems usually exhibit volatiles cloud combustion,rather than single-particle combustion.

3. Char Oxidation

The rate of char oxidation is determined by a combina-tion of chemical kinetics and mass transport. Heteroge-neous char oxidation can proceed simultaneously with orafter vaporization and devolatilization, depending on re-action conditions. The time required for char and cokecombustion is substantially greater than that required fordevolatilization. There is also much emphasis on reduc-ing the amount of carbon in the ash, which means thatthe burnout of the char or coke needs to be more ef-fective. The physical structure of the char, such as porestructure, surface area, particle size, and inorganic con-tent, controls the reaction processes of the char. However,the structure of char is determined by the devolatilizationprocess as well as by the parent fuel. Models of char com-bustion can include the following processes: (1) diffusionof gaseous reactants and products through the boundarylayer surrounding the particle, (2) diffusion of reactantswithin the porous structure of the particle, (3) conductionof heat through the particle, (4) adsorption of oxygen,(5) chemical reaction of the oxidizer on the particle sur-face, (6) desorption of products (e.g., CO) from the solidsurface, (7) the homogeneous reaction of carbon monox-ide with oxidizer within the pores and in the boundarylayer, and (8) the evolution of particle and pores during thereactions.

Global models of heterogeneous reaction typically in-clude only processes 1 and 5 described above as resis-tances in the reaction rate expression. The resistance dueto film diffusion is usually calculated through the useof diffusional coefficients corrected for the mass leav-ing the particle surface (i.e., blowing factor). The effec-tive diffusivity of the gases into the particle and the por-tion of the particle that participates in the reaction canbe estimated by using methods that have been published.




The char reacts with oxygen to form CO primarily, andCO2, depending on conditions. The measured oxida-tion rate is correlated through reaction rate expressionssuch as

r1 = ηξkx T mp pn

O2,s = ηξkx Rng T m−n

p CnO2

, (4)

where Tp is the particle temperature, kx is the kineticrate coefficient, which varies exponentially with surfacetemperature, and pO2,s is the oxygen partial pressure atthe solid surface. The parameter ξ is the ratio of activesurface area to external surface area and can be used toincorporate intrinsic kinetics in the model. The parame-ter η is the fraction of the surface area that is availablefor reaction. As the ash fraction of the char particle in-creases, rates of reaction of the entire particle based ontotal surface area generally decrease. The particle diame-ter changes as the particle burns and is typically modeledas a shrinking sphere with constant density, a constant-diameter sphere with changing porosity, or some inter-mediate relationship. Some chars fragment during com-bustion, a process most pronounced in high-rank, initiallylarge coal particles. Since the oxygen partial pressure isnot known at the particle surface, the molecular diffusionequation is used in conjunction with the oxidation equa-tion, Eq. (4), to solve for reaction rate, assuming quasi-steady state as the char particle reduces in size and/ormass. Other complications during char oxidation includepossible changes in reactivity during reaction becauseof char annealing and graphitization, effects of impuri-ties on surface reaction rates, and the ratio of CO/CO2

formed on the char surface. More recent char oxidationmodels include oxygen adsorption and product desorptionsteps through the classic Langmuir expression, which dif-fers in form substantially from the empirical expressionabove.

Descriptions of pore development and structure requiremicroscopic models of the particle. These models in-clude intrinsic kinetics and pore structural changes duringburnoff. Three of the most popular microscopic modelsare a random capillary pore model, one in which the poresare considered spherical vesicles connected by cylindri-cal micropores, and one in which the pores have a treelikestructure. These models allow for pore growth and coales-cence in their respective fashions and provide estimates ofreactive surface area. Parameters required for these mod-els are obtained from experimental measurements of thevarious chars.

Other microscopic models have been developed thatare based on percolation or other char transformation the-ories. These microscopic models currently serve more asscientific descriptors than practical tools.

FIGURE 9 Schematic illustration of the fate of inorganic materialin coals during combustion.

C. Fate of Inorganic Material

1. Ash Formation

Figure 9 schematically illustrates the fate of inorganic ma-terial in coal leading to the production of fly ash. Inorganicgrains may be imbedded in the particle or may be extra-neous to the particle itself. The fate of this second class ofinorganic material differs substantially from that of the in-herent material. The minerals undergo chemical reactionsand phase changes determined by their thermochemistryas well as interacting with other inorganic components andthe organic material. Components of the minerals may bereleased from the fuel by either thermal decomposition orvaporization during combustion.

The transformations are divided into two types: releasemechanisms and the fate of the residual ash. Release mech-anisms are indicated as vaporization, thermal or chemicaldisintegration of the inorganic material (inorganic reac-tion), or convection during rapid devolatilization or otherorganic reactions. These mechanisms tend to producesmall (<0.8 µm) particles or vapors. The residual ash mayundergo fragmentation either as an inorganic grain or inconjunction with fragmentation of burning char particles,may coalesce with some or all of the remaining inorganicmaterial, and may undergo significant chemical or physi-cal transformations. This material tends to produce largerash particles. Depending on the type of inorganic materialand the combustion conditions, the ash produced duringcombustion is composed of varying amounts of vapor,fume (<1-µm-diameter particles), and larger particles.

Many inorganic materials occur as hydrates in coal,all of which dehydrate through endothermic reactions,generally at temperatures from 100◦C to 200◦C. Sul-fates typically thermally decompose as they are heated,




as do carbonates and sulfides. In cases such as mostof the alkaline-earth-containing materials, these salts de-compose prior to melting and appreciable vaporization,while in cases such as most of the alkali-containing ma-terial, the salts melt, partially vaporize, and react fur-ther in the gas phase. Chlorides and hydroxides aremore stable than the salts with bivalent anions and gen-erally do not completely decompose or vaporize. Atflame temperatures, these species represent the most sta-ble form of the alkali metals and of chlorine and hy-droxyl. Phosphates and silicates undergo transformationswhen heated but generally do not completely decom-pose or vaporize. Their melting behavior depends ontheir composition and structure, and they can generallybe grouped into high- and low-temperature-melting ma-terials. High-temperature-melting materials include phos-phates, kaolinite, and their decomposition products. Low-temperature-melting materials include illite, salts, andalkali-containing species. Mathematical models have re-cently been developed to predict the decomposition andreaction rates of the inorganic species in coal.

2. Ash Deposition

A great deal of information is available on rates andmechanisms of ash deposition. Five major mechanisms ofdeposit formation include (1) inertial transport includ-ing impaction and sticking, (2) eddy impaction, (3) ther-mophoresis, (4) condensation, and (5) chemical reaction.Rates of inertial impaction in most environments are wellestablished, and inertial impaction commonly accounts formost of the accumulated mass. Rates of eddy impaction areless well known. The capture efficiency, a measure of thepropensity of material to stick to a surface upon impaction,is far less well established. The rates of thermophoreticdeposition are reasonably well established when localtemperature gradients and the functional form of the ther-mophoretic force on a particle (or the thermophoretic ve-locity) are known. Condensation rates can be predictedreasonably well, given accurate vapor pressure and con-centration data. The accuracy to which rates of chemicalreaction are known is often inadequate, especially thoseinvolving sulfation and alkali adsorption in silicates.

3. Deposit Properties

Properties of deposits that are important to boiler behav-ior and management include emittance and absorbance,thermal conductivity, strength, tenacity and thermal shockresistance, viscosity, and porosity. Methods for measure-ment of each of these properties are available, while tech-niques for prediction or correlation of these properties arein various states of development. Figure 10 illustrates the

FIGURE 10 Deposit surface temperature and heat flux as a func-tion of porosity and emissivity for a specified set of assumptions.

dramatic impact of deposit thermal conductivity and emis-sivity on a boiler’s deposit surface temperature and heatflux.

V. COAL-GENERATED POLLUTANTS

A. Criteria Pollutants

1. Pollutant Species

The most challenging problem associated with coal con-sumption is the control of pollutants. Increasing power de-mand will lead to increased production and use of fossilfuels with the potential for increased pollutant emissions.

In the United States, burning of coal accounts for 93%of the SO2 emissions from electric generating units and85% of the nitrogen oxides (NOx) with total emissions ofabout 14 million tons of SO2 and nearly 9 million tons ofNOx per year (figures for 1997). Total SO2 emissions fromelectric power generation declined by 16% from 1989to 1997, while total NOx emissions increased by about3%. During the same period, electric power generationincreased by 22%.

Concern for the atmospheric quality can be dated toseveral early episodes when coal began to replace woodas the primary source of domestic heating and industrialfuel. By the mid-20th century, public support and govern-ment resolve were strong enough to regulate air pollutantemissions from stationary combustors.

The combustion of coal produces both primary and sec-ondary pollutants. Primary pollutants include all species




of the combustor exhaust gases that are consideredcontaminates to the environment. The major primarypollutants include carbon monoxide (CO), hydrocarbonfragments, sulfur-containing compounds (SOx), nitrogenoxides (NOx), particulate material, and various trace met-als. Secondary pollutants are defined as environmentallydetrimental species that are formed in the atmospherefrom precursor combustion emissions. The list of sec-ondary pollutants includes particulate matter and aerosolsthat accumulate in the size range of 0.1–10 µm in di-ameter, NO2, O3, other photochemical oxidants, and acidvapors. The connection between combustion-generatedpollutants and airborne toxins, acid rain, visibility degra-dation, and stratospheric ozone depletion is well estab-lished. The detrimental impact of these contaminants onecosystems in the biosphere has been the impetus forstricter standards around the world.

Studies have shown that anthropogenic (i.e., human-generated) emissions of sulfur-containing pollutants de-creased in the United States during the 1970s and early1980s, whereas nitrogen oxides continued to increase. In1985, stationary fuel combustion in electric utilities andindustry accounted for approximately 50% of all NOx

emissions and more than 90% of all SOx emissions. Theincrease in NOx emissions can be correlated with the con-struction of many coal-fired, power-generating facilitiesthroughout the U.S. midwest and west. Coal naturally con-tains nitrogen in the range 0.5–1.5 wt%, while sulfur typi-cally varies from 0.5 to 10 wt%, with 1–2 wt% being morecommon.

2. Air-Quality and Emissions Regulations

The National Ambient Air Quality Standards (NAAQS)set by the United States specify the maximum allow-able concentrations for the following criteria pollutants:CO, hydrocarbons, particulate matter, NO2, and SO2

(see Table V). The last three are of most concern in theburning of coal. The NAAQS define Primary Standards,which are intended to protect public health with an ade-quate margin of safety. Secondary Standards are also spec-ified by the NAAQS, which are intended to protect publicwelfare (e.g., soils, vegetation, wildlife). Other countrieshave similar, sometimes more stringent, requirements, asshown in Table V. Regions are classified as either in “at-tainment” or “nonattainment” depending on whether theambient level of pollutant concentrations exceeds or doesnot exceed the standards established by the regulations.While these standards specify the maximum allowableambient concentrations for key pollutants, they do not ef-fectively control point-source emissions rates.

Two common provisions dictate the type and level ofpollutant abatement required by existing and new combus-tion facilities in the United States. These include the New

Source Performance Standards (NSPS) and the Preven-tion of Significant Deterioration (PSD) regulations. NSPSestablish nationwide emission standards for new and up-graded facilities regardless of source location or regionalair quality. The NSPS are set at levels that reflect the degreeof control achievable through the application of the bestsystem of continuous emission reduction that has beenadequately demonstrated for each category of sources.Under PSD regulations, the permitting of most new fa-cilities requires the installation of best available controltechnology (BACT) for the control of nitrogen and sul-fur oxides. BACT is defined as “an emissions limitationbased on the maximum degree of reduction for each pol-lutant . . . which the reviewing authority, on a case-by-casebasis, taking into account energy, environmental, and eco-nomic impacts and other costs, determines is achievable.”A BACT analysis must be performed for each new, mod-ified, or reconstructed combustion source.

The U.S. Clean Air Act and its 1990 amendment pro-vide the framework for controlling major sources, in-cluding stationary combustors. The regulations estab-lish NSPS which apply to units of more than 73 MWe

(250 × 106 Btu hr−1) of heat input and are directed to-ward the control of SO2, particulates, and NOx.

The 1980 Clean Air Act regulation of SO2 concen-trations for gaseous and liquid fuels is 96 mg/106 J(0.20 lb/106 Btu). The standard for coal-fired unit emis-sions is somewhat complicated. A unique maximum emis-sion rate and a unique minimum reduction of potential (un-controlled) emissions is based on the sulfur content andheating value of the coal. For high-sulfur coals (>1.5%sulfur content), a 90% reduction of uncontrolled SO2

emissions is required with a maximum allowable SO2

emission rate of approximately 520 mg/106 J (1.2 lb/106 Btu). For low-sulfur coals, sometimes called compli-ance coals, the standard requires at least a 70% reductionin uncontrolled SO2 emissions and limits the emission rateto 287 mg/106 J (0.60 lb/106 Btu).

The 1990 Clean Air Act amendment (CAAA) furtherreduced SO2 emissions to 50% of their 1980 levels bythe year 2000, and caps SO2 emissions at that level. Thenew standards not only affect new sources, but also re-quire a constant total emissions allowed from all electricutility generation units of 8.1 × 106 metric tons of SO2

per year. The Act contains provisions for trading of emis-sion allowances among units in order to achieve the mosteconomical application of control technologies.

In Germany, NOx emissions limits for new coal-firedboilers are almost half the currently prescribed level in theUnited States. Japan has placed nearly as stringent limitson NOx emissions. However, under CAAA, by 2003–04,emissions limits in the United States will drop by morethan half. Strict limits on sulfur oxides have also been setin Japan, where owners of SO2 emitters pay a proportional




TABLE V Ambient Air Quality Standards or Recommendations for CombustionPollutants in Various Countriesa

Country/organization Concentration Time

NOx (as NO2) Germany 0.05 ppm 2–12 mo

Japan 0.04–0.06 ppm 24 hr

United States 0.05 ppmb,c,d Annual arithmetic mean

Former USSR 0.05 24 hr

SO2 Germany 0.06 ppm 24 hr

United States 0.03 ppmb,c Annual arithmetic mean

0.14 ppmb,c 24 hr

0.5 ppmb,d 3 hr

Former USSR 0.02 ppm 24 hr

World Health Organization 38–37 ppbe 24 hr

15–23 ppbe Annual arithmetic mean

CO Canada 13 ppm 8 hr

Germany 26 ppm 0.5 hr

Japan 20 ppm 8 hr

World Health Organization 25 ppm f 24 hr

United States 9.0 ppmc 8 hr

35 ppmd 1 hr

Former USSR 1.3 ppm 24 hr

NMHCg Canada 0.24 ppm —

United States 0.24 ppmc,d Average from 6 to 9 A.M.

Particulate matter Japan 200 µg m−3 1 hr

United States 150 µg m−3 h 24 hr

50 µg m−3 h Annual arithmetic mean

World Health Organization 60–90 µg m−3 Annual arithmetic mean

a From Smoot, L. D., (Ed.) (1993). “Fundamentals of Coal Combustion,” Table 6.1, ElsevierScience Publishers, New York.

b NAAQS in effect in 1984.c NAAQS primary standard.d NAAQS secondary standard.e Based on observation of community populations exposed simultaneously to mixtures of SO2

and “smoke,” that is, particulate matter.f Based on recommended limit in blood of 4% carboxyhemoglobin (COHb) or less.g Nonmethane hydrocarbons; expressed as ppm carbon (ppmC).h Primary and Secondary Standards effective July 1, 1987. Includes only particles having an

equivalent spherical diameter of 10 �m or less. Referred to as the PM10 standard.

release levy. Some European communities have low lim-its on nitrogen and sulfur oxide emissions. These strictstandards have been the impetus for developing and im-plementing BACT, which includes flue gas treatment forboth nitrogen and sulfur oxides. In fact, Japan and Ger-many are the first to use catalyst beds in the flue ductsystem to control nitrogen oxide emissions. Similar tech-nologies are widely expected to be deployed in the UnitedStates in the near future as power plants prepare for PhaseII reductions of NOx required by the 1990 CAAA.

In the United States, the 1990 CAAA called on the Envi-ronmental Protection Agency (EPA) to analyze the healthimpacts of over 180 specific chemicals and all chemicalscontaining any of 12 hazardous metals. Mercury, arsenic,and cadmium are among the metals listed and are also of

specific concern for coal power plants. The Act requiredthe EPA to submit a report to Congress on the healthhazards of power plant (and other industrial) hazardousair pollutant emissions and to describe possible controlstrategies. The EPA must then regulate those hazardousair pollutants found to be significant in power plant stackexhausts. Currently, utilities in the United States are not re-quired to control the emission of hazardous toxins. Smallparticles, mercury, arsenic, and cadmium are among theemissions of greatest concern from power plants.

3. NOx Formation and Reduction

A detailed understanding of nitrogen oxide formation dur-ing the combustion of fossil fuels has been achieved by




years of research. Most of the nitrogen oxides emitted tothe atmosphere by combustion systems are in the formof nitric oxide (NO), with smaller fractions appearing asnitrogen dioxide (NO2) and nitrous oxide (N2O). Othernitrogenous pollutants include ammonia and hydrogencyanide.

Nitric oxide is formed in flames predominantly by threemechanisms: thermal NOx (the fixation of atmosphericmolecular nitrogen by reaction with oxygen atoms at hightemperatures), fuel NOx (the formation of NOx from ni-trogen contained in the fuel), and prompt NOx (the attackof a hydrocarbon radical on molecular nitrogen, produc-ing NOx precursors). N2O can be formed by a number ofpaths in gaseous and coal-laden reactors and is more preva-lent at lower temperatures. N2O levels in coal combustorsare generally less than 5 ppm (except for fluidized-bedsystems). NO2 is typically significant only in combus-tion sources such as fuel-lean gas turbines. Neither N2Onor NO2 reaches high concentrations in pulverized coalflames or in the unquenched effluent of most gas combus-tors. Hence, NO is normally the most significant nitrogenoxide species emitted from coal-fired furnaces.

It is generally agreed that the reaction network forthermal NO is qualitatively described by the modifiedZel’dovich mechanism:

N2 + O ←→ NO + N, (5)

O2 + N ←→ NO + O, (6)

N + OH ←→ NO + H, (7)

and where O atom concentrations are often assumed to bein equilibrium.

The first step is rate-limiting and requires high temper-atures to be effective due to the high activation energybarrier with the designation Thermal NO. The third stepis important in fuel-rich environments.

Prompt NO occurs by the collision and fast reaction ofhydrocarbons with molecular nitrogen in fuel-rich flames.This mechanism accounts for rates of NO formation in theearly flame region that are much greater than the rates offormation predicted by the thermal NO mechanism justdescribed. This mechanism is much more significant infuel-rich hydrocarbon flames. Two such reactions are be-lieved to be particularly significant:

CH + N2 → HCN + N, (8)

C + N2 → CN + N. (9)

The cyanide species is further oxidized to NO.Fuel NO typically accounts for 75–95% of the total

NOx accumulation in coal flames. Nitrogen in coal is pre-dominantly found as a heteroatom in aromatic rings. Ni-trogen primarily evolves from tars as HCN, with much

lower concentrations of NH3. The conversion of fuel ni-trogen to HCN is independent of the chemical nature ofthe initial fuel nitrogen (assuming that it is in-ring aro-matic nitrogen) and is not rate limiting. Once the fuelnitrogen has been converted to HCN, it rapidly decaysto NHi (i = 0, 1, 2, 3), which reacts to form NO and N2

as illustrated by the following simplified global reactionscheme, though the process is far more complicated:

fuel-N HCN NHi

NO

O2

N2

NO

. (10)

In fuel-rich combustion systems, NO can also be reducedby hydrocarbon radicals leading to the formation of HCNand then molecular N2. Based on such observations anNOx abatement technology termed reburning has beendeveloped by injecting and burning secondary fuel in thepostburner flame to convert NO back to HCN. It is gen-erally accepted that light hydrocarbon gases are the mosteffective staging fuels, although pulverized coal, particu-larly lignite, has been shown to work.

In a typical coal flame, char retains a significant fraction(30–60%) of the initial nitrogen. NO formation duringchar reactions often accounts for 20–30% or more of thetotal NO formed.

Heterogeneous destruction of gas-phase NO is signifi-cant during fuel-lean char burnout, but homogeneous de-cay reactions appear to dominate NO removel in fuel-richzones of coal combustion. Soot formation and subsequentconversion of soot nitrogen to nitrogen pollutants repre-sents yet another way to produce NOx from coal burning.The mechanism of NO destruction by soot particles is alsowell documented, with results indicating that NO removalrates on carbon black particles are much greater than het-erogeneous destruction of NO by char.

Commercially significant NOx abatement technologiesinclude installation of low-NOx burners, fuel/air staging,reburning, nonselective catalytic reduction (NSCR), andselective catalytic reduction (SCR). These are listed in ap-proximate order of their implementation. NOx reductionsas high as 70% can be expected by incorporating all ofthe noncatalytic processes, and nearly all power stationscurrently do so. However, the next phase of NOx regu-lations incorporated in the CAAA require many powerplants to reduce NOx emissions further than is practical us-ing burner and simple boiler modifications, and SCR rep-resents one of the few commercially demonstrated tech-nologies capable of such high reduction. Therefore, SCRsystems are expected to become much more prevalent inthe United States in the near future. Such systems are al-ready installed in Germany and Japan, where NOx emis-sion limits are more stringent than in the United States.




4. Sulfur Oxides Formation and Capture

Sulfur oxides are formed in stationary combustors fromthe sulfur entering with the coal. The sulfur content of coaltypically varies from 0.5 to 10 wt%, depending on rankand the environmental conditions during the geologicalcoal formation process. Sulfur oxides emissions can bereduced substantially by simply converting from high- tolow-sulfur coal. Such coal switching or partial switching(blending) is responsible for the large shift in coal pro-duction from the eastern and midwestern coal seams inthe United States to the low-sulfur coals predominantlyfound in Wyoming and Montana. For certain coals, thesulfur content can be reduced by removing some of thesulfur from the coal through washing or magnetic separa-tion before burning it.

There are three forms of sulfur in coal: (1) organic sulfursuch as thiophene, sulfides, and thiols, (2) pyritic sulfur(FeS2), and (3) sulfates such as calcium or iron salts. Bothorganic and inorganic sulfur phases undergo significantchemical transformations during coal devolatilization andcombustion. Sulfur-containing pollutants formed by burn-ing coal include SO2, SO3, H2S, COS, and CS2. Undernormal boiler operating conditions, with overall excessoxygen being used, virtually all of the sulfur is oxidizedto SO2, with small quantities of SO3. The SO2 is the ther-modynamically favored product at high temperatures. Atthe stack temperatures, SO3 is the most stable species,but kinetic limitations prevent significant amounts of SO3

from forming, leading to SOx emissions being dominatedby SO2.

Natural minerals such as limestone can be pulverizedand added directly to the furnace to capture the sulfur.The limestone calcines to form the calcium oxide withformation of pores in the particle:

CaCO3 ←→ CaO + CO2. (11)

The sulfur dioxide penetrates the pores and reacts withthe calcium oxide to form solid calcuim sulfate that canbe removed with the ash. Dolomite (CaCO3 · MgCO3) andhydrated lime [Ca(OH)2] are also used as sorbents. Sul-fur scrubbers based on a variety of chemical reactionshave become more common since 1990. Such systemsproduce by-products with some commercial value, suchas elemental sulfur, sulfuric acid, and gypsum. Scrub-bers have added benefits of removing some NOx, mer-cury, arsenic, and other pollutants that either currentlyor in the future may fall under formal regulation. How-ever, scrubbers add about 25% to the capital and op-erating costs of a power station, leading most powerstations to switch to low-sulfur coals rather than buildscrubbers.

B. Non-Criteria Pollutants

Coal contains significant amounts of many trace elements.Most trace elements are associated with the inorganic por-tion of the coal, although some are also chemically andphysically bound with the organic material. Trace elementconcentrations vary significantly among coal seams andeven among mines within the same seam. During coalcombustion, these trace elements become partitioned invarious effluent streams, depending on their chemical be-havior. Classification of trace element emissions has beenpartitioned into three general groups.

� Group 1. Elements concentrated in coarse residues,such as bottom ash or slag, or partitioned betweencoarse residues and particulates that are trapped byparticulate control systems (e.g., Ba, Ce, Cs, Mg, Mn,and Th).

� Group 2. Elements concentrated in particulatescompared with coarse slag or ash and tend to beenriched in fine particulate material that may escapeparticle control systems (e.g., As, Cd, Cu, Pb, Se, andZn).

� Group 3. Elements that volatilize easily duringcombustion, are concentrated in the gas phase, anddeplete in solid phases (e.g., Br, Hg, and I).

Emission of elements found in group 1 is controlled bystandard particle handling systems. Because group 2 el-ements are concentrated in fine particles, the release ofthese elements depends largely on the efficiency of the gas-cleaning system. The emission of volatile trace elements,such as those in group 3, can be controlled by combus-tion conditions or downstream processing. For example,trace element release may be lower from a fluidized-bedcoal furnace than from entrained-flow systems becauseof the lower temperatures in fluid beds. Furthermore, wetscrubbers used for flue gas desulfurization allow volatileelements to condense and therefore provide an effectivemethod of reducing emissions of certain trace elements.Trace elements that are of particular concern from coalcombustion are Hg, As, Cd, and Pb.

C. Greenhouse Gas Emissions

Greenhouse gas emissions represent a major and com-plex concern for the future use of fossil energy in generaland coal in particular. Greenhouse gases are atmosphericcomponents that are generally transparent in the visibleregion but absorb longer wavelength radiation, therebycontributing to a higher atmospheric temperature. Theyinclude water vapor (H2O), carbon dioxide (CO2), ni-trous oxide (N2O), methane (CH4), hydrofluorocarbons




(HFCs), perfluorocarbons (PFCs), and sulfur hexafluoride(SF6). Water vapor plays a large role in the greenhouse ef-fect and, in the form of clouds, introduces much of theuncertainty in global circulation models. However, wa-ter is rarely discussed in terms of anthropogenic sourcesbecause of its very high background concentration and vi-tal role in weather and precipitation. Most analyses, andthis discussion, are presented on a water-vapor-free ba-sis. Because of differences in the absorption efficiency ofthese gases and wide differences in their existing concen-trations, they vary widely in potency as greenhouse gases.For example, methane is about 21 times as potent as car-bon dioxide as a greenhouse gas. However, CO2 is by farthe largest potential contributor to emissions due to muchhigher concentration in both the atmosphere and in netemissions from most sources. In globally averaged num-bers, CO2 accounts for 80–90% of the total anthropogenicgreenhouse gas effect, CH4 between 5% to 10%, and NOslightly less than CH4, with all other gases making up theremaining few percent. Therefore, greenhouse gas emis-sions are often almost synonymous with carbon dioxideand methane emissions.

The complexity of the issue can be partially reducedby analyzing its aspects, starting at the most certain andprogressing to the less certain but more important. Thereis no doubt that the average concentration of greenhousegases in the earth’s atmosphere increased during the 20thcentury. There is also no doubt that much of this increaseis anthropogenic as opposed to natural fluctuations. Thelarge increase in fossil energy utilization during the sameperiod is responsible for much of this increase. There isdebate among scientists as to whether these global at-mospheric chemistry changes are driving a measurableincrease in the average world temperature, i.e., globalwarming, although a consensus seems to be growingthat measurable warming is occurring. There is consid-erable uncertainty regarding the current and long-termmagnitude of this change and its impact on global cli-mates, national and international economies, and globaland regional environments. The most radical predictedchanges from credible sources suggest very large impactson both environments and economies, with potentiallymassive destruction of property and loss of life asso-ciated with changes in sea levels, regional precipitationand drought, crop yields, and ecosystems. Others predictmore modest but still measurable changes. Most scien-tists acknowledge that such changes are difficult to pre-dict with confidence. However, the possible consequencesare such that some proactive measures may be war-ranted. The difficulty and magnitude of lifestyle changerequired to affect a decrease or even to significantly slowthe increase in global CO2 concentrations is not widelyacknowledged.

Coal figures prominently in this discussion. Coal con-tributes a disproportionately high amount of CO2 frompower production. For example, in the United States about54% of the electric power production is derived from coal,whereas about 88% of the CO2 emissions from electricpower production are derived from coal. Furthermore, thelargest increases in CO2 emissions in the future are an-ticipated to arise in association with increased industri-alization of large population centers, specifically Chinaand India. Both countries have large coal reserves andare actively developing them to support their improvingeconomies. If the per capita energy consumption of suchcountries becomes comparable to that of the United Statesor western Europe, or even if it becomes a significant frac-tion of it, the impact on greenhouse gas emissions willincrease significantly.

Technologies exist that effectively reduce greenhousegas emissions. Most are renewable energy options (solar,wind, biomass, and hydrogen-based power, among oth-ers), which generally do not produce a net change in green-house gases. Another effective and proven, but highly un-popular and nonrenewable option is nuclear power. Thosefamiliar with the practical implementation of renewablesystems recognize the enormous change in infrastructurerequired to implement these at large enough scale to im-pact greenhouse gas emissions. Most renewable optionsare more expensive, less proven, and less convenient thanfossil energy systems. Many suggest that the increasedcosts of renewable or sustainable power production aremore accurate indicators of the true cost of power produc-tion than are current market costs based mainly on fossilenergy.

Government programs are actively exploring increasedefficiency, renewable energies, energy conservation, andother systems to reduce greenhouse gases. Nevertheless,several prominent nations, including the United States,have yet to commit to specific reduction targets such asthose described in the Kyoto Protocol and many nationsthat have committed to these targets are well behind sched-ule in meeting these commitments. Greenhouse gas emis-sions may well become one of the major issues in coalutilization in the coming years.

VI. NUMERICAL SIMULATION OF COALBOILERS AND FURNACES

A. Foundations

Development and application of comprehensive, multidi-mensional, computational combustion models for coal-fired power stations is increasing across the world. Whileonce confined to specialized research computer codes,




these combustion models are now commercially avail-able. Simulations made with such computer codes of-fer substantial potential for use in analyzing, designing,retrofitting, and optimizing the performance of coal com-bustion during power generation.

Development of fossil-fuel combustion technology inthe past was largely empirical in nature, being basedprimarily on years of accumulated experience in theoperations of utility furnances and on data obtained fromsubscale and full-scale test facilities. Empirically basedexperience and data have limited applicability, however,when considering changes in process parameters, such asevaluating firing strategies for improving combustion ef-ficiencies or mitigating pollutant formation. Large-scalefurnace data are typically limited to effluent measure-ments. Subscale data provide valuable information andinsights into controlling phenomena and are less expen-sive to obtain. However, the results of subscale tests canbe difficult to extrapolate to large-scale systems becauseof the complex nature of turbulent, reactive flow processesin furnace diffusion flames. Combustion modeling tech-nology bridges the gap between subscale testing, whichtends to be phenomenological in nature, and the operationof large-scale furnaces typically used for power generationby providing information about the combustion processesthat experimental data alone cannot practically provide.

When modeling pulverized coal combustion, the frame-work for the gas-phase solution approach is most rigor-ously based on computational fluid dynamics (CFD) usingnumerical solutions of multidimensional differential equa-tions describing mass, energy, and momentum transport.Source terms and transport coefficients in the CFD mod-els are derived from particle models, gaseous turbulentcombustion models, radiation models, and similar coupleddescriptions. Information available from model predic-tions includes temperature, composition, and velocity forgases and particles, ash/slag accumulation, and so forth.Most models of this type provide spatial variation of suchquantities, while some also provide variation with time.

Comprehensive combustion models offer many advan-tages in characterizing combustion processes that caneffectively complement experimental tests. Code pre-dictions are typically less expensive and take less timethan experimental programs, and such models oftenprovide additional information that cannot always bemeasured. Computational modeling thereby becomes acost-effective, complementary tool to testing in designing,retrofitting, analyzing, and optimizing the performance ofcoal-fired power stations. Typical objectives of these appli-cations have been to determine key information requiredfor program planning, study the effects of various firingalternatives or different fuels on process performance, oraid in process design and optimization. Other objectives

include trend analysis, retrofitting, identification of testvariables, or scaling of measurements. Examples includereducing carbon loss, optimization of low-NOx burners,definition of regions of particle impact on walls of fur-naces, effects of changes in fuel feedstock quality, effectsof changes in burner configuration, or impacts of largerand smaller coal particle sizes.

Computer capacity for running these codes is also be-coming much more acceptable due to improved numer-ical methods and more advanced computer hardware.Three-dimensional code solutions for full-scale furnacesutilizing hundreds of thousands of grid nodes are beingobtained on advanced, high-powered engineering work-stations within a few hours to a few days.

While combustion modeling technology offers great po-tential, a major challenge in the use of such technology isestablishing confidence that such models adequately char-acterize, both qualitatively and quantitatively, the combus-tion processes of interest. This is typically accomplishedby making comparisons of code predictions with experi-mental data measured from flames in reactors that embodythe pertinent aspects of the turbulent combustion of coal.Consequently, data from a range of different-size facilitiesare necessary to validate the adequacy of code predictionsand establish code reliability in simulating the behavior ofindustrial furnaces. Such detailed data also provide newinsights into combustion processes and strategies.

B. Equations and Solutions

The significant physical and chemical phenomena that aretypically required to classify a combustion model as com-prehensive include (see Fig. 11) (1) gaseous, turbulentfluid mechanics with heat transfer, (2) gaseous, turbulentcombustion, (3) radiative energy transport, (4) multiphase,turbulent fluid mechanics, (5) water vaporization from par-ticles, (6) particle devolatilization, (7) particle oxidation,(8) soot formation, (9) NOx and SOx pollutant formationand distribution, and (10) fouling/slagging behavior. Theequations that constitute the mathematical model of re-acting gaseous and particulate turbulent fluid flow withheat transfer make use of the fundamental transport equa-tions for mass, momentum, and energy. For the gas phase,the equations are generally cast into fixed-coordinate (i.e.,Eulerian) form:

∂ρuφ

∂x+ ∂ρvφ

∂y+ ∂ρwφ

∂z− ∂

∂x

(�φ

∂φ

∂x

)

− ∂

∂y

(�φ

∂φ

∂y

)− ∂

∂z

(�φ

∂φ

∂z

)= Sφ, (12)

where φ refers to the conserved scalar (mass, momen-tum, enthalpy), x , y, and z represent the three coordinate




FIGURE 11 Major physical and chemical mechanisms in apulverized-coal combustion process. [Reproduced with permis-sion from Eaton, A. M., Smoot, L. D., Hill, S. C., Eatough, C. N.(1999). “Components, Formulations, Solutions, Evaluation andApplication of Comprehensive Combustion Models,” ProgressEnergy Combustion Sci. 25, 387–436.]

directions (in a Cartesian coordianate system), u, v, and w

are component velocities, ρ is the density, �φ is the turbu-lent transport coefficient, and Sφ is the source term. Substi-tutions of algebraic approximations for the derivatives insuch equations produces solvable equations that describethree-dimensional systems. To complete the model for-mulation, mathematical expressions are required for (1)gaseous reactions, including pollutant formation, (2) ef-fects of turbulence on chemical reaction rates, (3) coalparticle reaction rates, (4) radiative heat transfer, (5) tem-perature and composition dependence of physical prop-erties, (6) particulate flow within the gas continuum, (7)behavior of inorganic matter, and (8) initial and boundaryconditions. Recent reviews discuss these components indetail including code solution methods.

A list of typical dependent and independent variablesfor a furnace simulation is shown in Table VI. Coalparticles typically are treated in a procedure that couplesthe continuum (Eulerian or fixed reference fame) gas-phase grid and the discrete (Lagrangian or moving ref-erence frame) particles. Numerical solutions are repeateduntil the gas flow field is converged for the computed par-ticle source terms, radiative fluxes, and gaseous reactions.Lagrangian particle trajectories are then calculated. Aftersolving all particle-class trajectories, the new source terms

TABLE VI Variables Typically Considered in ComprehensiveSimulations of Coal Boilersa

Reactor parameters Independent variables

Configuration (e.g., shape, upflow, Physical coordinatesdownflow) (x, y, z) (x, r, θ )

Time (t)Inlet configurations (e.g., location,Dependent variablespresence of quarl)

Inlet locations Gas species composition

Dimensions Gas temperature

Wall properties Gas velocity

Wall thickness Pressure

Wall temperature Turbulent kinetic energy

Boundary/initial conditions Turbulent energy dissipation

Gas velocity rate

Gas composition Mixture fraction (mean and

Gas temperature variance)

Gas mass flow-rate Particle composition

Pressure Particle temperature

Particle velocity Particle velocity

Particle composition Particle diameter

Particle temperature Gas density

Particle size distribution Gas viscosity

Particle loading

Particle bulk density

a Adapted from Smoot, L. Douglas, and Kramer, Stephen, K. (1995).“Combustion Modeling (Table 2)” Energy Technology and the Envi-ronment, 2, (Bisio, A. and Boots, S. eds.) John Wiley and Sons, pp.863–897.

are compared with the old values and the entire procedureis repeated until convergence is obtained. Use of compre-hensive combustion models typically requires a technicalspecialist with appropriate training and experience.

C. Code Evaluation—Field Tests

Data for validating pulverized coal combustion predic-tions requires accurate information for the reactor param-eters shown in Table VI. Data measured in the combustionchamber typically include (1) locally measured values ofthe gaseous flow field velocity, temperature, and speciescomposition, (2) coal particle burnout, number density,velocity, temperature, and composition, and (3) wall tem-peratures and heat fluxes. Evaluation should include com-parisons with measurements from a wide variety of com-bustors and furnaces that range in scale from very smalllaboratory combustors (0.01–0.5 MW) and industrial fur-naces (1–10 MW) to large utility boilers (up to 1000 MW).

Laboratory-scale data are very important to compre-hensive model validation for several reasons. First, thenonintrusive, laser-based instrumentation needed to char-acterize the turbulent behavior in a flame is used moreconveniently in smaller laboratory reactors. Second, a




wider range and better control of reactor operating condi-tions are possible for a smaller facility. Third, the entirereactor volume can be traversed with small incrementalsteps, so that much more detailed datasets can be obtained.Other advantages of small laboratory facilities are rela-tively low operating cost, flexibility and accessibility, andability to control and to define carefully the boundary andinlet conditions. They can be large enough to give suffi-cient spatial resolution and to create a near-burner furnaceenvironment, and small enough to utilize the advancedmeasurement techniques which are essential to providingaccurate and complete data for model evaluation and fordetailed understanding of combustion processes.

In large-scale facilities, detailed measurements are dif-ficult to make, inlet conditions are often not well defined,operating costs for obtaining data can be high, and thefacility may have instrumentation limitations. For exam-ple, detailed model evaluation measurements of speciesand temperature within or across large flames cannot eas-ily be made, although such spatially resolved profile dataare often best for comparing flame characteristics, near-field burner performance, and jet mixing behavior withpredictions. Such measurements are essential for the eval-uation of comprehensive combustion models. These de-tailed measurements provide important information con-cerning flame response to parametric variation. Effluentmeasurements (measurements at the outlet of the systemor process) can be useful, particularly when effects of keysystem variables, such as excess air percentage, firing rate,or burner tilt angle, are measured. In some facilities, ef-fluent data are the only kind available because of furnacesize or access constraints, or because that was the only ob-jective for making the measurements. Three-dimensionaldata at the utility furnace scale obtained with the spe-cific intent of evaluating and validating model predictionshave been collected at large-scale 85-MWe and 160-MWe

corner-fired boilers of one U.S. utility.

VII. THE FUTURE

Energy production from coal will face at least three ma-jor challenges in the future: Legislative reduction of al-lowable pollutant emissions, effects of deregulation of thepower industry, and consideration of greenhouse gas emis-sions. The coal community has always faced pressure to re-duce pollutant production; however, constraints relating toderegulation and greenhouse gas emissions are relativelynew.

Emissions from coal-fired power plants have decreasedsubstantially in the last decades due to improved under-standing of pollutant formation mechanisms and combus-tion science. NOx is routinely reduced by 50–70% in majorpower stations, well over 99% of the particulate matter is

filtered from the stack gases, and low-sulfur coals or sulfurscrubbers reduce SO2 emissions by 40–90% compared touncontrolled emissions. However, newer and tighter pol-lutant regulations portend continued pressure to reduceemissions. In particular, NOx emissions have not been re-duced nearly as far as can be accomplished by advancedbut proven technologies or as far as is necessary to pre-vent environmental harm such as acid rain, ozone reac-tions, or smog. Impending environmental standards willrequire installation of treatment systems such as selec-tive catalytic reduction (SCR) of NOx. SCR systems areestablished technology but have not been widely imple-mented in commercial operation in the United States dueto the success of alternative and generally cheaper burneror boiler modifications. Germany and Japan are currentlyimplementing SCR systems and the coming years shouldsee large increases in the number of SCR installations orsimilarly effective NOx controls on commercial coal-firedpower plants in most countries. Catalyst lifetime exten-sion and catalyst regeneration are not as well understoodas are SCR NOx reduction mechanisms, and these issuesmay play a prominent role in near-future plant operation.Proposed particle cleanup regulations targeted at parti-cles less than 2.5 µm in size and mercury, arsenic, andother heavy or toxic metal regulations will also be majorpollutant-related issues in the near future.

Many world power markets are deregulating, withpower producers now able to compete for customersthroughout the grid. As one consequence of this deregu-lation, power production cost is likely to become increas-ingly more important than production reliability or con-sistency. This has the potential to fundamentally changethe way in which power plants operate and are managed.The intended consequence of deregulation is an overalldecrease in the cost of power, but some regions with inade-quate local production capacity have experienced rapid in-creases in cost and dramatic decreases in reliability duringtheir early efforts to deregulate power systems. Regionswith adequate supply from competing interests have madesmoother transitions to deregulated systems. Issues associ-ated with greenhouse gases have been discussed above. Ifnuclear power remains a socially or technically unaccept-able form of power generation, it is difficult to foresee theavailability of sufficient capacity for inexpensive powergeneration without coal continuing to play a major role.

To accomplish substantial reductions in greenhouse gasemissions, the efficiency of power production from coalwill necessarily increase, the cost of power will in-crease as more renewable energies are developed,significant conservation of energy will be practiced, ormethods of controlling/capturing CO2 emissions will bedeployed. A combination of these efforts is likely to bemost effective. Efficiency increases as high as 50% (onpaper) are achievable through advanced cycles such as




integrated gasification combined cycles and fuel cells.However, capital costs associated with these technologiesmay be twice those of conventional power systems andtechnical risks are very high. Since approximately half ofthe cost of coal-derived power is associated with capitalinvestment, there is likely no amount of fuel savings thatcould economically compensate for a doubling of capitalcosts. Global warming concerns may provide the motiva-tion for conservation, sequestration, and advanced powersystems that are otherwise difficult to justify on economicterms alone.


BIOMASS UTILIZATION, LIMITS OF • ENERGY FLOWS IN

ECOLOGY AND IN THE ECONOMY • ENERGY RESOURCES

AND RESERVES • GREENHOUSE EFFECT AND CLIMATE

DATA • POLLUTION, AIR • POLLUTION, ENVIRONMEN-TAL • RENEWABLE ENERGY FROM BIOMASS • THERMO-DYNAMICS

BIBLIOGRAPHY

Abbas, T., Costen, P. G., and Lockwood, F. C. (1996). “Solid fuel uti-lization: From coal to biomass,” In “Twenty-Sixth Symposium (In-ternational) on Combustion,” pp. 3041–3058, Combustion Institute,Pittsburgh, PA.

Bartok, W., and Sarofim, A. F. (eds.) (1991). “Fossil Fuel Combustion:A Source Book,” Wiley, New York.

Baxter, L. L. (1993). “Ash deposition during biomass and coal combus-tion: A mechanistic approach,” Biomass Bioenergy 4(2), 85–102.

Baxter, L. L., and DeSollar, R. W. (1993). “A mechanistic description ofash deposition during pulverized coal combustion: Predictions com-pared to observations,” Fuel 72(10), 1411–1418.

Baxter, L. L., and DeSollar, R. W. (eds.) (1996). “Application of Ad-vanced Technology to Ash-Related Problems in Boilers,” PlenumPress, New York.

Cengel, Y. A., and Boles, M. A. (1998). “Thermodynamics: An Engi-neering Approach,” 3rd ed., McGraw-Hill, New York.

EIA. (1999). “Annual Energy Review 1998,” Energy Information Ad-ministration, U.S. Department of Energy, Washington, DC.

Gupta, R. P., Wall, T. F., and Baxter, L. L. (eds). (1999). “Impact ofMineral Impurities in Solid Fuel Combustion,” Plenum Press, NewYork.

Hart, R. H. (1998). “Structure, properties, and reactivity of solid fu-els,” In “Twenty-Seventh Symposium (International) on Combustion,”pp. 2887–2904, Combustion Institute, Pittsburgh, PA.

Hinds, W. C. (1999). “Aerosol Technology: Properties, Behavior andMeasurements of Airborne Particles,” 2nd ed., Wiley, New York.

Leckner, B. (1996). “Fluidized bed combustion: Achievements and prob-lems,” In “Twenty-Sixth Symposium on Combustion,” pp. 3231–3241,Combustion Institute, Pittsburgh, PA.

Niksa, S. (1996). “Coal Combustion Modeling,” IEA PR/31, IEA CoalResearch, London.

Patankar, S. V. (1980). “Numerical Heat Transfer and Fluid Flow,” Hemi-sphere, Washington, DC.

Rosner, D. E. (2000). “Transport Processes in Chemically Reacting FlowSystems,” Dover, New York.

Seinfeld, J. H., and Pandis, S. N. (1998). “Atmospheric Chemistry andPhysics,” Wiley, New York.

Smith, K. L., Smoot, L. D., Fletcher, T. H., and Pugmire, R. J. (1994).“The Structure and Reaction Processes of Coal,” Plenum Press,New York.

Smoot, L. D. (ed.) (1993). “The Fundamentals of Coal Combustion,”Elsevier, Amsterdam.

Smoot, L. D. (1998). “International research centers’ activities in coalcombustion,” Prog. Energy Combustion Sci. 24, 409–501.

Solomon, P. R., and Fletcher, T. H. (1994). “Impact of coal pyrolysis oncombustion,” In “Twenty-Fifth Symposium (International) on Com-bustion,” pp. 463–474, Combustion Institute, Pittsburgh, PA.

Tree, D. R., Black, D. L., Rigby, J. R., McQuay, M. Q., and Webb, B. W.(1998). “Experimental measurements in the BYU controlled profilereactor,” Prog. Energy Combustion Sci. 24, 355–384.

Wall, T. F. (1992). “Mineral matter transformations and ash depositionin pulverised coal combustion,” In “Twenty-Fourth Symposium (In-ternational) on Combustion,” Combustion Institute, Pittsburgh, PA.

Willeke, K., and Baron, P. A. (eds.). (1993). “Aerosol Measurement:Principles, Techniques and Applications,” Wiley, New York.

Williams, A., Pourkashanian, M., Jones, J. M., and Skorupska, N. (2000).“Combustion and Gasification of Coal,” Taylor and Francis, New York.

P1: GPJ/GHE P2: FYK/FFV QC: FYK Final Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology En006g-276 July 12, 2001 14:4

Gas Hydrate in theOcean Environment

William P. DillonU.S. Geological Survey

I. The Nature of Gas HydrateII. Environment of StabilityIII. Presence in NatureIV. Effect on Seafloor Slides and SlumpsV. Effect on Climate

VI. Energy Resource

GLOSSARY

Biogenic methane Methane formed by microbialprocesses.

BSR An abbreviation for the so-called “bottom simulat-ing reflection.” A reflection recorded in seismic reflec-tion profiles that results from an acoustic velocity con-trast produced by the decrease in sound speed causedprimarily by the presence of gas trapped beneath thegas hydrate stability zone. BSRs provide a remotelysensed indication of the presence of gas hydrate.

Dissociation The breakdown of gas hydrate into its com-ponents, water and gas, commonly resulting from somecombination of pressure reduction and temperatureincrease.

Gas hydrate or clathrate A solid crystalline structureformed of cages of water molecules; a cage commonlyencloses and is supported by a gas molecule.

Gas hydrate stability zone The region within deep ocean

water and seafloor sediments where gas hydrate is sta-ble. Gas hydrate commonly occurs in the zone from theseafloor down to some depth, perhaps several hundredmeters below the seafloor, where temperature/pressurechanges result in conditions where gas hydrate isunstable.

Geothermal gradient The rate at which temperature in-creases downward into the solid Earth.

Phase boundary The interface in temperature/pressurespace where a phase change occurs. In the case of gashydrate, the phase change is from stability of solid gashydrate to stability of gas + water.

Seismic reflection profile An image of the structure be-low the seafloor along a ship trackline that is createdby using sound pulses and recording the time of returnfrom various reflecting layers.

Thermogenic methane Methane formed from organicmatter by heating at sediment depths exceeding about1 km.

473

P1: GPJ/GHE P2: FYK/FFV QC: FYK Final

Encyclopedia of Physical Science and Technology En006g-276 June 29, 2001 21:13

474 Gas Hydrate in the Ocean Environment

A GAS HYDRATE, also known as a gas clathrate, is agas-bearing, icelike material. It occurs in abundance in ma-rine sediments and stores immense amounts of methane,with major implications for future energy resources andglobal climate change. Furthermore, gas hydrate controlssome of the physical properties of sedimentary depositsand thereby influences seafloor stability.

I. THE NATURE OF GAS HYDRATE

Gas hydrate or gas clathrate is a crystalline solid; its build-ing blocks consist of a gas molecule surrounded by a cageof water molecules. The water molecules are connectedby hydrogen bonds, but the gas molecule is not bonded;rather it is simply trapped within the cage and held by Vander Waal forces (the term “clathrate” refers to a cage orlattice-like structure). Because the material is formed ofhydrogen-bonded water molecules, it looks like ice and isvery similar to ice, although of different crystallographicform, having its crystal structure stabilized by the guestgas molecules. A conceptual model of a gas hydrate cageis shown in Fig. 1, containing a methane molecule. This isthe most common gas hydrate structure, which is classifiedas structure I. Structure I gas hydrate has the smallest cav-ity sizes and is comfortable holding methane molecules orother gas molecules of similar size, such as nitrogen, oxy-gen, carbon dioxide, and hydrogen sulfide. The structure ofthe clathrate cages varies depending on the size of the guestmolecule, and some molecules are too small (smaller thanargon) or too big to exist in any of the gas hydrate struc-tures. If larger gas molecules are available, at least twoother gas hydrate structures are known. These are struc-

FIGURE 1 Diagrammatic concept of a structure I gas hydratecage holding a methane molecule. H, O, and C represent hydro-gen, oxygen, and carbon atoms. Double lines, filled or unfilled,represent chemical bonds.

ture II, holding molecules up to the size of propane andisobutane, and structure H, holding even larger moleculeslike isopentane and neohexane. Actually, more than onecavity size exists in a structure, and thus, a mixture ofgases can be accommodated. Because the gas moleculesare held within a crystal lattice, they may be held closertogether than in the gaseous state, at least at lower pres-sures, so gas hydrate can act as a concentrator of gas. Forexample, a unit volume of methane hydrate at one atmo-sphere (and 0◦C) can hold 163 volumes of methane gas atthe same conditions.

Many gas hydrates are stable in the deep ocean condi-tions, but methane hydrate is by far the dominant type,making up >99% of gas hydrate in the ocean floor.The methane is almost entirely derived from microbialmethanogenesis, predominantly through the process ofcarbon dioxide reduction. In some areas, such as the Gulfof Mexico, gas hydrates are also created by other thermo-genically formed hydrocarbon gases and other clathrate-forming gases such as hydrogen sulfide and carbon diox-ide. Such gases escape from sediments at depth, rise alongfaults, and form gas hydrate at or just below the seafloor,but on a worldwide basis these are of minor volumet-ric importance compared to methane hydrate. Methanehydrate exists in several forms in marine sediments. Incoarse-grained sediments it often forms as disseminatedgrains and pore fillings, whereas in finer silt/clay depositsit commonly appears as nodules and veins. Gas hydratealso is observed as surface crusts on the seafloor.

II. ENVIRONMENT OF STABILITY

Gas hydrate forms wherever appropriate physical condi-tions exist—moderately low temperature and moderatelyhigh pressure—and the materials are present—gas nearsaturation and water. These conditions are found in thedeep sea commonly at water depths greater than about500 m or somewhat shallower depths (about 300 m) inthe Arctic, where bottom-water temperature is colder. Gashydrate also occurs beneath permafrost on land in arcticconditions, but, by far, most natural gas hydrate is stored inocean floor deposits. A simplified phase diagram is shownin Fig. 2A, in which pressure has been converted to waterdepth in the ocean (thus, pressure increases downward inthe diagram). The heavy line in Fig. 2A is the phase bound-ary, separating conditions in the temperature/pressure fieldwhere methane hydrate is stable to the left of the curve(hatched area) from conditions where it is not. In Fig. 2B,some typical conditions of pressure and temperature inthe deep ocean were chosen to define the region wheremethane hydrate is stable. The phase boundary indicatedis the same as in Fig. 2A, so methane hydrate is stable



Gas Hydrate in the Ocean Environment 475

FIGURE 2 (A) Phase boundary of methane hydrate in the ocean (solid line). The pressure axis has been convertedto depth into the ocean, so pressure increases downward. (B) The same phase boundary as shown in (A) with aseafloor inserted at 2 km depth and a typical temperature curve (dashed line). Hatched region shows the verticalextent of the gas hydrate stability zone under these assumed conditions.

at locations in the pressure/temperature field where con-ditions plot to the left of the phase boundary. The dashedline shows how temperature conditions typically vary withdepth in the deep ocean and underlying sediments. In thiscase typical western North Atlantic Ocean thermal con-ditions were chosen and a seafloor at 2 km water depth(Fig. 2B) was selected. Near the ocean surface, tempera-tures are too warm and pressures too low for methane hy-drate to be stable. Moving down through the water column,temperature decreases and an inflection in the temperaturecurve is reached, known as the main thermocline, whichseparates the warm surface water from the deeper cold wa-ters. At about 500 m, the temperature and phase boundarycurves cross; from there downward, temperatures are coldenough and pressures high enough for methane hydrate tobe stable in the ocean. This intersection would occur at ashallower depth in colder, arctic waters.

If methane is sufficiently concentrated (near satura-tion), gas hydrate will form. However, like ice, the densityof crystalline methane hydrate is less than that of water(about 0.9), so if such hydrate formed in the water (e.g., atmethane seeps) it would float upward and would dissoci-ate when it crossed the depth where the curves intersect.However, if the gas hydrate formed within sediments, itwould be bound in place. Minimum temperature occurs

at the seafloor (Fig. 2). Downward through the sediments,the temperature rises along the geothermal gradient to-ward the hot center of the Earth. At the point where thecurve of conditions in the sediments (dashed line) crossesthe phase boundary, one reaches the bottom of the zonewhere methane hydrate is stable.

The precise location of the base of the gas hydrate sta-bility zone under known pressure/temperature conditionsvaries somewhat depending on several factors, most im-portant of which is gas chemistry. In places where the gasis not pure methane, for example, in the Gulf of Mexico,at a pressure equivalent to 2.5 km water depth, the baseof the gas hydrate stability zone will occur at about 21◦Cfor pure methane, but at 23◦C for a typical mixture ofapproximately 93% methane, 4% ethane, 1% propane,and some smaller amounts of higher hydrocarbons. Atthe same pressure (2.5 km water depth) but for a possiblemixture of about 62% methane, 9% ethane, 23% propane,plus some higher hydrocarbons, the phase limit will be at28◦C. These differences will cause major shifts in depth tothe base of the gas hydrate stability zone as would be im-plied by Fig. 2B. Such mixtures of gases essentially makethe formation of gas hydrate easier and therefore can resultin the formation of gas hydrate near the seafloor at shal-lower depths (lower pressures) than for methane hydrate




FIGURE 3 Assumed thickness of the gas hydrate stability zone for changing water depth (pressure) and a constantgeothermal gradient. Factors that change the chemistry (changes in gas composition, salinity variations, etc.) or thataffect the thermal structure (landslides, salt diapirs) can complicate this simple picture. [From Kvenvolden, K. A., andBarnard, L. A. (1982). Hydrates of natural gas in continental margins, In “Studies in Continental Margin Geology”(J. S. Watkins, and C. L. Drake, eds.), Am Assoc. Petroleum Geologists Memoir No. 34, pp. 631–640.]

at equal temperatures. Below the base of the gas hydratestability zone (500 m in our example in Fig. 2B), methaneand water will be stable and methane hydrate will not befound.

The geothermal gradient tends to be quite uniformacross broad regions where sediments do not vary. Thus,for a given water depth, the sub-bottom depth to the baseof the gas hydrate stability zone will be quite constant.However, because a change in water depth causes changein pressure, we anticipate that the base of the gas hydratestability zone will extend further below the seafloor aswater depth increases (Fig. 3).

III. PRESENCE IN NATURE

Two basic issues that concern us about gas hydrate are: (1)where does it exist and (2) how much is present? Thesequestions have turned out to be very difficult to answerin a precise manner because gas hydrate exists beneaththe ocean floor in conditions of temperature and pres-

sure where human beings cannot survive, and if gas hy-drate is transported from the sea bottom to normal Earth-surface conditions, it dissociates. Thus, drilled samplescannot be depended upon to provide accurate estimates ofthe amount of gas hydrate present, as would be the casewith most minerals. Even the heat and changes in chem-istry (methane saturation, salinity, etc.) introduced by thedrilling process, including the effect of circulating drillingfluids, affect the gas hydrate, independent of the changesbrought about by moving a sample to the surface. Gas hy-drate has been identified in nature generally from drillingdata or by using remotely sensed indications from seismicreflection profiles.

A. Identifying Gas Hydrate in Nature

1. Measuring Gas Hydrate in Wells

Drilled samples of gas hydrate have survived the trip to thesurface from hydrate accumulations below the seafloor,despite the transfer out of the stability field, just becausethe dissociation of gas hydrate is fairly slow. Such samples




have been preserved, at least temporarily, by returningthem to the pressure/temperature conditions where theyformed or, more commonly, by keeping them at surfacepressure, but at the ultra-cold temperature of liquid ni-trogen. Such preserved samples are valuable for manystudies, but certainly they do not provide quantitativelyaccurate indications of the amount or distribution of gashydrate or its relation to sediment fabric that existed in theundisturbed seafloor sediments.

Several indirect approaches have been used to gain in-dication of the amount and/or distribution of gas hydratein the sediments. Some are as obvious as making temper-ature measurements along a core to indicate where gashydrate existed, because the dissociation of gas hydrate,being an endothermic reaction, will leave cold spots.

The dissociation of gas hydrate leaves another indirectmarker of its former existence, because when gas hydrateforms it extracts pure water to form the clathrate struc-ture, excluding all salts as brine. Therefore, when hydratedissociates in a core, the interstitial water becomes muchfresher, and the amount of gas hydrate present before dis-sociation can be calculated. An example is shown in Fig. 4from Ocean Drilling Program hole 997 off the South Car-olina coast. Measured chloride content is shown in the leftpanel. The values near zero depth (depth at the seafloor)represent seawater chloride concentration (chlorinity). Itis assumed that a smooth curve of chlorinity vs depth,following the main trend of data points (solid curve), rep-

FIGURE 4 Chloride content, chloride anomaly, and volume of gas hydrate calculated from this information for OceanDrilling Program hole 997, Blake Ridge, off the southeastern United States. [From Paull, C. K., Matsumoto, R., andWallace, P. J. et al. (1996). “Proceedings of the ODP Initial Reports,” Vol. 164, Figure 47, p. 317, College Station, TX.]

resents the undisturbed (predrilling) chlorinity and thatspikes of low chlorinity to the left of the curve representthe result of hydrate dissociation. The base of hydrate sta-bility here is at 450 m below the seafloor, and the top ofsignificant gas hydrate concentrations is apparently about200 m. The estimation of the base curve is still a controver-sial issue in using this method. The second panel shows thebase curve straightened out (dashed line) so that the plot isconsidered to represent the chlorinity anomaly that resultsfrom dissociation of previously existing hydrate. The thirdpanel shows the calculated results in terms of proportionof sediment that had been occupied by gas hydrate.

A third approach that is important for identifying andquantifying gas hydrate in wells is the downhole loggingmethods in which sensors are pulled through the hole andmeasurements are made. These measurements can giveporosity information about the sediments, using gamma-rays or neutron flux, and provide indications about gas hy-drate concentration, using electrical resistivity or acousticvelocity measurements.

2. Sensing Gas Hydrate in Seismic ReflectionProfiles

Beyond the issue of precisely quantifying gas hydrate inthe few spots on the seafloor where wells have been drilled,we would like to have a technique of remotely sensing gashydrate without drilling a hole and by some means that




would allow us to survey large areas. The only effectiveapproaches that seem to fulfill these requirements are theuse of acoustic imaging of the near-bottom sediments.Most commonly, this means using seismic reflection pro-filing to image the sediments in the zone of gas hydratestability just below the seafloor. A seismic reflection pro-file is created by steaming a ship along a line and triggeringa sound source in the water every few seconds. The soundsource for gas hydrate studies can be a fairly small one,as the need is to provide penetration only down to thebase of the gas hydrate stability zone, which is unlikely tobe more than 1000 m below the seafloor. Use of a smallseismic source, which provides higher frequency sound,is also consistent with obtaining the highest possible res-olution. The sound source is often an “airgun,” which is asmall high-pressure air chamber (commonly 100–160 in.3

volume, which is equivalent to 1.6–2.6 liters), from whichthe air can be abruptly dumped by an electrical signal.Hydrophones trailed in the water behind the ship receivethe echoes of each sound pulse from the seafloor and be-low, and the echoes are recorded on the ship for computerprocessing into a profile. In a sense, any seismic profilerepresents a cross section through the seabed, but keepin mind that the image is created by reflections of soundfrom reflectors formed by density and acoustic velocitychanges that are inherent to sedimentary layers and gashydrate accumulations. The vertical axis is imaged in thetravel time of sound.

Fortunately, the base of the the gas hydrate stability zoneis often easy to detect in seismic reflection profiles. Free

FIGURE 5 Seismic reflection profile near the crest of the Blake Ridge off South Carolina in the southeastern UnitedStates. Note the strong reflection marked BSR that defines the base of the gas hydrate stability zone. The verticalaxis is in two-way travel time of sound, which varies with sound velocity in the medium; thus, this axis is a variablescale with respect to distance.

gas bubbles commonly accumulate, trapped just beneaththe base of the gas hydrate stability zone, where free gasis stable and gas hydrate will not exist. Presence of bub-bles in intergranular spaces reduces the acoustic velocityof the sediment markedly. Conversely, in the gas hydratestability zone the velocity is increased slightly by the pres-ence of gas hydrate, which in the pure state has twice thevelocity of typical deep sea sediments, and furthermore,bubbles generally cannot be present because any free gaswould convert to gas hydrate as long as water is present.A large velocity contrast generates a strong echo whenan acoustic pulse impinges on it. Thus, we can receivean image of the base of the gas hydrate stability zone ina seismic reflection profile. The base of gas hydrate sta-bility, as disclosed by this reflection, generally occurs atan approximately uniform sub-bottom depth throughouta restricted area because it is controlled by the tempera-ture. Thermal gradients across an area tend to be consis-tent, so isothermal surfaces have consistent depth. Hence,the reflection from the base of the gas hydrate stabilityzone roughly parallels the seafloor in seismic profiles andhas become known as the “Bottom Simulating Reflec-tion” (BSR) (Fig. 5). The BSR is a sure sign that gasexists trapped beneath the base of the gas hydrate stabil-ity zone and strongly implies that gas hydrate is present,because free gas, which has a tendency to rise, exists justbelow and in contact with the zone where gas would beconverted to gas hydrate. The coincidence in depth of theBSR to the theoretical, extrapolated pressure/temperatureconditions that define the base of gas hydrate stability and




TABLE I Global Organic Carbon Distribution in Giga-tons (Gt = 1015 g)a

Gas hydrate (methane) 10,000 Land

Biota 830

Detritus 60

Fossil fuels 5,000 Soil 1400

Peat 500

Atmosphere (methane) 3.6

Ocean

Marine biota 3

Dispersed carbon 20,000,000 Dissolved 980

a Data from Kvenvolden (1988).

the sampling of gas hydrate above BSRs and gas belowgive confidence that this seismic indication of the base ofgas hydrates is dependable.

A second significant seismic characteristic of gas hy-drate cementation, called “blanking,” is also displayed inFig. 5; blanking is the reduction of the amplitude (weaken-ing) of seismic reflections, which appears to be caused bythe presence of gas hydrate. Many observations of blank-ing in nature have been associated with gas hydrate accu-mulations, but the reality of blanking has been questionedbecause an undisputed physical model for it has not yetbeen identified.

B. Estimated Quantities ofGas Hydrate Worldwide

Many estimates of the global amount of gas hydrate havebeen made. Keith Kvenvolden of the U.S. Geological

FIGURE 6 Global distribution of confirmed or inferred gas hydrate sites, 1997 (courtesy of James Booth, U.S.Geological Survey). This information represents our very limited knowledge. Gas hydrate probably is present inessentially all continental margins.

Survey recently did a survey of these and looked at theirvariability over time. In 1980 the estimates covered a verybroad range, but since 1988 the range of disagreement hasbeen considerably reduced, even though the methods usedby ten different sets of workers have varied widely. From1988 to 1997 the estimates of worldwide methane con-tent of gas hydrate range from 1 to 46 × 1015 m3, with a“consensus value” of 21 × 1015 m3.

How does this amount of methane compare to otherstores of carbon in the natural environment? Table I at-tempts to answer that question by comparing the amountof organic carbon in various reservoirs on Earth. Obvi-ously, by far, the largest amount of organic carbon existsdispersed in the rocks and sediments. However, considerthe first three numbers (Table I). The organic carbon ingas hydrate is estimated to be twice as much as in allfossil fuels on Earth (including coal). Organic carbon ingas hydrate is also estimated to be about 3000 times theamount in the atmosphere, and here, we are directly com-paring methane to methane. As methane is a very pow-erful greenhouse gas, this reservoir of methane, whichmight be released to the atmosphere, may have significantimplications.

C. Environments of Gas HydrateConcentration

The presence of gas hydrate, identified on the basis ofdrilling and seismic reflection profiling, has been reportedall around the world, but most commonly around theedges of the continents (Fig. 6). Methane accumulates in




continental margin sediments probably for two reasons.First, the margins of the oceans are where the flux of or-ganic carbon to the seafloor is greatest because oceanicbiological productivity is highest there and organic detri-tus from the continents also collects to some extent. Sec-ond, the continental margins are where sedimentation ratesare fastest. The rapid accumulation of sediment serves tocover and seal the organic material before it is oxidized,allowing the microorganisms in the sediments to use itas food and form the methane that becomes incorporatedinto gas hydrate. In addition to sites in the marine envi-ronment, the map (Fig. 6) also shows some sites wheregas hydrate has been reported associated with permafrostin the Arctic, the one location where it has been foundin fresh water in Lake Baikal, Siberia, and in intermedi-ate salinity (between fresh and oceanic salinities) in theCaspian Sea.

Gas hydrate concentration commonly increases down-ward through the gas hydrate stability zone in marine sedi-ments, based on the evidence of well and seismic reflectionprofiling data. For example, note the gas hydrate distribu-tion described by chlorinity in Fig. 4 and the increase inblanking downward toward the BSR (base of the gas hy-drate stability zone) in Fig. 5. The increase in concentra-tion toward the base of the gas hydrate stability zone sug-gests that methane is being supplied below the base of thegas hydrate stability zone by some process and that the gasis probably being trapped there before entering the lowerpart of the gas hydrate stability zone. The presence oftrapped gas just beneath the base of the gas hydratestability zone is demonstrated by the common occur-rence of BSRs and has been observed in drilling. Thismethane may rise from below, either because it is be-ing generated by microbes at greater depth in the sed-iment or because of solubility considerations. Alterna-tively, it may be recycled from above. Many continen-tal margin settings have ongoing sediment deposition.As the seafloor builds up, the geothermal gradient tendsto remain constant, so the isothermal surfaces must risewith the accreting seafloor. A sediment grain or bit ofgas hydrate in the shallow sediments effectively sees thegas hydrate stability zone migrate upward past it as theseafloor builds up. Eventually, that bit of gas hydrate endsup far enough below the seafloor that it is beneath thebase of the gas hydrate stability zone and thus outsidethe range of gas hydrate stability. Then the gas hydratedissociates and releases its methane, which will tend torise through the sediments because of the low densityof gas and accumulate at the base of the gas hydratestability zone, ultimately to work its way up into thegas hydrate stability zone, where it forms more gas hy-drate.

IV. EFFECT ON SEAFLOOR SLIDESAND SLUMPS

When gas hydrate forms or dissociates within seafloor sed-iments, this has major effects on sediment strength. Disso-ciation converts a solid material, which sometimes at highconcentrations acts as a cement, to gas and water. Onlyone triaxial shear strength test of a natural gas hydrate-bearing sediment has been made at in situ conditions (atthe U.S. Geological Survey, Gas Hydrate And SedimentTest Laboratory Instrument—GHASTLI; Booth, Winters,and Dillon, 1999), but this confirmed the tremendous in-crease in strength that was expected from gas hydrate inhigh concentrations. However, in most locations in naturalseafloor sediments, hydrate concentrations are fairly low(<5%), so the cementing effect is minimal. The effectof dissociation of gas hydrate on sediment strength canstill be immense, though, because when hydrate breaksdown at shallower depths, the products of breakdown, gas+ water, occupy greater volume than the gas hydrate theywere derived from, and thus, the dissociation will increasethe internal pressure in the pore space. Such pressures arecalled “overpressures,” which are pressures greater thanthe column of water plus sediment above the spot. Gener-ation of overpressures weakens the sediment significantlyand is likely to initiate sediment slides and slumps on con-tinental slopes and rises.

Dissociation of gas hydrate that leads to weakening ofsediments can result from pressure reduction or warming(Fig. 2). Obviously, warming will occur if ocean bottomwaters warm up. However, gas hydrate will only disso-ciate at its phase boundary. In Fig. 2B, for example, thegas hydrate at the seafloor exists at pressure/temperatureconditions some considerable distance within its zone ofstability, so a few degrees of warming will not cause it todissociate. The deep limit of gas hydrate stability at condi-tions shown in Fig. 2B is about 500 m below the seafloor. Ifbottom water became abruptly warmer, the warming frontwould have to propagate downward through the sedimentto the depth where the gas hydrate is at the stability limit,which might take hundreds or thousands of years. Thechange would have to occur as a conductive heat flow, asdownward flow of water that could transfer (advect) heatis extremely limited in ocean sediments. Obviously, theplace where warming of bottom water will have a rapidinfluence is where the base of the gas hydrate stabilityzone is very close to the seafloor (see Fig. 3).

Atmospheric warming is presently occurring. Globalsurface air temperature has probably increased by roughly0.8◦C over the last century. This warming is probablybeing transferred to the ocean in a manner comparableto chemical tracers that have been observed, which means




that warmed surface water can be expected to circulatedown to depths of the shallower gas hydrates in severaltens of years. In specific cases in the Gulf of Mexico,where warm bottom currents sometimes sweep throughthe region, and off northern California, active dissocia-tion of gas hydrate at the seafloor has been observed (ev-idence is the absence of previously observed gas hydrateand release of methane bubbles). Some of this activity hasbeen related to identifiable water temperature changes, sopresent atmospheric warming may be leading to hydratedissociation.

Dissociation of oceanic gas hydrate by pressure reduc-tion clearly must also occur because a lowering of sealevel must cause an instantaneous reduction of pressureat the seafloor and down into the sediments. Pressure isdependent on the weight of a column of water and sedi-ment above a spot, and if that changes there is no delayin changing pressure at all depths, as there is in changingtemperature. Thus, a lowering of sea level, as occurredduring buildup of continental icecaps that removed waterfrom the ocean, will immediately reduce the pressure at thebase of the gas hydrate stability zone and cause gas hydrateto dissociate. The most recent glacial sea level loweringof about 120 m ended about 15,000 years ago and musthave caused significant dissociation of gas hydrate.

The result of hydrate dissociation on a continental slopeis diagrammed in Fig. 7. The initial hydrate breakdown re-leases a slide mass. When the slide takes place, the removalof sediment reduces the load on the sediment that was be-

FIGURE 7 Conceptual drawing of the triggering of a landslide and release of methane due to gas hydratedissociation.

low it and, thus, creates another pressure reduction thatmay cause further gas hydrate dissociation, resulting incascading submarine slides. Circumstantial evidence thatgas hydrate processes and submarine landslides are relatedis provided by a map of the shallow limit of gas hydrate sta-bility compared to the tops of known landslide scars on theU.S. Atlantic margin that is shown in Fig. 8. Clustering ofslope failures within the zone of gas hydrate stability, justbelow its upper limit, is suggestive of a relationship com-parable to that diagrammed in Fig. 7. Even on relatively flatslopes where slides are not triggered, evidence for buildupof pressure due to gas hydrate dissociation; mobilization ofgas + water-rich sediments; and escape of methane, water,and sediment has been interpreted. This evidence consistsof apparent blowout structures (craters with chaoticallydisturbed sediments beneath, sometimes with associatedejecta) in the Gulf of Mexico and the U.S. Atlantic margin.

The dissociation of gas hydrate, resulting in gas and wa-ter release and sediment deformation, can also be a safetyissue in petroleum exploration and production. Through-out the world, oil drilling is moving to deeper water wheregas hydrate can be anticipated. In the marine environment,gas hydrate commonly is concentrated near the base ofthe gas hydrate stability zone. If the gas hydrate is dis-sociated by warm drilling fluid, gas can enter the well-bore and circulating fluid. This reduces the density ofdrilling muds, which are placed in the hole to maintainhigh pressure at depth, and the reduced pressure encour-ages further dissociation of hydrate. Muds have foamed




FIGURE 8 Theoretical shallow limit of gas hydrate stability (dashed line) and distribution of mapped slope failureson the Atlantic continental margin of the United States. [From Booth, J. S., Winters, W. J., and Dillon, W. P. (1994).“Circumstantial evidence of gas hydrate and slope failure associations on the United States Atlantic continentalmargin.” Ann. N.Y. Acad. Sci. 715, 487–489].




and even been completely lost from the hole in gas hy-drate areas. However, anticipation of the problems allowsproper engineering of the drilling, and such issues can beavoided by controlling temperatures and pressures. Dur-ing the production phase, flow of warm fluids from be-low the gas hydrate stability zone upward through theconductor pipe could warm adjacent sediments, whichcould create an expanding dissociation front around thepipe, possibly resulting in such problems as subsidencearound the conductor pipe, escape of gas, loss of supportfor the platform foundation, or erosion by rafting awayof sediment with newly formed gas hydrate at the sedi-ment surface. The same considerations of dissociation bywarming and loss of support can affect pipelines. Again,control is possible as long as the situation is recognized inadvance.

V. EFFECT ON CLIMATE

The gas hydrate reservoir in the ocean sediments hassignificant implications for climate because of the vastamount of methane situated there and the strong green-house warming potential of methane in the atmosphere.It has been suggested that ancient climate shifts, such asthe Late Paleocene Thermal Maximum, might have beencaused by the release of methane associated with gas hy-drate. A reasonably conservative estimate (see Table I)suggests that there is roughly 3000 times as much methanein the gas hydrate reservoir as there is in the present atmo-sphere. Methane absorbs energy at wavelengths that aredifferent from other greenhouse gases, so a small addi-tion of methane can have important effects. If a mass ofmethane is released into the atmosphere, it will have animmediate greenhouse impact that will slowly decreaseas the methane is oxidized to carbon dioxide in the air.The global warming potential of methane is calculated tobe 56 times by weight greater than carbon dioxide over a20-year period. That is, a unit mass of methane introducedinto the atmosphere would have 56 times the warming ef-fect of an identical mass of carbon dioxide over that timeperiod. Because of chemical reactions (oxidation) in theatmosphere, this factor decreases over time; for example,the global warming potential factor is 21 for a 100-yeartime period. Oxidation of methane can also occur in theocean water when methane or methane hydrate is releasedfrom the seafloor. The methane that reaches the atmo-sphere can be gas released by dissociation of gas hydrateand/or gas that escapes from traps beneath the gas hydrateseal. The slides and collapses, which were discussed in theprevious section, are probably the mechanisms that allowgas to escape from the gas hydrate/sediment system to theocean/atmosphere system.

In the long term, methane from the gas hydrate reser-voir might have had a stabilizing influence on global cli-mate. When the Earth cools at the beginning of an iceage, expansion of continental glaciers binds ocean waterin vast continental ice sheets and thus causes sea levelto drop. Lowering of sea level would reduce pressure onseafloor hydrates, which would cause hydrate dissocia-tion and gas release, possibly in association with seafloorslides and collapses. Such release of methane would in-crease the greenhouse effect, and so global cooling mighttrigger a response that would result in general warming.Thus, speculatively, gas hydrate might be part of a greatnegative feedback mechanism leading to stabilization ofEarth temperatures during glacial periods. Currently, re-searchers are just beginning to analyze the potential effectsof this huge reservoir of methane on the global environ-ment and to consider many hypotheses.

VI. ENERGY RESOURCE

The very large amount of gas hydrate that probably ex-ists in nature suggests the idea that methane hydrate mayrepresent a significant energy resource. However, most ofthe reports of gas hydrate in marine sediments are onlyindications that gas hydrate exists at some place (Fig. 6).Almost every natural resource that we extract for humanuse, including petroleum, is taken from the unusual siteswhere there are natural high concentrations. Much of thelarge volume of gas hydrate may be dispersed materialand, therefore, may have little significance for the extrac-tion of methane from hydrate as an energy resource. Buteven if only a small fraction of the estimated gas hydrateexists in extractable concentrations, the resource couldbe extremely important. The goals for the future use ofmethane hydrate as an energy resource are to be able topredict where methane hydrate concentrations exist anddevelop methods to safely extract the gas.

A. Concentration

Most of the gas hydrate in the world (commonly statedas 99%) is formed of biogenic methane. Therefore, un-derstanding the processes that can cause this gas, whichis formed by bacteria dispersed through the sediment, tobecome concentrated into a hydrate accumulation is a crit-ical issue. A small amount of gas hydrate is formed fromthermogenic gas rising from great depths along fault chan-nelways that apparently are kept warm (thus outside thegas hydrate stability field) as a result of circulation ofwarm fluids. The gas-bearing warm fluids reach the oceanfloor, where temperature abruptly drops and gas hydrateis formed. This process is common in the Gulf of Mexico.




The accumulations are likely to be small, geometricallycomplex, and very near the seafloor. Thus, they are likelyto be difficult to extract, although, at this point, categoricalstatements are premature.

In normal, microbially formed, gas hydrate deposits,the common increase in gas hydrate concentration down-ward through the gas hydrate stability zone toward its baseand probable causes were discussed in Section III.C. Di-rect evidence from drilling and seismic reflection profiling(presence of BSRs) indicates that free gas bubbles typi-cally accumulate just beneath the base of the gas hydratestability zone, and, presumably, this gas is feeding the gashydrate accumulation directly above. The concentration ofgas beneath the gas hydrate stability zone can vary signif-icantly because the gas can migrate laterally, rising alongthe sloping sealing surface formed by the base of the gashydrate stability zone. When it does so, the gas will oftenreach a site where it becomes trapped at a culmination(shallow spot) and then probably will produce high con-centrations of gas hydrate above that spot. Such gas trapscan take several forms (Fig. 9). The simplest is formed ata hill on the seafloor, where the base of the gas hydratestability zone parallels the seafloor and forms a broad archor dome that acts as a seal to form a gas trap (Fig. 9, toppanel). Such a hill can be a sedimentary buildup, such asat the Blake Ridge off South Carolina shown in Fig. 5,where the BSR appears much stronger near the crest, pre-sumably indicating higher concentrations of trapped gasbeneath the gas hydrate stability zone there. Alternatively,the seafloor hill could be a fold in an active tectonic setting.

In some cases, a culmination that traps gas at the baseof the gas hydrate stability zone can form independent ofthe morphology of the seafloor. This happens over salt di-apirs (plastically rising bodies of lower density salt) as aresult of control of the depth of the base of the gas hydratestability zone by two parameters. First, salt has a higherheat conductivity than sediment, so a warm zone will existabove a salt diapir. Second, the ions that are dissolved outof the salt act as inhibitors (antifreeze) to gas hydrate, justas salt lowers the freezing temperature of water ice. Thisdouble effect of chemical inhibition and disturbance of thethermal structure causes the base of the gas hydrate stabil-ity zone to be warped upward above a salt diapir, creatinga gas trap (middle panel in Fig. 9). A seismic reflectionprofile across an isolated salt diapir is shown in Fig. 10.Notice that the deep strata appear to be bent up sharplyby the rising salt. The base of the gas hydrate stabilityzone, as indicated by the BSR, also rises over the diapir,and the BSR cuts through reflections that represent sedi-mentary layers. This does not actually represent a physicalbending of the base of the gas hydrate stability zone, butrather a thermal/chemical inhibition that prevents gas hy-drate from forming as deeply as it would if the salt diapir

FIGURE 9 Conceptual diagrams showing cross sections throughthree of the many types of traps that confine gas beneath gashydrate seals.

were not present. In this case, the BSR certainly does notfulfill its name by “simulating” the seafloor because of thestrong influence of variable temperature and chemistry, butit does show us where the base of gas hydrate is located.The extremely strong reflections just below the BSR to theleft of the diapir indicate that considerable gas is trappedthere and is likely to cause concentration of gas hydrateabove the trap. Of course, there are innumerable ways inwhich gas can be trapped beneath the gas hydrate-bearingzone. A common, simple trap (Fig. 9, bottom panel) isformed where dipping strata of alternating permeabilityare sealed at their updip ends by the gas hydrate-bearinglayer, forming traps in the more permeable layers.




FIGURE 10 Seismic reflection profile across a salt diapir beneath the continental slope off North Carolina in thesoutheastern United States. GHSZ indicates gas hydrate stability zone within the sediments. The BSR, denoting thebase of the gas hydrate stability zone, rises markedly over the diapir because of the thermal and chemical effectscreated by the diapir. This doming of the base of gas hydrate stability forms a trap for gas. Compare this to the middlediagram of Figure 9.

B. Extraction

The best places to look for gas hydrate deposits formethane extraction are the continental slopes, which, asdiscussed in Section III.C, seem to have the greatest con-centrations of gas hydrate because they contain higherconcentrations of organic matter for methane generationby microbes and have greater rates of sediment accu-mulation, which buries the organic matter and allowsmicrobial decomposition to form gas. Furthermore, theshallower depths of the upper slopes allow the greatestconcentrations of methane compared to conventional gastraps. From the minimum depth of hydrate stability downto total depths of about 1600 m, gas is held in hydratein concentrations greater than in conventional traps. Thisconcentration effect happens because gas molecules areheld closer together in the crystal lattice than they wouldbe by simple pressurization of free methane. However,the maximum concentration of methane in hydrate is es-sentially fixed, in contrast to free gas in reservoirs, wherepressure increase will cause increased concentration, soat greater depths, free gas reservoirs will be more concen-trated as a result of the higher pressure. The fixed concen-tration of gas in gas hydrate also means that shallower gashydrate deposits will release gas that will expand moreand fill a greater proportion of the pore volume than com-

parable deposits at greater depths (at greater pressures).For example, a 5% pore concentration of gas hydrate at1000 m will provide the same gas expansion as a 15%concentration at 3000 m. Therefore, there is greater like-lihood at shallow depths of exceeding the minimum limitof gas volume needed to generate spontaneous gas flow.Both the concentration factor at shallow depths and theexpansion considerations suggest that the shallower partof the gas hydrate range may be more favorable for gasextraction—perhaps in the 1000–2000 m range of waterdepths. Methane probably will be extracted from hydrateby using a combination of depressurization (perhaps pri-marily), warming (perhaps using warm fluids from deeperin the sediments), and selective use of chemical inhibitorsto initiate dissociation at critical locations.

C. Environmental Concerns

As previously discussed, methane is a powerful green-house gas, so why would we wish to extract it, perhapsrisking its escape to the atmosphere, for purposes of us-ing it as a fuel? The primary environmental reason is thatmethane is considerably less polluting than other fossilfuels. Methane has the highest hydrogen-to-carbon ra-tio of all fossil fuels (4:1), and therefore, its combustion




produces a minimum amount of carbon dioxide per en-ergy unit—about 34% less than fuel oil and 43% lessthan coal. Furthermore, methane does not produce anyother pollutants such as particulate matter or sulfur com-pounds, so conversion to methane fuel would significantlyclean up our present polluting emissions. The extractionof methane from gas hydrate probably will entail reduc-ing pressure at the base of the gas hydrate stability zone(see Section VI.B). Therefore, overpressures would be re-duced, and the likelihood of slides and slumps with releaseof gas to the ocean/atmosphere system might actually bediminished.

D. The Future

The future of gas hydrate as an energy resource dependson the evolution of (1) geology (to identify concentrationsites and settings where methane can be effectively ex-tracted from gas hydrate); (2) engineering (to determinethe most efficient means of dissociating gas hydrate inplace and extracting the gas safely); (3) economics (thecosts are likely to be higher than for extraction of gasfrom conventional reservoirs); and (4) politics (increasedenergy security and independence for energy-poor na-tions are significant driving forces). We are using otherenergy resources, which are nonrenewable, at significantrates, and the use of methane as fuel is preferable from agreenhouse-warming point of view because of its favor-able hydrogen-to-carbon ratio and other pollution con-cerns. Therefore, methane hydrate holds promise as a fu-ture energy resource.


ENERGY RESOURCES AND RESERVES • FUELS • GEO-CHEMISTRY, LOW-TEMPERATURE • GEOCHEMISTRY, OR-GANIC • GREENHOUSE EFFECT AND CLIMATE DATA •OCEANIC CRUST • OCEANOGRAPHY, CHEMICAL • OCEAN

THERMAL ENERGY CONVERSION • PETROLEUM GEOLOGY

BIBLIOGRAPHY

Booth, J. S., Winters, W. J., and Dillon, W. P. (1999). “Apparatus inves-tigates geological aspects of gas hydrates,” Oil Gas J. Oct. 4, 63–69.

Buffett, B. A. (2000). “Clathrate hydrates,” Annu. Rev. Earth Planet. Sci.28, 477–507.

Henriet, J. P., and Mienert, J., eds. (1998). “Gas Hydrates: Relevance toWorld Margin Stability and Climate Change,” Spec. Publication 137,338 p., Geological Society, London.

Holder, G. D., and Bishnoi, P. R. eds. (2000). “Gas hydrates—Challengesfor the future,” Ann. NY Acad. Sci. 912, 1044 p.

Kvenvolden, K. A. (1988). “Methane hydrate—A major reservoir ofcarbon in the shallow geosphere?” Chem. Geol. 71, 41–51.

Kvenvolden, K. A. (1993). “Gas hydrates—Geological perspective andglobal change,” Rev. Geophy. 31, 173–187.

Max, M. D., ed. (2000). “Natural Gas Hydrates in Oceanic and Per-mafrost Environments,” Kluwer Science, New York.

Paull, C. K., and Dillon, W. P., eds. (2001). “Natural Gas Hydrates:Occurrence, Distribution, and Dynamics,” Monograph GM 124, 316pp., Am. Geophys. Union, Washington, DC.

Paull, C. K., Matsumoto, R., Wallace, P. J., and Dillon, W. P. (2000).“Proceedings of the Ocean Drilling Program,” Scientific Results Vol-ume 164, 459 p., College Station, TX (Ocean Drilling Program).

Sloan, E. D., Jr. (1998). “Clathrate Hydrates of Natural Gases,” 2nd ed.,705 p., Dekker, New York.

P1: LLL/GLT P2: GPJ Final Pages Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology EN011E-511 July 26, 2001 15:23

Ocean Thermal EnergyConversion (OTEC)

William H. AveryJohns Hopkins University (Retired) (Ref.)∗

I. IntroductionII. OTEC Solar Energy ResourceIII. OTEC Systems CharacteristicsIV. OTEC Closed-Cycle Systems EngineeringV. Energy Transfer from Floating OTEC Plants

VI. Energy Transfer from Land-Based OTECPlants

VII. OTEC Total Systems InstallationVIII. Position Control for OTEC Plantships

IX. Position Control for Floating Off-Shore OTECPower Plants

X. Open-Cycle OTECXI. Closed-Cycle OTEC Development StatusXII. OTEC Closed-Cycle Systems Cost EvaluationXIII. Capital Investments and Sales Prices

for OTEC ProductsXIV. OTEC EconomicsXV. Economic, Environmental, and Other Aspects

of OTEC CommercializationXVI. Economic Impact of OTEC Commercialization

GLOSSARY

Heat engine A system that transforms heat energy intomechanical, electrical, or other types of energy. Steamengines are typical examples.

Heat exchanger A device that transfers heat through amembrane from a liquid contained in a vessel(boiler,evaporator) maintained at one temperature, to a heatreceiver (condenser) held at a lower temperature.

Solar energy Radiation from the sun that impinges onthe oceans is strongly absorbed by the seawater, so thatonly a minuscule amount reaches a depth of 100 m. Theabsorbed heat warms the surface layers of the tropical∗Present address: 237 N. Main St. South Yarmouth, MA 02664.

oceans until an equilibrium temperature of 22–26◦C isreached, when cooling caused by surface evaporationand emission of long-wavelength radiation balances theabsorption rate of heat from the sun.

OTEC IS A TECHNOLOGY that converts solar heatstored in the upper layers of the oceans into mechanicaland electrical energy. OTEC is an important energy op-tion because it offers a practical way to provide clean,renewable fuels and electricity in quantities large enoughto replace all fossil fuel and nuclear energy sources. Shipsequipped with OTEC powerplants (OTEC plantships)could produce over 200 kW of electric power/km2 of

123

P1: LLL/GLT P2: GPJ Final Pages


124 Ocean Thermal Energy Conversion (OTEC)

tropical ocean area. If deployed throughout the oceanregions suitable for OTEC, over 14 million MWe(megawatts) of electric power could be generated con-tinuously. The power would be used, via on-board waterelectrolysis, to produce hydrogen or to form hydrogen-based fuels that could furnish 300–800 quads per year ofinexhaustible energy. The fuels would be stored temporar-ily on ship-board and then shipped at intervals to worldports. OTEC feasibility has been demonstrated throughextensive tests at moderate scale.

I. INTRODUCTION

OTEC is a heat-engine cycle that uses the difference intemperature between the surface waters of the tropicaloceans and the cold water stored at depth in the ocean toconvert solar energy to mechanical and electrical energy.Warm seawater is drawn into a heat exchanger (evapora-tor) containing a fluid with a low boiling point. Transferof heat from the warm water to the fluid causes it to boiland form a vapor, which is passed through a gas turbineto generate power. The vapor (working fluid) leaving theturbine is condensed in a second heat exchanger, by trans-ferring heat to cold water drawn from the ocean depth.The turbine is connected to an electric generator to pro-duce a continuous source of electric power. There are twobasic OTEC systems: closed cycle, in which the workingfluid, after condensation, is returned to the evaporator; andopen-cycle, in which the working fluid is discarded aftercondensation.

OTEC power plants can be sited anywhere in the equa-torial regions around the world extending from roughly10◦ north latitude to 10◦ south where warm water at thesurface and cold water at depth are always available in theocean. OTEC systems can be installed on ships or barges(plantships) that will cruise slowly on the tropical oceans.This will enable OTEC plantships to produce power andfuels anywhere on the tropical oceans, and provides a wayfor OTEC to make a vast potential contribution to worldenergy needs. OTEC plants can also be installed on land oron the continental shelf at tropical sites where warm anddeep cold water are available near shore. This option canadd a renewable energy source to many tropical islandsand other suitable tropical sites.

II. OTEC SOLAR ENERGY RESOURCE

A. Geographical Distribution of OceanSurface Temperatures

Radiant energy from the sun that falls on the oceans is ab-sorbed in the “mixed-layer” of seawater, 50–100 m thick,

at the surface, forming a reservoir of warm water near theequator that stays at 26–31◦C day and night year round.Below the mixed layer the water temperature falls gradu-ally with depth, reaching a temperature of about 4◦C. at800–1000 m. Below this depth the temperature decreasesonly slightly to the ocean bottom at an average depth of4000 m. Thus the ocean structure embodies a vast reser-voir of warm water at the surface and a vast reservoir ofcold water below 1000 m that differ in temperature by22–27◦C. The distributions of temperature differences inthe tropical oceans that are suitable for OTEC power pro-duction are shown in Fig. 1.

B. Solar Heat Stored in Tropical Oceans

The area shown in Fig. 1 in which the temperature differ-ence (�T) exceeds 22◦C is approximately 60 million km2.The solar energy absorbed in one square kilometer of thesurface layers of the tropical oceans is 235 mJ (mega-joules)/sec (235-MWt) (annual 24-hr average). By con-verting only 0.1% of the solar energy input to electricity (aconservative estimate), OTEC will produce 0.235 MWe of

FIGURE 1 Ocean temperature resource for OTEC (U.S. Depart-ment of Energy, Assistant Secretary Energy Technology, Divisionof Solar Technology).



Ocean Thermal Energy Conversion (OTEC) 125

net power per square kilometer of ocean area. Therefore, ifOTEC plantships were sited throughout the suitable oceanarea, they could produce more than 14 million MWe ofpower, at the onboard electric generators. This is 19 timesthe total installed electric power generation capacity thatwas available in the United States in 1997.

III. OTEC SYSTEMS CHARACTERISTICS

An OTEC installation comprises: a power generating sub-system; subsystems to convert the electric power producedto forms useful for power transmission and fuel produc-tion; energy storage and transfer equipment; equipment forposition control of floating systems; and facilities (ships,platforms, or on-land installations) to house and maintainthe total system. Closed-cycle OTEC systems are suitablefor installation on ships and, therefore, offer a potentialway for OTEC plantships to draw power from the entirearea of the tropical oceans and provide a major new sourcefor renewable fuels production.

A. Closed-Cycle OTEC

An OTEC power-generating subsystem comprises:anevaporator (boiler), a condenser, a turbine-electric gen-erator, pumps to circulate working fluid and warm andcold seawater, and a demister to remove droplets fromthe working fluid vapor before it enters the turbine. Aschematic of the system is shown in Fig. 2a. The operatingparameters for a typical OTEC closed cycle are shown inFig. 2b. In the OTEC closed cycle, a low-boiling liquid ina heat exchanger (boiler) is vaporized by heat from a warmseawater source. The vapor (working fluid) at a pressurebetween 700 and 1400 kPa (100–200 psi) is used to drivea turbine-generator to deliver electricity to the onboarddistribution system. The vapor exhaust from the turbine iscondensed in a second heat exchanger (condenser) cooledby cold water from depth. The condensed liquid is then

FIGURE 2a Diagram of closed-cycle OTEC plantship.

FIGURE 2b Typical process parameters for the 40-MWe OTECpower system. [From Dugger, G. L. et al. (1983). Solar EnergyHandbook, McGraw-Hill, New York, ch. 19.]

pumped to the boiler to complete the cycle. Closed-cycleOTEC systems have been tested and shown to be feasi-ble at a 1-MWe scale, and engineering designs are readyfor 40-MWe pilot plant tests that will provide detailed in-formation on the engineering design and costs of OTECcomponents and subsystems, as a prelude to developmentof commercial systems of 100- to 400-MWe power.

B. Open-Cycle OTEC Systems

In open-cycle OTEC systems, first proposed in 1928 byGeorges Claude, a French engineer, seawater is used asthe working fluid. A diagram of the Claude-cycle OTECsystem is shown in Fig. 3. Warm seawater is drawn intoa vessel where the pressure is maintained below the va-por pressure of the seawater, approximately 1/30 atmo-sphere. Flash evaporation of the water occurs, producingsteam, which is used as the working fluid. The vapor passesthrough a gas-turbine, which drives an electric generator,and is then condensed by a spray of cold seawater drawnfrom depth. Several other types of open-cycle OTEC sys-tems have been studied. Details of the various designs arediscussed in Section X.

FIGURE 3 Diagram of Claude open-cycle OTEC power system.




IV. OTEC CLOSED-CYCLESYSTEMS ENGINEERING

Between 1975 and 1980 OTEC development was vigor-ously supported by the U.S. Department of Energy (DOE)(total funding >$250 million) and by French and Japaneseengineers. During this period R & D funding was providedfor study of a variety of design concepts, for testing ofsubsystems and components, and for resolution of prob-lems that might prevent or make impractical commercialdevelopment and operation of OTEC systems. By 1980solutions to all of the basic questions of technical feasibil-ity had been demonstrated on a moderate scale. The nextstep was to conduct a pilot program to provide quantitativedata on costs and engineering details of subsystems andcomponents that would be incorporated in commercial-size systems.

To begin OTEC system development, DOE issued a pro-gram opportunity notice (PON) asking industrial teams topropose concepts for construction of OTEC pilot systemsof 40-MWe size, for which DOE would share funding.To furnish a basis for evaluation of the various options,the Johns Hopkins University Applied Physics Laboratory(JHU/APL) was given a contract by DOE to developbaseline conceptual engineering designs of floating andmoored OTEC powerplants that were to be used for eval-uation of the industry proposals. Eight conceptual designplans for OTEC systems construction were submitted byindustrial teams in response to the PON. The JHU/APLbaselines, and nonproprietary data from industry propos-als, provide engineering information that comprises theprimary source for the description of OTEC systems inthe following sections.

A. OTEC Closed-Cycle Power Generation

1. Cycle Analysis

The efficiency of an OTEC system in converting heatstored in the warm surface water of the tropical oceansinto mechanical work has a theoretical limit, called theCarnot efficiency (named after Sadi Carnot, a French en-gineer who first defined the thermodynamic restrictionson the efficiency of heat engines), where

η(max) = Carnot efficiency = (Tw − Tc)/Tw

Tw = absolute temperature of the warm water

Tc = absolute temperature of the cold water.

For the ocean regions most suitable for OTEC opera-tion, the annual average surface temperature is 27–30◦C(80.6–86◦F). The average temperature of the cold water is4.4◦C (40◦F). Thus, the maximum OTEC efficiency will

be 7.5 to 8.1%. However, this ideal efficiency, even withoutunavoidable reductions caused by friction and heat losses,could be attained only at infinitesimal rates of power pro-duction. Maximum power production per unit water flowrequires both a high rate of heat transfer in the heat ex-changers and a large pressure difference across the turbine.This implies that only about half of the available �T canbe used to produce a pressure difference across the turbine.Analysis shows that the maximum gross power productionefficiency will be (Wu, 1984)

ηP = 1 − (TC/TW )1/2 = 4.0–4.5%.

Power for pumping warm and cold water and workingfluid, and for station keeping and hotel loads will consume20–30% of the gross power. Therefore, the net efficiencyof OTEC conversion of solar heat to electric energy willhave a maximum value of approximately 3.5%.

2. Heat Exchangers

Commercial heat exchangers (HX) developed for powergeneration operate at high temperatures, use high operat-ing pressures and require special materials. Costs of mate-rials, and maintenance and replacement costs are high perunit of heat exchanger area for these systems. In contrast,the low-pressure, low-temperature regime of the OTECcycle can use materials, system components, and oper-ating parameters that are similar to those in commercialrefrigeration and air-conditioning equipment, and are ofmuch lower cost. Therefore, OTEC HX costs can be attrac-tive despite the low thermal efficiency of the OTEC cycle.Several aluminum alloys have been shown to be suitablefor use in OTEC heat exchangers. OTEC does require largeHX area per unit of heat transferred because of the smalltemperature difference available for heat transfer. It is par-ticularly important, therefore, that OTEC heat exchangersachieve an optimum balance among performance, cost,technical feasibility, use of industrial components andfacilities, development risk, and safety.

To address this issue, DOE conducted programs duringthe period 1977–1981, at total funding over $250 million,to produce design data for OTEC systems, including powersystem configurations and heat exchanger types and com-ponents that would be optimal for the special requirementsof OTEC. The programs investigated and defined suitableapproaches to aspects of OTEC systems technology thatwould be critical to successful construction, operation,and durability. Analytical and experimental studies wereconducted. Heat exchanger subsystems and componentswere tested and documented to supply design data for a40-MWe pilot plant, which would provide quantitative en-gineering data on component performance and costs priorto commercial development.




FIGURE 4 Schematic of ANL test facility for 1-MWt model heat exchangers. [From Lewis, L. G., and Sather, N. F.(1978). OTEC Performance Tests of the Union Carbide Flooded-Bundle Evaporator. Argonne National Lab., Dec.ANL/OTEC-PS1.]

Contracts were awarded by DOE for studies and doc-umentation of (a) heat exchanger designs for OTEC con-ditions, (b) heat exchanger performance for OTEC condi-tions, (c) working fluid evaluation, (d) materials for heatexchangers, (e) effects of the seawater environment onOTEC design and performance, and (f) subsystem char-acteristics. To determine the engineering design require-ments for heat exchangers, a test facility was constructedat the Argonne National Laboratory (ANL) to test anddocument the performance, at a 1-MWt (one-megawatt)thermal heat transfer rate for model heat exchangers sub-mitted by potential developers. The test facility character-istics are shown in Fig. 4. To provide a uniform basis forcomparison, the heat exchangers submitted by contractorswere all evaluated at a 1-MWt rate, with a 2◦C tempera-ture change in the water between the inlet and exit. Freshwater was supplied to the evaporators and condensers toavoid surface contamination that would invalidate accu-rate measurement of heat transfer rates.

3. Heat Exchanger Designs

Seven heat exchanger types were submitted in response tothe DOE solicitation. (Details of the designs are presentedin Avery and Wu (1994). For each type, measurementswere made at the ANL facility, of the heat transfer coeffi-cients, the pressure-drop in the fluids flow, the stability ofoperation, and other relevant behavior.

The rate at which heat is transferred from warm water tothe working fluid in a heat exchanger, or from the workingfluid to cold water, is:

q = UA�T

where U = Heat transfer coefficient, A = Area of heattransfer surface. �T = Temperature difference betweenthe water flow and the working fluid. From knowledge ofthe heat transfer coefficients in the evaporator and con-denser, the overall heat transfer coefficient (U ) can be cal-culated. The results for the heat exchangers tested at ANLare shown in Table I. Details of the complex calculationsare given in Avery and Wu (1994). The ANL tests didnot allow optimization of U , which involves a trade-offbetween heat-transfer rates, which increase with the flowvelocities of water and working fluid, and power lossesfrom pressure drops that also increase with flow velocity.Thus, two values of U are shown in Table I for each heatexchanger type. The value in the first column is calculatedfor the ANL test conditions. The number in the secondcolumn lists the flow conditions for a maximum value ofU , as calculated with the Owens formulas.

Since the cold water ducting system is more costly thanthe warm water system it is desirable to arrange the flowconditions so that there is a large �T (and small waterflow) in the cold water flow and a small �T (and largewater flow) in the warm water. An example of a choiceof flow conditions is shown in the third column listedfor the APL heat exchanger design, in which the aimis to achieve maximum performance with a cold waterpipe of minimum cost. The data in Table I demonstratethat OTEC heat exchangers can have values of heattransfer coefficients that are equal to or higher than valuestypical of commercial heat exchangers operating at hightemperatures.




TABLE I OTEC Power Plant Performance Comparison

APL HX ANL-SP3 CMU Trane Japan Alfa-Laval

(nom) (opt 1) (opt 2) (nom) (opt) (nom) (opt) (nom) (opt) (nom) (opt) (nom) (opt)

Heat loading (kWt) 937 1591 570 937 1332 938 2019 938 1615 4091 4664 938 938

Water flow area (evaporator) (m2) 0.26 0.26 0.26 0.10 0.10 0.26 0.26 0.089 0.089 0.20 0.20 0.31 0.31

Water flow area (condenser) (m2) 0.26 0.26 0.26 0.14 0.14 0.26 0.26 0.089 0.089 0.20 0.20 0.31 0.31

Mass flow, warm water (kg/sec) 202 202 75 202 202 202 202 202 202 409 409 202 202

Mass flow, cold water (kg/sec) 202 202 30 202 202 202 202 202 202 392 392 202 202

Working fluid (WF) NH3 NH3 NH3 NH3 NH3 NH3 NH3 NH3 NH3 R-22 R-22 NH3 NH3

Mass flow, WF (kg/sec) 0.76 1.28 0.46 0.76 1.09 0.77 1.66 0.78 1.35 21.47 23.50 0.77 1.81

�p (sat) (kPa) 402 271 268 378 283 446 276 412 279 320 320 440 266

�T (sat) (K) 15.83 10.61 9.64 14.80 10.98 17.42 10.59 15.94 10.59 10.66 9.73 17.63 10.66

�T (lmtd) (evaporator) 3.27 5.55 2.75 3.52 5.01 2.26 4.86 2.51 4.32 3.82 4.35 2.48 5.02

�T (lmtd) (condenser) 3.11 5.27 6.35 3.90 5.69 2.50 5.39 3.73 6.42 4.52 4.77 2.09 4.95

�T in WW flow through 1.16 1.97 1.90 1.16 1.65 1.16 2.50 1.16 2.00 2.50 2.85 1.16 2.75evaporator (K)

�T in CW flow through 1.13 1.91 4.61 1.13 1.60 1.13 2.43 1.13 1.94 2.43 2.76 1.13 2.67condenser (K)

U0 (evaporator) (W/m2 K) 1469 1469 1464 2844 2844 2709 2709 4030 4030 2593 2593 1956 2461

U0 (condenser (W/m2 K) 1503 1509 644 2678 2678 3423 3423 2811 2811 1886 1886 2406 2406

Net OTEC power (kWe) 25.69 32.31 12.86 19.65 23.24 23.92 38.97 21.49 28.33 80.27 91.77 17.52 37.68

Cycle net power/max cycle power 0.825 1.000 1.000 0.891 1.000 0.683 1.000 0.817 1.000 0.994 1.000 0.663 1.000

Net power/total heat 0.092 0.116 0.046 0.147 0.173 0.173 0.282 0.195 0.257 0.197 0.225 0.115 0.247exchanger area (kWe/m2)

CW flow for 1 MWe (net) 7.86 6.25 2.33 10.28 8.69 8.44 5.18 9.39 7.13 4.88 4.27 11.53 5.36power (m3/s)

CWP diam. for 40-MWe (net) 14.15 12.61 7.71 16.18 14.87 14.66 11.49 15.47 13.47 11.15 10.43 17.13 11.68power (m)

4. Heat Exchanger Materials

Heat exchanger materials used in commercial power gen-eration favor titanium, copper alloys, and stainless steel.These could be used for OTEC but their cost makes themunattractive. Aluminum alloys are used in heat exchangersbuilt for cryogenic systems, which operate at pressuresand temperatures in the same ranges as OTEC, are lowin cost, and are the preferred materials for OTEC. Beforealuminum could be selected for OTEC it was necessary toprove that aluminum heat exchangers would be durable inthe seawater environment. This was demonstrated in testsconducted at the Natural Energy Laboratory of Hawaii(NELHA) with warm and cold ocean water, which con-firmed the durability of several aluminum alloys (Georgeand Richards, 1980).

5. Working Fluid Requirements

Desirable characteristics of working fluids for closed-cycle OTEC include vapor pressure between 700 and1400 kPa (100–200 psi) at 25◦C, low volume flows of va-por per kilowatt of generated power, and high heat-transfercoefficient for the liquid films on the HX surfaces. The

thermodynamic properties of four working fluids that aresuitable for OTEC HX are listed in Table II. The fluoro-carbon, R-22, used in the ANL test program, was bannedbecause of its ozone depleting properties. The hydrofluo-rocarbon, R-134a, tetrafluoroethane, shown in Table II isconsidered to be a suitable substitute.

Ammonia was chosen as the preferred working fluidby most of the PON respondents because of its favor-able thermodynamic properties. However, ammonia canbe corrosive, and does not dissolve and wash away greasewhich can be deposited on condenser surfaces by vaporsemitted by lubricants used in pumps and other HX equip-ment. This problem can be eliminated by selection of suit-able subsystem components. Corrosion problems can beavoided by careful design. Hydroflurocarbons and hydro-carbons may also be used as working fluids. They dissolvegrease and are compatible with materials that are suitablefor OTEC heat exchangers and other cycle components Aworking fluid that is selected for OTEC must comply withstrict regulations established for safe production, trans-port, storage, and use of bulk chemicals. New regulationswill not be required for OTEC. The safety demonstratedin commercial handling of millions of tons of ammonia,




TABLE II Properties of Heat Engine Working Fluids of Interest to OTECa [15.56◦C (60◦F)]

TetrafluoroethaneProperty Ammonia Propane Butane (R 134a)

Formula NH3 C3H8 C4H10 C2F4H2

Molecular weight (M) 17.03 44.06 58.08 102.0

Densityb (1) (kg/1) 616.7 506.9 583.4 1241.0

Densityc (v) (kg/1) 5.83 16.06 4.64 23.77

Vapor pressure (sat) (kPa) 742.1 506.9 583.4 490.5

Heat of vaporization (kJ/kg) 1204.0 351.2 370.8 186.3

Specific heat (1) (kJ/kg·K) 4.713 2.643 2.397 1.389

Specific heat (v) (kJ/kg K) 2.942 1.929 1.739 0.971

Viscosity (1) (Pa s) 0.1446 0.1074 0.1741 0.2227

Viscosity (v) (Pa s) 0.00954 0.00836 0.00734 0.0139

Thermal conductivity (1) (W/m K) 0.5128 0.09756 0.1106 0.0852

Thermal conductivity (v) (W/m K) 0.0250 0.0178 0.01536 0.0129

a (Courtesy of NIST).b Saturated liquid (1).c Saturated vapor (v).

hydrocarbons and fluorocarbons, on land and at sea, indi-cates their acceptability for OTEC design and engineering.The cost of any working fluid selected for OTEC wouldbe a minimal fraction of the power system cost.

B. Water Ducting

1. General Requirements for FloatingOTEC Plants

Commercial OTEC plants sited in the tropical oceans willeach generate power in the range of 100–400 MWe (net)and will use modular construction. An OTEC plant willrequire a total water flow of about 3 m3/sec MWe (net).Thus, commercial plants of 100–400 MWe will have totalflow rates of 300–1200 m3/sec. Design studies indicatethat plants composed of 50- to 60-MWe modules will beoptimum and that the flow velocity in the water ductswill be limited to about 2.5 m/sec to avoid excessive draglosses. Therefore, the minimum total duct area will be60–75 m2/module.

2. Cold Water Pipe (CWP) Designand Construction

The CWP is costly because it must extend to a depth of800–1000 m and be designed to withstand ocean wavesand currents that could be generated in a tropical stormof once-in-a-hundred-years intensity. For minimum sizeand cost, the CWP diameter in each 50- to 60-MWe mod-ule will be 7.5–10 m (25–33 ft). Dependant on considera-tions, other CWP diameters may be chosen. Many designsand materials of construction of CWPs (including steel,

aluminum, and elastomeric materials) have been studied.Scale tests have shown that CWPs made of light-weightconcrete or fiber-reinforced plastics (FRP) are suitable forfull-scale OTEC floating power plants and plantships. Themanufacturing procedures and facilities for constructionof pipes made of sections of light-weight concrete, joinedend-to-end via a flexible joint, were demonstrated at the3-m scale. Construction of suspended post-tensionedlight-weight concrete pipes with diameters of 10 m isfeasible with state-of-the-art facilities. Deployment maybe implemented with existing technology available in theoff-shore oil industry. Dynamic analyses show that theseCWPs will withstand the maximum loads that are antic-ipated in the tropical ocean environment, in a 100-yearoperation of floating OTEC plants and plantships.

Analyses and tests done with plastic tubes of 2.5-mdiameter show that CWPs made of fiber-reinforced plas-tic (FRP) are feasible. Cylindrical storage vessels madeof FRP are in commercial use in diameters as large as25 m, and the technology can be adapted to CWP con-struction. For CWP deployment a double-walled pipe withinternal reinforcement will be required to give the plasticsufficient bending strength. The cost was estimated to beabout 20% above the estimated costs of concrete CWPs butthere could be advantages in deployment of the pipes. Thedeployment procedure was successfully demonstrated inthe Mini-OTEC and OTEC-1 at-sea test programs. TheOTEC-1 deployment involved assembly on shore of a2-m-diameter, 700-m-long CWP, followed by towing toan off-shore ship, attaching the top end to a support, andthen allowing the (loaded) other end to sink to orient thepipe in a vertical position. The upper end was then insertedfrom below into a gimbal connection on the platform.




TABLE III Collapsing Loads on CWP [Cold Water Flow3.4 m3/sec/MWe (net)]

Power MWe (net)

40 40 60 60 80 80

V (m/s) 2.00 2.50 2.00 2.50 2.00 2.50

D (m) 9.30 8.32 11.40 10.19 13.16 11.77

L (m) 1000 1000 1000 1000 1000 1000

P (kPa)

Friction 3.53 6.16 2.88 5.03 2.50 4.36

Density effect 7.65 7.65 7.65 7.65 7.65 7.65

Dynamic loss 1.03 1.60 1.03 1.60 1.03 1.60

Miscellaneous 1.00 1.00 1.00 1.00 1.00 1.00

P(total) (kPa) 13.20 16.40 12.60 15.30 12.20 14.60

P(total) (psi) 1.90 2.40 1.80 2.20 1.80 2.10

P(800 m) (kPa) 12.50 15.20 12.00 14.30 11.70 13.70

3. Cold Water Pipe Dynamics

In operation of an OTEC floating plant or plantship, theCWP must withstand collapsing loads created by suction,will be subjected along its length to fluctuating lateral andvertical loads that are caused by waves and currents actingon the ship that supports the pipe, and will be subjectedto under-water waves and currents acting along the lengthof the pipe. These loads determine the design strength re-quirements of the CWP. The collapsing strength require-ments are shown in Table III. Analysis of the dynamics iscomplex and requires sophisticated computer programs toprovide quantitative solutions. Fortunately, the off-shoreoil industry has developed analytical techniques to predictthe stresses and strains in long suspended pipes in oceanenvironments. The analytical programs were adapted topredict OTEC CWP dynamics, and were verified in at-sea and laboratory tests. Computer programs were devel-oped by Paulling, TRW, and others that predict strains andstresses as a function of platform movements in the seastate, and motions and moments of segmented and flexibleCWPs of various dimensions and materials strengths, asdependent on wave heights, wave frequencies and direc-tion, and underwater forces. The programs include effectsof internal and external water flow and are applicable tofixed or flexible connection of the CWP to the platform andinclude the dynamics of the platform–CWP combination.The predictions of the analysis programs were verified inseveral programs:

a. At-sea tests. With DOE support and JHU/APLsupervision, measurements of strain and stress weremade along the length of a segmented 1.52-m-diameter(5-ft) steel pipe (L/D = 100) suspended from a semisub-mersible, off-shore oil test platform provided by the

Deep Oil Technology Corporation. The tests were con-ducted near Catalina Island in California. The tests showedthat the analysis programs gave good predictions of theobserved pipe strains and stresses along the pipe, ifthe sea state was represented by equations developedby Breitweiser, incorporating a bidirectional wave pat-tern. With direction and support by the National Atmo-spheric and Oceanic Administration (NOAA), at-sea testswere also done near Kea-Hole Point in Hawaii, using a2.44-m-diameter (8-ft) fiber-reinforced FRP CWP, 122-m(400-ft)-long. Results of stress and strain measurementsalong the length of the pipe were in good agreement withanalytical predictions. The tests demonstrate that a satis-factory theoretical base exists for the design and construc-tion of concrete and FRP CWPs of diameter large enoughto meet the requirements of 50- to 60-MWe OTEC powerplant modules.

b. Water tunnel tests. A series of tests to evalu-ate the survivability of the 40-MWe OTEC platform–CWP assembly in 100-year tropical storms on the highseas off Brazil and near-shore Puerto Rico were made byJHU/APL using a 1/20th-scale OTEC barge model in awater tunnel operated by the Offshore Techology Corpo-ration in Escondido, California. The facility was able tosimulate sea states having waves that approached the bargein two directions, as was observed in the ocean tests nearCatalina Island. The test data showed that the 40-MWeOTEC barge, with a flexible attachment to the platform,would survive the open-ocean 100-year worst-storm con-ditions. At the Puerto Rico site, the moored platform wouldrequire mooring lines that would orient the vessel to headinto the waves, and would need some fairing of the bowto prevent water from coming over the deck.

4. CWP Support Design

Analysis and tests show that a CWP connected to a floatingplatform by a fixed joint will not survive in an ocean envi-ronment. Therefore, a flexible connection to the platformis required at the top of the pipe. The CWP connectionwith the platform must embody a sea-water seal and allowmovement at the joint through an angle of up to 30◦ at anyplatform orientation, to survive in the maximum 100-yearstorm. The connection must also support the weight of theCWP, including loads imposed by heave of the platform.

Several engineering designs for supporting the CWPin a flexible connection were studied in the 40-MWe-demonstration proposals for floating OTEC plants. Fea-tures of three designs are shown in Fig. 5. Consideration ofthe options led to the the ball and socket joint being incor-porated in the JHU/APL base-line 40-MWe plant design.




FIGURE 5 Types of flexible joints studied for OTEC CWPsupport. [From TRW (1979). Ocean Thermal Energy ConversionCold Water Pipe Preliminary Design Final Report. TRW EnergySystems Group.]

5. General Conclusions Regarding CWPRequirements for Floating OTECPlants and Plantships

Methods of construction for CWPs of 2.5-m (FRP) and3-m (light-weight concrete) were demonstrated that aredirectly applicable to CWPs with diameters of 10 m ormore. At-sea tests confirm that CWPs will be able to sur-vive storms of maximum severity expected in a 100-yearperiod in the tropical oceans. Lightweight concrete andFRP have been shown to be suitable for CWP construction.Further information is desirable on added-mass effects ofexternal water flow in lateral motions of large-diameterpipes and on effects of multidimensional sea states onCWP dynamics.

6. Cold Water Pipes for Shore-Based Plants

The design and construction of shore-based OTEC plantsinvolve significant CWP problems not present in float-ing plants, namely, the design is site-specific and surveysof the configuration and structure of the sea bottom atproposed sites must be made to define a suitable trajec-tory for the CWP. Furthermore, data must be availablefor three ocean regimes: the near-shore regime, whichinvolves high waves and currents, requires heavy struc-tures and/or emplacement below the sea floor to preventfailure of the CWP; a shallow zone on the continentalshelf where the water depth increases gradually from near

shore to a depth of 100 m or more; a final zone wherethe depth increases sharply to reach the cold water source.The total CWP length will be around more than 2 km atthe best tropical island sites and 5 km or more at aver-age sites. Decisions must also be made regarding supportof the CWP, with possibilities including anchoring thepipe to the sea floor, entrenching it, or using a buoyantpipe held by cables attached to weights on the sea floor.For the cold water supply pipes installed at the NELHAfacility in Hawaii, entrenching of the pipe near shoreand catenary support for the remaining length have beenadopted.

7. General Conclusions Regarding CWPRequirements for Shore-Based OTEC Plants

Buoyant polyethylene (HDPE) cold water pipes of ∼1 min diameter, anchored at intervals with ‘cables attachedto weights on the sea bottom, have been constructed anddeployed at NELHA and are operating successfully. FRPpipes suitable for shore-based plants of several megawattsseem feasible. A preliminary engineering design is avail-able of water ducts for a 50-MWe OTEC plant using con-crete ducts for cold deep water, warm surface water, andmixed exhaust water.

C. Warm and Mixed Effluent Ducting

Warm water from the mixed layer at the ocean surface ispumped into the OTEC vessel through screens at a flowvelocity below 1/2 m/sec to prevent fish being trappedagainst the screens. The warm water flows through theheat exchangers, and is discharged at the bottom of theplatform. It may be mixed with the cold water exhaustbefore leaving the vessel or may be discharged at thebottom where it may be interleaved with the cold wa-ter flow. The exhaust flow is denser than the surroundingocean water and forms a plume which sinks to a depthwhere the plume density matches that of the ambientocean water. For floating plants, no external inlet pipesare needed. Dependent on the site, shelf-mounted plantsmay or may not need external exhaust pipes. OTEC-1experience showed that the discharge plume descendedquickly with insignificant mixing with the ambient oceanlayer. This result indicates that external exhaust pipes willnot be needed. If they were needed, a new method fortheir attachment would have to be developed. Attachedexhaust pipes failed in some tests. Shore-based plants willrequire inlet pipes that extend along the bottom to a wa-ter depth low enough to minimize wave effects and farenough above the bottom to prevent entrance of sedimentdislodged by water flowing to the inlet or by under-seacurrents.




V. ENERGY TRANSFER FROM FLOATINGOTEC PLANTS

OTEC power may be supplied to users by direct trans-mission of electricity to on-shore customers at sites wheredeep cold water is available near shore, or used on-boardto produce a fuel or product that can shipped ashore andstored for later use in motor vehicles or electric powerplants. OTEC plantships sited on the tropical oceans willuse electric power generated onboard to electrolyze wa-ter and form hydrogen based fuels, which will be storedtemporarily and then shipped to world ports at monthlyintervals.

A. Water Electrolysis

Water electrolysis provides an efficient mechanism forconverting electrical energy to stored chemical energy.In principle, the conversion process can have 100% effi-ciency. R & D programs indicate that efficiencies in therange of 80–90% will be attainable. Therefore, hydrogenproduction is a highly efficient way to store electrical en-ergy produced on OTEC plantships.

Electrolysis systems are classed as unipolar or bipolar.Unipolar electrolyzers have positive and negative elec-trodes immersed in a conducting fluid (electrolyte). Inwater-based electrolytes an applied voltage causes hy-drogen gas to be liberated at the negative terminal (cath-ode) and oxygen gas at the positive terminal (anode). Theelectrodes are designed to channel the gases into separateducts.

Electrolyzer, Inc. has been a principal developer andsupplier of unipolar electrolyzers since the 1940s. Theelectrodes are supported in containers (cells) filled withconcentrated KOH. The cells are connected in parallel andoperate at voltages in the range of 1.5–2 V. Experimentalresults indicate that energy conversion efficiencies of 85%or above will be possible in commercial units. A layoutfor a planned 35-MWe installation is 29 m wide and 95 mlong. With these dimensions, a 350-MWe electrolyzer foran OTEC plantship would occupy an area of 2750 m2.

Bipolar electrolysis systems are characterized by thetype of electrolyte. The proton exchange membrane(PEM) system, developed by the General Electric Com-pany (GE), uses as the electrolyte a thin membrane ofsulfonated fluorocarbon (NafionTM) that conducts elec-tricity when saturated with water. Electrodes are formedby depositing a thin platinum film on opposite sides of themembrane to form a bipolar cell. An electrolyzer is madeby stacking 50–200 cells in series, with suitably formedseparators to direct the exhaust gases into channels at thesides. Since the membrane is the electrolyte, only purewater needs to be supplied to the cell. When the cell oper-

ates, water is protonically pumped through the membraneat about eight times the electrolysis rate. The water pro-vides the source of hydrogen and oxygen, and removesthe heat generated by ohmic resistance of the Nafion film.PEM electrolyzers, consisting of 60 cells 15 cm2, weredeveloped by GE and successfully tested at current densi-ties ranging from 2000 to 10,000 amperes/m2 (200–1000amperes per square foot (ASF)). Tests indicated that cells1 m2 would be feasible. When funding for the programwas terminated some problems with sealing the cellshad not been solved. At 1000 ASF, the cell efficiencywas 70%.

Alkaline electrolyzers, which use concentrated KOHas the electrolyte, are commercially available in mul-tikilowatt sizes from Teledyne Energy Systems in theUnited States and several foreign companies—Brown-Boveri (Switzerland), Norsk-Hydro (Norway), and Frenchand Belgian suppliers. The design employs thin mem-branes of asbestos (or other material) saturated with KOH,with nickel (or proprietary) screens pressed into the mem-branes, as the electrodes, to form an electroysis cell. Cellsare stacked in series, as in the PEM design, to achieve adesired input voltage. The KOH electrolyte provides thesource for hydrogen and oxygen production and is circu-lated for cooling. Cell efficiency improved steadily withR & D support after 1976 leading to a projected value of85–90% in 1984. As with the GE program, funding wasterminated in 1984.

It may be concluded that electrolyzers with efficienciesin the range 60–85% will be available when a significantcommercial demand arises.

B. Storage of OTEC Electrical Energyin Hydrogen-Based Energy Products

OTEC electric energy generated at on-land or near-shoreinstallations may be used to generate hydrogen at off-peaktimes, and stored as gas or liquid for later power gener-ation. The power requirements for hydrogen liquefaction(∼30% of the heat of combustion) make it inefficient tostore electric energy in liquid hydrogen unless there is a re-quirement for direct use of liquid hydrogen. Because of itslow density it is impractical to store or ship hydrogen pro-duced on OTEC plantships. However, the hydrogen maybe stored efficiently if it is combined with nitrogen on theplantship to form ammonia (NH3), or other hydrogen-richmolecules.

1. Ammonia

Ammonia is of particular interest as a way to store OTECelectric energy because hydrogen generated by OTECelectrolysis on a plantship can be combined with nitrogen




extracted from the air to form ammonia. The reaction isan equilibrium process that is a highly efficient method(∼90%) of storing OTEC energy. The physical propertiesof ammonia are nearly identical with those of propane,making it practical to synthesize, store, and transportliquid ammonia from plantships to on-land users. Tests ofammonia fuel in internal combustion (IC) engines yieldan octane number of 130. It would be a practical, inex-haustible fuel, which does not produce carbon dioxide,to replace gasoline in motor vehicles. It will not producepolluting exhaust products, if used in fuel cells or powerplants.

2. Methanol

Methanol (CH3OH) synthesis is also an attractive methodfor storing OTEC energy on plantships. Methanol is ahigh-octane (110) liquid fuel that may be used directlyin motor vehicles, and is the preferred feed-stock forfuel cells. Methanol synthesis is highly efficient becauseit uses both the oxygen and the hydrogen produced inwater electroysis, in the reactions: C (coal) + 1/2O2 =CO, CO + 2H2 = CH3OH. Coal for the process is trans-ported to the plantship in a conventional collier, and con-verted to CO by injecting oxygen and water into a moltencarbonate reactor. If commercialized to use the total tropi-cal ocean resource OTEC methanol synthesis could supplymore than 800 quads per year of fuel, roughly three timesthe total 1998 world fossil fuel consumption. Methanolcombustion produces approximately the same amount ofCO2 as gasoline. Therefore, OTEC methanol productionwill not offer a way to reduce global warming.

C. Electrical Energy Transfervia Electrochemical Cycles

Two electrochemical energy transfer concepts have beeninvestigated. The redox battery system uses electrodes im-mersed in an electrolyte that can exist in two ionic statessuch as Fe++ and Fe+++. Application of a voltage abovethe ionization potential converts Fe++ to Fe+++, storingthe energy in the electrolyte, which can be shipped toshore where the process is reversed to liberate the storedenergy. Costs of shipment of the electrolyte make the sys-tem impractical for OTEC energy transfer. The lithium–airbattery system uses a slurry of lithium hydroxide, whichis prepared on land and shipped to the plantship whereOTEC power is used electrolytically to prepare billets ofmetallic lithium. These are shipped to land where electricenergy is produced by reaction of the lithium billets withair in a battery, regenerating lithium hydroxide. Shippingcosts make this system impractical for OTEC.

D. Other Methods of Using Energy Producedon OTEC Plantships

Studies and some testing showed that aluminum refiningusing power generated on shore-based or floating near-shore OTEC plants at several sites would not be cost ef-fective at 1990 fossil energy costs. Deep-ocean miningof minerals using low-cost OTEC power could make thisoption economically attractive in the future.

VI. ENERGY TRANSFER FROMLAND-BASED OTEC PLANTS

At tropical sites where cold, deep water is available nearshore OTEC electric power can be transmitted directlyto on-land consumers. This can be economically attrac-tive if the power demand is large enough to justify con-struction of a multi-megawatt OTEC plant. The economicviability of land-based OTEC can be enhanced in favor-able locations by using the cold water for air condition-ing, mariculture, fresh water production, and coldwateragriculture. Programs conducted at the Natural EnergyLaboratory of Hawaii (NELHA) have demonstrated thefeasibility of these options.

VII. OTEC TOTAL SYSTEMS INSTALLATION

An OTEC system requires a structure to house the totalOTEC operating complex. It may be designed for directtransmission of power to consumers on land, or designedfor at-sea operation to produce a storable product for latershipment. A configuration to be selected will optimizeefficiency, ease of construction, cost, ability to withstandenvironmental stresses, and suitability for operation at se-lected sites. For floating platforms designed to produce astorable product while “grazing” on the tropical oceans, abarge installation is the best choice. For floating platformsdesigned for power transmission to shore, the optimumconfiguration is strongly dependent on the site selection.Designs that have been studied include spar configura-tions, semisubmersibles, ships, and barges. Problems ofconstruction, deployment, and demonstration of abilityto withstand waves and currents generated near shore by“100-year-storms” are involved in the selection, as well ascost.

VIII. POSITION CONTROLFOR OTEC PLANTSHIPS

OTEC plantships will use thrusters to propel the plantshipsat roughly 1/4 knot in normal operation, and at 1 knotat maximum speed. Rotatable thrusters mounted below




the platform will maintain direction and orientation inwaves and currents. (The low forward speed makes ruddercontrol not feasible.) Analyses of the sea-keeping needsfor the 40-MWe demonstration barge provided designinformation for other configurations and sizes. For thebaseline 40-MWe plant the estimated maneuvering poweris 5–7 MWe. A total power complement of 8 MWe wasfound adequate to withstand the maximum winds, waves,and current requirements of the 100-year tropical storm,and to supply the maximum navigation requirements.Thrusters that are commercially available will be prac-tical to install and will provide a force of 15–22 kg/kW ofpower supplied (1994 data).

IX. POSITION CONTROL FOR FLOATINGOFF-SHORE OTEC POWER PLANTS

Dynamic positioning and mooring have been investigated.A dynamic positioning system must keep the position ofthe OTEC platform within a circle of about 100-m radiusat the surface to allow the power cable to remain attachedto the underwater equipment that holds and protects thecable in its trajectory to the onshore facility. Continuouspower delivery requires that standby diesel power must beavailable. An extensive study of the requirements for pro-posed plant configurations—spar, barge, ship, and semi-submersible—was done to provide quantitative data onthrust requirements and costs; for a moderate environmentwith surface current of 1.5 m/sec, and for an extreme sur-face current of 3 m/sec. The data tabulated to show cost asfunction of plant requirements included vectoring of thewater flow in the inlet and exhaust ducts. The cost effec-tiveness of this option is not known. Mooring of OTECpower plants will require cables reaching to depths greaterthan 1000 m that can hold the platform within a small sur-face area without interfering with the power transmissionlines. The systems must be designed to operate in stormsof 100-year severity or be able to detach the power cablesand maneuver out of harm’s way when devastating stormsare imminent. The mooring requirements for spar designsdiffer from the other configurations and have received sep-arate design studies. For a platform floating at the surface,a multileg mooring system is required that will hold thevessel within the required area, and will be able to orientthe vessel in storms so that waves do not strike it broadside.

Documentation is available for feasible designs usingmooring lines made of wire ropes and/or chains. Sparconfigurations are designed to place the main structureat a depth where forces from waves and currents are rel-atively small. This makes it possible to hold the plant inposition using a single mooring line. For this case therelative merits of multi-anchor-leg (MAL) mooring ver-

sus tension-anchor-leg (TAL) mooring were investigated.In the TAL concept the mooring line is attached betweenthe bottom of the CWP and an anchor emplacement on theocean floor. The CWP must be designed to withstand thetension loads. Tension-anchor-leg (TAL) mooring offersseveral attractive features: (i) There is no need for on-board anchor-line handling equipment. (ii) Design sim-plicity reduces handling and operating needs. (iii) Powercable handling will be simpler. (iv) Life-cycle costs will belower. There are two serious disadvantages: (i) The tensionloads for commercial sized plants far exceed commercialdesign experience. (ii) Failure of the mooring system willcause catastrophic failure of the entire assembly. Furtherstudies are needed to determine whether the TAL or theMAL system will be the optimum choice.

X. OPEN-CYCLE OTEC

A. Claude Cycle

In the Claude open-cycle OTEC system warm ocean wateris drawn into a vessel in which the pressure is maintainedbelow the boiling point of the warm water—roughly—1/30 atm (3.8 kPa). This causes the seawater to boil, pro-ducing vapor (steam) that becomes the working fluid. Thesteam passes through a turbine, producing power, andthen is drawn into a cold chamber at lower pressure—1/125 atm (0.8 kPa)—where the steam is absorbed in aspray of cold water drawn from the ocean depth. The con-densate is then pumped into the ocean. As an alternative,the vapor may be condensed by being directed on to acold surface, and collected as a source of fresh (desali-nated) water. Because the system pressures are so lowin the Claude cycle, the turbines, ducts, and containmentvessels must be very large. Therefore, open-cycle OTECsystems of the Claude type are impractical for ship or bargeinstallations. Open-cycle 1- to 10-MWe OTEC systems arefeasible on shore and could be attractive for installations attropical sites where cold water near shore can be used byOTEC to furnish fresh water, air conditioning, maricultureand cold water agriculture, in addition to electricity andfuel production. However, closed-cycle OTEC can offerthe same options and can be built with less difficulty andcost uncertainties.

A schematic diagram of the Claude cycle system isshown in Fig. 3. Warm seawater passes through a deaer-ator, then enters a flash evaporator, where it flows intothe evacuated vessel through small nozzles, and emergesas a vigorous spray. This induces efficient extraction ofheat of vaporization from the surface area of the evapo-rating droplets, which collect at the outlet with little cool-ing of the unvaporized water. The vapor passes throughthe turbine and is then efficiently condensed by the cold




water spray. As a result, the temperature differences be-tween the vapor and water in both vessels can be small.Experiments by Kreith and Bharathan and associates atthe National Renewable Energy Laboratory of DOE haveprovided quantitative data that allow optimization of theheat transfer coefficient as a function of the evaporatordesign. The results confirm that the cycle efficiency ofopen-cycle OTEC can be significantly higher than that ofclosed-cycle OTEC. The open-cycle process eliminatesthe need for large heat exchangers and also eliminates con-cerns of environmental risks posed by working fluids usedin the closed cycle. The technical feasibility of the processwas demonstrated at 210-kW scale in 1998 at the NaturalEnergy Laboratory of Hawaii (NELHA). The net powerwas 40-kWe. Further development has been discouragedby problems of construction and operation of the evacu-ated vessels and process equipment that are foreseen forlarge-scale systems.

Four other open-cycle systems have received someR & D support:

B. Panchal–Bell Cycle

In this cycle steam from flash evaporation of warm wateris condensed on the surfaces of a heat exchanger contain-ing liquid ammonia, which vaporizes to form the workingfluid for the closed cycle discussed in Section II. The con-densed steam is discharged to the ocean or may be saved asa source of fresh water. The Panchal–Bell cycle has a num-ber of potential advantages, compared to the Claude cycle:(a) The large low-pressure turbine is eliminated. This al-lows a major reduction in system volume and eases prob-lems of constructing a vacuum-tight enclosure. (b) Theheat transfer coefficient for the steam condensing on theheat exchanger surfaces is high, with benefits in reducedheat exchanger area and improved efficiency. (c) Seawatercontact with the heat exchanger surfaces in the evaporatoris avoided, eliminating possible contamination and corro-sion from the warm seawater. (d) Steam condensation oc-curs at a pressure of ∼1/40 atmosphere rather than ∼1/80atmosphere that is required in the Claude cycle. Thissignificantly reduces the power required for the vacuumpumps, which can be over 10% of the gross power in theClaude system. (e) The cycle may produce fresh water andpower in various ratios, depending on needs at a particularsite. Further R & D is needed to define a complete system.

C. Mist-Lift Cycle (Beck)

In this cycle warm seawater is introduced near the bottomof a long vertical tube with a vacuum pump at the top.Flash evaporation of the water produces a mixture of vaporand water droplets which are drawn to the top of the tube(at a theoretical height of 280 m), where they strike a

surface chilled by cold water, coalesce and flow back intothe ocean, passing through a hydraulic turbine to generatepower. Analysis and R & D have confirmed the validity ofthe concept. Experiments demonstrated effective couplingof vapor and mist to a height of 50 m. Further designstudies and experiments are required to define a completesystem.

D. Foam-Lift Cycle (Zener–Fetkovitch)

In this cycle, cold seawater is drawn through a hydraulicturbine, connected to an electric generator, into a long,vertical, partially evacuated, cylindrical vessel submergedin the tropical ocean. Compressed air containing a foamingagent is introduced at the bottom of the vessel. The foambubbles reduce the density of the water column, creatingan upward flow that causes the mixture to spout from thetop of the tube, where a foam-breaker removes the foamingagent. The dense, cold water then flows out into the warmlower-density surface water and sinks. Further R & D isrequired to determine whether the concept is technicallyfeasible and can be cost effective. (Although the foamingagents may have a concentration as low as 10 ppm theircost will be significant.)

E. Uehara Cycle

Professor Uehara of Saga University in Japan has inventeda new hybrid OTEC cycle that uses a mixture of ammoniaand water as the working fluid. Analysis shows that it willhave higher efficiency than the Kalina cycle (Uehara et al.,1998), which also uses an ammonia–water mixture as theworking fluid. Tests of the concept have been conductedat Saga University. A flow diagram for the test setup isshown in Fig. 6.

FIGURE 6 Schematic diagram of the Uehara OTEC cycle.




In operation, a warm mixture of ∼90 wt% ammoniain water is drawn into a vessel maintained at a pressurebelow the saturation pressure of the mixture, which causesboiling. The vapor passes through the turbines, drivingelectric generators, and is condensed in the cold water heatexchanger. The test results show that the cycle can result inhigher efficiency than closed-cycle systems. Further testsat larger scale will be needed to show what system costbenefits will be possible.

XI. CLOSED-CYCLE OTECDEVELOPMENT STATUS

During the period 1974–1985, engineering design studies,R & D on subsystems and components, and tests of OTECsystems, including at-sea and on land programs, wereconducted under DOE sponsorship and in French andJapanese OTEC programs that had substantial funding.Total funding for the programs exceeded 250 milliondollars (1980 $). In 1980 there was a consensus amongOTEC investigators that the technical feasibility of OTEC

Team leader System concept Site Product Energy delivery method

General Electric Co. (GE) Shelf-mounted Hawaii Electric power Power cable

Ocean Thermal Corp (OTC) Land-based Hawaii Electric power Power cable

Puerto Rico Electric Shelf mounted Puerto Rico Electric power Power cablePower Authority (PREPA)

Solar Ammonia Company Plantship Tropical oceans Ammonia Ammonia(SOLARAMCO)

Virgin Islands Water and Land-based Virgin Islands Electric power Power cable

Power Authority (VIWPA) Land-based Virgin Islands Fresh water Pipe line

Ocean Solar Energy Shelf-mounted Puerto Rico Electric power Power cableAssociation (OSEA)

had been proved and that the next step in the develop-ment of OTEC technology should be construction of apilot plant that would provide firm data on system per-formance and costs. A goal for OTEC commercializationwas established to demonstrate 100 MWe of OTEC by1985 and 500 MWe by 1990. DOE was assigned respon-sibility to plan and carry out a development program toachieve the goals. As the first step, funding was appropri-ated by the U.S. Congress for DOE to begin a phasedprogram to develop a pilot OTEC system that woulddemonstrate production of 40 MWe (net) of OTEC power.The objective was to provide engineering data for a broadrange of OTEC system concepts, including alternativesin plant installation (mounted on a floating platform, oron a fixed platform on or near shore), geographical loca-tion, facility needs, capital and operating cost trade-offs,

development time, and financing options. A program op-portunity notice (PON) was issued that invited industrialteams to submit proposals to conduct a program that wouldconstruct, deploy, and demonstrate operation of a totalOTEC system. The respondents were asked to include amanagement scheme and proposed cost sharing to accom-plish the program objectives. The program was to be donein six phases: conceptual design, preliminary engineer-ing design, engineering design, construction, deploymentand operation, and finally, transfer of ownership to indus-try. After the first phase, two or three proposals would befunded, followed by a single choice for the last phases(or perhaps two choices if funds were available). Fundswere appropriated by the Congress to begin the program.Six industrial teams submitted conceptual OTEC systemdesigns and development proposals for the first phase ofthe 40-MWe demonstration programs. A DOE committeewas established to review the proposals and select those tobe funded. A seventh proposal for a 10-MWe installationin Saipan was not accepted because it did not conform tothe PON guidelines. The conceptual design proposals thatwere submitted in response to the PON were:

To provide a firm technical base for evaluation of theproposed systems, DOE awarded a contract to JHU/APLto conduct preliminary design and testing which woulddocument base-line designs for 40-MWe OTEC floatingplants, containing all of the subsystems, that would cruise(graze) near the equator producing an energy storage prod-uct, or deliver electric power via underwater cable fromOTEC plants moored near shore.

In the first phase of the PON program, funds were ap-propriated with the intent to support a variety of concep-tual designs of OTEC systems, by the six industrial teams.However, with the change in program management to theReagan administration in 1981, only the proposals by GEand Ocean Thermal Corporation (OTC) were selected forconceptual design funding. At completion of this phase,OTC was awarded the only second-phase contract to do




preliminary engineering design of an OTEC plant thatwould be installed on an artificial island near the warm wa-ter out-flow from an oil-fired power plant of the HawaiianElectric Corporation (HECO). The proposal envisionedmixing near-shore ocean surface water with the HECOplant outflow to raise the temperature of the warm waterinflow to the OTEC plant from about 25 to 29◦C, with animprovement in cycle efficiency.

Volumes of engineering data on OTEC subsystemand components requirements, performance, and costswere developed by the proposal teams. However, whenOTEC support was discontinued in 1982 the proposalmaterial was destroyed because it contained proprietaryinformation that DOE could not protect. Part of the ma-terial was made available to JHU/APL and is the sourceof the following discussion of the U.S. OTEC engineeringstatus. Information on French and Japanese programs istaken from Avery and Wu (1994) and Kalina (1987).

Plans were announced in 1997 by the Indian NationalInstitute of Ocean Technology (NIOT) for installation ofa 1-MWe floating OTEC plant at a site near Tuticorin,South India. The project is being conducted jointly with ateam from Saga University in Japan, headed by ProfessorUehara. Operation of the 1-MWe demonstration vesselwas scheduled to begin in 2001.

A. Standard Equipment Available for OTEC

OTEC can directly use technology in commercial use inthe shipbuilding industry, in the oil industry, and in powergeneration. Firm costs are available from these sourcesfor equipment that represents 40–60% of the total OTECsystem costs.

1. Shipyards

Facilities in the United States suitable for OTEC construc-tion include: Newport News Shipbuilding Co., AvondaleIndustries, Inc., and Concrete Technology Corporation.Concrete Technology Corporation in 1976 built a 146-m-long concrete barge (the ARCO barge) of roughly thesame dimensions as those proposed for the OTEC 40-Mwepilot plant. The design and construction of that vessel pro-vided detailed information on engineering requirementsand costs that were used in the base-line design of the40-MWe OTEC barge. Shipyards in Germany, Spain,Portugal, Japan, Singapore, and Korea have built super-tankers, and their facilities would be readily adaptable forcommercial construction of OTEC plantships.

2. Construction Procedure for Floating Systems

A complete OTEC system will be constructed in stages.The hull is laid down in a dry dock and then floated to

a nearby pier where water depth is 20 m or more. Themajor equipment items are installed. The vessel is thentowed to a tropical site where CWP construction facilitiesare available and nearby water depth exceeds 1000 m.CWPs are constructed, transported to the floating off-shore OTEC vessel, and installed. The complete assemblyis then towed, or navigated using onboard power, to theoperating site.

B. OTEC Land- and Shelf-Based Installations

These are site specific. Details of proposed systems aregiven in Section D.

C. Pilot-Scale OTEC Systems Tests

1. U.S. Mini-OTEC Program

A complete OTEC system, including a 670-m CWP, wasconstructed, deployed, and tested in the Mini-OTEC pro-gram, in 1978–1979. The program was conducted with pri-vate funding by an industrial team headed by the LockheedMissiles and Space Division. Mini-OTEC was constructedin 28 months, using off-the-shelf components. Net powerof 18 kW was produced, in close agreement with the pre-dicted performance. The Mini-OTEC program demon-strated OTEC floating system feasibility in 4 months ofat-sea operation, including first-time production at sea ofnet OTEC power.

2. OTEC-1 Program and Accomplishments

This program was conducted by DOE to demonstrateOTEC heat exchanger operation at 35-MWt scale in anat-sea environment. A moth-balled Navy T2 tanker wasretrofitted to house the OTEC system components. A sim-ulated OTEC power system was installed in which theturbine-generator was replaced by a throttle valve that sim-ulated the pressure drop through the turbine. The equip-ment included a 35-MWt shell and tube heat exchanger, a700-m-long, 2-m-diameter plastic CWP and pumps. Pro-vision was made for testing biofouling control methodssuch as Amertap and chlorine injection. Diesel generatorsprovided electric power to operate pumps and auxiliaries.The CWP was assembled on shore then towed to a siteroughly 10 miles off-shore from Kailua-Kona, Hawaii,where it was upended and inserted into a gimbal on theship. The gimbal allowed movement through an angle of30◦C.

Significant accomplishments demonstrated during the4-month at-sea operation were (a) The system performedwell with a maximum wave height of 4 m, maximum sub-surface currents of 3 knots, wind gusts up to 45 knots,and ship roll angle of 18◦. (b) The heat exchanger cy-cle operated in close accord with predicted performance.




(c) Deployment and operation of the CWP were success-ful. The pipe survived a sea state that caused the pipe tomove to a 30◦ angle at the gimbal. Detachment of the shipfrom the mooring at this point caused the pipe angle todrop to 10◦. This indicated that survival of floating OTECsystems in extreme storm conditions would involve sig-nificantly lower risk in the grazing mode than in mooredoperation. (d) Biofouling control was accomplished withinjection of 60 parts per billion of chlorine, 1 hr perday, well below EPA requirements for drinking water. (e)The test provided assurance that the marine operationalrequirements were well understood. The test programinvolved no surprises.

3. Japanese Shore-Based 100-kWOTEC Pilot Plant at Nauru

In cooperation with the government of the Republic ofNauru, engineers of the Tokyo Electric Power Co. andthe Toshiba Corporation in 1980 constructed a 100-kW(gross) OTEC plant at a site on the island of Nauru, whichis on the equator 40◦ east of the Philippine Islands. Power-generating operation was performed from October 1981to December 1981. Heat-loop operation was continued toJuly 1982 (Mitsui et al., 1983; Ito and Seya, 1985). TheNauru test was designed to provide accurate experimen-tal data on the performance of the complete power cycleas well as demonstrate total pilot-plant construction andoperation. The goals were all successfully accomplished.The plant provided the first demonstration of land-basedOTEC net power generation and established a record fortotal net power production, at 31.5 kW. It also was thefirst demonstration of OTEC power connection to a utilitygrid.

4. Japanese Mini-OTEC Program

A Mini-OTEC plant was constructed for at-sea tests offthe coast of Shimane Prefecture, as an initial phase ofthe Japanese program directed to demonstration of a100-MWe moored OTEC system. Deployment took placeon October 11, 1979. The work was done by a team headedby Professor Uehara of Saga University. The tests includeddeployment of a steel CWP, 190 m long, made of 12-m-long tubular sections of 26.7-cm OD and 0.45-cm wallthickness, joined by flanges. The pipe was supported bya float and held in place by a steel cable attached to thepipe and anchored at a depth of 300 m. The OTEC powerplant (with plate-fin heat exchangers) and water pumpswere mounted on the research vessel, which was attachedto the float. A flexible pipe carried the cold water from thefloat to the vessel. The program demonstrated successfuldeployment and operation of the system, including heattransfer measurements.

D. Conceptual Designs of OTEC Systems

Responses to the DOE program opportunity notice (PON)are described here. Only the GE and Ocean Thermal Corp.proposals were funded for the Conceptional Design phase.

1. General Electric (GE) 40-MWeShelf-Mounted Hawaii Plant

GE proposed a tower-mounted OTEC plant which wouldbe installed 2200 m offshore at a depth of 100 m, atKahe Point, Oahu, Hawaii. Power produced at the OTECplatform is carried by underwater cable to the HawaiianElectric Company (HECO) plant for distribution to itscustomers.

2. Puerto Rico Electric Power Authority (PREPA)

PREPA proposed a tower-mounted 40-MWe (net) OTECplant to be sited at the edge of the continental shelf at75-m depth approximately 200 m offshore from PuntaTuna, Puerto Rico. Electric power is transmitted by un-derwater cable to the PREPA power distribution system.Cold water is drawn from a depth of 1000 m at a distanceof approximately 1500 m from the tower.

3. SOLARAMCO Concept of OTEC40-MWe (net) Grazing Ammonia Plantship

The SOLARAMCO group proposed a floating 40-MWe(net) OTEC ammonia plantship system, designed to grazein a region south of Hawaii where a �T of 22–23◦C wouldbe available. The plantship would use the net OTEC elec-tric power for water electrolysis to produce hydrogen foron-board ammonia synthesis. The proposal included pro-visions for testing four 10-MWe power modules equippedwith different heat exchanger types.

4. OSEA 100-MWe OTEC Power Plant

A consortium named OSEA, that included SSP (SeaSolar Power, Inc.), GE, and General Dynamics, proposeda 40-MWe floating OTEC plant of the SSP design thatwould supply power to the Puerto Rico Water and PowerAuthority. (PREPA). The technical details are proprietary.The following discussion is based on the published re-ports of the SSP 100-MWe OTEC plant conceptual de-signs and analyses. Features include (a) design of thepower plant components for submerged operation so thatthe major power plant components can be placed in theocean outside the hull; (b) turbine-driven pumps for thewarm and cold water, which eliminate efficiency losses inconventional designs that result from conversion of turbinepower to electric power for operation of electrically driven




pump motors; (c) a new heat exchanger design with pro-jected heat transfer rate four times that of conventionalheat exchanger designs; (d) a four-stage vapor cycle toimprove the average temperature difference between thewater flows and working medium flow. Placing the evap-orators 25 m below the condensers allows the condensedliquid, R-22, to flow by gravity to the evaporators, elim-inating the need for a working fluid liquid pump and theassociated power losses; (f) a “stockade” construction ofthe CWP designed to permit CWP assembly on the OTECplatform from sections of steel pipe that would be easy totransport. With appropriate valves and seals, the hollowpipes could be used to control the buoyancy of the CWP.Details of the construction, deployment, and operation arenot available.

5. 200-MWe (nom) OTEC Methanol Plantship

As a follow-up to the preliminary design of the APL40-MWe OTEC baseline ammonia plantship, DOE sup-ported, a conceptual design study of a 200-MWe (nom)OTEC plantship that would use electrolytic hydrogen, andoxygen reacting with coal, for on-board methanol synthe-sis. Pulverized coal for the process is transported to theplantship by conventional coal colliers. The methanol isaccumulated in on-board tanks for approximately 1 monthand then shipped to continental ports in conventionaltankers. The process efficiently uses both oxygen and hy-drogen produced by water electrolysis in combination withlow-cost coal to produce a low-pollution liquid vehiclefuel that can replace gasoline for transportation at compet-itive costs. The platform configuration is based on scale-upof the JHU/APL 40-MWe plantship to 200 MWe to providepower and space for a 1750-tonne/day methanol plant.Layout diagrams are shown in Figs. 8a,b. The methanolplant is placed at the center of the platform with OTECpower modules at either end. Power for sea-keeping is pro-vided by rotatable thrusters beneath the hull. The powerplant includes four power modules at each end of thebarge. Each module includes two evaporators and twocondensers, supplied with seawater by gravity flow fromoverhead ponds. CWPs, 7–10 m diameter made of FRPor lightweight concrete, are supported by a universal jointor ball-and-socket mounting. Energy transfer is done byproduction of methanol on the OTEC plantship with sub-sequent shipment to world ports. The system will produce1750 tonnes/day of methanol.

6. Conceptual Designs of Japanese100-MWe Floating OTEC Plants

During the period 1974–1977, a program, named the“Sunshine Project,” was established in Japan to explore

FIGURE 7a Sketch of JHU/APL 40-MWe baseline OTEC ammo-nia plantship [From George, J. F., and Richards, D. (1980). JohnsHopkins Appl. Phys. Lab. SR-80-A&B.]

the feasibility and cost of supplying power- or energy-intensive products to Japan via OTEC. The program be-gan with a first-phase study of a 1-MWe, land-based plantand proceeded to a study of a 100-MWe floating plant.The design included platform structure, station keeping,riser cable design, and heat exchanger improvements. TheOTEC system employs moored floating surface ships orsubmerged discs. The power plant uses ammonia as theworking fluid. Plate heat exchangers are incorporated in25-MWe power modules. An estimate of the total OTECpower that could be produced from the thermal resourcesof the sea near Japan indicated a potential output of3.75 × 105 GWh/year from an area 1◦ wide in longitudeand 1◦ in latitude. This output exceeds the annual electricpower consumption in Japan.

7. Conceptual Designs of French5-MWe OTEC Plants at Tahiti

A feasibility study of shore-based and floating OTECplants to be sited in Tahiti was conducted during1978–1980 under sponsorship of the French Centre pourl’Exploitation des Oceans (CNEXO). The program in-cluded detailed site studies, evaluation of closed- andopen-cycle options, methods of constructing and deploy-ing the cold water pipe and estimates of costs of favoredalternatives. Preliminary surveys led to selection of a site atthe port of Papeete for the projected plant. The design doesnot call for a platform or containment building. Insteadthe power plant components are mounted separately. Plantlayouts for closed-cycle and open-cycle systems were de-veloped. A pond adjoining the power plant installationreceives nutrient-rich water leaving the condensers that




FIGURE 7b Subsystems layout of baseline 40-MWe OTEC barge. [From George, J. F., and Richards, D. (1980).Johns Hopkins Appl. Phys. Lab. SR-80-A&B.]

are to be used for mariculture programs. The water efflu-ent streams are finally directed to a lagoon from whichthe water flows to the ocean. Conceptual designs of bothclosed- and open-cycle OTEC installations were prepared,based on selection of the most attractive of a wide rangeof alternative designs. Investment comparisons indicatedthat the costs of the two systems would be within 5% ofeach other. The closed-cycle plant uses horizontal shelland tube heat exchangers with smooth titanium tubes inboth evaporators and condensers. Ammonia was chosenas the working fluid. The open-cycle design uses spoutevaporators and direct-contact condensation of the steamwith the cold sea water. A variety of pipe designs wereevaluated. The preferred overall design that emerged in-cluded a double-walled CWP of FRP, enclosed in a 15-m-deep concrete pipe entrenched in the shallow water zoneat the shore, extending approximately 100 m to a pointwhere the sea floor drops sharply. The deep-water pipe issupported above the sea floor in a trajectory defined by“hawses” anchored by piles in the sea floor and supportedby buoys designed to float at 50 m depth. The neutrallybuoyant CWP, 3 m in diameter, assembled on shore as de-ployment proceeds, is deployed by pulling it through thehawses. Direct transmission of electric power, prepara-tion of fresh water, and mariculture could be combined tomake operation of the 5-MWe Tahiti plant of commercialinterest.

8. Conceptual Design of Taiwan 50-MWe(net) Shore-Based OTEC Plants

In 1984 the Taiwan Power Company initiated a programto determine the feasibility and to prepare conceptual de-signs of shore-based 50-MWe OTEC plants in Taiwan.Three sites on the east coast of Taiwan were selected forevaluation. Studies led to selection of shore-based plantsat Ho-Ping and Chang-Yuan as preferred sites for Concep-tual Design. It was decided that the platform (containmentstructure) should be constructed in a shipyard, where fa-cilities for installing heavy equipment would be available.The platform would then be floated and towed as a unitto the plant site on the beach. Here it would be loweredby flooding the sumps to rest on a prepared bottom foun-dation behind an appropriate breakwater. There are four12.5-MWe (net) power modules that use ammonia as theworking fluid in horizontal shell-and-tube heat exchang-ers of conventional overall design. Titanium tubes are em-ployed. The design permits use of either chlorine or Amer-tap for biofouling control. Warm and cold water enter theplatform through pipes that pass through the breakwaterand emerge into sumps below the platform floor. Wateris forced through the heat exchangers by vertical pumpsthat draw water from the sumps. The heat exchanger efflu-ents discharge into a common sump from which the waterflows out by gravity. A single CWP of 10 m diameter,




made of lightweight concrete, provides cold water forall four power modules. Studies of the variation of netpower versus water flow velocity showed that maximumnet power of 50 MWe would be possible with a 10-m pipediameter. Consideration of various methods of pipe con-struction and deployment led to the conclusion that thepipe should be constructed in sections, 10 m in diameterand 15 m long, with elastomeric bell and spigot configu-rations at the ends. The pipe sections would be barged tothe plant site where they would be joined end to end toform the cold water pipe. Installation of the power planton the beach allows the generated power to be transferreddirectly to the onshore utility.

E. Preliminary Design of OTEC Plants

1. Baseline 40-MWe (net) Moored andGrazing OTEC Plants

In 1977, The Applied Physics Laboratory of the JohnsHopkins University (JHU/APL) was awarded a contract byDOE to prepare baseline preliminary engineering designsof moored and grazing 40-MWe (net) OTEC plants thatwould furnish basic data on systems requirements, perfor-mance, and costs. The information would be used to judgethe responses to the PON. A barge-type hull made of post-tensioned concrete was selected for the baseline OTECgrazing and moored system designs because of suitabilityfor both missions, ease of installation and replacement ofsubsystems and components, minimal cost, compatibilitywith aluminum heat exchangers, and long-term durability.Provisions are made for alternative types of heat exchang-ers, water ducting including the CWP, pumps, energy de-livery systems, station-keeping systems, support facilities,and personnel accommodations. The configuration andsusbsystems layout are shown in Figs. 7a,b. The platformlength is 135 m. Analysis of the maximum loads that wouldbe imposed by the 100-year storm defined the structuraldesign. The analysis included consideration of the sur-vival requirements for grazing plantships stationed in theequatorial Atlantic Ocean (significant wave height ((Hs)8 m, period (To) 18.0 sec)) and for moored plants stationednear Puerto Rico ((Hs) 10.9 m, (To) 13.1 sec) or near theisland of Oahu in Hawaii ((Hs) 8.4 m., (To) 11.7 sec). TheOTEC barge accommodates alternative 20-MWe powersystems of designs proposed by JHU/APL and LockheedSpace Sysytems. The dimensions and weight of the twosystems are nearly the same. Thus, the hull configurationis suitable for 40-MWe (net) power systems of either de-sign. Other compact power systems such as the Trane orAlfa-Laval designs could also be installed without majorchanges in the hull dimensions.

The APL power system was designed with two majorobjectives: (a) to minimize HX cost by using low cost, roll-

welded aluminum tubing in a compact folded-tube con-figuration with internal ammonia flow. This eliminates thebulky containment structure needed in conventional shelland tube HX designs, thereby reducing packaging spaceand cost; (b) to minimize platform cost by making thewater ducting structures compact, integral, load-bearingelements of the hull structure. The power system employstwo 5-MWe (net) evaporator and condenser modules com-bined with a single turbine to form a complete 10-MWe(net) power module. Each 5-MWe (net) unit is housed in aseparate bay with vertical sides that form an integral part ofthe hull structure. Two 10-MWe (net) modules, each hav-ing two evaporators and two condensers, are located at thefore and aft ends of the baseline platform. The practicalityof fabrication, assembly, installation, and deployment ofheat exchangers of the folded-tube type was the subjectof a separate design study conducted by the Trane Corpo-ration, which verified the suitability of the JHU/APL HXdesign for low-cost, mass production. The platform spaceand HX bays allotted to the APL design are well suited topackaging the Lockheed module, thus no modifications ofthe baseline hull are indicated. The plate heat exchangersare mounted in two tiers in a heat exchanger bay. Seawaterfrom the overhead pond enters the heat exchangers at thesides, flows horizontally across the heat exchanger plates,and then exits at the bottom of the HX bay. Liquid ammo-nia enters the evaporator from below and flows verticallyupward as it vaporizes. Vapor leaving the turbines entersthe condensers at the top and flows vertically downward tothe sump from which it is pumped to the evaporators. Anal-ysis by Lockheed showed that this cross-flow arrangementwas superior to one using parallel flow.

Cold water is supplied to the OTEC power systemsthrough a 9.1-m-diameter, 930-m-long segmented con-crete pipe, supported by a ball-and-socket structure incor-porated in the hull at approximately the center of gravity.The joint allows the pipe to pivot freely through an angleof 18◦ to the vertical. It also allows the pipe to rotate aboutthe vertical axis. The dynamics of the coupled system, an-alyzed as discussed in Section IV, show that the coupledpipe-hull system will survive the most extreme conditionsof the 100-year-storm in grazing operation at the tropicalAtlantic site, with a good safety margin, and in mooredoperation at the Hawaii site. The analysis predicts that thedesign is marginal for the extreme hurricane condition atthe Puerto Rico site, where the combination of extremevalues of surface currents and waves leads to a predictedmaximum pipe angle of 19.7◦ (sum of extreme values dueto hull deflections and surface currents). The calculationdoes not include mitigating effects of mooring forces onthe platform motions or potential benefits of bow-fairingof the rectangular platform. Further study is required toindicate detail changes in design that might be required




for moored operation at this site. Warm water enters theevaporators through screens mounted at the sides of thehull at a mean depth of 16.6 m. The screen area is about125 m2 for each 10-MWe evaporator module. In the graz-ing mode of operation, the warm and cold water effluentsfrom the heat exchangers are discharged at the bottom ofthe barge, where the higher density of the effluents thanthose of the surrounding water causes them to sink. Sincethe grazing barge is always moving into fresh water, thereis no recirculation of the discharged water into the warmwater inlets. The moored baseline plant is designed withdischarge pipes to carry the effluent to a depth below thethermocline to prevent reingestion and possible impactson the near-shore environment. Eight pipes 4.6-m in di-ameter and 76-m long are used to carry the mixed effluentbelow the thermocline, where it is discharged and allowedto sink. A detailed structural analysis of these pipes wasnot funded but will be necessary if they are to be retainedin the design. The OTEC-1 experiment indicated that thesepipes may be unnecessary because the mixed plume de-scends rapidly and may not pose any reingestion problem.However, termination of the program did not permit a con-clusion to be reached.

The OTEC on-board electric power generation and dis-tribution system is based on the use of standard com-mercial equipment for primary and secondary applica-tions and is designed to comply with the requirements ofIEEE standard No. 45 and relevant regulations of the U.S.Coast Guard, ABS, NEMA, NFPA, and UL. There arefour 16-MWe (gross) ammonia turbine generators, eachhaving its own set of auxiliaries and switch gear. Ammo-nia storage, liquid nitrogen, circulating cooling water, andgenerator loading systems are used in common. All of thegenerators may be used in parallel, and in normal operationall power will come from them. During start up or whenthe OTEC plant is shut down during storms, reserve poweris provided by diesel-generator sets. Tie feeders betweenthe OTEC generators and the reserve power units allowthis interchange. The on-board power utilization systemsdiffer for the grazing and moored OTEC options; how-ever, the systems are the same up to the final phase. Inthe grazing plant, step-down/DC rectifiers are installed tofurnish DC current to the electrolyzer cells. In the mooredplant, transformers step up the voltage to 115 kV for ca-ble transmission to shore. In OTEC plantships, electricpower generated on board is used for water electrolysis,which provides efficient conversion of electrical energy tochemical energy stored in hydrogen-based fuels that canbe transported to land sites.

The baseline 40-MWe (net) OTEC plantship uses thepower produced by the turbo-generator to produce ammo-nia (NH3). The on-board ammonia plant includes waterelectrolyzers to generate gaseous hydrogen (along with

gaseous oxygen that is vented to the air), an air separationplant to produce pure nitrogen, a catalytic converter tocombine the hydrogen and nitrogen to form ammonia, am-monia liquefaction and storage equipment, and facilitiesfor transferring the liquid ammonia to tankers for shipmentto world ports. Details of the ammonia synthesis processare described in Section IV.A.3. Commercial ammoniaplants typically produce 1000 tons/day (909 tonnes/day)of liquid ammonia. The output of the 46-MWe (net) base-line plantship would be 125 tons/day.

The feasibility and practicality of installing an ammo-nia plant on an OTEC platform were investigated in detailin the baseline study, with favorable results, includingpossible effects of ship motions on the process opera-tions. Smaller and less efficient plant components wouldbe used in the baseline vessel in comparison with commer-cial plants. The OTEC ammonia plant would be similarto the commercial ammonia plant designs but would besimpler because hydrogen would be produced by waterelectrolysis rather than by the expensive steam-reformingof methane, which also generates carbon dioxide. A flowdiagram of the process is shown in Fig. 8. An installationdiagram of a 113-tonnes/day ammonia plant on the OTECplatform is shown in Fig. 9.

It was recognized early in the OTEC program that atfavorable sites moored OTEC plants could deliver OTECpower directly to onshore facilities via underwater powercables (Winer, 1977). The requirements for this mode ofOTEC operation were also determined in the 40-MWe(net) baseline design study. The site selected for detailedstudy was near Punta Tuna, Puerto Rico. Power transferrequires on-board equipment to convert the OTEC bus-baroutput to suitable voltage and frequency for transmissionthrough the selected power cable, and equipment to sup-port the power cable that must connect to the OTEC plat-form through a flexible junction. A proposed arrangementfor the aft cable installation on the platform is shown inFig. 10. Experiments with power cables under simulatedOTEC conditions indicated satisfactory durability couldbe achieved, however, a complete design of the power ca-ble system had not been completed when the program wasterminated. A design of the electrical system, which satis-fies the total requirements of both moored and and grazingOTEC plants, was produced. The principal features of theJHU/APL 40-MWe (nom) OTEC baseline plant designsare listed in Table IV.

2. Preliminary Design of Shelf-Mounted OTCOTEC Power Plant for Hawaii

At the completion of the DOE Phase 1 conceptual designcompetition, a contract was awarded to the Ocean Ther-mal Corporation (OTC) for preparation of a preliminary




FIGURE 8a Plan view of 1750 mt/d methanol plantship [BARDI 1982. Coal-to-methanol study using OTEC technol-ogy. For Johns Hopkins University Applied Physics Laboratory. Brown and Root Development, Inc., Houston, Texas.]

design of their proposed shelf-mounted plant. The plantwould be sited on the continental shelf 600 m offshorefrom the Hawaiian Electric Company’s (HECO) oil-firedpower plant at Kahe Point on the island of Oahu. In theOTC concept, shown in Fig. 11, the warm water outflow

FIGURE 8b Side view of OTEC methanol plantship [BARDI 1982. Coal-to-methanol study using OTEC technology.For Johns Hopkins University Applied Physics Laboratory. Brown and Root Development, Inc., Houston, Texas.]

from the HECO plant is mixed with the warm seawaterinput to the OTEC evaporator, thereby raising the temper-ature of the OTEC inlet water from 25.6 to 27.9◦C, witha significant increase in OTEC performance. The goal ofthe OTC contract was to develop, with government cost




sharing, an engineering and economic evaluation of anOTEC power plant design that would provide a firm basisfor detailed engineering and construction of a shore-based40-MWe demonstration OTEC electric plant. In the con-duct of the contract, the following topics were evaluatedin some depth:

OTEC system performance and Preliminary design systemoperational requirements engineering

Codes, standards, and OTEC plant system characteristicsrecommended practices OTEC plant system overall layout

HECO requirements Land-based containment systemPossible impact on military (LCBS) design

operations near OAHU Power system designAesthetic considerations Water ducting designSite-specific design conditions Asset acquisitionSite description Project acheduleWind, wave design criteria OTEC plant system deploymentTsunamis LBCS deploymentTides Inspection, maintenance and repairCurrents Safety analysisSeawater temperatures Environmental monitoring andHECO condenser outflow assessment

conditions Cost and commercializationSeismicity prospects

Climatology Risk assessment

Biofouling and corrosion Pipeline deployment

The LBCS comprises a concrete building 71.8 m wide(236.5 ft), 93.3 m long (306 ft), and 33.5 m high (110 ft),10 m above sea level mounted on the sea bottom at a

FIGURE 9 Flow diagram for OTEC ammonia plant. [From George, J. F., and Richards, D. (1980). Johns HopkinsAppl. Phys. Lab. SR-80-A&B.]

depth of 22.2 m (73 ft), at 550 m (1800 ft) offshore. TheLBCS houses the OTEC power plant, water and ammoniapumps, biofouling control equipment, and power conver-sion equipment. A breakwater structure extending fromthe sea floor to just above sea level is attached on the sea-ward end. Design of the CWP system required an exten-sive study of the physical and hydrological features of theKahe point site. The studies included detailed bathymetricsurveys with a resolution of 50 m covering approximately5.5 km2 bordering Kahe Point. Bottom core samples andsubsurface data representative of the area were also ac-quired. With this information it was possible to define thepipe structural requirements and to lay out a pipe trajectorythat would minimize risks in deployment and long-termoperation, and would be the most cost effective. In the fi-nal preliminary design of the CWP, water enters the LBCSthrough a bottom mounted pipe 3660 m long that drawswater from an offshore depth of 670 m (2200 ft). The CWPis divided into two major segments. The seaward sectionof the pipe is of FRP. It is 2560 m long (8400 ft), of 6 mdiameter (19.8 ft), and 0.072-m wall thickness (2.83 in.).It is composed of segments 75 m to 90 m long, joined byflexible bellows, and supported on a steel support struc-ture that rests on the bottom. This structure protects thepipe from bottom erosion, relieves tension loads on theFRP, provides weight to ensure stability against hydrody-namic loads, and allows flexibility to accommodate seis-mic loads, bottom scour, and differential settling. At thetop of the escarpment, the FRP pipe connects to a triple-pipe concrete structure that combines a pipe to conduct the




FIGURE 10a Schematic diagram of 40-MWe OTEC ammoniaplant installation. [From George, J. F., and Richards, D. (1980).Johns Hopkins Appl. Phys. Lab. SR-80-A&B.]

cold water to the LBCS, with two parallel pipes that carrythe mixed effluent (flowing in the opposite direction) fromthe OTEC plant to the discharge point. A profile of thissection and the concrete triple-pipe structure is shown inFig. 11. Near-surface seawater enters the LBCS through astructure on the landward side where it is mixed with warmwater conducted from the HECO condenser discharge. Af-ter passing through the OTEC power system, the exhauststreams from the evaporators and condensers are mixedand then conducted through the bottom-mounted concreteducting structure to the edge of the escarpment where thecombined flow is discharged at a depth of approximately115 m (380 ft). All support facilities not directly involvedin plant power production are sited onshore. These includeoffices, laboratories, maintenance facilities, ammonia andequipment storage, and related equipment. The principalfeatures of the OTC design are listed in Table V. Costinformation is listed in Section XII.

F. Conclusions about Closed-CycleOTEC Development Status

The review in this chapter of the OTEC programs con-ducted in the United States and abroad shows that the tech-nology required for high-confidence pilot plant demon-

FIGURE 10b Schematic diagram of power cable equipment in-stallation on 40-MWe moored OTEC plant. [From George J. F.,and Richards, D. (1980). Johns Hopkins Appl. Phys. Lab. SR-80-A&B.]

stration of sea-based and land-based OTEC systems wasavailable in 1984. Satisfactory solutions to feasibilityquestions were demonstrated. It should be noted that acomplete OTEC plantship assembly will incorporate com-mercially available turbines, generators, pumps, powerdistribution equipment, control equipment for power sys-tems and sea-keeping, hotel and shop facilities, and mis-cellaneous items. Only the construction of cold water pipesand advanced types of heat exchangers will involve newtechnology. These items will contribute about 20% of thetotal estimated cost.

Commercial applications of OTEC have been proposedin three principal categories. The first includes OTECplantships that would generate a 50- to 400-MWe (net) ofonboard electric power. The need to minimize plant sizemakes it mandatory to use closed-cycle OTEC for theseapplications. The second category includes land-based orshelf-mounted plants designed to supply power in the 50-to 400-MWe range to municipal utilities. Closed-cyclesystems would be suitable. The third category comprisessmall (5- to 20-MWe) land-based or shelf-mounted OTECplants designed for island (and some other) applicationswhere electric power generation, mariculture, fresh wa-ter production, supply of cold water for air-conditioningsystems, cold water agriculture and fuel production couldbe combined to offer an economically attractive OTEC




TABLE IV JHU/APL Preliminary Design of Baseline 40-MWe (nom) OTEC Plants

Hull

Rectangular barge

Dimensions in (ft): L 135 (443.5), W 42.7 (140), D 31.4 (103)

Hull material: post-tensioned concrete

Power Plant

Closed Rankine cycle; ammonia working fluid

Folded tube HXs of JHU/APL design with nested 0.072-m (3-in.) tubes, or plate HXs of Lockheed design using titanium plates

Internal ammonia flow; 4 10-MWe (net) modules

Nominal operating characteristics

Nominal �T (◦C)

Annual average 23.65

Seasonal �T (◦C) (ATL-1): Feb. 23.6, May 24.6, Aug. 23.4, Nov. 23.0

JHU/APL power Lockheed powermodules modules

Power [MWE (nom)](optimized for ATL-1)

Gross turbo-gen output 52.0 52.9

Cold water pump 5.76 9.66 (both pumps)

Warm water pump 4.08

Ammonia pumps 1.51 1.18

Noncondensible purge 0.16

Ship services 0.17 2.07

Net busbar power 40.3 40.0

JHU/APL Lockheed

Evap. Cond. Evap. Cond.

Heat load (MWt) 1,747 1,689 1,680 1,629

Water flow (kg/sec) 132,300 136,720 105,600 102,900

NH3 turbine flow (kg/sec) 2,824 2,824 1,722 1,722

U0 (W/m2 K) 2,537 2,463 3,259 3,118

Foul. coeff. (W/m2 K) 3.5 × 105 3.5 × 105 5 × 105 5 × 105

H2O in/out temp. (◦C) 27.89/25.27 4.56/7.46 27.78/23.81 4.44/8.36

NH3 in/outlet temp. (◦C) 23.94/21.94 9.50/10.00 13.37/21.71 10.37/10.47

Seawater �p (m) 2.58 2.95 3.21 3.44

Inlet duct 0.51 0.88 0.51 0.88

HX 2.07 2.07 2.70 2.56

Seawater velocity (m/sec)

Inlet duct 0.3 2.0 n.a. n.a.

HX 0.76 0.76 n.a. n.a.

Seawater flow [m3/s/MWe (net)] 3.24 3.33 2.59 2.51

system despite the relatively high cost of power from smallOTEC installations. Open-cycle OTEC plants could be thepreferred choice at some sites for the third category. Sincethe importance of OTEC as a renewable energy sourcewill depend on its ability to furnish significant quantitiesof fuels that will be economically and environmentallyattractive, the major emphasis in the following discus-sion of OTEC economics is placed on the OTEC plantshipoptions.

The estimated investment costs of installed completeOTEC systems, measured in dollars per kilowatt of

net OTEC electric power generated, differ significantlyamong the three categories. Large cost benefits are gainedby siting OTEC systems where �T is a maximum, byscaling to optimum size, by large volume production ofsubsystems and components, and by integrated systemdesign. Small land-based systems are relatively expen-sive in dollars per kilowatt of power generated but thecost can be offset in part by using the cold water forair-conditioning, mariculture, cold water agriculture, andfresh water production. Experiments at reasonable scaledemonstrated practical solutions for all of the technical




FIGURE 11 Artist’s sketch of the OTC 50-MWe OTEC powerplant at Kea-hole Point, Hawaii. [From Ocean Thermal Corporation(1984). Development of the Land-Based Containment System forthe 40 Megawatt Ocean Thermal Corporation OTEC Plant. MakaiOcean Engineering, Inc., Honolulu, Hawaii.]

problems that were identified as major barriers to OTECcommercial operation in early reviews of OTEC. Theseinclude heat exchanger efficiency, heat exchanger biofoul-ing, aluminum heat exchanger durability, OTEC heat en-gine performance, cold water pipe fabrication, cold waterpipe deployment, OTEC plant and plantship ocean dura-bility and survival in tropical 100-year-storm conditions,and OTEC environmental impacts.

XII. OTEC CLOSED-CYCLE SYSTEMSCOST EVALUATION

A. Cost Uncertainties and Perceived Riskversus Engineering Status

Reliable estimates of the costs of complete systems canbe made by dividing the total cost into categories forwhich the cost uncertainty of subsystems developmentis low, moderate, or high. A weighted average cost un-certainty can then be calculated for the total cost of thecomplete system. For OTEC cost estimation, a low uncer-tainty of −/+10–20% was assigned to cost estimates forsystem components or construction technologies that arein standard industrial use and for which firm quotes wereobtained. The fraction of the total cost in this categoryranges from 45% of the total for the ammonia plantship to70% for the methanol plantship. Moderate uncertainty of−/+20–35% was assigned to components or constructiontechnologies that are commercially available at a smallerscale or under different working conditions but must bemodified for OTEC use. The corresponding fractions ofthe total cost in this category were 41 and 23% for thetwo cases. High-cost uncertainty of −35 to +100% was

TABLE V OTC Design of a Shore-Mounted 40-MWe OTECPower Plant

LBCS dimensions m (ft)

Above water: L 93 (306), W 71 (234), H 11.3 (37)

Under water: L 116 (380), W 71 (234), H 22.2 (73)

Power plant

Closed Rankine cycle; ammonia working fluid; shell-and-tubeHXs with titanium tubes 0.01905 m OD, 0.01765 mID; 4 10-MWe (net) modules

Nominal operating characteristics

Nominal �T (K) (mixed seawater and HECO flow) 26.83

Power (MWe)

Gross turbine-generator power 53.8

Warm water pump power 4.1

Cold water pump power 8.4

Ammonia pumps power 0.9

Other 0.4

Net power output to busbar 40.0

Evaporator Condenser

Heat load (MWt) 1,765 1,711

Water flow (kg/s) 13,747 11,510

Ammonia flow (kg/s) 1,434 1,434

Number of HX tubes 40,128 37,671

Tube length (m) 13.65 15.03

HX area (m2) 321,024 301,368

Water side 243,197 251,239

Ammonia side 262,442 271,121

Overall heat trans. coeff. (W/m2 K) 2,516 2,601

Fouling coeff. (m2 K/W) × 105 1.76 1.76

Water inlet (outlet) temp. (◦C) 26.8 (22.8) 5.0 (9.7)

Ammonia inlet (outlet) temp. (◦C) (21.44) 10.66

Log mean temp. diff. (LMTD) (+◦C) 2.95 2.69

Seawater �p (m) 2.39 2.25

Sum of inlet & outlet �p 0.60 2.91

Heat exchanger �p 2.25 2.39

Seawater velocity (m/s) 1.37 1.22

CWP flow (m3/[s MWe (net)]) 2.25

Net power/total HX area (kWe/m2) 0.081

assigned to fabrication and deployment of the cold waterpipe. The uncertainties in this category contribute 13 and7% to the total estimate. The weighted average risks of costunderestimates or overruns in constructing and deployingOTEC systems were thus judged to be between 20 and30%. Scaling factors and experience factors [Industrialcost surveys show that unit costs of items in commercialproduction tend to decrease by a constant factor with eachdoubling of the number produced, i.e., the second unitcosts “x” times the cost of the first, the fourth “x” timesthe cost of the second, and so on.] The factor “x,” called the“experience factor,” is used in deriving the final estimatesshown in Tables VI and VII.




TABLE VI Cost Estimates for Floating 40-MWe OTEC Subsystems, Components, Facilities, and Procedures ($M 1990)a

46 MWe (net) 40 MWe (net)

Est. cost Est. costGrazing plantship uncertainty Moored plant uncertainty

Subsystem Mid-1980$ Mid-1990$ −% +% Mid-1980$ Mid-1990$ −% +%

Platform 47.3 64.8 13 13 66.6 91.2 14 28

Hull 20.4 27.9 15 20 20.4 27.9 15 20

Water pumps 9.7 13.3 15 15 9.7 13.3 15 15

Position control 7.6 10.4 10 10 24.2 33.1 15 50

Other 9.6 13.1 10 10 12.3 16.8 10 10

Power systems 42.0 57.5 24 17 42.0 57.5 24 18

Heat exchangers 28.3 38.7 30 20 28.3 38.7 30 20

NH3 systems 3.5 4.8 15 15 3.5 4.8 15 15

Power generation 7.6 10.4 15 15 7.6 10.4 15 15

Other 2.6 3.6 10 10 2.6 3.6 10 10

Water ducts 9.4 12.9 25 99 13.0 17.8 22 76

WW & disch. pipes 0 0.0 15 15 3.5 4.8 15 15

CWP 6.6 9.0 25 100 6.7 9.2 25 100

CWP/hull joint 2.7 3.7 25 100 2.7 3.7 25 100

Other 0.1 0.1 15 15 0.1 0.1 15 15

Energy transfer

NH3 production 27.7 37.9 23 25

Liquid N2 plant 3.2 4.4 15 15

Electrolysis 12.0 16.4 30 35

Plant NH3 synthesis 8.7 11.9 20 20

Other 3.8 5.2 15 15

Elec. power to shore 23.1 31.6 25 76

Control and bus 2.2 3.0 15 15

Conditioning 1.4 1.9 15 15

Riser cable 9.0 12.3 30 100

Bottom cable 4.1 5.6 30 100

Buoys 2.4 3.3 25 25

Misc. hardware 2.4 3.3 15 15

Potheads, gimbals, 1.6 2.2 15 15disconnects

Deployment 11.6 15.9 31 84 20.7 28.3 34 90

Platform 1.3 1.8 15 15 0.9 1.2 15 15

Power system 1.1 1.6 20 40 0.7 1.0 20 40

CWP 9.2 12.5 35 100 9.4 12.9 35 100

Discharge pipe 1.7 2.3 35 50

Electric cable 8.0 11.0 35 100

Acceptance, ind. 6.4 8.8 20 20 8.4 11.5 20 20fac., E& A

Weighted cost 22 30 26 54uncertainty (%)

Direct cost ($ M) 144.4 197.7 170.2 233.0

Interest dur. constr. 22.1 30.2 26.0 35.6

(15% of direct cost)

Tot. investment ($M)

Nominal 228 269

Minimum 182 199

Maximum 296 414

a 1990$ = 1980$ × 1.369 = 1983$ × 1.128. [From Avery, W. H., and Wu, C. (1994). Renewable Energy from the Ocean: A Guide to OTEC.Oxford University Press, New York, p. 361.]




TABLE VII Cost Estimates for Grazing OTEC Methanol and Ammonia Plantships ($M 1990)a

First Subsystem cost First Subsystem cost200-MWE uncertainty 368-MWe uncertaintymethanol ammonia

Subsystem plantship −% +% plantship −% +%

Platform 190.7 14.0 16.9 247 13.8 16.3

Hull 109.7 15 20 124 15 20

Water pumps 42.7 15 15 65 15 15

Position control 20.7 10 10 31 10 10

Other 17.6 10 10 27 10 10

Power systems 180.4 23.4 17.6 273 24.5 17.6

Heat exchangers 124.3 30 20 188 30 20

NH3 system 12.2 15 15 18 15 15

Power generation 30.5 10 10 46 10 10

Other 13.4 15 15 20 15 15

Water ducts 30.8 24.9 99.0 47 24.9 99.0

WW & disch. pipes 0.0 0

CWP 21.9 25 100 33 25 100

CWP/hull joint 8.5 25 100 13 25 100Other 0.4 15 15 1 15 15

Energy transfer

NH3 production 175 18.7 18.7

Liquid N2 plant 23 15 15

Electrolysis plant 86 15 15

NH3 synthesis 62 20 20

Other 27 15 15

CH3 OH production 311.2 16.5 16.5

Coal prep. molten 47.0 25 25.0Na2CO3 gasifier

Gas clean-up, CO shift, 104.6 15 15.0sulfur recov.

CH3OH synthesis 68.0 15 15.0

Steam syst. 25.3 15 15.0

H2O, waste treatment 27.1 15 15.0Na2CO3 regen.

Serv. & maint. 39.2 15 15.0

Electrolysis 58.4 30 35.0

Deployment 39.0 30.6 81.4 59 30.6 81.4

Platform 4.9 15 15 7 15 15

Power system 3.7 15 15 6 15 15

CWP 30.5 35 100 46 35 100

Discharge pipe 0.0

Electric cable 0.0

Acceptance, ind. fac., E& A 45.1 20 20 68 20 20

Weighted cost uncertainty (%) 13 24 20 26

Direct cost ($M) 855.6 869

Interest dur. constr. 131 133(15% of direct cost)

Tot. investment ($M)

Nominal 986 1002

Minimum 789 802

Maximum 1292 1313

a 1990$ = 1980$ × 1.369 = 1983$ × 1.128. [From Avery, W. H., and Wu, C. (1994). Renewable Energy from theOcean: A Guide to OTEC. Oxford University Press, New York, p. 361.]




All complete OTEC installations will incorporate aplatform system or land-based containment system tosupport and house the OTEC subsystems, including waterpumping and distribution, platform station-keeping, andauxiliary platform services; a power plant(s) includingheat exchangers, turbines, electric generators, waterand working-fluid pumps, and control equipment; waterducting equipment; an energy transfer system which mayinclude a process plant and related equipment for fuelproduction, storage and transfer; equipment for routingelectric power to consumers, mariculture installations,fresh water production, and use of the cold water effluentfor air conditioning and/or cold water agriculture; supportfacilities and equipment including shops, personnelaccommodations, etc.

B. Estimated Costs of OTEC Systems

In the OTEC R & D programs sponsored by DOE costestimates were made for many alternative designs of sub-systems and components. In the United States, the mostdetailed OTEC development cost studies were conductedin the JHU/APL and OTC preliminary design programs,which drew heavily on industrial experience and costs, andwork by other DOE contractors. These studies have beenused as the primary basis for the estimates listed in the ta-bles presented in later sections of this chapter. Cost studiesin less detail have also been reported in other Americanstudies, and by Japanese, French, and Taiwan investiga-tors. Some of these are also listed. Because of unknowndifferences in estimating procedures the results may notbe strictly comparable but are deemed useful to include.

1. Baseline 40-MWe (nom) OTEC GrazingPlant-ship and Moored Plan

A cost assessment of floating OTEC systems at thePreliminary Design level has been completed for thebaseline 40-MWe (nom) grazing ammonia plantship andfor moored electric power plant designs. Optimizationstudies show that the grazing plantship stationed near theequator where the water average annual temperature dif-ference (�T ) is 23.3◦C will deliver 46–MWe (net) at theonboard busbar. The moored plant at Punta Tuna, PuertoRico, with the same power system but with inlet water �Tof 22.3◦C, will deliver 40 MWe (net). These values of thegenerated net power are used in Table VI. The estimatedcost of net power at the onboard bus-bar is $3200/kW forthe plantship and $3700/kW for the Puerto Rico plant.

2. SOLARAMCO 40-MWe (nom) OTECDemonstration Plantship

Solar Ammonia Company (SOLARAMCO) prepareda Conceptual Design of a 40-MWe (nom) ammonia

plantship, to be sited on the ocean south of Hawaii, thatwould produce 125 mt/d of liquid ammonia. The estimatedcost of this system was $250 M (1990$). ($6250/kW, in-cluding the ammonia plant)

3. Commercial Sized OTEC Grazing Plantships

Conceptual Design studies have been completed for a200-MWe methanol plantship. The data have been usedalso for an estimate of the cost of a 320-MWe (nom),(368-MWe des. est.) ammonia plantship. The data pro-vide a basis for estimating the costs and cost uncertaintiesof OTEC systems that would enter commercial produc-tion. Cost information available from conceptual designstudies of other OTEC systems and preliminary designsof subsystems, as well as prices of standard items andcomponents, have been used in these estimates. The costdata are based on industrial estimates made in the period1978–1982. Termination of OTEC development fundingprevented the acquisition of comparable estimates of cur-rent costs. Therefore, the quoted estimates of total systemcosts in 1990$ represent an updating of earlier data to ac-count for inflation in the costs of industrial equipment andconstruction. The inflation factors are taken from the peri-odic surveys of construction costs published by ChemicalEngineering magazine. (McGraw-Hill) Estimated costsfor the Conceptual Design of commercial-size OTECmethanol and ammonia plantships are listed in Table VII.

4. Sea Solar Power, Inc. (SSP) 100-MWe FloatingOTEC Power Plant

The cost of the 100-MWe plant was estimated as $250 M(1985$) Anderson, J. H. (1985). “Ocean thermal power-the coming energy revolution,” Solar and Wind PowerTechnology 2, 25.

5. Japanese Floating OTEC Plants

Conceptual Designs of moored OTEC plants sited in theSea of Japan were discussed in Section VI. The costs esti-mated by the Japanese engineers are listed in Table VIII.

TABLE VIII Estimated Costs of 100-MWe (net) JapaneseMoored OTEC Power Plants (1980$)

At Osumi At Toyama

Type of platform Submerged disc Surface ship

Net power output (MWe) 100 100

Unit construction cost (yen/kW) 78 × 105 59 × 105

($/kW) 3550 2690

Unit power cost at busbar (yen/kW h) 12.21 12.36($/kW h) 0.0555 0.0562

Power transmission cost (yen/kW h) 0.66 (AC) 0.66 (AC)($/kW h) 0.003 0.003




C. Estimated Costs of Shore-BasedOTEC Systems

1. Ocean Thermal Corporation (OTC)Shore-Based Power Plant

During 1981–1983 a Preliminary Design of a 40-MWe(nom) shelf-mounted OTEC power plant to be sited atKahe Point on the island of Oahu, HI, was completed byOcean Thermal Corporation (OTC) under DOE auspices.The program is described in Section VII.F. Details of thesystem design costs are proprietary but the estimated costsof the major subsystems are available and are presented inTable IX. Performance analysis by OTC in 1983 showedthat the nominal 40-MWe plant would produce a designpower output of 50 MWe (net), with state-of-the-art com-ponents. The cost uncertainties listed in the table are thoseof the authors of this article, based on a review of the OTCpreliminary design report and comparison with other rele-vant information. The much larger system costs, compared

TABLE IX Estimated Costs of OTC 40-MWe (nom) Land-Based OTEC Power Plant[50-MWe (net) Final Design] ($M 1985)

Land-based containment system 60.3 Energy transfer system 5.0

Structure 38.1 Power control & bus 1.4

Pumps and motors 12.1 Power conditioning 0.5

Facility support 7.7 Cable system 1.0

Outfit and furnishings 2.4 Shore-grid station 1.8

Cold water pipe system 73.1 Biofouling corrosion control 0.3

Pipe 67.7 Deployment services 54.2

Intake 0.1 LBCS deployment 6.8

CWP/LBCS transition 0.7 CWP deployment 27.4

CWP embedment/anchoring 1.7 WWP deployment 6.8

Biofouling corrosion control 2.9 Effluent pipe deployment 11.2

Warm water pipe system 3.7 Power system deployment 2.0

Pipe 3.4 Energy transfer syst. deployment 0.1

WWP/HECO basin transition 0.2 Industrial facilities 12.0

WWP/LBCS transition 0.2 Construction facilities 11.5

Mixed effluent pipe system 5.0 Logistic support facilities 0.5

Pipe 2.3 Acquisition, acceptance, testing 2.0

Effluent pipe/LBCS transition 1.1 Engineering & detail design 5.2

Outfall & discharge channel 0.5 Total direct cost 327.8

Embedment/anchoring 1.1 Construction interest 50.1

Power system 107.2 Plant investment (M 1984$) 378

Evaporators 39.7 Plant investment (M1990$) 419

Condensors 37.1

Turbine/generators 15.4

Working fluid loop 4.8

Nitrogen purge 0.3

Instrument & control 5.4

Biofouling corrosion control 3.0

Auxiliary systems 1.6

with those of the floating baseline moored power plant,are caused by the use of titanium heat exchangers insteadof aluminum in the OTC design, and by the expensiveconstruction and deployment of the 3-km-long, bottommounted, CWP. Estimated costs of the OTC OTEC plantare shown in Table IX.

2. General Electric Conceptual Design of aTower-Mounted 40-MWe OTEC PowerPlant for Hawaii

Costs estimated in the design study, discussed inSection XI , are listed in Table X.

3. Puerto Rico Electric Power Co. (PREPA)Conceptual Design of a 40-MWe (nom)Tower-Mounted OTEC Plant

The costs quoted for the PREPA installation are presentedin Table XI.




TABLE X Estimated Costs of GE 40-MWe (nom)Tower-Mounted OTEC Plant

System cost estimates $M (1982$)

GE tower-mounted designTower system 90

Fabrication 63

Transportation 3

Installation 17

Engineering 4

Insurance 4

Cold water pipe system 64

Pipe material 49

Templates 1

Deployment 12

Weather allowance 2

Power plant 166

Heat exchangers 34

Ammonia charge 1

Water pumps 6

Turbine-generators 20

Deploy, install, misc. 105

Engineering, misc. 8

Total system cost 320

$/kWe (net) 8000

4. French (CNEXCO) (ONERA) Designs of5-MWe OTEC Land-Based Plant for Tahiti

Conceptual designs of alternative 5-MWe closed-cycleand open-cycle plants, including equipment for the pro-duction of fresh water, were completed. The cost compari-son for the two options, is shown in Table XII. (IFREMER1985. Ocean thermal energy conversion, the French ver-sion, a pilot electric power plant for French IndonesiaMarch.

5. Taiwan 100-MWe (gross) Land-Based OTECPower Plant

Estimated costs are listed in Table XIII. Liao, T., Giannoti,J. G., van Mater, P. R., and Lindman, R. A. (1986). “Feasi-bility and design studies for OTEC plants along the coastof Taiwan, Republic of China,” Proc. International Off-shore Mechanics and Arctic Engineering. Symp. 2, 618.

6. Design of OTEC Land-Based OTEPlant for Jamaica

Plans for a 1-MWe OTEC pilot plant to be sited nearKingston, Jamaica, yielded a cost estimate of $6000/kWto $9000/kW, for a 40-MWe demonstration plant. The �Tis 21.3◦C at the site. Details of the cost estimate are not

TABLE XI Estimated Costs of PREPA 40-MWe(net) Pilot Plant ($M 1982)

Tower system 21

Fabrication 13.9

Transportation 0.6

Installation 6.5

Power plant system 117.4

Heat exchangers 62.8

Seawater systems 12.9

Turbine-generator 10.0

Biocontrol 7.4

Ammonia systems 5.0

Startup/standby power 7.0

Control and elect. 10.6

Miscellaneous 1.5

Water ducting 22.2

CWP incl. deployment 18.6

Warm and mixed systems 3.6

Energy transfer system 66.0

Acceptance testing 2.0

Deployment services 20.0

Industrial facilities 2.0

Eng. and detail design 2.0

Total 259.4

$/kW 6485

available. [For a �T equal to that in Puerto Rico (22.3◦C),the estimated cost range would be $5300–8000/kW.]

XIII. CAPITAL INVESTMENTS AND SALESPRICES FOR OTEC PRODUCTS

The cost of supplying energy from a facility is determinedby the operating costs plus the cost to amortize the cap-ital investment. Thus, a knowledge of the plant invest-ment and operating costs allows an estimate to be madeof the sales price of power or products that will returnan attractive profit to a private investor, or an allowable

TABLE XII Estimated Costs of French Land-Based 5-MWe(net) OTEC Electric Power and Fresh-Water Plants for Tahiti(FF 1985)

Closed cycle Open cycle

General 23 23

Power system 144 167

Land structures (incl. land partof CWP and civilian work) 221 221

CWP 135 136

Total 523 547

FF/kW 105,200 106,900




TABLE XIII Power and Costs of Taiwan 50-MWe (nom) Land-Based OTEC Plants (1980$)

Optimized plant

Module Total Gross Net power PowerCold-water Warm-water pump pump turb/gen output output

Site pump pump power power power (nom) (ann.avg) (Investment M$)

Ho-Ping (long CWP) 1.54 3.07 4.62 18.46 68.46 50.0 53 301.5 ($5690/kW)

Ho-ping (short CWP) 1.50 3.07 4.57 18.27 68.27 50.0

Chang-Yuan (long CWP) 1.71 3.13 4.84 19.37 69.37 50.0 74 364.5 ($4920/kW)

Chang-Yuan (short CWP) 1.60 3.13 4.73 18.93 68.93 50.0

return to a publicly owned enterprise. OTEC systems canbe financially competitive if funding can be supplied tobring OTEC to a commercial status, and if sales prices aredetermined by rules that establish a “level playing field.”This requires that prices quoted for products from com-petitive systems be based on replacement costs, and in-clude the costs of environmental impacts, waste disposalcosts, and expenses of maintaining the security of nuclearpower sources and imported oil supplies. With these re-quirements, OTEC product prices can be compared in avalid way with prices of products from existing commer-cial and proposed energy delivery systems.

Project financing in general requires two types of in-vestors: (a) the project sponsors who build and operatethe system, sell the product, and receive the profits; and(b) lenders who provide the funds for construction andimplementation, and receive interest on the money lent.The first types of investors are “venture capitalists,” thesecond, bankers and other financial institutions. The totalinvestment consists of equity and debt.

A. Financial Analysis

A financial analysis procedure specifically devised byMossman (1989) for comparison of OTEC price estimateswith other energy sources allows a valid comparison to bemade of the relative commercial attractiveness of energyproduction methods. It establishes uniform guidelines toestimate the sales price that would return an acceptableprofit for investors in alternative energy production sys-tems. The Mossman procedure eliminates distortions andmisunderstandings caused by failure to include the effectsof government subsidies or hidden costs that differ widelyamong the various options.

The Mossman financial analysis of alternative fuel andpower production options assumes 2 plus years of financedconstruction, followed by an operating life of 25 years, andassumes that the project will be undertaken by a firm orconsortium that will use all possible tax shields providedby depreciation and interest payments. The analysis cal-culates a quantity termed “The Adjusted Present Value”

(APV) that measures the profitability of a potential ven-ture. If the APV has a positive value, the venture will returna profit sufficiently high to attract entrepreneurs to investin it rather than to select alternative profit-making opportu-nities. If the APV has a negative value the project will notreturn a satisfactory profit or will be too risky to be consid-ered. The APV methodology calculates the price at whichat which the product must be sold if it is to achieve a posi-tive APV, by following procedures described by MossmanFull details are presented in Avery and Wu (1994). With theabove guidelines, a financial analysis computer programwas prepared that enables one to calculate the profitabilityof alternative energy production ventures, as they dependon the values of the capital investments, plant life, inter-est rates for construction and amortization, operating costsand other parameters. Product sales prices are varied in theprogram to find the price that will give a positive numberfor the APV. The financial analysis program has been usedto estimate sales prices of methanol and ammonia pro-duced on OTEC plantships and delivered to U.S. ports, andof electric power generated on moored OTEC plants andtransmitted via underwater cable to land-based utilities.The same financial analysis is used for estimating salesprices of fuels and electric power from alternative energysources. A valid comparison can then be made of the prof-itability of OTEC products versus the alternative sources.The results are discussed briefly in the next section andare displayed in the accompanying tables and graphs.

XIV. OTEC ECONOMICS

The economic contribution that OTEC will make to the so-lution of the nation’s energy problems depends on whetherits virtues relative to existing and alternative sources ofenergy and fuels are recognized, and whether it can elicitfunding for commercialization. By using the Mossman fi-nancial analysis method to estimate the prices that must becharged for profitable operation, a decision to fund OTECcommercialization can be based on an unbiased compar-ison of prices of OTEC fuel or electricity, with prices




for profitable construction and operation of the proposedalternatives.

A. Commercial Viability of AlternativeEnergy Options

The information needed to compare the profitability of al-ternative energy production systems in the Mossman anal-ysis includes:

1. Capital investment. Costs of related facilities as wellas the plant investment

2. Annual payments to retire plant investment.3. Fuel costs for electric plants and refineries. These

costs vary with time in an unpredictable manner. Pastforecasts have been in error by large factors.

4. Operating and Maintenance Costs (O&M). For theoptions considered, these costs vary annually from afew percent to about 10% of plant investment.

5. Federal, state, and local taxes, licensing fees,insurance, and incidental costs. These costs are a fewpercent of plant investment. State and local propertytaxes will be zero for OTEC plantships.

6. Demand for the energy option, and potentialrevenues. Many options may be profitable for specialapplications but the resource that could be developedis too small to have a major role in satisfying nationalneeds.

7. Environmental charges. Environmental impacts ofenergy production are typically ignored or charged tothe public through state or local taxes. These costsshould be paid by the energy producer. Theenvironmental costs can be large but the financialimplications are difficult to quantify, particularly ifthe effects are cumulative. Compliance by existingnuclear power installations with environment andsafety regulations are estimated to requireexpenditures in excess of $200B, and are federallyfunded. Environmental costs will be large also for oilshale and tar conversion systems and for coalconversion plants. Solar-based options are generallynonpolluting and are expected to involve minimalenvironmental costs. Because OTEC plantshipoperation will enhance fish production, the overallenvironmental effects are expected to be beneficial.

8. Vested interests of public, industry, and governmentorganizations. Commercialization of new energysources is dependent on the support that can bedeveloped from the existing energy producing anddistributing organizations and their powerful politicalsupport groups. The absence of a strong political basefor solar energy options is a major impediment totheir commercialization.

XV. ECONOMIC, ENVIRONMENTAL,AND OTHER ASPECTS OF OTECCOMMERCIALIZATION

As noted earlier, the plant investments and market prices ofenergy sources involve subsidies, tax breaks, and other fac-tors, including R & D expenditures and environmental im-pacts, that are paid by the government or the general pub-lic. These factors disguise the true costs. An objective eval-uation will determine the desirability of new energy op-tions based on the price of the fuel or power that can be of-fered profitably to consumers. The price comparison mustalso include the replacement costs of conventional powersources. The Mossman financial analysis allows compar-ison of all of the alternatives on a uniform basis, and isused in evaluating the options discussed in the followingand expressed quantitatively in Tables XIV and XV.

A. Production of Liquid Fuels from Existingand Proposed Fuel Sources

The material in this section is based mainly on price quota-tions from industrial suppliers of components and equip-ment that were obtained in the DOE-funded programsthat ended in 1985. The costs were updated to 1990 inpreparation of Section XV.A.1. Information on fuel andelectricity prices was available from reports issued by theEnergy Information Administration of the U.S. Depart-ment of Energy (EIA/DOE). Other sources are listed inthe references.

1. Petroleum-Based Fuels

Petroleum must be refined to make products suitable foruse in transportation, in industrial or residential boilers,and in miscellaneous applications. The refining processesadd to the cost. In November 1999 a barrel of crude oilcosting $25 ($0.595/gal) was associated with a wholesaleprice of $0.745/gal ($31.3/bbl) for gasoline and $0.686/gal($28.8/bbl) for diesel fuel. Thus, the wholesale gasolineprice exceeded the quoted oil price ($/gal) by the factor74.5/59.5 = 1.25.

2. Synthetic Fuels

a. Fuels from oil shale. Two programs initiated withsupport by the now defunct Synfuels Corporation receivedmajor funding in the early 1980s to demonstrate technolo-gies for extracting motor fuel from oil shale: above-groundretorting, and underground gasification. The OccidentalOil Shale Company proposed a demonstration program,involving a plant requiring an investment of $225M, thatwould produce 1200 bbl/day of motor fuel. The process




TABLE XIV Alternative Motor Fuels Price Comparison ($ 1990)a (ITC = 10%; APV = 0)

Number of InvestmentSales Gasoline modules for Average Sales Gasoline for 105 bbl/d

Annual price mpg 105 bbl/d plant price equivalent gasolinePI Output cost (APV = 0) equivalent gasoline investment (APV = 0) price equivalent

Plant type ($M) Product (gal/y) ($M/y) ($/gal) ($/gal) equivalent ($B) ($/gal) ($/gal) ($B)

OTEC CH3OH 960 CH3OH 1.99 × 108 51.1 0.70 1.27 13 0.72 0.620 1.12 9.3(coal $50/t)

OTEC NH3 975 NH3 1.70 × 108 25.6 0.74 1.98 24 0.70 0.540 1.44 16.9

Shale oil 225 HC fuel 1.47 × 107 7.5 1.51 1.51 83 0.14 0.662 0.66 12.0

Shell-FT coal 4400 HC fuel 1.28 × 109 450 0.76 0.76 1 4.32 0.762 0.76 5.2liquefaction(coal $22.7/t)

NG to CH3OH 237 CH3OH 2.44 × 108 117 0.58 1.04 6.3 0.19 0.461 0.83 1.7(NG $3/mBtu)

NG to NH3 100 NH3 1.53 × 108 44 0.37 1.02 26 0.074 0.337 0.93 1.9(NG $3/mBtu)

a Does not include carbon tax or automobile modification costs.

would also generate 35 MW of electric power that wouldadd to the profitability.

b. Liquid fuels from tar sands. This approach seemsto be worth developing but cost information on U.S. tarsands fuel production is not available.

c. Liquid hydrocarbons from coal. Approachesthat have been supported include (i) coal pyrolysis, whichproduces a liquid fuel (+ gas) and coke; (ii) coal gasifica-tion and liquefaction, involving reaction of coal with steamfollowed by catalytic combination of the products to formliquid hydrocarbon fuels; (iii) hydrogen addition to coal,

TABLE XV Alternative Electricity Sources Price Comparison (Does Not Include Carbon tax) (ITC = 10%; APV = 0)

Number of InvestmentPlant Annual Break-even modules Sales for

Fuel cost investment Output cost sales price for price 1010 kWh/yPlant type Fuel ($/GJ) ($M) (kWh/y) ($M/y) ($/kWh) 1010 kWh/y ($/kWh) ($B)

Coal steam with FGD Coal 2.2 141 6.1 × 108 20.1 0.055 16 0.047 1.7

Nuclear Uranium 7.0 2350 5.3 × 109 84.0 0.069 2 0.063 4.5

Photovoltaics Solar actinic 0.0 125 5.9 × 107 1.25 0.229 169 0.137 12.4energy

Solar–thermal–electric

Central receiver Solar heat 0.0 70 3.1 × 107 2.30 0.288 326 0.164 12.5

Focused dish Solar heat 0.0 200 8.8 × 107 12.0 0.258 114 0.160 13.9

Cylindrical mirror Solar heat 0.0 225 2.5 × 108 6.8 0.117 46 0.071 6.3

Wind Wind 0.0 18 1.6 × 107 0.16 0.118 633 0.064 5.8

Methanol fuel cell CH3OH 10.5 50 3.9 × 108 28.5 0.098 25 0.073 0.89a

6.7 50 3.9 × 108 16.4 0.056 25 0.051 0.89a

Ammonia fuel cell NH3 12.7 50 3.9 × 108 29.9 0.090 25 0.085 0.89a

8.4 50 3.9 × 108 20.1 0.065 25 0.060 0.89a

a Includes only the cost of the fuel-cell fuel plant.

heavy oils or tars to form a liquid fuel. Research and devel-opment by industry with government participation showedthat the product cost for the hydrogen addition processesand the coal gasification processes were nearly the same.The pyrolysis approach is not considered to be compet-itive with the other processes if used alone but could bepromising in combination with electric power generationusing the residual coke.

d. Liquid fuels formed by natural gas reforming.Methanol and ammonia are both produced at lowcost in large quantities (∼20 million tons/year in theUnited States) by natural gas reforming, followed by




catalysis. Ammonia plant costs are similar to thosefor methanol. Coal gasification to produce fuel-grademethanol could be competitive with natural gas reform-ing if large use of methanol or ammonia in transportationcaused the price of natural gas to increase.

e. Synthesis of methanol or ammonia by naturalgas reforming followed by purification and catalyticsteps to produce a liquid fuel. Coal gasification canalso provide the feed stock,

f. Production of hydrogen by water electrolysisfollowed by further reactions to produce a liquid fuel.In addition to ammonia and methanol, hydrogen can becombined with heavy oils or tar to produce a liquid fuel.

B. Production of Electricity from Existing andProposed Power Plants

Electric power may be supplied to consumers from a va-riety of existing and proposed sources.

1. Coal-powered steam electric plants with flue gasdesulfurization (FGD). The estimated plantinvestment for a 1000-MWe electric power plantbased on coal-powered steam generation withflue-gas desulfurization (FGD is quoted as $1.41 B. O& M costs are estimated as $166 M/year.

2. Nuclear steam–electric power plants. Costs of1000-MWe nuclear power plants, estimated by Starr(1987), are plant investment (PI) $1700–$3000 M,fuel cost/year $32 M, O & M cost/year $42 M,delivery cost/year $37 M.

3. Hydroelectric power. Hydroelectric power is thecheapest source of electricity and providesapproximately 10% of U.S. electric power. Nosignificant increase in output is foreseen.

4. Natural-gas-fueled steam electric power. Natural gascan substitute for coal in conventional coal-firedelectric plants. The capital cost is estimated at$780/KW (design capacity) and O&M at 2.6% ofplant investment.

5. Oil-fueled steam electric plants. Fuel oil is aconvenient source for steam-powered electric plantsbut has been more expensive than coal and thereforeused only in cases where the convenience outweighedthe added cost.

6. Photovoltaic (PV) electric power generation.Photovoltaic cells convert solar energy directly intoelectricity and could supply a significant fraction ofelectric power needs in the southwestern regions ofthe United States if system costs can be reduced.Energy conversion efficiencies of 10% or more, of

insolation are now being achieved. If PV installations,in, e.g., Arizona, were designed that intercepted onefourth of the incoming radiation the peak electricpower output would be 7 MW/km2 or a daily averagepower of 1.25 MWe for clear sky conditions. Thisimplies that the total electric power needs of Arizona(15,000 MW in 1996) could be supplied by PVinstallations occupying 12000 km2 (4630 squaremiles), or 4% of the state area. Installed costs forgrid-connected PV systems have been predicted tofall to $2500 per peak kilowatt after the year 2000.

7. Solar–thermal–electric systems. Three types ofsolar–thermal–electric concepts have receivedsubstantial development support: (a) Central receiversystems. These systems employ an array of mirrors(heliostats) that focus solar energy on a central heatsink connected to a heat-engine electric-generatorcombination for electric power production. For a10-MWe installation, the plant investment wasestimated to be $70 M. (b) Focused-dish array. Thissystem employs an array of parabolic mirrors with asmall heat-engine electric generator at the focus ofeach mirror. It has demonstrated 15% efficiency.Predicted peak power efficiencies are in the range16–28%. Plant investment was estimated to be $200M for a 100 MWp plant. (c) Cylindrical mirrorassembly. This system has cylindrical one-axisparabolic mirrors that concentrate solar energy on anaxial pipe at the focus that contains a heat absorbingliquid. This is pumped to a central station where thehot liquid is used to generate power. The estimatedpower cost was $225 M for 80 MWe output.

8. Electric power generation by wind turbines. Electricpower generation by wind turbines has achievedcommercial status. The estimated installation cost(PI) is $(900–1000)/kWp.

9. Fuel-cell electric power generation. Ammonia,methanol, natural gas, or hydrogen could becomesignificant, environmentally attractive, sources ofelectric power if used in fuel cells as local sources ofelectricity. Whether methanol or ammonia producedby OTEC will be used as fuel sources will depend onthe cost relative to the costs of the fuels derived fromnatural gas.

C. Comments

The financial analysis program can estimate fuel or powersale prices for a range of values of plant investment and op-erating cost. Comparisons can also be made of the effectson profitability of improvements or deficiencies in per-formance or cost, which might result from changes ininterest rates, capital investments, or from further R & D




or manufacturing experience. In commercial productionof mechanical systems, construction costs are reduced ascomponent sizes and numbers increase. Therefore, esti-mates of projected product prices should be based on thesame total output of fuel or electric energy for each sys-tem compared. Typically, the product cost is reduced bya constant factor, called the experience factor, when thenumber produced is doubled. When the unit size is in-creased, the ratio of costs depends on the ratio of sizes,with an exponent, called the “scaling factor.” Conservativevalues of the exponents have been selected for all of theoptions to provide a means for judging the importance ofthe scaling and experience factors in comparing the op-tions. Values of 0.93 for the experience factor and 0.6 forthe scaling factor have been chosen, drawn from indus-trial experience in the shipbuilding, petroleum, and utilityindustries.

XVI. ECONOMIC IMPACT OFOTEC COMMERCIALIZATION

As a means of judging the economic impact of a nationalplan that would ensure a sustainable energy future be-yond the year 2050, the plant investments and fuel orelectricity prices have been estimated for profitable op-eration of new alternative systems that would each installenough plants to yield a total energy equal to the fuelor electric power delivered to U.S. consumers in 1999.Examples of the economic features of proposed futureenergy systems are compared in Tables XIV and XV.The pertinent data, including plant investment (PI), plantoutput, and estimated delivered energy prices, are shownfor each proposed system. The capital costs estimated in1990 for the systems listed, based on DOE supported pro-grams, are used in the estimates in lieu of unavailable up-to-date values. The number of plants (modules) neededto supply the total energy demands is shown for eachsystem.

The tables show that huge capital investments will benecessary to install any new system designed to supplygasoline or electric power in the amounts consumed inthe United States in 1999. Annual capital investments ofbillions dollars per year will be required to satisfy the na-tional energy demands by 2050, whether existing systemsare replaced or new types are installed.

The commercial prices of fuels and electricity are de-termined by the market and by the effects of governmentactions on capital investments, interest rates, and otherfactors. The data presented in Tables XIV and XV areestimates of sales prices for fuels and power, based ondata developed in the DOE OTEC program that are use-ful in comparing the relative merits of alternative energy

options. OTEC methanol and ammonia are shown to bemarginally profitable in comparison with other options.When plans are developed for particular systems, the anal-ysis program can be used to provide more accurate esti-mates, which will use up-to-date information for all of therelevant variables.

Tables XVI and XVII show examples of the depen-dence of sales prices, and profits to investors, of OTECmethanol and ammonia production, as the number of in-stalled plantships increases, and when the effects on saleprices of the environmental costs associated with fossilfuel combustion and costs to the national economy ofimporting crude oil are included. [The rationale for theestimates is explained in Avery and Wu (1994).] For theassumed values of the variables, production and sale ofOTEC methanol would be profitable beginning with thefirst plantship. Production of ∼128 plantships would benecessary before OTEC ammonia sales would be prof-itable. If each system were developed in sufficient num-bers to replace all U.S. gasoline consumption, and gasolinetaxes were imposed to pay for economic impacts of im-porting oil and of CO2 additions to the atmosphere, substi-tution of OTEC methanol for gasoline at the 1990 averagewholesale price per gallon would save $152B/year; andOTEC ammonia, $174B/y.

OTEC commercialization has not occurred becauseindustry and government have concentrated funding onprojects that have strong political support or a short-termpayoff. The large capital investment required for OTECcommercialization has been regarded as an impregnablebarrier despite the fact that equivalent funds to supportother less attractive developments have been spent bygovernment and industry. Wise consideration of the long-term energy needs of the United States is needed to makeit evident that OTEC development should be supported.OTEC has unique features that merit giving it high prior-ity in future plans: (i) It uses stored solar energy that isinexhaustible. (ii) OTEC energy is available 24 hr a day.(iii) OTEC ammonia fuel generates no greenhouse gases.(iv) OTEC produces no hazardous wastes in constructionor operation. (v) OTEC is environmentally friendly, andwould eliminate the “not-in-my-backyard” roadblock toon-land energy installations. (vi) With commercial devel-opment, OTEC will be financially competitive if the envi-ronmental and national security costs of the of alternativeoptions are included in the financial comparisons.

The commercialization of OTEC to its full potential willhave a significant impact on the U.S. economy throughelimination of oil imports and of the huge sums requiredto protect the imported oil sources, through eliminationof air pollution caused by combustion of fossil fuels, andby abating hazards associated with power production andwaste disposal from nuclear sources. Inexhaustible OTEC




TAB

LE

XV

IP

rofi

tto

Inve

sto

rsS

ellin

gM

eth

ano

lP

rod

uce

do

nO

TE

CP

lan

tsh

ips

ToR

epla

ceG

aso

line

Use

din

U.S

.A

uto

mo

bile

s(D

ebt

Inte

rest

11%

)(C

on

stru

ctio

nIn

tere

st9%

)(I

TC

10%

)(A

PV

0)($

1990

)

CH

3O

HC

H3O

HN

etpr

ofit

Gas

olin

em

ileag

eSa

ving

sm

ileag

eN

etpr

ofit

nth

“rea

lpri

ce”

equi

vale

ntus

ing

Gas

olin

eeq

uiv

pric

eto

inve

stor

sC

H3O

HC

H3O

HTo

tal

CH

3O

HE

quiv

.G

asol

ine

wit

hpr

ice

incl

CH

3O

Hpr

ice

incl

incl

auto

repl

acin

gpl

ant

tota

lA

vgC

H3O

HSa

les

gaso

line

mar

ket

Gas

olin

eim

port

auto

mob

ileto

repl

ace

impo

rtan

dco

stan

dga

solin

eP

lant

Rel

ativ

eP

IP

IP

Ipr

od/n

pric

epr

ice

pric

ere

plac

edim

pact

cost

s(8

)ga

solin

eca

rbon

tax

carb

onta

xw

ith

CH

3O

Hno

.co

stSu

m($

M)

PI

($B

)($

M)

(Bga

l/y)

($/g

al)

($/g

al)

($/g

al)

(Bga

l/y)

($/g

al)

($/g

al)

($B

/y)

($/g

al)

($/g

al)

($B

/y)

11.

000

1.00

934

0.93

934

0.20

0.68

41.

230.

790.

111.

361.

340.

001.

731.

500.

025

20.

800

1.80

747

1.68

841

0.40

0.64

21.

160.

790.

221.

361.

260.

021.

731.

420.

068

30.

800

2.60

747

2.43

809

0.60

0.62

71.

130.

790.

331.

361.

240.

041.

731.

390.

110

40.

744

3.34

695

3.12

781

0.80

0.61

41.

110.

790.

441.

361.

210.

061.

731.

370.

157

80.

692

6.32

646

5.90

738

1.59

0.59

41.

070.

790.

881.

361.

180.

161.

731.

330.

347

160.

643

11.9

601

11.1

692

3.18

0.57

31.

030.

791.

771.

361.

140.

381.

731.

300.

76

320.

598

22.2

559

20.7

647

6.37

0.55

31.

000.

793.

541.

361.

100.

901.

731.

261.

65

640.

557

41.3

520

38.6

603

12.7

0.53

20.

960.

797.

081.

361.

062.

061.

731.

223.

56

128

0.51

876

.948

371

.856

125

.50.

513

0.92

0.79

14.2

1.36

1.03

4.61

1.73

1.19

7.61

256

0.48

114

345

013

452

250

.90.

495

0.89

0.79

28.3

1.36

1.00

10.1

1.73

1.46

7.4

423

0.45

722

442

720

949

484

.20.

482

0.87

0.79

46.8

1.36

0.97

18.0

1.73

1.47

12.3

512

0.44

826

641

824

948

610

20.

479

0.86

0.79

56.6

1.36

0.97

21.9

1.73

1.46

14.9

965

0.41

946

939

143

845

419

20.

464

0.84

0.79

107

1.36

0.94

44.1

1.73

1.46

44.3

1024

0.41

649

638

946

345

220

40.

463

0.83

0.79

113

1.36

0.94

47.0

1.73

1.46

29.8

Tota

lU.S

.cru

deoi

lcon

sum

ed(B

BL

/D)

(199

0)1.

30E+0

7

Tota

lim

port

edcr

ude

oilc

onsu

med

(BB

L/D

)5.

69E+0

6

U.S

.gas

olin

eco

nsum

ed(1

990)

(BB

L/D

)6.

96E+0

6

Rat

ioga

solin

eco

nsum

edto

crud

eoi

lcon

sum

ed0.

533

U.S

.gas

olin

eco

nsum

ed(1

990)

(GA

L/Y

)1.

08E+1

1

Equ

ival

entC

H3O

Ham

ount

tom

atch

gaso

line

road

mile

age

(gal

/yea

r)1.

92E+1

1

OT

EC

CH

3O

Hpl

anto

utpu

t(ga

l/yea

r)(3

45D

)1.

99E+0

8

Num

ber

ofpl

ants

tore

plac

eal

lU.S

.gas

olin

eco

nsum

ed96

5

Rat

ioim

port

edoi

lto

tota

loil

cons

umed

0.43

8

Tota

lgas

olin

em

ade

from

impo

rted

oil

4.67

E+1

0

Num

ber

ofpl

ants

tore

plac

ega

solin

eeq

uiva

lent

ofim

port

edoi

l42

3




TAB

LE

XV

IIP

rofi

tto

Inve

sto

rsS

ellin

gA

mm

on

iaP

rod

uce

do

nO

TE

CP

lan

tsh

ips

ToR

epla

ceG

aso

line

inU

.S.

Au

tom

ob

iles

(Co

nst

ruct

ion

Inte

rest

9%)

(Deb

tIn

tere

st11

%)

(IT

C10

%)

(AP

V0)

(199

0A

vera

ge

Wh

ole

sale

Gas

olin

eP

rice

wit

ho

ut

Taxe

s)

Net

profi

tto

nth

Gas

olin

eG

asol

ine

Savi

ngs

Gas

olin

eN

H3

Savi

ngs

inve

stor

sN

H3

Tota

lTo

tal

NH

3E

quiv

.m

arke

t“r

ealp

rice

”us

ing

NH

3pr

ice

incl

equi

vpr

ice

usin

gN

H3

repl

acin

gpl

ant

NH

3A

vgN

H3

Sale

sga

solin

epr

ice

Gas

olin

ew

ith

carb

onto

repl

ace

impo

rtan

din

clau

toto

repl

ace

gaso

line

Pla

ntP

IP

IP

Ipr

od\n

pric

epr

ice

w/o

tax

repl

aced

impa

ctga

solin

eca

rbon

taxe

sco

stga

solin

ew

ith

NH

3

no.

($M

)($

B)

($M

)(B

gal/y

)($

/gal

)($

/gal

)($

/gal

)(B

gal/y

)($

/gal

)($

B/y

)($

/gal

)($

/gal

)($

/gal

)($

B/y

)

194

80.

9594

80.

170.

661.

770.

790.

064

1.36

−0.0

31.

652.

07−0

.42

−0.0

27

275

81.

7185

30.

340.

611.

620.

790.

131.

36−0

.03

1.65

1.92

−0.2

7−0

.034

375

82.

4682

20.

510.

585

1.56

0.79

0.19

1.36

−0.0

41.

651.

86−0

.21

−0.0

41

470

53.

1780

60.

680.

571.

530.

790.

251.

36−0

.04

1.65

1.83

−0.1

8−0

.047

865

65.

9974

91.

360.

541.

430.

790.

511.

36−0

.04

1.65

1.73

−0.0

8−0

.043

1661

011

.270

22.

720.

511.

350.

791.

021.

360.

011.

651.

650.

000.

00

3256

721

.065

65.

440.

481.

270.

792.

01.

360.

181.

651.

570.

080.

16

6452

839

.261

210

.90.

451.

190.

794.

11.

360.

671.

651.

490.

160.

63

128

491

72.9

570

21.8

0.42

1.12

0.79

8.1

1.36

1.98

1.65

1.42

0.23

1.90

256

456

135.

753

043

.50.

391.

050.

7916

.31.

365.

091.

651.

350.

304.

9

512

424

253

493

87.0

0.37

0.98

0.79

331.

3612

.41.

651.

280.

3712

.0

894

400

415

464

152

0.35

0.93

0.79

571.

3624

.51.

651.

230.

4223

.9

1024

395

470

459

174

0.34

0.91

0.79

651.

3629

.11.

651.

210.

4428

.5

1676

375

727

434

285

0.33

0.88

0.79

107

1.36

51.6

1.65

1.18

0.47

50.6

2048

367

874

427

348

0.32

0.87

0.79

130

1.36

64.4

1.65

1.17

0.48

63.2

Tota

lU.S

.cru

deoi

lcon

sum

ed(B

BL

/D)

(199

0)1.

30E+0

7

Tota

lim

port

edcr

ude

oilc

onsu

med

(BB

L/D

)5.

69E+0

6

U.S

.gas

olin

eco

nsum

ed(1

990)

(BB

L/D

)6.

96E+0

6

Rat

ioga

solin

eco

nsum

edto

crud

eoi

lcon

sum

ed0.

533

U.S

.gas

olin

eco

nsum

ed(1

990)

(gal

/yea

r)1.

07E+1

1

Equ

ival

entN

H3

amou

ntto

mat

chga

solin

ero

adm

ileag

e(g

al/y

ear)

2.85

E+1

1

OT

EC

NH

3pl

anto

utpu

t(ga

l/yea

r)(3

45D

)1.

70E+0

8

Num

ber

ofpl

ants

tore

plac

eA

llU

.S.g

asol

ine

cons

umed

1676

Rat

ioim

port

edoi

lto

tota

loil

cons

umed

0.43

8

Tota

lgas

olin

em

ade

from

impo

rted

oil

4.67

E+1

0

Num

ber

ofpl

ants

tore

plac

ega

solin

eeq

uiva

lent

ofim

port

edoi

l73

4




energy could be produced in amounts exceeding worlddemands. This would remove the need to develop fusionpower, which receives substantial funding from govern-ment and industry, because it is regarded as a future sourceof inexhaustible energy that could replace power now pro-duced from fossil fuels and from nuclear fission, both ofwhich can become depleted in the future.

A detailed discussion and analysis of the overall eco-nomic and social impacts of full commercialization ofOTEC is given in Avery and Wu (1994).


ENERGY FLOWS IN ECOLOGY AND IN THE ECON-OMY • ENERGY RESOURCES AND RESERVES • OCEAN-ATMOSPHERIC EXCHANGE • OCEAN SURFACE PROCESSES

• OFFSHORE STRUCTURES

BIBLIOGRAPHY

Abraham, R., Jayashankar, V., Ikegami, Y., Mitsumori, M., and Uehari,H. (1999). Analysis of Power Cycle for 1 MW Floating OTEC Plant.Proceedings of the International OTEC/DOWA Conference ‘99 Imari,Japan, p. 123.

Avery, W. H., and Wu., C. (1994). “Renewable Energy from the Ocean:A Guide to OTEC,” Oxford Univ. Press, New York.

George, J. F., and Richards, D. (1980). “Baseline designs of moored andgrazing OTEC pilot plants,” Johns Hopkins Appl. Phys. Lab. SR-80-A&B.

Ito, F., and Seya, Y. (1985). “Operation experience of the 100-kW OTECand its subsequent research,” Japanese–French Symp. on Ocean De-velopment, Tokyo, Sept.

Kalina, A. I. (1987). Regeneration of the working fluid and generationof energy. Japanese Patent Sho62-39660.

Lin, G., Lee, T. T., and Hwang, K. K. (1983). “Ocean thermal energyconversion technology development program in Taiwan, Republic ofChina,” Proc. Oceans ‘83 (Aug. 2), 723.

Marchand, P. (1985). “French thermal energy conversion program,” Proc.French-Japanese Symp. Ocean Devel. Tokyo, Sept. 1.

Mitsui, T. F., Soya, Y., and Nakamoto, Y. (1983). “Outline of the 100-kWOTEC pilot plant in the island of Nauru,” IEEE/PES 1983 Wintermeeting, New York, Jan. 212.

Owens, W. L. (1983). “Optimization of closed-cycle OTEC plants,” Proc.ASME-JSME Joint Conf. Honolulu 2, 227.

Starr, C. (1987). “Technology innovations and energy futures,” Proc.Energy Technology Conf. XIV Govt. Institutes Rockville, MD, 21.

Uehara, H. (1982). “Research and development on ocean thermal energyconversion in Japan,” Proc.17th IECEC Conf. 3, 1454.

Uehara, H., Ikegami, Y., Sakaki, K., and Nogamim, R. (1998). “Theexperimental research on ocean thermal energy conversion using theUehara cycle,” Proc. International OTEC /DOWA Conf., Imari, Japan.

Wu, C. (1987). “A performance bound for real OTEC heat engines,”Ocean Eng. 14, 349.

P1: GTM/GRD P2: GNHFinal Pages Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology EN011C-555 July 14, 2001 21:51

Petroleum GeologyRichard C. SelleyUniversity of London

I. Field of Petroleum GeologyII. Properties of PetroleumIII. Origin of PetroleumIV. Generation and Migration of PetroleumV. Petroleum Reservoirs and Cap Rocks

VI. Petroleum TrapsVII. Conclusions: Exploration Methods

GLOSSARY

Anticline Upfold of strata, a common form of petroleumtrap.

Cap rock Layer of impermeable strata sealing petroleumwithin a reservoir.

Evaporites Group of rock-forming minerals resultingfrom the evaporation of seawater.

Kerogen Solid hydrocarbon, insoluble in petro-leum solvents, that generates petroleum whenheated.

Petroleum Fluid hydrocarbons, both liquid (crude oil)and gaseous.

Reservoir Porous permeable rock, generally a sediment,capable of storing petroleum.

Seal Layer of impermeable strata that caps a petroleumreservoir.

Sedimentary basin Part of the earth’s crust containing athick sequence of sedimentary rocks.

Source rock Fine-grained organic-rich sediment capableof generating petroleum.

Trap Configuration of strata, such as an anticline, capableof trapping petroleum.

Unconformity Break in deposition, often marked by anangular discordance between two sets of strata.

PETROLEUM GEOLOGY is concerned with the wayin which petroleum forms, migrates, and may become en-trapped in the crust of the earth. It studies the generationof organic matter on the surface of the earth, the wayin which this may be buried and preserved in sediments,and the physical and chemical processes that lead to theformation of petroleum and its expulsion from sedimen-tary source beds. Petroleum geology also embraces thedynamics of fluid flow in porous permeable rocks. It isconcerned with the depositional environments and post-depositional history of the sandstones and limestones thatcommonly serve as petroleum reservoirs. Petroleum ge-ology also studies the ways in which tectonic forces maydeform strata into configurations that may trap migratingpetroleum.

729

P1: GTM/GRD P2: GNHFinal Pages


730 Petroleum Geology

I. FIELD OF PETROLEUM GEOLOGY

Petroleum is the name given to fluid hydrocarbons, boththe gases and the liquid crude oil. It is commonly noted thatpetroleum occurs almost exclusively within sedimentaryrocks (sandstones, limestones, and claystones). Petroleumis seldom found in igneous or metamorphic rocks. Thuspetroleum geology is very largely concerned with thestudy of sedimentary rocks. It deals with processes oper-ating on the surface of the earth today to discover how or-ganic detritus may be buried and preserved in fine-grainedsediments. It considers where and how porous sands be-come deposited, because these may subsequently serve asreservoirs to store petroleum. It is also important to knowhow porosity and permeability are distributed in sedimen-tary rocks. This is essential to understand the way in whichpetroleum emigrates from the source beds, how it may sub-sequently be distributed in reservoir beds, and how it mayflow once that reservoir is tapped.

Petroleum geology is also concerned with the struc-tural configuration of the crust of the earth. This inter-est includes both large- and small-scale crustal features.On the broadest scale, petroleum geology is concernedwith the formation of sedimentary basins. This is becausesome basins are highly favorable for petroleum generation,while others are less suitable. The deposition of petroleumsource beds varies with basin type, as does the heat flowthat is responsible for generating oil and gas from thesource rock.

On a smaller scale, petroleum geology is concernedwith the way in which strata may be deformed into folds,and how they may be ruptured and displaced by faults.Such structural deformation may form features capable oftrapping migrating petroleum.

II. PROPERTIES OF PETROLEUM

Petroleum, as previously defined, is the name given to thefluid hydrocarbons. A closer examination of the physicaland chemical properties of petroleum is perhaps necessaryas a prerequisite to an analysis of the geology of petroleum.

The term kerogen is applied to solid hydrocarbons thatare insoluble in the normal petroleum solvents, such ascarbon tetrachloride, yet which have the ability to gener-ate fluid petroleum when heated. One of the best-knownexamples of kerogen is coal and its associates peat, lig-nite, and anthracite. Kerogen is also found disseminated inmany fine-grained sediments and sedimentary rocks (theclays, claystones, and shales). This is seldom of directeconomic value itself, but is of great interest to petroleumgeologists because it is the source material from whichcommercial quantities of petroleum are derived.

Liquid petroleum is generally termed crude oil, oil, orsimply “crude.” In appearance, oils vary from colorless toyellow-brown or black. The former are generally of lowviscosity and low density, while the black oils tend to beheavier and more viscous. The gravity of petroleum iscommonly expressed by reference to a scale proposed bythe American Petroleum Institute (API). The API gravityis calculated as follows:

API gravity = (141.5/specific gravity at 60◦F) − 131.5

Chemically, petroleum consists principally of hydrogenand carbon, but also contains small percentages of oxy-gen, nitrogen, sulfur, and traces of metals, such as vana-dium, cobalt, and nickel. The common organic compoundsinclude alkanes (paraffins), naphthenes, aromatics, andheterocompounds. The specific gravity of oil is variable.Heavy oils are defined as those with a specific gravitygreater than 1.0 (10◦API). These are rich in heterocom-pounds and aromatics. The heavy oils grade with increas-ing density and viscosity into the plastic hydrocarbons(generally termed pitch, tar, or asphalt). The light oils arerich in alkanes and naphthenes. They grade into “conden-sate.” This is petroleum that is gaseous in the earth’s crust,but which condenses to a liquid at the surface.

Petroleum gas is generally referred to as “natural gas,”though this ignores the nonhydrocarbon natural gases ofthe atmosphere. Natural gas is generally composed ofmethane and varying amounts of ethane, propane, andbutane. Petroleum gas composed almost exclusively ofmethane is termed “dry gas.” Petroleum gas with sub-stantial amounts of the other gaseous alkanes is termed“wet gas.” Petroleum gases are also associated with tracesof other gases, notably hydrogen sulfide, carbon dioxide,and nitrogen.

Gas hydrates are compounds of frozen water that con-tain gas molecules. Naturally occurring gas hydrates con-tain substantial amounts of petroleum gas, principallymethane. These are only stable under certain critical pres-sure/temperature conditions. They occur today in the sur-face sediments of deep ocean basins and in permafrost.Vast reserves of petroleum are trapped in gas hydrates butthey cannot yet be extracted safely and commercially.

III. ORIGIN OF PETROLEUM

The origin of petroleum has long been a matter for de-bate. From the days of Mendeleev, some chemists and as-tronomers have argued that petroleum is of inorganic abio-genic origin. Petroleum geologists have seldom doubtedthat it is of organic origin. It is known that hydrocarbonscan form in space. Jupiter and Saturn are believed to becomposed of substantial amounts of methane. A particular



Petroleum Geology 731

type of meteorite (the stony meteorites, or chondrites) con-tain complex hydrocarbons up to the level of amino acidsand isoprenoids. Petroleum is found in some igneous rocks(rocks formed from the cooling of molten magma). Inmost of these occurrences the petroleum infills fractureand have obviously migrated into the rock long after itcooled. Rarely, however, it occurs in gas bubbles withinthe rock itself. In some cases it can be demonstrated thatthe magma has intruded organic-rich sedimentary rocksfrom which the petroleum may have been acquired, butthis is not universally true. Methane occurs in many riftvalley floors, both on land (as in the East African rifts) andin the mid-ocean ridge rift systems and in permafrost anddeep ocean gas hydrates. Carbon isotope analysis is ableto show that this is not always of shallow biogenic origin,but that it is often of deep thermal origin.

These observations have caused some people to ar-gue that petroleum is of inorganic abiogenic origin; thatmethane occurs in the mantle of the earth from which itseeps to the surface, being intermittently expelled up faultsystems aided by earthquakes. The fundamental reactionis believed to be between iron carbide and water (anal-ogous to the reaction whereby acetylene is produced bycalcium carbide and water):

CaC2 + 2H2O = C2H2 + Ca(OH)2

Once it was generally believed that all life forms beganwith photosynthesizing algae, and that these formed thebase of ecosystems. Hence petroleum must be of organicorigin. The discovery of hot vents on the deep ocean floortermed “smokers,” shows that ecosystems can be based onmethanogenic bacteria.

An abiogenic origin for petroleum is popular with someastronomers, but, except in the Commonwealth of Inde-pendent States (CIS), these views are seldom held by geo-logists in general, and petroleum explorationists in par-ticular. Most geologists, and certainly those engaged inthe commercial quest for petroleum, believe that it is oforganic origin.

There are many lines of evidence to support the thesisthat commercial petroleum accumulations are of organicorigin. Some of the evidence is chemical and some geolog-ical. Oil exhibits levorotation, the ability to rotate polar-ized light; this property is peculiar to biogenically synthe-sized organic compounds. Oil contains traces of complexorganic molecules that are common in living plant and an-imal tissues. Examples include porphyrins, steroids, andderivatives of chlorophyll. In many instances oil is foundin porous permeable reservoirs totally enclosed by im-permeable shale. It is hard to see how petroleum couldhave migrated into the reservoir, either from the earth’smantle or from outer space. In such instances the enclos-ing shales generally contain kerogen. Gas chromatography

demonstrates the similarity of petroleum generated fromthe kerogen in the shale with that found in the reservoir.

Finally there is the ubiquitous observation thatpetroleum is commonly found in sedimentary rocks andnot in igneous or metamorphic ones. These observationslead to the conclusion that although petroleum may forminorganically in space and in the earth’s mantle, commer-cial accumulations of petroleum are formed from the ther-mal maturation of biologically synthesized organic com-pounds in the earth’s crust. The foregoing analysis of theorigin of petroleum points us toward the study of sedimen-tary rocks and their contained fluids. It is now generallyrecognized that the four basic requirements for a commer-cial petroleum accumulation are (1) a thermally maturesource rock, (2) a porous permeable reservoir to retain thepetroleum, (3) an impermeable cap rock, or seal, to retainthe petroleum within the reservoir, and (4) a configurationof rock, such as an anticline, to trap the petroleum withinthe reservoir. These items will now be considered in turn.

IV. GENERATION AND MIGRATIONOF PETROLEUM

A. The Production and Preservationof Organic Matter

Photosynthesis is the process on which all organic matteris based. In this reaction, carbon dioxide and water areturned into sugar and oxygen in the presence of sunlight,according to the following reaction:

6CO2 + 12H2O = C6H2O6 + 6H2O + 6O2

The sugar that is produced by photosynthesis is the startingpoint for the construction of the more complex organicmolecules found in plants and animals. In the ordinaryway, when life forms die, they are eaten or broken downby the normal processes of bacterial decay. Thus organicmatter is seldom preserved in sedimentary rocks.

There are, however, certain depositional environmentsthat favor the preservation of organic matter (Fig. 1). Inlakes, a thermal stratification of water sometimes devel-ops. Sunlight warms the shallow water and encouragesblooms of algae. As these give off oxygen they providethe base for animal food chains. In the deeper waters,sunlight does not penetrate. The water is thus cooler anddenser. Lacking photosynthesizing algae this zone may be-come anoxic. The density stratification inhibits the mixingof oxic and anoxic waters. Organic detritus falling fromabove may be preserved in the stagnant sediments of thelake floor.

Thus ancient lacustrine deposits are often petroleumsource beds. A similar stratification, with an upper




FIGURE 1 Geophantasmograms to illustrate four common de-positional models for organic-rich petroleum source bed deposi-tion. (A) Thermally stratified lake. (B) Hypersaline lagoon, withbriney deep water recharged by normal-salinity open sea water.(C) Anoxic conditions occur on the edge of the continental shelfon the eastern sides of oceans. Shallow water is oxygenated byphytoplankton. Cold polar oxygenated waters sink to ocean floors.(D) Pan-oceanic anoxic event. This occurs when polar glaciationis absent during episodes of uniform warm global temperature,such as may have occurred during the Cretaceous period. [FromSelley, R. C. (1998). “Elements of Petroleum Geology,” 2nd ed.,Fig. 5.8, Academic Press, San Diego.]

oxygenated and a lower anoxic layer of water, sometimesdevelops in arid climate lagoons. In this environment thestratification is due not to thermal layering, but to evap-oration. Dense anoxic brines concentrate on the lagoonfloor and favor the pickling of organic detritus in the un-derlying muds. Organic sediments may also be depositedin ocean basins. In modern oceans, the shallow waters areoxygenated by plankton photosynthesis. The deep oceanbasins are seldom anoxic. But an oxygen-deficient zonecommonly occurs between 200 and 1500 m on the easternsides of the major ocean basins. The upper limit of thiszone commonly coincides with the edge of the continentalshelf. Thus organic matter may be preserved in sedimentsdeposited on the continental slope and rise.

Global rises of sea level cause marine transgressionsand favor the deposition of blanket petroleum source bedsacross continental shelves. Today the flow of cold denseoxygenated polar water into the ocean basins of lower lat-

itudes prevents their floors from becoming stagnant. It hasbeen argued that when the earth has had a uniform equableclimate, this mixing would not occur. Global anoxic eventsmay then favor the ubiquitous preservation of organic mat-ter on the ocean basin floors. These are the four types ofsedimentary environments that favor the preservation oforganic detritus in fine-grained sediments.

B. Generation of Petroleum from Kerogen

As muds are buried they compact, lose porosity, and arelithified into shales or mudstones. The organic matter thatthey contain undergoes many changes. These can be con-veniently grouped into three stages: diagenesis, catagen-esis, and metagenesis. Diagenesis occurs in the shallowsubsurface at near normal temperatures and pressures.Methane, carbon dioxide, and water are driven off fromthe organic matter to leave a complex hydrocarbon termedkerogen (already defined as hydrocarbon that is insolublein normal petroleum solvents, but expels petroleum whenheated).

There are four main types of kerogen. Type I, sapro-pelic kerogen, is formed from algal matter. It has a highpotential for generating oil. Type II, liptinic kerogen, cangenerate both oil and gas. Type III, humic kerogen, is pre-dominantly gas prone. This type of kerogen forms as peaton the surface of the earth. As this is buried and matured itevolves through lignite, bituminous coal, and anthracite tographite (pure carbon). With increasing burial the sourcebed is subjected to rising temperature and pressure. Diage-nesis merges into catagenesis. This is the principal phaseof petroleum generation. Initially, oil is given off once tem-peratures are above about 60◦C. The peak of oil generationis at about 120◦C. Above this temperature, oil generationdeclines, to be replaced by gas generation and ceases atabout 200◦C. This temperature marks the transition be-tween catagenesis and metamorphism. At this stage, thekerogen has yielded up all its petroleum. The residue ispure carbon, graphite.

The fourth type of kerogen is reworked from older sedi-mentary formations. It is essentially inert, and has nopetroleum-generating potential.

C. Petroleum Migration

The migration of petroleum has long been a matter for de-bate. It is normal to differentiate primary from secondarymigration. Primary migration refers to the emigration ofpetroleum from the source rock into a permeable car-rier bed. Secondary migration refers to the migration ofpetroleum from the carrier bed into the reservoir, and tosubsequent flow during petroleum production. Secondarymigration is relatively simple to understand. Oil and gas




move in response to buoyancy and pressure differentialswithin large pore systems.

Primary migration presents a different problem. Shalesare impermeable and normally provide excellent seals thatinhibit upward petroleum movement (see later). How thencan petroleum emigrate from an impermeable source bedinto a permeable carrier formation? It is not enough to in-voke simple compaction of clay during burial. When clayscompact, most of the porosity is lost in the first kilome-ter or so of burial. Temperatures are still too low to per-mit kerogen maturation. At the depths at which petroleumgenerally forms (some 3 to 4 km) little porosity loss takesplace.

Many different mechanisms have been proposed to ex-plain primary migration. These invoke solution of oil andgas in water, improbably high temperatures, abnormallyhigh pressures, and the use of soapy micelles as petroleumsolvents which then vanish once they have done their task.

It is often noted that petroleum is commonly producedfrom source rocks that are, or once were, overpressured,that is to say, whose fluid pressure is/was above the hy-drostatic pressure. There is evidence that the pressure inan overpressured shale decreases over time to hydrostaticpressure, not gradually, but episodically. A trigger, such asan earthquake, causes a sudden expulsion of fluids throughfaults and fractures. These immediately close up once thehot fluid flush has passed through. This mechanism couldwell explain how petroleum emigrates from impermeableorganic-rich shales.

Though no single theory has received universal accep-tance, the last mentioned is one of the most popular. Per-haps an understanding of the mechanism is not too im-portant. The significant fact is that gas chromatographycan correlate kerogen in source beds with the trappedpetroleum that once emigrated from them.

The first requirement for a commercial accumulation ofpetroleum is the existence of a thick sequence of organic-rich shale, with an excess of 1.5% organic carbon. Whetherthe kerogen yields oil or gas, or both, is dependent onthe type of kerogen and its level of thermal maturation.These parameters are commonly displayed on graphs thatplot the ratio of hydrogen:carbon against oxygen:carbon(Fig. 2).

V. PETROLEUM RESERVOIRS ANDCAP ROCKS

A. Porosity and Permeability

The second requirement for a commercial accumulationof petroleum is the existence of a reservoir capable ofstoring the oil or gas. There are two basic requirements

FIGURE 2 Graph of hydrogen/carbon ratio plotted againstoxygen/carbon ratio. This shows the composition of the differenttypes of kerogen. Type I kerogen generates principally oil withminor gas. Type II kerogen generates both oil and gas. Type IIIkerogen generates only gas. Type IV kerogen, not shown, is inert.[From Selley, R. C. (1998). “Elements of Petroleum Geology,” 2nded., Fig. 5.13, Academic Press, San Diego.]

for a reservoir: porosity and permeability. Porosity is thestorage capacity of the rock. This is sometimes expressedas the void ratio, or more commonly as a percentage:

Porosity = (void volume/rock volume) × 100%

Permeability is the rate of flow of fluid through the rock.It is expressed by Darcy’s Law. This states that the rate offlow of a homogeneous fluid in a porous medium is propor-tional to the pressure gradient and inversely proportionalto the fluid viscosity. This is generally expressed mathe-matically thus:

Q = [K (P1 − P2)A]

mL

Where Q is the rate of flow, K a constant, (P1 − P2) thepressure drop, A the cross-sectional area of the sample, Lthe length of the sample, and m the viscosity of the fluid.Permeability is measured in Darcies (D). One Darcy is thepermeability that allows a fluid of 1 cP viscosity to flow at avelocity of 1 cm/sec for a pressure drop of 1 atm/cm. Sincemost petroleum reservoirs commonly have permeabilitiesof less than 1 D the millidarcy (1/1000 D) is usually used.Permeability can be measured in the laboratory from acore or a plug cut from a core. It can also be calculated




from a production well test of a petroleum reservoir. Somegeophysical borehole logs can also measure permeabilityin a roundabout way.

Theoretically, any rock can act as a petroleum reservoirso long as it has these two essential properties of porosityand permeability. Igneous and metamorphic rocks are gen-erally composed of a tight interlocking mosaic of mineralcrystals. They thus seldom have porosity and permeabilityunless they have been fractured or weathered beneath anunconformity (an old land surface). Only about 10% of theworld’s oil reserves have been found in igneous and meta-morphic basement. Most petroleum reservoirs are found inthe sedimentary rocks (sandstones, limestones, and shales)because, by their very nature, they form from the deposi-tion of granular particles with microscopic pores betweenthe grains. Shales (compacted mud) are commonly porous.But, because of the small diameter of their pores, shalesare generally impermeable and seldom act as reservoirs.About 45% of the world’s petroleum reserves have beenfound in sandstones, and about 45% in carbonates (lime-stones and dolomites).

Table I lists the various types of pores systems encoun-tered in petroleum reservoirs. There are two main typesof pores: primary (or syndepositional) and secondary (orpostdepositional). Primary pores are those that form whena sediment is deposited. They may be either intergran-ular, occurring between grains, or intragranular, occur-ring within grains. The former is common in terrigenoussediments and is therefore the major type of pore foundin sandstone reservoirs (Fig. 3). Intraparticle pores occurwithin shells in skeletal lime sands. During burial, how-ever, these pores are often destroyed by compaction or bycementation.

There are several different types of secondary porosity.Solution due to migrating fluids may generate moldic orvuggy porosity. Moldic porosity is fabric selective. Thatis to say, only one element of the rock is leached out, suchas fossil shell fragments. Vuggy pores cross-cut the fabricof the rock, grains, matrix, and cement. Vugs are thus

TABLE I A Classification of Reservoir Porosity

Primary or syndepositional porosity

Intergranular or interparticle (characteristic of sandstones)

Intragranular or intraparticle (occur within fossils)

Secondary or postdepositional porosity

Solution porosity (common in limestones)

Moldic (fabric selective)

Vuggy–cavernous (cross-cuts rock fabric)

Intercrystalline (characteristic of dolomites)

Fractures (develop in any brittle rock, sedimentary,igneous, or metamorphic)

commonly larger than moldic pores. They grade up withincreasing size into cavernous porosity. Solution porosityis characteristic of limestone reservoirs. This is becausecalcium carbonate is less stable in the subsurface than thequartz (silica) of which sandstones are largely composed.Unlike primary intergranular porosity, however, solutionpores are often isolated from one another. Thus, suchrocks may have considerable porosity but negligiblepermeability.

The other main type of porosity is that which is causedby fracturing. Fractures can develop in any brittle rock,including igneous or metamorphic basement, limestone,or cemented sandstone. Fracturing is generally caused bytectonic processes, such as folding and faulting. Once ini-tiated, fractures may be enlarged by solution or infilled bycementation. Fracturing may not increase the porosity ofa rock very much, but it may cause a dramatic increase inpermeability. Thus if a well has drilled into a petroleumreservoir of low permeability, it is common to fracture therock adjacent to the borehole by explosive charges.

The last main type of porosity to consider is theintercrystalline variety. This is the type of pore sys-tem that is typical of dolomite reservoirs. Dolomite isa mixed calcium magnesium carbonate (CaMg[CO3]2).Some dolomite forms at the surface of the earth. This isgenerally fine grained, and, though porous, generally tight.In the subsurface, however, dolomite may form by the re-placement of limestone. The reaction is a reversible one:

2CaCO3 + Mg2 = CaMg(CO3)2 + Ca2

When calcite is replaced by dolomite, there is an overallvolume shrinkage of as much as 13%. This resultstherefore in a considerable increase in porosity if rockbulk volume remains unchanged. The resultant dolomiterock commonly has a friable saccharoidal appearance.These secondary dolomites may thus serve as veryeffective petroleum reservoirs because they exhibit bothporosity and permeability.

Figure 4 shows the relationship between porosity andpermeability for some of the different pore systems thathave just been described. This shows the very complexrelationship that exists between the texture of a rock andits reservoir potential. A major part of petroleum geologyis concerned with understanding the depositional environ-ments of sediments, so as to predict their distribution inthe subsurface. Similar interest is shown in the postde-positional geochemical reactions that take place betweensediments and migrating pore fluids. These cause sedi-ments to be cemented and turned into solid rock, thusdestroying primary porosity. But selective leaching cansubsequently give rise to secondary solution pore systemsof erratic distribution.




FIGURE 3 Photomicrograph of a thin section through a sandstone petroleum reservoir showing primary intergranularporosity between the quartz framework grains. (Simpson Group (Ordovician) Arbuckle Mountains, Oklahoma.) Widthof photograph is 3 mm.

B. Cap Rocks or Seals

The third essential requirement for a commercial accu-mulation of petroleum is a cap rock or seal. This is a sed-imentary stratum that immediately overlies the reservoirand inhibits further upward movement. A cap rock needhave only one property: It must be impermeable. It canhave porosity, and may indeed even contain petroleum,but it must not permit fluid to move through it. Theo-retically any impermeable rock may serve as a seal. Inpractice it is the shales and evaporites that provide mostexamples. Shales are probably the commonest, but evap-orites are the more effective. We saw earlier how mud iscompacted during burial into mudstone, or shale. Theserocks are commonly porous, but because of the narrowdiameter of the pore throats, they have negligible perme-ability. Thus shales generally make excellent seals to stoppetroleum migration. When strata are folded or faulted,however, brittle shales may fracture. As described earlier,fractures enhance permeability most dramatically. In suchinstances, petroleum may leak from an underlying reser-voir and ultimately escape to the surface of the earth.

The evaporites are a group of sedimentary rocks thatwere originally thought to have formed from the evapora-tion of sea water. They are composed of crystalline layersof halite, or rock salt (NaCI), anhydrite (CaSO4), carnal-

lite (KCI), and other minerals. The evaporites exhibit anumber of anomalous physical properties. Unlike mostsedimentary rocks, when they are subjected to stress theydo not respond by fracturing, but by plastic flow. Thus,a petroleum saturated reservoir may undergo all sorts ofstructural deformation and may even be fractured, but anoverlying evaporite may still provide an effective unfrac-tured impermeable seal.

VI. PETROLEUM TRAPS

A. Terminology and Classification

The last major requirement for a commercial accumula-tion of petroleum is the existence of a trap. Source rock,reservoir, and seal must be arranged in such a geometrythat migrating petroleum is trapped in some configurationof strata and cannot escape to the surface of the earth. Thesimplest type of trap is an upfold of strata termed an an-ticline. Figure 5 illustrates such a structure and some ofthe terms applied to traps. A trap may contain oil, gas, orboth. This will be a function of the pressure/temperatureconditions in the reservoir, and will also be determined bythe type of kerogen that exists in the source bed, and in itslevel of thermal maturation.




FIGURE 4 Graph of porosity plotted against permeability show-ing the reservoir characteristics of the different types of pore sys-tems. [From Selley, R. C. (2000). “Applied Sedimentology,” 2nded., Academic Press, San Diego.]

Gas is lighter than oil and oil lighter than water. Thereis thus a gravity segregation of the reservoir fluids. Thetop may consist of a gas zone, or cap, separated by thegas:oil contact from the oil zone. The oil zone is separatedfrom the water zone by the oil–water contact. The reservoirbeneath the oil zone is termed the bottom water, and thatadjacent to the field area is termed the edge water.

FIGURE 5 Cross section through a simple anticlinal petroleum trap to illustrate terminology. [From Selley, R. C.(1998). “Elements of Petroleum Geology,” 2nd ed., Fig. 7.1, Academic Press, San Diego.]

TABLE II A Classification of Petroleum Traps

Structural traps

Anticlines

Fault traps

Diapiric traps

Salt diapirs

Mud diapirs

Stratigraphic traps

Unconformity related (truncation and onlap)

Unassociated with unconformities (channels, bars, and reefs)

Diagenetic traps (due to solution or cementation)

Hydrodynamic traps (due to water flow)

Combination traps (due to a combination of any 2 or moreof the above)

The vertical interval from the crest of the reservoir tothe petroleum:water contact is termed the pay zone. Notall of this interval may be productive. It may also containimpermeable strata. Thus it is usual to differentiate be-tween the gross pay and the net effective pay. The verticalinterval from the crest of a reservoir to the lowest closingcontour on a trap is termed the “closure.” The lowest clos-ing contour is termed the spill plane. The nadir of the spillplane is termed the “spill point.” Depending on the amountof petroleum available a trap may or may not be full to thespill point. The term “field” is applied to a petroleum-productive area. An oil field may contain several separate“pools.” A pool is a petroleum accumulation with a singlepetroleum–water contact.

There are many different types of petroleum traps. Theyare commonly classified into five main categories and as-sociated subcategories as shown in Table II. These willnow be defined and then described in turn. Structural trapsare those caused by tectonic forces in the earth’s crust.They thus include anticlines and fault traps. Diapiric trapsare due to density contrasts in sedimentary rocks, normallyevaporites and overpressured clays. Stratigraphic traps aredue to depositional, erosional, or diagenetic processes.




Hydrodynamic traps are caused by water flow. Combina-tion traps are due to a combination of any two or more ofthe above processes.

Statistical studies have shown that about 75% of theworld’s petroleum comes from structural traps, principallyanticlines, about 5% from diapirs, and about 6% fromstratigraphic traps. The rest are of combined origin. It mustbe remembered, however, that these figures pertain to thepetroleum that has been discovered to date. No one knowswhat the actual proportions are. These figures thus reflectour ability to find petroleum, and probably indicate ourobsession with drilling anticlinal prospects.

B. Structural Traps

Structural traps are those that are caused by earth move-ments such as folding or faulting. Folds may be caused bycompression or tension of the earth’s crust. Compressioncauses a shortening of the crust, deforming it into a se-ries of upfolds, or anticlines, and downfolds or synclines.Compressive anticlines are the main types of trap for theoil that comes from the foothills of the Zagros Mountainsof Iran.

When the earth’s crust is subjected to tension it be-gins to subside, developing into a basin several hundredkilometers or more in diameter. This basin may be sub-sequently infilled by sediments. The floor of such a basincommonly splits into a mosaic of fault blocks. The upliftedblocks are termed “horsts” and the downdropped blocks“grabens.” As the basin floor is buried by sediments, thestrata are draped over the horsts. Subsequent compactionenhances the vertical closure. As the topography is man-tled by successive layers of sediment, the closure of the an-ticlines diminishes upward. These anticlines are thereforedifferent in origin from those due to compression. Theyare commonly termed drape or compactional anticlines.The Forties field of the North Sea is a classic example ofa compactional anticlinal trap.

Fault traps are of several varieties, but in all cases it isessential for the fault to be impervious and sealing. Thisis by no means always the case, and many faults are openconduits to fluid movement. It is seldom possible to estab-lish whether a fault is open or closed without drilling totest. Tension of the crust generates “normal” faults acrosswhich part of the sedimentary section is absent (Fig. 6B).The Fahud field of Oman is a good example of a normalfault petroleum trap. When a normal fault moves, the strataon the downthrown side often flop down into the space cre-ated by the fault movement. These are called “rollover”anticlines. It is often noted that strata thicken when tracedfrom the upthrown to the downthrown side of the fault, andthat the throw of the fault increases when measured down-ward from sediment increment to increment (Fig. 6C).

FIGURE 6 Cross sections to illustrate the different types of faulttraps. (A) Reverse fault caused by compression. (B) Normal faultcaused by tension. (C) Growth fault with petroleum trapped inadjacent rollover anticline.

This demonstrates that the fault is removed repeatedlythrough time. Such faults are thus termed growth faults.Growth faults host petroleum traps, both where reservoirbeds are truncated and sealed by the fault, and in rolloveranticlines on the downthrown side. Examples are commonin the Tertiary deltas of Nigeria and the Gulf Coast of theUnited States.

Compression causes reverse faults that give rise to thevertical repetition of strata (Fig. 6A). Thrust faults, as theseare often called, are commonly associated with compres-sive anticlines. Such traps are found along the edges ofmountain fronts. The Turner Valley and Winterton fieldsof the Rocky Mountains are examples of such traps.

Transverse movements of the crust give rise to wrenchfaults. These commonly consist of a single fault at depththat bifurcates upward into several branches. These oftengenerate an anticline coincident with the wrench fault.These are commonly termed “flower” structures. Oil fieldsin such traps are found, for example, in the Los Angelesbasin of California.

C. Diapiric Traps

Diapiric traps are caused by differences in the densityof sediment layers. In the ordinary way as sediments areburied they become compacted due to the overburden pres-sure. They lose porosity and their density increases. Thereare two exceptions to this general rule in which low-density sediments may be overlain by denser sediment.This situation is inherently unstable. The deeper less densematerial is displaced upward as the overburden bears




down. The upward movement is localized into domes ordiapirs only a few kilometers across. These structures canbe foci for petroleum entrapment.

The two rocks that can generate diapirs are the evap-orites and overpressured clay. The evaporites commonlyhave a density of about 2.03 g/cm3. This is greater thanthat of freshly deposited sand and clay. But as normal sed-iments compact, their density exceeds that of evaporitesat about 800 m of burial. Below this depth, therefore, di-apiric deformation may be expected to commence. Move-ment may be initiated by structural forces, but can alsoapparently occur spontaneously.

The second type of diapir is produced by overpressuredclays. In some environments, especially deltas, sedimenta-tion is so rapid that some sediments are unable to dewaterbefore they are buried and sealed by younger detritus. Withincreasing burial, the situation can occur in which densecompacted sediment overlies less dense undercompactedclay. This may be squeezed upward in cylindrical diapirsanalogous to those produced by salt movement. There aremany ways in which petroleum may become trapped overand adjacent to a diapir (Fig. 7). This can range from sim-ple domal traps over the crest to cylindrical fault trapsin which the dome has pierced through the surroundingstrata. Salt dome oil fields include the Ekofisk and associ-ated fields of the North Sea. Clay diapirs trap petroleumin the Beaufort Sea of Arctic Canada. Both salt and muddiapirs generate petroleum traps in the Gulf of Mexicoarea of Texas and Louisiana.

D. Stratigraphic Traps

As previously defined, stratigraphic traps are due to de-positional, erosional, or diagenetic processes (Table II).Though the beds involved may be tilted from the horizon-tal, folding and faulting are absent in a pure stratigraphic

FIGURE 7 Cross section through a salt dome to show the dif-ferent types of associated petroleum traps. (A) Crestal dome.(B) Flank fault traps. (C) Flank pinchout traps.

trap (Fig. 8). Depositional, erosional, and diagenetic strati-graphic traps will now be described in turn.

The three main types of depositional stratigraphic trapsare reefs, bars, and channels. Reefs are a major type ofdepositional stratigraphic trap. Modern coral reefs arehighly porous and permeable. They may, in the fullnessof time, be buried beneath organic-rich marine muds, thusbecoming potential petroleum traps (Fig. 8A). There aremany ancient petroliferous reefs. Today reefs thrive onlyin warm clear shallow seawater. The evidence of ancientreefs suggests a similar ecology, though corals have onlybecome important reef-building organisms in recent geo-logical time. Calcareous algae have always been importantreef builders, as have bryozoa, bivalves, and several ex-tinct groups of lime-secreting colonial organisms. Recentreefs are highly porous and permeable. But they are com-posed of calcium carbonate which is highly unstable in thesubsurface. Acid rain water may leach lime from the upperpart of a reef and reprecipitate it as a cement in the primarypore spaces deeper down. Subsequent uplift and weather-ing may generate secondary solution and fracture porosity.Thus ancient reefs are not always porous and permeable.Even where they possess these properties their distributionmay be unrelated to the initial arrangement when the reeffirst formed. Despite these problems, there are many an-cient reefs that serve as stratigraphic traps for petroleum.Notable examples occur in the Devonian rocks of Alberta,and the Cretaceous rocks of Mexico and the Arabian Gulf.

Beaches, barrier islands, and offshore bars are all sedi-mentary environments that deposit clean well-sorted sandsof high porosity and permeability. Where these are en-veloped in organic-rich marine sales, they may becomestratigraphic traps (Fig. 8B). Notable examples occur inthe Cretaceous basins of the Rocky Mountain foothillsfrom Alberta in the north to New Mexico in the south.

There are several sedimentary environments in whichchannels become infilled with sand that may serve as apetroleum reservoir (Fig. 8C). These range from alluvialflood plains, through deltas, to deep-water submarinechannels at the foot of the continental slope. In all ofthese situations, sand channels may have cut into andbeen overlain by impermeable muds. These muds mayact as source beds and seals, generating petroleum thatis then trapped within the channel sands. There are manyexamples of alluvial channel sand traps in the Cretaceousbasins of Colorado and Wyoming and western Canada.Petroliferous deltaic distributary channel sand traps arecommon in the Pennsylvanian sediments of Illinois andOklahoma. Channel and bar stratigraphic traps both havelinear geometries. They are thus colloquially referred toas “shoestring” sands. It is important to note, however,that channels will tend to trend down the old depositionalslope into the center of a sedimentary basin. Barrier bar




FIGURE 8 Block diagram to illustrate the different types of stratigraphic traps. (A) Fossil reef limestone. (B) Coastalbarrier-bar sand. (C) Sandstone infilled channel. (D) Onlap of sand above unconformity. (E) Truncation of limestonebeneath unconformity.

and beach sands, however, will tend to be elongated atright angles to channels, being aligned parallel to thepaleoshoreline.

The second main group of stratigraphic traps to considercomprises those that are associated with unconformities.An unconformity is a surface that marks a major break inthe depositional history of an area. The underlying rocksmay be of any type, igneous, metamorphic, or sedimen-tary. Where the latter is the case, the strata may have beenfolded or tilted before the deposition of the overlying sedi-ments. There is often evidence of weathering and fractur-ing in the rocks beneath the unconformity, though sub-aerial exposure has not always occurred. Unconformitiesplay an important role in petroleum entrapment for twomain reasons. Weathering beneath the unconformity maygenerate secondary porosity and permeability in all sortsof rocks, both sedimentary and basement. These weath-ered zones may serve both as reservoirs and as conduitsto permit petroleum migration. Second, unconformitiespermit the juxtaposition of source beds and reservoirs.Sometimes source overlies reservoir and sometimes thereverse occurs. Thus there are two types of stratigraphictraps associated with unconformities: onlap or pinchouttraps and subcrop or truncation traps.

At its simplest, an onlap trap may be a blanket sand thatpinches out up-dip (Fig. 8D). It is sealed by impermeablerocks beneath and by an onlapping shale (generally thesource rock, as well as the cap). Many unconformities areold land surfaces, however, and sands may be deposited inold topographic lows. Alternating hard and soft sedimentsmay have been weathered and eroded to form scarps, dipslopes, and strike valleys. Fluvial or shallow marine sandsmay have been deposited along the old valleys and sealedby marine muds. Stratigraphic traps of this type are knownas the Mississippian: Pennsylvanian unconformity of Ok-lahoma. Alternatively, the unconformity may have beena planar land surface that was locally incised by alluvialvalleys. These may have been sand filled and drowned by

an advance of the sea that deposited impermeable sourcebeds above them. Such alluvial valley sands may thus actas stratigraphic petroleum traps.

Subcrop truncation traps can be developed in sand-stones or limestones (Fig. 8E). Some sort of closure, eitherdepositional or structural, is required to seal the reservoiroff along the strike of the truncated reservoir. Some of theworld’s largest fields occur in truncation traps. The EastTexas field and the Messla field of Libya are two suchexamples.

The third type of stratigraphic trap is due to diagen-esis. This term encompasses the physical and chemicalchanges that take place in a sediment after it has been de-posited. Diagenesis generally leads to a gradual destruc-tion of porosity as the sediment is compacted due to theweight of the overlying detritus, and as primary porosity isdestroyed by the precipitation of minerals from percolat-ing groundwater. Sometimes, however, diagenesis leadsto the generation of secondary porosity. This is either byleaching, giving rise to molds, vugs, and caverns, or bymineral replacement, such as dolomitization, giving riseto intercrystalline porosity. There are few petroleum trapsthat can be attributed solely to diagenesis. There are, how-ever, many traps in which diagenesis has played a majorrole in the distribution of porosity and permeability. This ismore true of carbonate than sandstone reservoirs, becausecalcium carbonate is chemically less stable than silica inthe subsurface environment. We have already observed theeffects of diagenesis on the porosity evolution of reefs andsubunconformity traps.

E. Hydrodynamic and Combination Traps

The last two types of traps to consider are those due tohydrodynamic flow, and those caused by a combinationof factors. In some sedimentary basins, groundwater flowsdown toward the basin center. In so doing it may encounteroil migrating up toward the surface. Oil is lighter than




FIGURE 9 Cross section through a hydrodynamic trap.Petroleum moving up through permeable strata is trapped in a mi-nor flexure by downward flowing water. [From Selley, R. C. (1998).“Elements of Petroleum Geology,” 2nd ed., Academic Press, SanDiego.]

water, so within a given permeable bed, oil will tend to oc-cupy the upper part and water the lower part. Strata some-time contain flexures which lack vertical closure. That isto say, a regional dip may have local horizontal wrinklesto it. In the ordinary way, oil could not be trapped beneathsuch benches; but where there is hydrodynamic flow, en-trapment can take place in such situations. The oil:watercontact will, of course, be tilted, and the petroleum will re-main trapped in the flexure only as long as the downwardflow of water continues (Fig. 9). Very few pure hydro-dynamic traps have been discovered to date. There are,however, many oil fields around the world in which theoil:water contact is not a horizontal plane, but is in facttilted. The lateral flow of water beneath such traps is oneof several causes of tilted oil:water contacts.

This leads logically to a discussion of combinationpetroleum traps. These were previously defined as thosedue to a combination of two or more trap-forming pro-cesses: structural, diapiric, stratigraphic, or hydrody-namic. There is a multiplicity of ways in which these sev-eral processes may combine to form traps. The interplayof hydrodynamic and diagenetic processes on the morecommon trap types has been already noted. Probably thecommonest type of combination trap is where a structuralhigh has been growing during sedimentation. The upliftitself may have been initiated by a fold, fault, or diapir.Such highs are likely to be subject to erosion that cancause unconformity-related traps. These will include cre-stal subunconformity truncations as well as depositionalpinchouts around the edges. Carbonate reefs may also belocated on syndepositional structural highs. Large fieldsthat are formed by a combination of structural and strati-graphic processes include the Brent and associated fields

of the North Sea, and the Prudhoe Bay field on the NorthSlope of Alaska. The permutations of ways in which com-bination traps may form are infinite.

VII. CONCLUSIONS: EXPLORATIONMETHODS

The preceding account has described the fundamentalfacts and concepts concerning the generation, migration,and entrapment of petroleum in the earth’s crust. It doesnot tell you how to find and oil field. In the old days,petroleum was found by wandering about with a nakedlight, a spirit of optimism, and a sense of adventure.

The first exploration method involved no more thanlooking for an oil seep and collecting the gunck in a bucket,as in the days of Noah (Genesis 6: 13–14). Then wells weresunk adjacent to petroleum seeps, as in China and Japan 2millenia ago. The first application of science to petroleumexploration came toward the end of the nineteenth centurywith the realization that petroleum was commonly trappedunderground in anticlinal folds. Then the main explorationtool was to send geologists out and about mapping anti-clines. Wells were then drilled on the anticlinal crests.

Nowadays, however, geophysics provides the main ex-ploration tool. Gravity and magnetic surveys define theoverall architecture of sedimentary basins. Seismic sur-veys can map traps, not only anticlines, but also channels,barrier bars, and reefs. Seismic surveys can now be usedto image petroleum:water contacts within a reservoir. Ifseismic surveys are shot at regular intervals during theproductive life of a field, it is even possible to image themovement of gas:oil and oil:water contacts as the fielddepletes.


COAL GEOLOGY • EVAPORITES • GEOLOGY, EARTH-QUAKE • PETROLEUM REFINING • SEDIMENTARY PETROL-OGY • STRATIGRAPHY

BIBLIOGRAPHY

Hunt, J. M. (1996). “Petroleum Geochemistry and Geology,” 2nd ed.,Freeman, New York.

Selley, R. C. (1998). “Elements of Petroleum Geology,” 2nd ed.,Academic Press, San Diego.

Selley, R. C. (2000). “Applied Sedimentology,” 2nd ed., Academic Press,San Diego.

Stoneley, R. (1995). “An Introduction to Petroleum Exploration for Non-geologists,” Blackwell, Oxford.

Tissot, B. P., and Welte, D. H. (1984). “Petroleum Formation and Occur-rence,” 2nd ed., Springer-Verlag, Berlin.

P1: GPA/GWT P2: GTY/GVX P3: GSS/FFV QC: GSS Final Pages Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology EN014K-59 July 30, 2001 17:6

Renewable Energyfrom Biomass

Martin KaltschmittDaniela ThranInstitute for Energy and Environment, Leipzig

Kirk R. SmithCenter for Occupational and Environmental Health,University of California, Berkeley

I. Biomass SourcesII. Biomass Production and SupplyIII. Thermo-Chemical ConversionIV. Physico-Chemical ConversionV. Bio-Chemical Conversion

VI. Outlook

GLOSSARY

Bio-chemical conversion This involves biological pro-cesses. The most important to date are alcohol pro-duction from biomass containing sugar, starch and/orcelluloses and biogas production from organic wastematerial (like animal or human waste). Both technolo-gies are state of the art and widely used for fuel mak-ing in developing as well as in industrialized countries.Additionally heat can be produced during compostingof biomass waste, although this heat is difficult to tapeconomically.

Biofuel Biomass used as fuel. Sometimes used to refer

to processed liquid, solid, or gaseous fuels derivedorginally from biomass.

Biomass This term is used somewhat more narrowly thanits common use in ecology, where it refers to all livingmatter. Biomass refers here to organic material pro-duced initially by photosynthesis with or without hu-man support. In practice, this includes:

� Recently living plant material∗

∗Although dried fish and other animal remains have been used occa-sionally as fuels, they do not make up a significant portion of energy useand are not discussed here.

203

P1: GPA/GWT P2: GTY/GVX P3: GSS/FFV QC: GSS Final Pages


204 Renewable Energy from Biomass

� Agricultural residues including animal manure andcrop wastes

� All material made directly or as a by-product duringbiomass processing (e.g., residues from food, feed,fiber, lumber, and organic fertilizer production,slaughterhouse wastes, black liqueur, paper residues,organic fraction of municipal solid waste, sewagesludge, plant oil, and alcohol).

We do not regard peat as biomass, although it is so definedin some Scandinavian countries.

Herbaceous biomass refers to all types of biomass witha grassy nature, such as cereal and hay. Such biomassis provided mainly by agriculture. Compared to woodherbaceous biomass it is more difficult to burn suchtypes of biomass in an environmentally sound way dueto a higher amount of trace elements like chlorine anda lower ash melting point.

Lower heating value (LHV) The LHV (net calorificvalue) characterizes the energy released during com-plete oxidation of dry biomass without counting thelatent heat of the water produced. If the water vapor inthe flue gas produced from fuel hydrogen during ox-idization is condensed within the conversion device,additional energy is released. The higher heating value(HHV—gross calorific value) includes this energy aswell.

Physico-chemical conversion This may be used to pro-duce liquid fuel by physical (e.g., pressing) or chemical(e.g., esterification) means. The only important processwith a certain market potential so far is vegetable oilproduction from oil seed and esterification of this veg-etable oil to fatty acid methyl ester as a substitute fordiesel fuel. This technology is used to some extent inEurope.

Thermo-chemical conversion This is used to convertbiomass into a solid, liquid, or gaseous fuel by heat-ing under various conditions with some access toair. Gasification, pyrolysis, and charcoal productionare the main examples. Of these, only charcoal pro-duction is widely used to date, but gasification inconnection with electricity production seems to bepromising in both developed and developing coun-tries. Biomass gasification also shows advantages indeveloping countries for production of process heat insmall-scale industrial applications such as food drying,silk manufacture, and cremation. Pyrolysis for pro-duction of liquid fuel useable in engines is a futureoption.

Woody biomass This can be produced by forestry (i.e.,firewood from natural or intensively managed forests),agriculture (i.e., short-rotation plantations of fast-growing trees such as poplar or willow), or as residues

from wood production for nonfuel purposes such aslumber. Compared to herbaceous biomass, it is a morepromising source of energy. Besides the fact that thecontent of trace elements in wood is lower comparedwith herbaceous biomass, the ash melting point is alsousually higher. Therefore from an environmental pointof view wood is more promising as a solid biofuel com-pared to straw or hay.

ENERGYfrom biomass currently contributes 10–12% ofgross worldwide energy. Due to geographical, economic,and climatic differences, the share of biomass energy in re-lation to total consumption differs widely among differentcountries, ranging from less than 1% in some industrial-ized countries like the United Kingdom and The Nether-lands to significantly more than 50% in some developingcountries in Africa and Asia. It is by far the most impor-tant renewable energy source, being significantly larger inenergetic terms than the second largest, hydropower. Theenergy from these, the oldest fuels utilized by humanity,is much larger in absolute terms than the energy fromthe newest, nuclear fuels (International Energy Agency,2000). In terms of numbers of people, biofuels dominatethe world picture for it is probably true to say that most ofthe people of the world still rely on biomass fuels for mostof their energy, a situation that has not changed since themastery of fire a hundred thousand years ago. This is dueto the relatively low costs and easy access of biofuels, suchas wood, dung, and crop residues, in the poor rural areasof developing countries where about half of humanity stilllives (Smith, 1987). In addition, however, biofuels such aswood chips and biogas commonly play an important rolein many developed regions where their cost and accesscompare favorably with alternatives. The potential sizeof total biofuel resources is considerable throughout theworld. Depending on how utilized, biofuels also have theadvantage of being relatively environmentally sound com-pared to many other sources of energy, a characteristic thathas led to a resurgence in interest throughout the world.

Cooking with biofuel is extremely rare today in richcountries although there is significant use of biofuel forspace heating in regions with easy access to forests. Spe-cific space heat demand is generally decreasing becauseof improved insulation of old and new buildings due toregulations and other incentives. Average living area percitizen in most industrialized countries is still increasing,however, while population itself is not growing signifi-cantly. The net result is that space heat demand is likelyto remain more or less stable in industrialized countriesfor the next decade. The household market in developingcountries is still increasing due to increasing populationand economic growth. The rapid urbanization common in



Renewable Energy from Biomass 205

many developing countries acts as a countervailing trend,however, because urban households tend to rely less onbiofuels than do rural households. The net result is prob-ably similar to that in developed countries, i.e., overallconsumption of household biofuels is relatively stable,although in this case with large regional variations. Chinamay well be an important exception because there seemsto be an ongoing trend to substitute coal for biomass inrural household applications.

The demand for electric energy is increasing in indus-trialized as well as in developing countries. Therefore,electricity production from biomass is often seen as one ofthe most important future markets for biomass worldwide.The same is also true, in principle, for the production oftransportation fuels from biomass. The need for mobility isstill increasing worldwide, particularly in the developingcountries of Asia. Therefore finding ways to provide cheapand environmentally more benign fuels based on renew-able energy in general and biomass in particular is an ex-tremely active endeavor. To date, however, only Brazil andthe United States have processed significant amounts ofbiomass into liquid fuel (ethanol) for transport.

More broadly, expanded use of biomass in clean ap-plications offers a way toward more sustainable energysystems in all countries, an issue of growing internationalconcern (Kaltschmitt and Reinhardt, 1997 and Kaltschmittet al., 1997). Because many renewable energy sources of-fer the potential to provide useful energy with reducedemissions of greenhouse gases and be additionally envi-ronmental advantageous compared to fossil fuel energy,their attractiveness has been increasing during the lastdecade. In particular, biomass is considered environmen-tally and climatically sound because

� If operated with high combustion efficiency and withrenewable harvesting, biomass fuel cycles aregreenhouse neutral, i.e., the carbon is completelyrecycled and there is no net increase of greenhousegases in the atmosphere.

� Biomass combustion normally has low sulfur andnitrogen emissions with consequent smallcontributions to secondary particle formation and acidprecipitation downwind.

� Biomass has few other intrinsic contaminants that cancontaminate the environment such as the toxicelements, mercury, lead, arsenic, fluorine, etc., foundin some fossil fuels.

� In many circumstances, the ash produced duringcombustion can be usefully recycled back onto theland from which the biomass has been grown.

Among the various possibilities for exploiting solar en-ergy, biomass shares with a hydropower reservoir the char-

acteristic of being a form of “stored” solar energy. Thisoffers significant advantages over wind or direct solar en-ergy, which need to be linked with expensive storage sys-tems to be available reliably throughout the day, month,and year. In addition, being a chemically based fuel, theenergy density of biomass is fairly high, although not ashigh as that of most fossil fuels.

Biofuel is rarely directly affected by energy crises, be-ing an indigenous energy carrier (i.e., biomass is, in mostcases, produced close to the place where it is used) char-acterized by short supply chains with low risks of failure.Also the production and utilization of biomass is widelyaccepted (unlike nuclear and coal power in some coun-tries) and offers benefits for rural areas related to employ-ment, rural infrastructure, the conservation of cultivatedareas, and hence the attractiveness of rural regions.

Most biomass used for energetic purposes is directlycombusted to produce heat and/or power, but a hugevariety of additional possibilities are available to provideenvironmentally sound heat and/or electricity as wellas transportation fuels from organic material. The mostimportant conversion routes available now or in the nearfuture will be discussed according the framework shown inFig. 1.

It is useful to consider two major forms of biomassfuel: unprocessed and processed. For unprocessed biofuelin which the material is used essentially in its natural form(as harvested) direct combustion usually supplies cooking,space heating, or electricity production needs, althoughthere are also small- and large-scale industrial applica-tions for steam raising and other processes requiring low-to-medium temperature process heat. Processed biofuelsin the form of solids (mainly charcoal), liquids (mainlyalcohols), or gases (mainly mixtures with methane or car-bon monoxide), can be used for a wide range of applica-tions, including transport and high-temperature industrialprocesses.

The purpose of biomass conversion is to provide fuelswith clearly defined characteristics that meet given fuelquality standards. Such defined solid, liquid, or gaseousfuels can then be used to meet a specific supply task ef-ficiently. To ensure that these fuel quality standards aremet and these biomass-based fuels can be used with ahigh efficiency in conversion devices (like engines, tur-bines) upgrading is needed. Here, a distinction is madebetween thermo-chemical, physico-chemical, and bio-chemical conversion processes to accomplish this aim (seeGlossary).

These upgraded biomass fuels can be used in speciallyadapted engines, turbines, boilers, or ovens to providethermal and/or mechanical energy (i.e., heat and power),which in turn can be converted into electrical energy.Additionally, liquid and (potentially) gaseous fuels can




FIGURE 1 Possibilities to provide heat and/or power as well as fuels from biomass. [From Kaltschmitt, M. et al. (2000).International Workshop on Integrating Biomass Energy with Agriculture, Forestry and Climate Change Policies in Eu-rope, Impact and Opportunities from Agenda and Agriculture, Forestry and Climate Change Policies in Europe, Impactand Opportunities from Agenda 2000 and the Kyoto Protocol, Imperial College London; Kaltschmitt, M., and Wiese,A., eds. (1997). “Renewable Energies, Systemtechnology, Economics, Environmental Aspects,” Springer, Berlin.]

be used directly or after treatment as transportationfuels.

Heat production and electricity generation are the mostimportant uses worldwide for biomass fuel so far. Directcombustion devices are widely distributed with thermalcapacities ranging from a few kW in household stoves upto heating plants with several tens of MW. The conversionefficiencies vary from 8 to 18% for simple stoves up toapproximately 90% and above for modern heating unitswith high-end technology. Electricity production has beenbased mainly on the conventional steam cycle with effi-ciencies around 30% and capacities of several 100 kW andabove.

The production of transportation fuels is of less impor-tance on a worldwide scale but has been important in somecountries. In Brazil, for example, production of ethanolfrom sugar provided nearly 14 billion liters per year forautomobiles during the mid-1990s. In the United States,ethanol from corn has provided as much as about 4 billionliters per year, although production depends on financialsubsidies. There is potential for increased demand as anagent to reduce air pollution emissions from vehicle fuel,however (Goldenberg, J. et al., 2000).

I. BIOMASS SOURCES

A. Terms, Definition, and Description

Biomass consists primarily of the elements carbon (C),hydrogen (H), and oxygen (O) (Table I). Additionally,significant amounts of trace elements can be found insome types of biomass. For example, straw can containfairly high amounts of chlorine (Cl) and/or silicon (Si) andrapeseed shows a relatively high amount of nitrogen (N).These trace elements can sometimes cause problems. Forexample, during combustion Cl is responsible for hightemperature corrosion in boilers, Si can lead to boiler slag-ging, and nitrogen can cause high NOx emissions.

In most practical applications the energy content of bio-fuel is best described by the LHV (see Glossary). The LHVis greatly influenced by the water content of the biomass aswell as fuel hydrogen content. The actual LHV of biomasscontaining a known percentage of water can be calcu-lated from the LHV of the absolute dry biomass, which isavailable from the literature. In Eq. (1) Hu(w) describes theLHV (in MJ/kg) of the biomass at a specific water content,Hu(wf) the LHV of the fully dry biomass, and w the water




TABLE I Energy Content and Concentrations of Some Elements in Untreated Biomass Compared with Coal

Heatingvaluea Ash Volatile

in content compounds C H O N K Ca Mg P S Cl

Type of biomass MJ/kg in % in % in % of dry substance

Sprucewood (with bark) 18.8 0.6 82.9 49.8 6.3 43.2 0.13 0.13 0.70 0.08 0.03 0.015 0.005

Beech-wood (with bark) 18.4 0.5 84.0 47.9 6.2 45.2 0.22 0.15 0.29 0.04 0.04 0.015 0.006

Poplar wood (short rotation) 18.5 1.8 81.2 47.5 6.2 44.1 0.42 0.35 0.51 0.05 0.10 0.031 0.004

Willow wood (short rotation) 18.4 2.0 80.3 47.1 6.1 44.3 0.54 0.26 0.68 0.05 0.09 0.045 0.004

Bark (softwood) 19.2 3.8 77.2 51.4 5.7 38.7 0.48 0.24 1.27 0.14 0.05 0.085 0.019

Rye straw 17.4 4.8 76.4 46.6 6.0 42.1 0.55 1.68 0.36 0.06 0.15 0.085 0.40

Wheat straw 17.2 5.7 77.0 45.6 5.8 42.4 0.48 1.01 0.31 0.10 0.10 0.082 0.19

Triticale straw 17.1 5.9 75.2 43.9 5.9 43.8 0.42 1.05 0.31 0.05 0.08 0.056 0.27

Barley straw 17.5 4.8 77.3 47.5 5.8 41.4 0.46 1.38 0.49 0.07 0.21 0.089 0.40

Rape straw 17.1 6.2 75.8 47.1 5.9 40.0 0.84 0.79 1.70 0.22 0.13 0.27 0.47

Corn straw 17.7 6.7 76.8 45.7 5.3 41.7 0.65 0.12 0.35

Sunflower straw 15.8 12.2 72.7 42.5 5.1 39.1 1.11 5.00 1.90 0.21 0.20 0.15 0.81

Hemp straw 17.0 4.8 81.4 46.1 5.9 42.5 0.74 1.54 1.34 0.20 0.25 0.10 0.20

Rice straw 12.0 4.4Husk 14.0 19.0

Rye whole crop 17.7 4.2 79.1 48.0 5.8 40.9 1.14 1.11 0.07 0.28 0.11 0.34

Wheat whole crop 17.1 4.1 77.6 45.2 6.4 42.9 1.41 0.71 0.21 0.12 0.24 0.12 0.09

Triticale whole crop 17.0 4.4 78.2 44.0 6.0 44.6 1.08 0.90 0.19 0.09 0.22 0.18 0.14

Rye grain 17.1 2.0 80.9 45.7 6.4 44.0 1.91 0.66 0.17 0.49 0.11 0.16

Wheat grain 17.0 2.7 80.0 43.6 6.5 44.9 2.28 0.46 0.05 0.13 0.39 0.12 0.04

Triticale grain 16.9 2.1 81.0 43.5 6.4 46.4 1.68 0.62 0.06 0.10 0.35 0.11 0.07

Rape grain 26.5 4.6 85.2 60.5 7.2 23.8 3.94 0.10

Miscanthus 17.6 3.9 77.6 47.5 6.2 41.7 0.73 0.72 0.16 0.06 0.07 0.15 0.22

Switch grass

Sugar cane stalk (bagasse) 8.0 4.0 80 45 6.0 35 0.0 0.0

Hay from various sources 17.4 5.7 75.4 45.5 6.1 41.5 1.14 1.49 0.50 0.16 0.19 0.16 0.31

Road side green 14.1 23.1 61.7 37.1 5.1 33.2 1.49 1.30 2.38 0.63 0.19 0.19 0.88

Hard coal 29.7 8.3 34.7 72.5 5.6 11.1 1.3 0.94 <0.13

Lignite 20.6 5.1 52.1 65.9 4.9 23.0 0.7 0.39 <0.1

a LVH.From Kaltschmitt, M., and Hartmann, H., eds. (2001). “Energy from Biomass,” Springer, Berlin, Heidelberg. (In German.)

content (in%). The constant 2.44 results from the heat ofevaporation of water.

Hu(w) = Hu(wf )(100 − w) − 2.44w/100 (1)

Figure 2 shows that the LHV of wood decreases fromapproximately 18.5 MJ/kg with increasing water content.The LHV is zero at water content of approximately 88%.Normally the water content of air-dried wood is between12 and 20% yielding a heating value of 13–16 MJ/kg.Freshly harvested wood is characterized by a water contentof about 50% or more. A low LHV is the result.

For a discussion of different conversion routes, it isuseful to subdivide the biomass into three categories:woody biomass, herbaceous biomass, and other biomass(see Glossary).

FIGURE 2 LHV of wood depending on the water content. [FromKaltschmitt, M., and Hartmann, H., eds. (2001). “Energy fromBiomass,” Springer, Heidelberg, Germany. (In German.)]




FIGURE 3 Material flow of wood within the overall economy. [From Kaltschmitt, M., and Hartmann, H., eds. (2001).“Energy from Biomass,” Springer, Heidelberg, Germany. (In German.)]

1. Woody Biomass

In addition to being grown directly for fuel, wood fuel canderive from production and consumption of wood andwood products as a residue, a by-product and/or a wastematerial all along the overall supply chain of wood, andpulp and paper products (Rosch, C., et al., 2000) (Fig. 3).For example, thinning wood and forest residues areproduced as a by-product during the production of stemwood as a raw material for paper and pulp production orfor cupboard and furniture production. Industrial residualwood, bark, and wood dust are by-products and/or wastematerials resulting from the production of timber andthe manufacturing of wood structures or products basedon wood as a raw material. At the end of the productlifetime and after recycling of some wood fractions (suchas old furniture or roof timbering, and chipboards), partlycontaminated woody material and waste wood remain,which can be used as fuel. If the contamination withheavy metal due to coloring is too high, such materialshave to be treated first.

2. Herbaceous Biomass

Here, a differentiation is made between energy cropsgrown specifically as fuels (like whole crop cereal, rape orsunflower seed, miscanthus, switch grass, or sugar cane)and by-products of other production (like straw). The lat-

ter arise primarily from the production processes for theprovision of food, fiber, and fodder (e.g., the productionof grain is necessarily connected with the production ofstraw). Additionally herbaceous biomass is also produced,for example, in landscape conservation areas, parks, ceme-teries, gardens, and on roadside areas.

3. Other Biomass

Various other organic materials usable as energy sourcesare available within the overall economy, includingby-products and waste materials from agriculture (e.g.,animal manure), industrial food and fodder processingprocesses (e.g., slaughterhouse residues, vegetable waste,olive and fruit pits), food and fodder trading (e.g., spoiledfood and fodder), households (e.g., the organic fractionof solid municipal waste), as well as from waste waterprocessing (i.e., sewage sludge).

B. Use and Potential

These different biomass sources can be used to meet en-ergy demand in industrialized as well as in developingcountries. A substantial fraction of biomass fuel is nottraded in markets, particularly in developing countries,but is gathered and used within individual households orenterprises. Thus, biomass fuel use is poorly representedin official energy statistics and the data available are often




based on sample surveys of uncertain validity and timeli-ness. Only in the late 1990s, for example, has the Interna-tional Energy Agency (IEA) included biomass fuel in itsofficial statistics and then only in fairly aggregated forms(International Energy Agency, 2000). More and more de-veloping countries, however, are starting to include fueluse questions on census forms and other national surveysthat will help reduce these uncertainties.

Recognizing these data constraints, it seems that bio-mass contributes about one-third of the primary energyconsumption in developing countries but varies from over90% in less developed countries such as Uganda, Rwanda,Tanzania, and Nepal to about 45% in India, 30% in Chinaand Brazil, and 10–15% in Mexico and South Africa.By comparison, the share of primary energy providedby biomass within industrialized countries is estimatedto be only about 3%. Importantly, however, the absoluteconsumption per capita varies by a much smaller amountworldwide. Indeed, cross-sectional studies seem to indi-cate that economic development does not usually resultin less overall absolute use of biomass fuel, although itsfraction of total energy declines and use shifts from house-holds to other sectors. Overall, current commercial andnoncommercial biomass fuel supplies about 20–60 EJ/yworldwide. Recent IEA estimations, for example, indicateapproximately 40 EJ/y (Table II).

Use of biofuel differs significantly throughout the world(Table II ). Residues concentrated at industrial processingsites (like bark and sawdust in saw mills) are currentlythe largest commercially used biomass source. For exam-

TABLE II Biomass Energy Potentials and Current Use in Different Regions

North Latin Middle FormerAmerica America Asiab Africa Europe East Russia World

Biomass potentials in EJ/a

Wood 12.8 5.9 7.7 5.4 4.0 0.4 5.4 41.6

Straw 2.2 1.7 9.9 0.9 1.6 0.2 0.7 17.2

Dung 0.8 1.8 2.7 1.2 0.7 0.1 0.3 7.6

(Biogas)c (0.3) (0.6) (0.9) (0.4) (0.3) (0.0) (0.1) (2.6)

Energy crops 4.1 12.1 1.1 13.9 2.6 0.0 3.6 37.4

Sum 19.9 21.5 21.4 21.4 8.9 0.7 10.0 103.8

Use in EJ/a 3.1 2.6 23.2 8.3 2.0 0.0 0.5 39.7

FPEC and HEa in EJ/a 104.3 15.1 96.8 11.0 74.8 15.4 37.5 354.9

Shares in %

Use/potential 16 12 108 39 22 7 5 38

Use/PFEC and HEa 3 17 24 76 3 <1 1 11

Potential/PFEC and HEa 19 143 22 195 12 5 27 29

a Primary fossil energy consumption (PFEC, including nuclear power) and hydroelectricity (HE).b In Asia, current use exceeds available potential (i.e., biomass is used nonsustainably).c Potential of biogas production from the available animal dung.From Kaltschmitt, M., and Hartmann, H., eds. (2001). “Energy from Biomass,” Springer, Berlin, Heidelberg, Germany. (In

German.)

ple, bagasse, the fiber remaining after juice extraction insugar cane processing, often provides energy for extract-ing the juice and processing to pure cane sugar or alcoholat the sugar mill and, in addition, surplus bagasse is usedto supply electricity for the local grid.

In the United States, some 8000 MWe of biomass-fueledelectricity generating capacity are installed primarily ascombined heat and power production systems. In Europebiomass is mainly used for the provision of heat and, toa minor extent, for electricity production. In rural areasbiomass is used traditionally for heating purposes. Elec-tricity production has become more important in recentyears due to environmentally motivated policy measures(e.g., the Feed-in Law for electricity from renewables inGermany, the CO2-tax in Sweden). In average, however,only the use of firewood in households as well as industrialapplication in the wood processing industry are of majorimportance in Europe (for example, firewood covers al-most 75% of biofuel use in Europe). The use of wood andother organic material for district heating is important inonly a few countries (e.g., Denmark, Finland, Austria).

In developing countries, biomass is used mainly in opencombustion devices for cooking and, to a lesser extent,space heating. The poor combustion conditions in thesedevices results in high emissions of health-damaging sub-stances as products of incomplete combustion and lowfuel-use efficiencies. Additionally, biomass is sometimesused in a nonsustainable way in that more is removedfrom the natural biomass resources (e.g., forests, grass-lands) than is regrown. Therefore, the available biomass




resources are degraded, with negative environmental im-pacts on land, soil, and water resources. Research hasshown, however, that fuel demand is rarely the sole oreven primary driver of deforestation in developing coun-tries, but it can be a contributing factor in some areas,particularly where semi-legal commercial forms of forestexploitation such as charcoal production are undertaken.

The total sustainable worldwide biomass energy po-tential seems to exceed 100 EJ/y (Table II), which isjust below 30% of total global energy consumption today(Kaltschmitt, 1999 and Goldenberg et al., 2000).

II. BIOMASS PRODUCTION AND SUPPLY

Biomass feed to the conversion facility is accomplishedthrough a supply chain. Because biomass grows in a dis-tributed manner, supply chains are usually characterizedby a significant transport step to gather sufficient biomassat one site to take advantage of economies of scale in con-version. Additionally, depending on the biomass source,processing or treatment steps, as well as storage and/ordrying activities, may be connected with this transportchain.

For household and other small-scale use in developingcountries, supply chains are often operated by householdmembers using solely human and/or animal labor. This isespecially so for low-quality biomass fuels, such as dung,crop residues, and brush that are typically gathered nearthe point of use. Wood fuel can be transported economi-cally further and is often provided commercially to citiesand towns from rural areas by trains, trucks, or buses. Thecharcoal fuel cycle that is important for many Africancities involves charcoal made in simple kilns in remoteareas that is then transported by truck, sometimes manyhundreds of kilometers.

A. Industrialized Countries

Here a differentiation can be made between woodybiomass and herbaceous biomass (see Glossary).

1. Residual Wood

Wood log production in conventional forests requiresperiodic thinning to ensure the availability of high-qualitystem wood. The wood produced during such thinning canbe a source of energy. Trees removed from the forest witha breast height diameter (BHD) from 8 to 20 cm arehighly suited for the production of wood chips. To operatea chain like this, a variety of different technologies isapplied, including chain saws, mechanical harvesters,or forwarder (large machines that move whole trees) to

move the wood to the main forest road where mobilechippers are operated.

After felling and moving the trees or wood pieces to theforest road, the wood may be stored in the forest until theend of summer to dry out by respiration and evaporation.During the following season with higher heating demand,the wood pieces are chipped at the forest road and trans-ported to the conversion plant using trucks or tractors.

2. Woody Energy Crops

Wood-chip production from relatively fast-growing poplaror willow trees is especially promising for many indus-trialized countries, although eucalyptus can also be at-tractive in some areas. For a plantation, 12,000–20,000plants per hectare are needed. An average yield of around10 t/(ha/y) of dry matter is typical in countries like theUnited Kingdom or Germany. During the first year, weedcontrol is necessary. In the following years, only fertil-izer is needed in most cases. Harvesting is typically doneduring winter to early spring every third year. After the bi-ological/economic lifetime of such plantations, removingroots from the soil is necessary to make the land availableagain for planting annual crops.

For such short-rotation plantations, the so-called “greenchain” is used in most cases in which trees are harvestedaccording to demand at the conversion plant. Thus, theharvested wood chips are transported directly to the plantwithout any additional storage.

3. Straw

During grain harvesting, straw is often left on the field.For transport to the conversion facility, straw can be col-lected from the field and pressed into bales by a tractor-driven straw press. With such locally available machinery,the loose straw material can be compressed into small orlarge rectangular bales, cylindrical bales, straw pellets, orcompact rolls. Depending on the requirements of the con-version technology used to transform the straw into heatand/or electricity, storage of bales within a barn or on thefield wrapped with plastic film is possible.

Handling of bales in and out of storage is done withequipment already available on the farm in most cases(e.g., front loader for a tractor). On demand, the straw istransported from the storage to the conversion plant witha truck or a tractor-driven trailer depending on distance.

4. Herbaceous Energy Crops

A distinction is made between annual and perennial crops.The former are sown and harvested within one year (e.g.,triticale) and the latter are used for a couple of years.

Annual crops are mainly cereal crops produced withavailable technology for the production of grain for feed




and fodder. Biomass is then harvested and pressed intobales like straw. After a storage period, it can be used insuitable combustion devices. Compared to straw the lossesare higher because cereal grains can not be used fully withavailable technology. Also pest losses during storage areoften fairly high.

Perennial crops like switch grass, miscanthus, or sugarcane are grown on plantations with a technical/economiclifetime of approximately a decade and even more.Here these crops are produced like grass on pasture forfeeding cattle. The harvest for some crops is like that forconventional hay production. Alternatively, some cropsare harvested with a specially developed harvester in earlyspring when the biomass from the last growing season isfairly dry and the soil is frozen. In most cases bales areproduced and the further provision chain is comparableto the provision of straw. Sugar-rich biomass (sugar cane,sugar beet) is difficult to store because of the high intensityof biodegradiation, so that such biomass is treated season-ally without a long time storage. Like for annual crops,provision chains for perennial crops are not very muchdeveloped. Only very few chains are realized so far on acommercial basis (i.e., cultivation sugar cane for ethanolproduction).

B. Developing Countries

Logs, brush, crop residues, and large-animal dung arethe principal fuels used by the majority of rural house-holds living in developing countries. They are generallyharvested locally, although in some areas people mustspend an average of 1–3 hours per day or more to walkto and from the areas where it is available. Being gath-ered directly by the users, a large fraction of this fueldoes not pass through markets and is thus “noncom-mercial.” Thus, it tends to be poorly accounted by theauthorities.

By definition, agricultural wastes (dung and crop resi-dues) are harvested renewably; i.e., the carbon releasedthrough burning is recaptured from the atmosphere in thenext crop cycle. In contrast, in some areas brush and woodare sometimes harvested nonrenewably putting pressureon local forests and other biomass resources and directlycontributing to greenhouse gas emissions. In general, how-ever, it has been difficult to find examples of large-scaledeforestation that is driven mainly by fuel demand. Inmost cases, deforestation is driven by demand for newagricultural land, clear-cutting for lumber, road-building,and other land-use changes.

Examples of deforestation significantly driven by fueldemand mostly stem from urban demand for wood or char-coal. In such cases, trucks, buses, and trains are used tosupply these markets from some distance. In the case of

charcoal in East Africa, for example, such “wood sheds”can extend hundreds of kilometers from large cities. Al-though not as prevalent as in rural areas, there is, never-theless, significant biomass use as household fuels in de-veloping country cities, particularly among the one-thirdof the urban population thought, on average, to live ininformal settlements (slums or shanty towns).

III. THERMO-CHEMICAL CONVERSION

The most widely use of biomass is the provision of heatreleased during combustion of solid biofuels. But combus-tion is only one conversion process among others based ona heat-induced chemical conversion of the organic mate-rial. Therefore within the following explanations we firstdiscuss the basics of such thermo-chemical conversionprocesses. Then we focus on the technical implementa-tion of these principles in conversion plants.

A. Basics

During combustion, biofuels are oxidized primarily to car-bon dioxide, water, and heat. Equation (2) shows theapproximate process for wood, which can described asCnHmOp.

CnHmOp +(

n + m

4− p

2

)O2 → nCO2 + m

2H2O + heat

(2)

1. Basic Processes

Depending on conditions, biomass combustion consists ofthe steps: Heating and drying, pyrolytic decomposition,gasification, and oxidation (Fig. 4).

a. Heating and drying. Before any chemical reac-tion of the organic material can take place, physicallybound water is evaporated at temperatures up to 200◦C.The water may leave the reaction region with the flue gasor become involved in reduction to be converted to H2.This process requires energy (i.e., is endothermic).

b. Pyrolytic decomposition. Within this step thebiomass macromolecules are decomposed by heat inabsence of oxygen and the volatile compounds are drivenout of the biomass material due to thermal effects. Atabout 200◦C and more, such pyrolysis processes start totake place and the hemicellulose and lignine which areparts of solid biomass-like wood are decomposed. Start-ing at about 300◦C is decomposition of cellulose, whichis another component of wood. The volatile componentsof the biofuel are vaporized at temperatures below 600◦C




FIGURE 4 Example of thermo-chemical conversion of wood. [From Kaltschmitt, M., and Hartmann, H. (2001). “En-ergy from Biomass,” Springer, Berlin, Heidleberg, Germany. (In German.)]

by a set of complex decomposition reactions not fullyknown and understood. As a result of such decompositionprocesses, the following components are formed:

� Volatile compounds, such as hydrogen (H2), carbonmonoxide (CO), methane (CH4), carbon dioxide(CO2), nitrogen (N2) and steam (H2O)

� A carbon-rich solid fraction (char)� Low molecular weight organic compounds and high

molecular weight (condensable) compounds (i.e.,liquid products)

Fixed carbon (char) and ash are the by-products that arenot vaporized.

The pyrolytic decomposition of biomass (representedfor wood by the formula CH1.4O0.6) can be characterizedby the following equations:

CH1.4O0.6 → 0.6CO + 0.7H2 + 0.4C (3A)

CH1.4O0.6 → 0.6CO + 0.35H2 + 0.225C + 0.175CH4

(3B)

The reactions taking place during a specific conversionprocess are dependent on process conditions. Therefore,a broad variety of different processes take place, oftenconcurrently in different parts of the reaction region.

c. Gasification. It is obvious, that solid carbon nec-essarily remains as a product of the pyrolytic decompo-sition [Eqs. (3A) and (3B)]. In order to convert this solid

carbon into a gas, an oxygen-containing agent such as airor pure oxygen is required. At a temperature range be-tween 700 to 1500◦C the solid carbon as well as gaseousproducts (CO, H2, CH4) are oxidized as follows:

Partial oxidation C + O2 → CO �H = −111 kJ/mol (A3)of solid carbon

Complete C + O2 → CO2 �H = −197 kJ/mol (A4)oxidation ofsolid carbon

Oxidation of CO + O2 → CO2 �H = −283 kJ/mol (A5)carbonmonoxide

Oxidation of H2+ O.2 → H2O �H = −242 kJ/mol (A6)

hydrogen

Oxidation of CH4 + 2O2 → CO2 �H = −802 kJ/mol (A7)methane + 2H2O

Additionally, balance reactions take place. The most im-portant are the reduction of CO2 to CO (Boudouard reac-tion) and of H2O to H2 (heterogeneous water gas reaction).Simultaneously, carbon can be gasified to CH4.

Boudouard C + CO2 → 2CO �H = 173 kJ/mol (A8)reaction

Heterogeneous C + H2O → CO + H2 �H = 131 kJ/mol (A9)water gasreaction

Heterogeneous C + 2H2 → CH4 �H = −87 kJ/mol (A10)methaneproduction




During the complicated mix of these and other reactionsmaking up gasification, sometimes energy is needed andsometimes energy is released, depending on conditions.Additionally these reactions can also take place duringthe pyrolytic decomposition at higher temperatures be-cause the biomass itself contains oxygen. Therefore theseparation between these two conversion steps is not al-ways clearly defined.

d. Oxidation. Within this last step of the thermo-chemical conversion, the gaseous products producedduring the steps already performed are fully oxidizedto carbon dioxide and water releasing energy (i.e.,exothermic).

2. Technical Implementation

If Eq. (2) occurs in one step, we speak of full oxidation.Under these conditions the excess air ratio is 1.0 or above[the excess air ratio is defined as the ratio between theamount of oxidizing agent fed to the conversion processand the amount of oxidizing agent needed to fully oxidizeall reaction products; per definition the excess ratio is 1.0if the conversion process is realized exactly as shown inEq. (2)].

Oxidation can also be realized in two steps in which theexcess air ratio of the first step is below 1.0. Under theseconditions the reaction products can be further oxidizedat a second step releasing the rest of the available energy.Carbon monoxide and/or hydrocarbons are typically pro-duced at the first step and transported to another devicefor full oxidation. Within such processes the proceduredescribed above is paused, for example, after the gasifica-tion step (e.g., within a gasifier) and the oxidizing step isrealized at another time and at another place (e.g., withinthe engine).

If in such a two-stage process the intermediate productis a liquid we call the process pyrolysis (the procedure de-scribed above is than suspended after the pyrolytic decom-position). Under these conditions the excess air ratio of the

TABLE III Excess Air Ratio and Reaction Product Composition for Thermo-Chemical Conversion Processes

Excess Temperature Pressure Productsair ratio in ◦C in bar Gas Liquid

Oxidation λ ≥ 1 800–1,300 1–30 Hu = 0

Gasifcation 0 < λ < 1a 700–900 1–30 Hu > 0

Pyrolysis, λ = 0 350–550 1–30 Hu ≥ 0 Hcoalification

a In most cases 0.2 < λ < 0.5.From Kaltschmitt, M., and Hartmann, H., eds. (2001). “Energy from Biomass,” Springer,

Berlin, Heidelberg, Germany. (In German.)

pyrolysis process is zero. Solid, liquid, and gaseous prod-ucts are formed in varying amounts depending on the pro-cess conditions (e.g., temperature, heating rate, pressure,water content). If a gas is to be produced at the first step,the excess air ratio is between zero and one (Table III).The gas, which often mainly contains carbon dioxide, iscalled producer gas.

B. Direct Combustion

The thermo-chemical conversion of biomass or biomassproducts into heat is called combustion. This heat releasedduring the oxidation of organic material mainly into car-bon dioxide and water can be used directly at the con-version plant (e.g., for cooking or space heating) or canbe transported by means of a heat carrier (i.e., hot wateror steam) to the place of consumption (e.g., district heat-ing systems). The thermal energy can also be convertedvia a steam turbine or other combined heat and power(CHP) process into electricity and low temperature heat.Because of technical and economic limits, the temperatureand pressure of steam processes based on biomass cannotapproach those using fossil fuels. Therefore, only fairlylow conversion rates of biomass fuel energy into electri-cal energy (maximum of 25–30%) are possible at present.The technology required to optimize combustion dependson the capacity, fuel consistency, water content, ash melt-ing behavior, trace contaminants, and other factors. Wherethese conditions vary, for example, with changeable mix-tures of biomass species in the feed, performance willinevitably suffer.

As noted above, due to the relatively high volatile con-tent and other characteristics of biomass, spatial separa-tion is usually provided in modern combustion devices be-tween the fuel gasification process and the full oxidationof the producer gas into CO2 and H2O. The former occurswith primary air fed into the glowing fire and the latterthrough secondary air fed into the burning gas, prefer-ably in an after-burning chamber. To achieve low pollu-tant emissions, good mixing of air and combustible gas




TABLE IV Typical Combustion Technologies and Their Characteristics

Typical thermalCombustion technology capacity Biofuels

Manually fed systems Open/closed chimney 2–15 kW Wood logs, briquette

Single stove 3–12 kW Wood logs, briquette

Tiled stove 2–15 kW Wood logs, briquette

Pellet stove 3–10 kW Pellets

Wood log stove 10–500 kW Wood logs, briquette

Automatic Gasification firing system 20 kW–2 MW Wood chips

Under feed system 20 kW–2 MW Wood chips, wood shavings and filings

Grate firing system for wood 150 kW–15 MW Wood, bark

Fluidized bed system From 10 MW Wood, bark, sewage sludge, black liqueur

Grate firing system for 50 kW–20 MW Bales, chipped herbaceous biomassherbaceous biomass

Blow in firing system 200 kW–50 MW Dust, shavings and filings

is necessary. Additional design features assist in reducingparticle emissions, including those for keeping flue gas ve-locities low so that ash particles are not entrained. Table IVgives an overview of combustion technologies primarilyused in industrialized countries.

1. Household Cookstove

A typical household stove used in developing countriesfor cooking utilizes a semi-enclosed combustion chambermade from mud, bricks, or rocks with no flue or chimney(Baldwin, 1986). It basically operates as a single stagecombustor, i.e., there is no secondary air or other sys-tem to separate gasification from the oxidization. As aresult, combustion efficiency can be low because gasifica-tion products escape from the combustion zone before be-ing fully oxidized. Typical biomass stoves in India, for ex-ample, have nominal combustion efficiencies of 80–95%,which refers to the percentage of the fuel carbon convertedto carbon dioxide. The remainder is released as productsof incomplete combustion (PIC), which consist mainlyof carbon monoxide, hydrocarbons, and airborne soot. Alarge range of other products have also been identified,however, including various carbonyls and polyaromatichydrocarbons.

A cookstove, like other combustion devices designedto produce useful heat, has two internal efficiencies thatdetermine total efficiency (ηt ). The combustion efficiency(ηc) indicates how much chemical energy in the fuel is con-verted to heat. The heat-transfer efficiency (ηh) indicateswhat fraction of the heat produced goes into the heatingvessels. Thus

ηt = ηc × ηh

Figure 5 shows the carbon balance for a well-functioning open wood cookstove with 18% total effi-ciency. Many stove/fuel combinations in common usehave total efficiencies of less than 10%, however. Note thata significant amount of the energy is lost to PIC, which,in turn, consists of many health-damaging species includ-ing respirable soot particles, carbon monoxide, and haz-ardous organic compounds such as formaldeyhde, benzo-α-pyrene, and benzene. In poorly ventilated conditionssuch as a typical village kitchen, the resulting concentra-tions of pollutants often greatly exceed levels required toprotect health. Because of the widespread use of such sim-ple stoves in developing countries, it has been estimatedthat more than a million premature deaths due to asso-ciated respiratory and other diseases are caused annuallyworldwide from exposures to smoke from such stoves,mostly in women and young children who spend mosttime near the fire (Smith, 2000). In addition, methane andother greenhouse gases are also a component of PIC andin inefficient stoves may actually comprise a significantwarming potential. Because so much carbon is divertedto PIC, most of which have higher warming potentialsthan carbon dioxide, stoves with low combustion efficien-cies can actually produce greenhouse gas warming evenif the biomass itself is harvested renewably (Smith et al.,2000).

Improved biomass stoves have been promulgated inmany countries, with mixed results. China has introducedsome 180 million since 1980, commonly with flues andceramic combustion chambers. Total efficiencies have ap-parently risen and smoke exposure fallen, although nosystematic review seems to have been done since 1990.The second largest stove program, that of India, has intro-duced about 30 million over the same period, but has haddifficulty with quality control, lifetime, and maintenance,




FIGURE 5 (a) Energy flows in well-operating traditional wood-fired cookstove. PIC, traditional productions of incom-plete combustion. (b) Indoor health-damaging pollutants (HDP) concentrations from typical wood-fired cookstove.Many dozen other HDPs are also known to be in woodsmoke. (c) Greenhouse warming commitment for typical wood-fired cookstove. Figures in lowest line indicate global warming commitments of each of the non-CC equivalents.

partly because of very strict cost limits. As a result only afew million are thought to be still in use. Dozens of othercountries have local programs.

The basic approach to improving simple biomass stovesis to enclose the combustion zone and vent to the outsidethrough a flue/chimney. Unfortunately, such designs cansometimes actually lower total efficiency because of thehigh airflow induced by natural draft in the flue reduce heattransfer to the pot. Shutting back the airflow with damperscan lead to a rise in total efficiency, but at the same timemay reduce combustion efficiency sufficiently to actuallyincrease PIC (pollutant) emissions per meal. If vented out-

doors, of course, the emissions are less damaging to health,but in crowded villages or urban slums can lead to a sig-nificant degree of “neighborhood” pollution. Longer termsolutions would seem to entail improving combustion ef-ficiency through secondary air or other means.

A promising approach, for example, is to run the stove ina downdraft mode in which all gasification products passthrough the flame zone before exiting and may be cleanenough to vent directly into a living space. It is difficultto maintain stable combustion conditions in these devices,however, without use of a small electric fan, which wouldnot be practical in many parts of poor countries. With




such a fan, however, reliably high efficiencies and lowemissions can be achieved (Reed and Larson, 1996) and,as electrification proceeds, such devices may come to havewidespread potential.

2. Pellet Combuster

Automatically fed combustion devices have been devel-oped for standardized wood pellets. On the back of such adevice (Fig. 6), a container is located to store fuel for auto-matic operation over a few days. The fuel pellets are trans-ported with a screw to a pipe from where the pellets fallinto the combustion chamber. Primary air is fed by nozzlesthrough the bottom of the combustion bowl. Secondary airis fed above the burning fuel via ring-shaped nozzles toensure full conversion of the flue gas. Additional air isfed into the system via the fall pipe to prevent fire frombacking into the fuel container. This well-developed air-feeding system produces low emissions, even to the levelof burners fed by natural gas. The ash produced duringthe combustion of the wood pellets is normally removedmanually. This combustion residue normally contains lessthan 1% carbon and can be used as a fertilizer or taken toa landfill.

The thermal efficiency of such systems can be 90% ormore and the emissions comparably low when used with astandardized and well-defined biofuel. Such heaters can be

FIGURE 6 Example of a combustion device fired with wood pellets. [From Kaltschmitt, M., and Hartmann, H., eds.(2001). “Energy from Biomass,” Springer, Berlin, Heidelberg, Germany. (In German.)]

operated over a wide range of power settings, i.e., betweenapproximately 30 and 100% of the rated capacity.

3. Tiled Stove

Tiled ceramic stoves are used traditionally in Europe. Dueto the large mass of ceramic, a significant amount of heatcan be stored. This allows the fire to be set at high power fora relatively short period (2–3 hr) and still provide steadyheat into the room overnight or even up to 24 hr. This isconvenient and thermally efficient, as well as producingrelatively low emissions because combustion efficiency isbetter during high-power burning.

A variant is the tiled stove with a heating insert madefrom cast iron or metal sheets. Cold room air enters the gapbetween the metal insert and ceramic stove body and isheated before returning to the room. Unlike the high-massceramic stoves, such insert stoves store heat for relativelyshort periods, providing more flexibility for varying heatdemand but less convenience.

4. Oven for Wood Chips

A throw-charging system is used where wood chips aretransported from fuel storage with a stoker scroll andthrown with the help of a centrifugal wheel into a com-bustion chamber equipped with a stiff grate. Such a fullyautomatic feeding system allows the smaller fuel particles




to be combusted during flight while more coarse fuel par-ticles are burned on the grate. Such a feeding system alsohas the advantage that the fuel is fed gently on the fire,thus reducing airborne particle emissions.

The primary air is blown with an automatic ventilatorthrough holes in the grate into the glowing fire. The sec-ondary air is blown at the top of the burning fuel. Theair feeding system adjusts itself automatically accordingto information from a sensor fixed in the flue gas outlet,minimizing pollutant emissions. No additional flue gastreatment is needed and ash can be used as a fertilizer orput on a landfill site.

The heat is extracted from the flue gas via a heat ex-changer located on top of the combustion device. Withinthe heat exchanger, the flue gas is cooled down and wateris heated up—depending on the heat utilization system—either close to the boiling point (for hot water systems) orto steam.

Such combustion systems are characterized by a fullyautomatic operation and available on the market in therange of thermal capacities of 300 up to approximately3500 kW. They can be used either for the heat provisionfor a single- or multifamily house or for industrial processheat.

5. Blow-in Combustion Units

Blow-in combustion units have been developed for usingdusty biomass residues (e.g., sawdust, shavings, filings)from the wood processing industry (Fig. 7). The fuel par-ticles are transported pneumatically from the fuel silo into

FIGURE 7 Example of a blow-in combustion unit for wood dust. [From Kaltschmitt, M., and Hartmann, H., eds. (2001).“Energy from Biomass,” Springer, Berlin, Heidelberg, Germany. (In German.)]

the combustion chamber itself or—as shown in Fig. 7—into a precombustion chamber. Additionally, primary airis blown into the precombustion or combustion chamberto ensure a full conversion of the fuel into gas. Secondaryair is blown into the combustion chamber to ensure a com-plete combustion. To guarantee that the fuel particles ig-nite by themselves during their flight through the combus-tion chamber, fuel water content has to be below 15–20%.To first ignite such a unit, a burner based on light oil or nat-ural gas is used to heat the precombustion or combustionchamber to 450–500◦C.

Such a combustion system is characterized by relativelylow emissions of gaseous pollutants. Because fuel is fedin very small pieces, however, the particle content in theflue gas is relatively high. To meet particle emission reg-ulations in most countries, the use of a flue-gas cleaningis necessary. Therefore such systems are often equippedat least with a cyclone, but often this is not sufficient. Insome cases, therefore, a fabric or bag house filter or evenan electrostatic dust removal system is also used. Suchcombustion units are characterized by fully automatic op-eration and low emissions at thermal capacities of severalMW up to some 10 and more MW and are mainly locatedat places where wood dust is produced as a residue, suchas the furniture industry.

6. Cigar Burner

Compared to wood it is much more demanding to com-bust herbaceous biomass in an environmentally sound waydue to its poorer fuel characteristics. One approach is




FIGURE 8 Example of a cigar burner for straw bales. [From Kaltschmitt, M., and Hartmann, H., eds. (2001). “Energyfrom Biomass,” Springer, Berlin, Heidelberg, Germany. (In German.)]

illustrated by a device developed in Denmark for heatprovision from straw bales (Fig. 8). The bales are fedcontinuously without any preparation into the combus-tion chamber within a horizontal feeding canal via ahydraulic cylinder. They are transported from a storagefacility (barn) close to the plant via a fully automatic roof-mounted crane system to the end of this feeding canal.Such feeding arrangements allow a fully automatic oper-ation of the overall combustion plant. To ensure an easystorage without rotting of the fuel (or self-ignition) as wellas an environmentally sound combustion, the water con-tent of the straw has to be below 20%.

Within the combustion chamber the primary air is blowndirectly on the forehead of the bale where the combus-tion takes place. Secondary air to ensure complete oxi-dation is blown into the after-burning zone located closeafter the flame front (Fig. 8). Incompletely burned partsof the fuel sometimes fall down at the forehead of theburning bale. Such fuel particles are fully burned after-wards on a moving grate system located right below theflame front. A backburning of the flames through thefeeding system is prevented by a water-cooled system tolower the temperature, one-way valves, and maintenanceof a minimum velocity of the bales being fed into thecombustion chamber. The coarse ash deposits within thecombustion chamber and is removed by an automatic ashremoval system.

The flue gas flows through a heat exchanger and is thencleaned via a flue gas treatment system if necessary. De-pending on regulations, a cyclone with or without fabricfilter is used. The removed fine ash is in most cases taken tolandfill and the coarse ash can be used as a fertilizer. Such

combustion devices are adjustable between 30 and 100%of installed capacity. Operation at the lower end of thisadjustment range, however, often leads to relatively highpollutant emissions. Due to the fixed dimensions of thefeeding canal, only bales with clearly defined dimensionscan be used. Additionally with the necessary minimumfeeding velocity of such a plant to avoid back-burning, thethermal capacity of one unit is fixed to 2–3 MW. Smallerthermal capacities are not possible. Larger thermal capac-ities can only be realized by adding additional feedingcanals (i.e., building up a modular system). Such systemsare mainly used for the provision of heat for district heat-ing systems in some European countries.

C. Coalification

In this process, woody biomass is heated in a nearlyoxygen-free environment. Up to about 200◦C, drying oc-curs, followed at higher temperatures with pyrolytic de-composition of the organic compounds. Substantial liquidand gaseous residues such as tar and carbon monoxideare released in the course of creating the final product,charcoal.

Such processes can be realized with quite different tech-nologies. In developing countries for example, charcoalkilns are usually made from earth or brick, although simplemetal kilns are sometimes found. The efficiency of suchdevices is low (less than 25%) and the airborne emissionsof volatile organic compounds are high, and there can besubstantial production of toxic liquid residues. In indus-trialized countries, charcoal is produced in large fully au-tomatic industrial devices. Here a differentiation is made




between retort and flush gas processes. Within the formerthe charcoal is produced in a batch mode in a closed con-tainer (i.e., retort) where the thermo-chemical conversionfrom wood to charcoal takes place. The heat necessary toallow this process to take place is obtained from the com-bustion of the gaseous and liquid products produced as aby-product. There are also continuous charcoal produc-tion systems where the wood is fed constantly througha big reaction container. Within this container are dif-ferent zones with various settings of reaction conditionsto ensure that the charcoal is produced during the mi-gration of the material through this container. Also herethe gas and liquids produced as a by-product are usedas a source of energy to keep the coalification processgoing.

In general the importance of such processes is relativelylow in industrialized countries because charcoal plays onlya minor role within the energy systems. But charcoal isused in such countries to a certain extend as a raw material(e.g., to produce activated charcoal for filtering) and as fuelfor leisure. In developing countries charcoal is used as aclean fuel for heating and cooking. Additionally in somecountries charcoal is also used for industrial purposes asin Brazil where charcoal is used for steel-making. Theuse of charcoal in these markets is dependent on politicalsupport and economic conditions.

D. Pyrolysis

The decomposing processes realized during pyrolysis aresimilar to those during charcoal production, but here theprocess conditions are set to ensure that the main productis liquid rather than solid. There have been a broad varietyof technologies developed during the last decades attempt-ing to make pyrolysis oil for use in engines without anyadditional processing.

Most promising is flash pyrolysis in which fuel parti-cles are heated very rapidly (more than 1000◦C/sec), butremain in the hot zone for a short time (in general less than1 sec). After this short time, the liquid compounds pro-duced from the solid biomass by decomposing the organiccompounds of the biomass have to be removed and cooledrapidly to avoid further decomposition into gases. To date,flash pyrolysis reactors have been developed mainly to thelaboratory stage.

Reactors with ablative impact designs decompose thebiological raw material primarily into liquid componentson the surface of a hot rotation wheel. On the surface ofthis wheel the solid biomass is “melted.” To avoid furtherdecomposition, the produced liquid components have tobe removed quickly from the hot zone close to the rotatingwheel. Necessarily, gaseous and solid components are alsoproduced, which are used to heat the wheel.

Pyrolysis oil is a mixture of different hydrocarbons,many partly oxygenated, along with charcoal, ash, andwater. The actual composition is strongly dependent onthe pyrolysis process as well as the specific process condi-tions. This is especially true for the average compositionand structure of the hydrocarbons (e.g., chain length,degree of double bonding). This oily liquid can be toxicand is in most cases not stable in air. Additionally the con-version efficiency of the available fast pyrolysis processesis relatively low. The average heating value of pyrolysisoil is approximately 40% that of petroleum-based fuels.

The goal of pyrolysis is the provision of liquid that canbe used directly in engines, but the current state of tech-nology does not achieve it. Upgrading of the produced oilis necessary to ensure easy use in existing engines. Forexample, charcoal and ash particles need to be removedto ensure a sediment-free fuel and, to increase stability,double bonds need to be broken by adding hydrogen. Ad-ditionally other measures have to be taken to ensure therequired viscosity and combustion behavior. Like the fastpyrolysis process itself, these upgrading technologies arenot fully developed yet. Because of these technical andeconomic constraints, there are no applications known forthe provision of pyrolysis oil as a source of energy on afully commercial basis.

E. Gasification

Gasification describes the complete conversion of solidbiomass at high temperatures to a gaseous fuel by addinga small amount of oxidizing reactants to the gasificationprocess. Unlike coalification and pyrolysis, gasificationof biomass is realized in the presence of some oxygen.The main objective of gasification is to transfer the max-imum possible share of the chemical energy within thefeedstock to the gaseous fraction remaining by produc-ing a high yield of low molecular weight products. Thisso-called “producer gas” can be used as a fuel for the pro-vision of heat through direct combustion or used in enginesor turbines. Even use in fuel cells might be possible. Elec-tricity production seems to be a promising option becauseelectricity production via gasification allows much higherefficiency in principle compared to processes driven onlyby direct combustion.

Biomass gasification consists of the following, more orless spatially distributed, steps: heating and drying of thebiomass, pyrolytic decomposition of the biomass (i.e., ex-tracting the volatile components by heating), and gasifica-tion (i.e., partial oxidation of the biomass, partial reductionof the oxidation products (CO2 and H2O to CO and H2),and simultaneous transformation of solid carbon to CO).

The physical and chemical processes of biomass gasi-fication are carried out in a variety of different forms of




equipment and technical concepts. Each of them offerscertain advantages and disadvantages concerning feed-stock possibilities, plant size, and gas quality.

The gasification techniques can be distinguished relatedto different criteria such as reactor type (fixed bed or flu-idized bed), gasifying agent (air, oxygen, or steam), heatsupply (directly or indirectly heated), and reactor pressure(atmospheric or pressurized). Over the years, a consider-able variety of biomass gasifiers have been developed.

� In fixed-bed reactors, the feedstock is exposed to thegasifying agent in a packed bed that slowly movesfrom the top of the gasifier to the bottom, where theash is discharged. Fixed-bed gasifiers are dense-phasegasifiers characterized by a relatively large amount offuel exposed to a limited amount of reactive gas. Infixed-bed reactors the feedstock occupies most of thereactor volume.

� Fluidized-bed reactors are lean-phase gasifiers havinga low ratio of solid to reactor volume. Typically, thefeedstock is occupying only a small fraction of thetotal reactor volume. Fluidized-bed gasifiers areclassified, depending on the intensity of fluidization,as bubbling fluidized-bed gasifiers, circulatingfluidized-bed gasifiers, or entrained flow gasifiers.

� Gasifiers have also been developed that cannot easilybe classified as belonging to one of the former groupsbut have features of both.

Commonly, the goal of gasification is not to providegaseous fuel itself but to provide an easy to handle andenvironmentally sound intermediate energy carrier withclearly defined characteristics that can be converted easily

FIGURE 9 Schematic of an updraft gasifier (left) and downdraft gasifier (right). [From Kaltschmitt, M., and Hartmann,H., eds. (2001) “Energy from Biomass,” Springer, Berlin, Heidelberg, Germany. (In Germany.)]

into another, more valuable, energy carrier, for example,electricity or methanol. To reach that goal, gas cleaningis usually necessary to ensure a long lifetime of the con-version device (e.g., engine or turbine) because the gasproduced within the gasifier does not usually match thefuel requirements in terms of condensable organic com-pounds and/or particles (cf. Kaltschmitt and Bridgwater,1997 and Kaltschmitt, Roesch, and Dinkelbach, 1998).

1. Gasifier

a. Fixed-bed gasifier. In general, fixed-bed gasifiersrely upon gravity feed for the fuel and the extraction of thecoarse ash at the bottom. Due to its continuous decompo-sition, the biomass slowly moves downward passing thezones of drying, degasification, oxidation, and reductionin the reactor. Fairly distinct zones are established withinthe gasifier for each of these gasification steps. The heatnecessary to keep the gasification process going is sup-plied by partial combustion of the feedstock (Fig. 9).

Although feedstock moves from the top to bottom, thegas stream can be co-current, countercurrent, or cross-flowleading to several types in all:

� In downdraft gasifiers, both fuel and the gas movedownward.

� In updraft gasifiers, fuel moves downward and gasupward.

� In co-current gasifiers, the fuel and gas move in thesame direction, usually downward (downdraftgasifiers), but the movement can also be upward.

� In countercurrent gasifiers, fuel and gas move inopposite directions; updraft if the fuel moves up but




flows can be reversed. In cross-draft or cross-currentreactors, fuel moves downward and gas moves at rightangles, i.e., horizontally.

Fixed-bed gasifiers were developed first in the 19th cen-tury and have been sized from a few 10–100 kW ther-mal capacity. Updraft gasifiers have been developed forhigher thermal capacities, but only for the provision ofheat. Fixed-bed gasifier technology is relatively well-developed, but sometimes losses (i.e., heat, unburnedmaterial) are fairly high. Unfortunately, gas quality of-ten does not match the requirements of gas conversionunits (e.g., engine and turbine) due to high particle andtar content in the producer gas. Also the environmentalimpact is sometimes fairly high.

b. Fluidized-bed gasifier. Fluidized-bed gasifierswere originally developed for coal gasification and havebeen adapted for biomass conversion. The gasifying agentis fed from the bottom into the gasifier at a speed highenough to “fluidize” or swirl the feedstock within theconversion reactor. Unlike fixed-bed gasifiers, no distinctreaction zones exist within this fluidized bed. All differentongoing thermo-chemical conversion processes are tak-ing place in the same volume. Fluidized beds are versa-tile and insensitive to fuel characteristics, except for fuelsize, which may require feedstock to pass through a size-reduction stage.

Fluidized beds have high rates of heat and mass transferand provide good mixing, meaning that reaction rates arehigh, the residence time of the fuel particles are short,and the temperature is more or less constant in the bed.Fluidized beds can be operated at low capacity, but thisis not economic in most cases. Therefore, such gasifiersare usually commercially viable with thermal capacities ofseveral megawatts. They can generally be scaled up easilyto several tens of megawatts or even higher.

Fluidized-bed gasifiers can be operated either under at-mospheric pressure or as pressurized systems. Pressurizedgasification has the advantage that downstream compres-sors are not needed to reach the pressure requirements ofthe turbine where the producer gas might be used. Further-more, high-pressure gasification plants can be constructedmore compactly and have, in principle, higher efficienciesthan atmospheric-pressure plants, but they are also morecostly in most cases. There is no clear judgment yet as towhich is best for power production from an overall tech-nical and economic point of view.

There have been several fluidized-bed gasifiers built re-cently. Their disadvantage is that the producer gas is quitedirty, requiring demanding gas-cleaning systems. Never-theless, this gasification approach seems to be the mostpromising for future large-scale applications.

2. Gas Cleaning

Depending on feedstock, process conditions, and gasi-fying agent, the producer gas consists mainly of differ-ent fractions of carbon monoxide (CO), carbon dioxide(CO2), methane (CH4), hydrogen (H2), and nitrogen (N2).In addition, it could contain steam (H2O), nonmethane hy-drocarbons (e.g., ethylene, propane), as well as unwantedcomponents, especially condensable organic compounds(i.e., tars) plus unburnt carbon, ash, and other particles.Biomass gasification with air results in a high N2 (60%)content in the producer gas, and thus in a gas with a lowheating value, i.e., 4–6 MJ/m3

n. The heating value can behigher if pure oxygen or water is used as a gasifying agent.

Each application of the producer gas requires a differentlevel of gas cleaning to prevent erosion, corrosion, depo-sition, and other problems downstream of the gasifier. Gascleaning usually starts with the separation of particulatesand condensable organic compounds from the raw pro-ducer gas. Depending on needs, additional removal of mi-nor gas impurities, e.g., alkali metals, can be required. Forparticle removal, several filter techniques are available,such as cyclones, fabric filters, electrostatic precipitators,granular filter beds, and ceramic candle filters as well aswashing towers. Technologies are often combined to takeadvantage of the individual strengths and avoid the weak-nesses of each.

In gasification processes, it seems to be impossible tototally prevent the formation of condensable organic com-pounds by the design of the reactor and modification ofgasification conditions. Reducing the amount of high- andlow-molecular-weight condensable organic compoundsby adsorption techniques, wet scrubbing, or catalytic de-composing into noncondensable organic compounds al-lows the gas to be cooled for entry into downstream clean-ing equipment preventing condensation of organics thatmay cause severe plugging and fouling.

Cheap, reliable gas cleaning is still the greatest need inbiomass gasification for electricity production. No avail-able single or combination technique is able to fulfill therequirements of gas engines or turbines in a commercialway, although there have been promising demonstrationprojects (e.g., Vermont Project, United States; VarnamoPlant, Sweden). This arena has a high priority for addi-tional research and development.

3. Gas Utilization

The gas generated from biomass gasification may be fireddirectly, e.g., for process heat within a retrofitted furnaceor boiler, or fed to an engine, turbine or fuel cell for theprovision of mechanical power or electricity (along withheat in a CHP system).




If producer gas is cleaned to the level of purity found innatural gas, there is a wide range of possible applications,as noted above. Since this is currently not economicallypossible, there are efforts to develop more robust enginesand turbines that might tolerate less clean gas.

The conversion of gas to liquid secondary energy carrieris also technically possible. The most widely known pro-cess converts the carbon monoxide in the gas to methanolby adding hydrogen. This hydrocarbon can be used as anadditive to conventional petrol as well as a raw material forthe chemical industry. To date, however, no commerciallyviable facility has been developed. In large plants it is alsofeasible to use the producer gas to produce chemicals suchas methanol, hydrogen, or ammonia.

IV. PHYSICO-CHEMICAL CONVERSION

The technology for producing fuel oils from biomassis similar to that for the production of vegetable oil asfood or fodder. Some biomass contains fatty and/or oilycomponents, for example, rape and sun flower seeds, co-conuts, peanuts, and corn. In some cases, the potentialyield is high. For example, rapeseeds contain 40–45%vegetable oil. These crops can only be cultivated in cer-tain regions and require relatively high agricultural ef-forts to produce and thus the potential for their use islocalized.

In most cases, vegetable oil from these sources is morevaluable as food or fodder, but use as a fuel is possible.Vegetable oil is easy to store, has a relatively high en-ergy content, and can be used without major problemsin transport. Additionally, use in small-scale CHP plantsfor rural electrification in developing and in industrializedcountries as well as in heating plants is possible. Becausethe fuel characteristics of conventional diesel fuel and rawvegetable oil are quite different, treatment of the vegetableoil is often needed. Therefore the vegetable oil is convertedinto fatty acid methyl ester (FAME).

FIGURE 10 Esterification of vegetable oil from rape seed to fatty acid methyl ester. [From Kaltschmitt, M., andHartmann, H., eds. (2001). “Energy from Biomass,” Springer, Berlin, Heidelberg, Germany. (In German.)]

A. Basics

Vegetable oil combusts differently from conventionaldiesel fuel because it is a triglycerid (Fig. 10). Thereforean adaptation of existing diesel engines to the specific fuelcharacteristics of vegetable oil is necessary. This adapta-tion is expensive and reduces the lifetime of the engine.Therefore, vegetable oil is sometimes chemically treatedto better match the combustion behavior of diesel fuel byconverting it into FAME. This is done by removing theglycerine from the triglycerid (Fig. 10). To saturate theester bindings instead of the glycerin, methanol is addedto the remnant. This reaction is realized with the help ofa catalyst.

B. Oil Production

Three approaches are available for producing oil: press-ing only, extraction only, and a combination of both. Dur-ing the pressing, the oil is squeezed out of the oil seedsbased on a one- or two-stage pressing process leaving veg-etable oil and oil cake. The latter still contains 4–10% oil.Sometimes this oil cake is used as a cattle feed wherethe remaining oil contributes to the nutrition value of thisproduct. In developing countries and within traditional oilproduction processes in industrialized countries (e.g., tra-ditional olive oil production), this simple technology isstill of considerable importance.

Extraction occurs by means of a solvent applied in acountermovement in large-scale units. This technologyhas been widely employed for many decades. Such de-vices produce two different material streams: the solventsaturated with oil and the oil-free extraction residue sat-urated with solvent. The solvent must then be removedfrom both streams by heating and reused. Compared tosimple pressing, a much higher share of the oil originallywithin the oil seed can be removed. After extraction themeal contains significantly less than 1% oil and can beused as cattle feed.




For biomass with high oil contents, like rapeseed, press-ing followed by extraction will maximize production andbe economic. The produced vegetable oil contains about0.5–6% oil-free solid residues originating from the seed.The amount of such residues mainly depend on the con-dition of the pressing aggregate, the flow-rate, the kind ofpressed seed, and the water content in the oil seeds. Thesolid residues have to be separated from the oil by filtrationand sedimentation to prevent possible problems during thefurther downstream processing of the vegetable oil. Evenafter this cleaning step, however, the vegetable oil is notqualified for direct utilization because it still contains un-wanted substances (fatty acids, ketones, wax, pigments,heavy metals, pesticides) reducing shelf-life and compli-cating subsequent processing and use. Further refining isneeded and typically leads to a loss of 4–8% of the us-able oil mass through de-acidification, de-coloring, andsteaming.

C. Esterification

Figure 11 shows a continuous two-stage esterification inwhich the mixture of cleaned vegetable oil together withthe catalyst and the methanol is pumped with a low veloc-ity through a vertical pipe. The low velocity of the liquidensures that the glycerin produced during the esterifica-tion process can settle in the reactor and removed for useas a raw material. Additionally glycerin is removed withina separation tank as well as by a separation unit. After theremoval of the remaining methanol, the liquid is cleaned

FIGURE 11 Example for a continuously working esterification process. [From Kaltschmitt, M., and Hartmann, H.,eds. (2001). “Energy from Biomass,” Springer, Berlin, Heidelberg, Germany. (In German.)]

by a multistage washing process. The produced FAME isnow ready to be used as a fuel.

There are FAME plants operating in Europe. InGermany, for example, plants with a capacity of 300,000t/y FAME are in operation and more are planned. Manufac-turing FAME is therefore state of technology, although theenergy balance is fairly poor compared to other biomassconversion routes. Accordingly, the costs are relativelyhigh.

D. Use

Pure vegetable oil can be used in some existing en-gines, but it lowers engine lifetime and increases main-tenance requirements. But under specific conditions thiscould be a promising option to provide power especiallyin rural areas also for rural decentralized electrification.This is true for developing as well as for industrializedcountries.

Fatty acid methylester can be used directly as a dieselsubstitute in conventional diesel engines (Table V). Large-scale demonstration projects with FAME based on rape oil(RME) have shown that there are no significant problems,although the more aggressive chemistry of RME com-pared to conventional diesel fuel requires fuel lines to bemade from a more resistant material like Teflon® (Heinzet al., 2000).

FAME has promising environmental advantages com-pared to conventional diesel fuel in certain niche applica-tions. It is more easily degraded in nature and shows lower




TABLE V Selected Properties of Diesel Fuel, FAME, andRape Oil

Diesel fuel FAME Rape Oil

Density (15◦C) in kg/m3 820–845 875–900 900–930

Viscosity (40◦C) in mm2/s 2.0–8.0 3.08 78.7

Flashing point in ◦C >55 Min. 110 Min. 220

CFPP-Value in ◦C Max. 0 Max. 0(summer/winter) Max. −20 Max. −20

Sulfur content in mg/kg Max. 350 Max. 100 Max. 20

Cetan-Value Min. 51 Min. 49

Calorific value in MJ/kg 42.7 37.9 Min. 35

particle emissions, for example, thus showing promise foruse in water conservation areas and congested cites.

V. BIO-CHEMICAL CONVERSION

Nearly all biomass is eventually decomposed naturallythrough biological processes. Some of these processes canbe harnessed to produce fuels.

� Composting occurs if oxygen is available. Duringcomposting the biomass is degraded by bacteriamainly to carbon dioxide (CO2) and water (H2O) whilereleasing low temperature heat that can, in principle beused via a heat pump. But for the time being thisoption is only of theoretical importance.

� Under anaerobic (oxygen-free) conditions, a variety ofdegrading processes are employed by microorganisms.Most relevant to energy are ethanol production viaalcohol fermentation and biogas (methane) productionvia anaerobic digestion.

A. Alcoholic Fermentation

Sugar (C6H12O6) is converted to ethanol (C2H5OH), car-bon dioxide (CO2), and low temperature heat by yeastunder anaerobic conditions. Because starch and even cel-luloses can be converted more or less easily into sugar,such biomass streams are also a potential resource forthe production of ethanol in addition to naturally sugar-containing crops like sugar cane and sugar beet. For fuel,ethanol is needed in a pure form and therefore further stepsare needed after fermentation.

1. Basic Processes

Sugar (C6H12O6) is converted to ethanol (C2H5OH), car-bon dioxide (CO2), and low temperature heat by yeastunder anaerobic conditions:

C6H12O6 → 2C2H5OH + 2CO2 + 88 kJ/mol

According to the above equation 100 kg of sugar isconverted into 51.14 kg of ethanol and 48.86 kg of car-bon dioxide and 400 kJ of heat. This heat is needed toallow the microorganisms to live and grow. To reach aquick transformation of the sugar into ethanol, additionalgrowth supporter and mineral nutrients, as well as opti-mal temperature and pH are needed. Most processes arerealized at a temperature level between 25 and 40◦C. Thisprocess is well known because of humanity’s long andextensive production of alcoholic beverages.

Because starch and even celluloses can be convertedmore or less easily into sugar, such biomass fractions arealso a potential resource for the production of ethanolin addition to naturally sugar-containing crops like sugarcane and sugar beet. For fuel, ethanol is needed in a nearpure form and therefore further steps are needed after fer-mentation.

2. Substrate Production

For fermentation, sugar is needed in a water solution.Therefore sugar has to be extracted from biomass con-taining sugar (like sugar cane, sugar beet) and/or has tobe produced from biomass containing starch (like grain,potatoes) or celluloses (like wood).

� Biomass containing sugar. Sugar can be directlyremoved from sugar cane and sugar beet and thus thesetwo species are the most widely used for this purpose.Sugar cane is a grass growing mainly under tropicalclimatic conditions containing a sweet liquid that iseasily removed by pressing and accounts for about30% of the weight of fresh cane. The fiber remainingafter pressing is called bagasse. It is used often as asolid biofuel for the provision of the energy needed torun a sugar mill and/or an alcohol production factory.After several preparation and processing steps thisliquid can be used as a feedstock for alcoholicfermentation. Sugar beet, which grows in temperateclimates, is first chipped into small pieces from whichthe sugar is extracted by means of water appliedcountercurrent to the flow direction of the chips. Theremaining products are the nearly sugarfree extractedbeet which can be used as a cattle feed.

� Biomass containing starch. Cereal, corn, potatoes,topinambur, maniok, and other similar crops containstarch, which is a polysaccharide consisting of longchains of glucose modules. To produce sugar fromstarch, this polysaccharide has to be decomposed insingle saccharid rings by means of biological processesbased on enzymes. Due to the variety of starchcontaining biomass crops, there is no typical chemicalprocess. For each raw material, a process has beendeveloped and optimized in the past because these




materials are often used also for the production ofalcoholic beverages. In most cases, however, water isadded to the chipped or milled material. Due toswelling, the starch becomes a paste to which is addedthe respective enzyme adapted to each type of biomass.Under the right process conditions, the starch is thenconverted into sugar, which can be used as a feedstockfor the alcoholic fermentation.

� Biomass containing celluloses. All biomass containscelluloses as this is one of the most important elementsof plants. Like starch, celluloses is built of sugarcomponents. Therefore celluloses can be split againinto sugar, in this case both through biological andnonbiological processes. The former are based onspecial enzymes, but are relatively slow and inefficient.The latter are realized with an acid catalyst-basedhydrolysis. Such a process is technologically verydemanding and thus alcoholic fermentation based onsugar from celluloses has no importance for the timebeing.

3. Alcoholic Fermentation

The sugar-containing feedstock is next inoculated withyeast. Continuous operation is usually done for small-scale production of high-quality spirits such as beveragesand discontinuous (batch) operations are used in large-scale systems for production of ethanol as a raw materialfor the chemical industry. After the fermentation, the yeastis removed from the slurry and recycled.

4. Production of Pure Alcohol

After fermentation the material contains 8–10% alcohol,the rest being primarily water and the residues from thesugar- or starch-containing organic material. Distillationor rectification is done to purify the product using a crudealcohol column. The result is an alcohol-water-mixturewith a high share of alcohol (more than 80%) and a slurrywithout significant alcohol content. The latter is used asan animal fodder and/or as a fertilizer. It is also suitablefor biogas production. This technology is in large-scaleoperation worldwide.

At best, distillation and rectification will only achieve95–96% pure ethanol. Engine fuel must be 99.9% pure,which can be achieved via a further step called absoluta-tion. Here, a third chemical (expident) is added to thealcohol-water-mixture producing a mixture of alcohol-chemical plus water-chemical. Afterward the chemical isremoved from both mixtures and reused.

5. Use

In engines, ethanol can substitute fully or partly for gaso-line although the LHV of ethanol is lower than that for

TABLE VI Selected Properties of Fuels

Ethanol Gas

Composition in %

Carbon 52 86

Hydrogen 13 14

Oxygen 35 0

LHV in MJ/kg 26.8 42.7in MJ/1 21.3 ca. 32.0

Density (15◦C) in kg/l 0.794 0.72–0.78

Viscosity (20◦C) in mm2/s 1.5 0.6

Boiling point in ◦C 78 25–215

Flame point in ◦C 12.8 −42.8

Ignition temperature in ◦C 420 ca. 300

Evaporation heat in kJ/kg 904 380–500

Minimum air volume in kg/kg 9 14.8

Octane-value 107 93

gasoline (Table VI). The combustion of ethanol, however,required a lower air volume. Therefore the heating valueof the mixture pressed into the cylinder is more or lessthe same for ethanol and gas. This is the reason why anengine powered by ethanol produces the same power asan engine driven by gas.

Internal combustion engines must be specially adaptedto ethanol, because it shows different combustion behav-ior compared to gasoline. Adaptation kits have been usedsuccessfully in several countries, for example, Brazil. Un-der current petroleum prices, pure ethanol is a high-pricedalternative as well as requiring modified distribution net-works and such problems as water retention during stor-age. In a number of countries, however, ethanol is mixedwith gas to a maximum of 10% in which form it can beused without any known problems in existing engines anddistribution networks. Additionally, ethanol can be con-verted into ethyl tertiary butyl ether (ETBE) for use as apollution-reducing gasoline additive.

B. Anaerobic Digestion

During anaerobic digestion, organic material is de-composed in an oxygen-free atmosphere by bacteriathat produce a gas containing approximately two-thirdsmethane (CH4) and one-third carbon dioxide (CO2) plussome impurities. Such decomposition occurs in nature,for example, at the bottom of lakes and moors within thesediments containing organic material, and takes place inlandfills and dumps where the organic fraction of the wastematerial is converted into landfill gas.

1. Basics

In addition to degradable organic input and freedom fromair, additional nutrient for the bacteria and absence of




FIGURE 12 Anaerobic process scheme. [From Neubarth, J., and Kaltschmitt, M., eds. (2000). “Renewables inAustria—Systemtechnology, Potentials, Economics, Environmental Aspects,” Springer, Vienna, Austria. (In German.)]

harmful, pathologic, and inhibiting substances are alsoneeded for success. Also the organic material has to be

bic digestion are agricultural residues (e.g., liquid manure,leaves from sugar beet), residues from the food processingindustry (e.g., slurry of fruit and potato processing, sometypes of slaughterhouse residues), and sewage sludge.Only lignin-containing biomass like wood cannot be de-graded by anaerobic bacteria.

During anaerobic digestion, organic matter is degradedby three different kinds of bacteria: fermentative, aceto-genic, and methanogenic. The first two bacteria familiesdegrade the complex organic compounds of biomass intosimpler intermediates (Fig. 12). These intermediate prod-ucts are then converted to methane and carbon dioxide bythe methanogenic bacteria.

Anaerobic digestion relies on a dynamic equilibriumamong the three bacterial groups, which is strongly af-fected by temperature. Most digesters operate in themesophilic temperature regime (with a peak in microbeactivity at around 35◦C). Some operate in the thermophilicregime (with a peak in microbe activity around 55◦C). Upto the temperature of peak microbial activity, higher op-erating temperatures produce greater metabolic activitywithin either regime. The pH value, the composition ofthe biomass in relation to easily degradable biomass com-pounds, and rate of loading of the feedstock affect the bac-terial balance also. For optimal fermentation, pH-valuesbetween 6.8 and 7.2 are often maintained by use of buffers.

TABLE VII Yield Targets of Biogas for Different Substrate

Yield of biogas in m3/t Yield of biogas in m3/tMaterial organic dry matter Material organic dry matter

Liquid manure from beef 250 Paunch content 420–520

Liquid manure from pork 480 Rey straw 300–350

Droppings from chicken 450 Potato herbs 560

Sewage sludge 400 Sugar beet leaves 550

Organic waste from households 170–220 Food residues 80–120

Waste fat 1040 Waste water from brewing industries 500

Roadside green 550 Waste water from sugar industries 650

The composition of the produced biogas is typicallyabout two-thirds methane and one-third carbon dioxide.Additionally the biogas may contain trace substances suchas H2S, depending on the composition of the biomass,the process conditions, and other parameters. The LHV ofbiogas, which depends mainly on methane content, rangesfrom 14 to 29 MJ/m3.

Expected biogas yields for different substrates aregiven in Table VII. Already digested substrates like liq-uid manure and droppings show, in general, smalleryields than fresh material, e.g., roadside green or wastefat. Economic viability, however, also depends on therate.

2. Biogas Production

The bacteria use relatively little of the biomass energyfor their own survival and only a small amount of heatis produced, making the conversion process of the or-ganic material into biogas fairly efficient. To ensure opti-mal process conditions with a good access of the bacteriato the organic material so that gas yields are maximized,the organic material needs to be in a slurry with a wa-ter content of more than two-thirds. To optimize produc-tion rates, the slurry should be kept at temperatures of28–35◦C or 50–70◦C, which correspond to the ranges ap-propriate to different bacterial species. Production willoccur at lower temperatures, but essentially stops at10◦C.




In Europe, the main task of this conversion route is totreat organic waste or residues to reduce waste disposalcosts; the production of energy is only a by-product inmost cases (like the treatment of sewage sludge).

Typical feedstock for a biogas production is organicmaterial available at low (or zero) costs. A big advantageoccurs if it is already available with high water content(i.e., in most cases significantly more than 90% water), aswith sewage sludge, animal manure, and organic residuesfrom some types of agricultural production and the foodprocessing industry. Within biogas plants, preparation ofthe feedstock is needed that might include short-time stor-age, sedimentation of mineral contaminants (like sand),reducing the material into small pieces, mixing of differ-ent types of feedstock to maximize the gas yield, and heat-ing to the needed temperature level. Then the feedstock ispumped into the biogas container or reactor where anaer-obic fermentation takes place. For successful operation,the bacteria must always be well-mixed with the organicmaterial. It is also important to realize a good temperaturedistribution within the reactor. The biogas is removed fromthe top of the plant and, after removal of impurities likewater, stored before use as a source of energy is realized.

The digested material is removed from the reactor andstored in a tank where a small amount of additional bio-gas is produced. This digested slurry is used as a fertilizer,because it contains the nitrogen originally part of the feed-stock. Since biogas technology helps to close the nutrientcycle, it is of increasing importance in environmentallysound waste management. In China, human as well as an-imal waste is often added to household biogas digestersas well, giving them a sanitation as well as an energyfunction.

FIGURE 13 Example of a biogas plant using animal manure as a feedstock. [Form Neubarth, J., and Kaltschmitt, M.,eds. (2000). “Renewables in Austria—Systemtechnology, Potentials, Economics, Environmental Aspects,” Springer,Vienna, Austria. (In German.)]

Figure 13 shows a biogas plant commonly used in coun-tries like Denmark, Germany, and Austria. The animalmanure from the stable is mixed, stored for a short time,heated, and pumped into the biogas reactor. This mixingfacility is driven by a slow-moving electric motor.

3. Use

Biogas can be used in a boiler, stove, or engine for theprovision of heat, mechanical power, and/or electricity. Apart of the heat and electricity produced is often partly ap-plied to warming up the feedstock and running of the plant.This is a system layout most widely used in industrializedcountries. Alternatively or additionally, biogas can also befed into the local natural gas system, to do so requires a de-manding upgrading of the biogas to fulfill the local naturalgas standards. This is done in Switzerland and Sweden,for example. Use of the biogas as a transportation fuel ispossible and has been shown in demonstration projects,but it is not clear at present if such options will becomemore important in the future.

In developing countries the use of the biogas for house-hold cooking and lighting is becoming common in areaswith appropriate climates, sufficient animals to providedung as feedstock, and access to low-cost capital fundsto build the digesters. The technology is simpler thanused in large-scale facilities, for example, not involving aheating system and relying on manual feedstock prepara-tion. This results in lower biogas yields, higher operatingcost per unit energy, and low or zero production duringthe cold season. Community-scale facilities in which oneplant supplies gas to many households are used in sev-eral Asian countries, although experience has been mixed.




Large-scale plants, similar to those used in industrial-ized countries, are also in use in developing countries atpig farms and other facilities with concentrated feedstockavailability.

VI. OUTLOOK

Biomass fuel will continue to be important locally aroundthe world. As oil and gas prices move upward over time,more efficient modern technologies will be employed toconvert biomass energy to these easier to use forms alongwith expanded use for power and co-generation (Karthaand Larson, 2000 and Hall et al., 2000). In developingcountries, although economic growth will lead to a shiftup the household energy ladder to liquid and gaseous fuelsfor hundreds of millions of people, for many decades therewill still be a billion or more of the poorest people who willbe unable to switch from using unprocessed traditionalfuels. Moving these people to cleaner and more efficientalternatives would take concerted efforts involving newtechnologies, subsidies, and education at the scale nowemployed to provide clean water and sanitation. Thereare also many attractive uses for biomass in commer-cial and small-scale industrial applications in develop-ing countries (Larson, 2000). There are prospects for ad-vanced biomass conversion technologies to liquids andgases, such as application of the Fischer-Tropsch processto create synthetic LPG from crop residues, that showpotential, but are at present too speculative for accurateevaluation. Should oil and gas prices remain high for longperiods, however, there may be sufficient incentive to de-velop such biomass alternatives. Finally, if serious long-term efforts are made to reduce carbon emissions throughtaxes, subsidies, and other incentives, biomass fuels willlikely become much more attractive. As much as half ofworld energy could be provided by 2050 with a major ef-fort according to some estimates (Goldenberg et al., 2000).At the scale required to significantly reduce global emis-sions from fossil fuel, however, competition with agricul-ture and other sectors for land, water, and other resourceswould become issues.


BIOMASS, BIOENGINEERING OF • BIOMASS UTILIZATION,LIMITS OF • ENERGY EFFICIENCY COMPARISONS AMONG

COUNTRIES • ENERGY FLOWS IN ECOLOGY AND IN

THE ECONOMY • ENERGY RESOURCES AND RESERVES •POLLUTION, AIR • WASTE-TO-ENERGY SYSTEMS

BIBLIOGRAPHY

Baldwin, S. (1986). “Biomass Stove Technologies: Engineering Design,Development, and Dissemination,” Volunteers in Technical Assistanceand Princeton University Center for Energy and Environmental Stud-ies, Princeton, NJ.

Goldenberg, J. et al. (2000). “World Energy Assessment,” United NationsDevelopment Programme, New York.

Hall, D. O., House, J. I., and Scrase. I. (2000). An overview of biomassenergy. In “Industrial Uses of Biomass Energy: the Example of Brazil,”(F. Rosillo-Calle, H. Rothman, and S. V. Bajay, eds.), Taylor & Francis,London.

Heinz, A., Kaltschmitt, M., Hartmann, R., and Hitzler, G. (2000). “TheFirst Year of Operation of the Biodiesel-Fuelled Energy Supply Sys-tem of the New German Parliament Building in Berlin—Analysis ofthe Technical Feasibility, Environmental Aspects and the Costs,” 11thEuropean Biomass Conference, Sevillia, Spain.

International Energy Agency, World Energy Outlook, Paris, 2000.Kaltschmitt, M. (1999). Utilization of Biomass in the German Energy

Sector. In “Strategies and Technologies for Greenhouse Gas Mitiga-tion” (J.-F. Hake, N. Bansal, and M. Kleeman, eds.), Ashgate, Alder-shot, U.K.

Kaltschmitt, M., and Bridgwater, A. V., eds. (1997). “Biomass Gasifi-cation and Pyrolysis—State of the Art and Future Prospects,” CPLScientific, Newbury, U.K.

Kaltschmitt, M., and Hartmann, H., eds. (2001). “Energy from Biomass,”Springer, Berlin, Heidelberg. (In German.)

Kaltschmitt, M., and Reinhardt, G. A., eds. (1997). “Regrowing En-ergy Carrier; Basics, Processes, Eco Balances,” Vieweg, Braun-schweig/Wiesbaden. (In German.)

Kaltschmitt, M., Reinhardt, G. A., and Stelzer, T. (1997). Life cycleanalysis of biofuels under different environmental Aspects. BiomassBioenergy 12, 121–134.

Kaltschmitt, M., Rosch, C., and Dinkelbach, L., eds. (1998). “BiomassGasification in Europe,” European Commission, DG XII, Brussels,Belgium.

Kaltschmitt, M., and Wiese, A., eds. (1997). “Renewable Energies,Systemtechnology, Economics, Environmental Aspects,” Springer,Berlin. (In German.)

Kartha, S., and Larson, E. D. (2000). “Bioenergy Primer, ModernizedBiomass Energy for Sustainable Development,” United Nations De-velopment Programme, New York.

Larson, E. D., ed. (2000). Modernized biomass energy, special issue ofEnergy Sustainable Deve. 4(3).

Neubarth, J., and Kaltschmitt, M., eds. (2000). “Renewables in Austria—Systemtechnology, Potentials, Economics, Environmental Aspects,”Springer, Vienna, Austria. (In German.)

Reed, T. B. and Larson, R. (1996). A Wood-Gas Stove for DevelopingCountries. In “Developments in Thermochemical Biomass Conver-sion,” (A. V. Bridgwater, ed.), Blackie Academic Press, London.

Rosch, C., Kaltschmitt, M., and Limbrick, A. (2000). “Standardisationof Solid Biofuels in Europe,” 11th European Biomass Conference,Sevillia, Spain.

Smith, K. R. (1987). “Biofuels, Air Pollution, and Health,” Plenum, NewYork.

Smith, K. R. (2000). National burden of disease in India from indoor airpollution. Proce. Nati. Acad. Sci. 97(24), 13286–13293.

Smith, K. R., Zhang, J., Uma, R., Kishore, V. V. N., Joshi, V., and Khalil,M. A. K. (2000). Greenhouse implications of household fuels: Ananalysis for India. Annu. Rev. Energy Environ. 25, 741–763.

P1: ZBU/GUB P2: GRB Final Pages Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology EN015J-700 August 2, 2001 11:53

Solar Energy in BuildingsS. V. SzokolayUniversity of Queensland

I. Solar CollectorsII. Hot Water SystemsIII. Space HeatingIV. Thermal Behavior of BuildingsV. Passive Systems

VI. Solar CoolingVII. Design Methods

VIII. User InfluenceIX. EconomicsX. Urban ProblemsXI. Recent Developments

GLOSSARY

Active systems Solar–thermal systems in which heattransfer takes place by mechanically driven fluid(water, air) flow, usually in closed channels (pipes orducts) where pumps or fans (blowers) use a significantamount of conventional form of energy, if CoP < 20(see definition of coefficient of performance CoP).

Auxiliary energy Ea Energy from conventional sources(boilers, furnaces, etc.) used to supplement the energyfrom the solar source should the solar source fall shortof the demand.

Coefficient of performance CoP Ratio of energy ofsolar origin Es delivered by the system as usefulheat (excluding Ea) to the parasitic energy used;CoP = Es/E p (this definition differs somewhat fromthat used for cooling machines or heat pumps).

Hybrid systems Essentially passive systems with somesmall amount of conventional energy used to assistthe otherwise spontaneous heat transfer processes, if50 > CoP > 20.

Low temperature systems Solar–thermal systems fordelivering heat at a temperature below 100◦C, mostoften between 40 and 80◦C.

Parasitic energy E p Energy from conventional sources(e.g., electricity) used to drive the heat transfer mech-anisms of the system, such as pumps, fans (blowers),and controls.

Passive systems Solar–thermal systems in which heattransfer takes place predominantly by spontaneous(natural) processes, if CoP > 50.

Selective surfaces Generally, surfaces having absorptionand emission characteristics varying as a function ofradiation wavelength; specifically, surfaces with high

111

P1: ZBU/GUB P2: GRB Final Pages


112 Solar Energy in Buildings

absorptance for solar wavelengths (∼6000◦K), but lowemittance at normal operating temperature (∼350◦K).

Solar fraction Fs Fraction of the total load L (e.g.,heating requirement) supplied from the solar source;the remainder is the auxiliary energy. Fs = Es /L =1 − (Ea /L).

Solar–thermal systems Systems for converting radiantsolar energy into thermal energy (heat) and deliveringthis heat where and when it is required for use.

SOLAR ENERGY can be utilized in many ways (e.g.,photovoltaics, solar–thermal–electrical systems, solarponds, and biomass conversion), but only its applica-tion to provide (or contribute to) the heat requirementsof buildings will be discussed here (with a brief men-tion of building-integrated photovoltaics). These are alllow-temperature heat requirements and may include theheating of water (for domestic hot water systems or swim-ming pool heating), the heating of the building space itself,and possibly also space cooling.

The three essential functions of such systems are

1. Collection. The conversion of radiant solar energyinto heat

2. Storage. The system used to keep the collected heatand make it available when required (i.e., distribution intime)

3. Distribution (in space). The way to supply this heatwhere it is required.

The heat transfer process must be driven and controlled,and an auxiliary heat source must be included should thesupply from the solar source be insufficient at times.

These systems may be mechanical installations, es-sentially independent of the building (although perhapsattached to or integrated with some elements of thebuilding), which are generally referred to as active sys-tems, or they may be constituted by the building itself(where elements of the building and their configurationmay be designed to perform the three functions), whichare known as passive systems. For the discussion of thelatter, it is essential to consider the thermal behavior ofbuildings because all buildings are to some degree passivesolar systems.

I. SOLAR COLLECTORS

If a surface is exposed to solar radiation, it will be heated.In this sense, any such exposed surface is a solar collector.The degree of this heating will depend on (1) the amountof radiation incident on the surface and (2) the absorp-

tance of the surface. Radiation incident on a surface canbe measured as

1. Irradiance, or the instantaneous density of radiant flux(energy flow rate or power density), in watts persquare meter (W/m2). The term intensity is often (buterroneously) used to mean this quantity.

2. Irradiation, or energy density, i.e., the amount ofenergy incident on the surface over a specified timeperiod (an hour, a day, etc.) in joules per square meter(J/m2) or watt-hours per square meter (Wh/m2).

Radiation from the sun can reach a surface three ways;thus the total irradiance (denoted G for global) of a sur-face may consist of three components: Gb, which is thedirect beam irradiance arriving along an unobstructedstraight path from sun to surface; Gd, which is the dif-fuse irradiance (i.e., radiation scattered by the atmosphere)apparently arriving from all directions within the skyhemisphere; and Gr, which is reflected irradiance from theground surface or from other objects in the environment.Thus, G = Gb + Gd + Gr.

Radiation falling on an opaque surface is either re-flected or absorbed. The proportions reflected and ab-sorbed are measured by the surface qualities reflectanceρ and absorptance α. The sum of the two must be unity:ρ + α = 1. For a surface exposed to radiation, the rate ofenergy input is G α. As the surface is heated and becomeswarmer than its surroundings, it will lose heat at a ratedepending on the surface conductance (ho) and the tem-perature difference between the surface (Ts) and the air(Ta): ho(Ts–Ta). The surface will reach an equilibriumtemperature when the heat input rate equals the heat lossrate, that is, when G α = ho(Ts–Ta), from which the sur-face temperature can be expressed as Ts = Ta + (G α/ho).This is the maximum temperature the surface could reachin the absence of any heat flow into the body of ma-terial behind that surface, and it is often referred to asthe sol-air temperature. (More precise definitions wouldinclude a radiant emission term, so that the numera-tor would become the net irradiance absorbed by thesurface.)

If we have an absorber plate that incorporates fluid chan-nels through which some liquid or air can be circulated,then the heat collected by the absorber plate can be re-moved and transported to storage or to a location where itis needed. Figure 1 shows some possible arrangements ofsuch an absorber plate.

The flat plate collector consists of an absorber platein a casing, incorporating back and edge insulation andone or two sheets of transparent cover (most often glass).The thermal processes in such a collector are illustratedin Fig. 2.



Solar Energy in Buildings 113

FIGURE 1 Some absorber plate constructions.

The instantaneous collector efficiency is

η = useful heat delivered

irradiance on collector surface= Quse

G Ac, (1)

where Ac is the collector surface (aperture) area (in squaremeters) and G the global irradiance, because flat platecollectors can utilize the diffuse and reflected components,as well as the beam component (as opposed to optical

FIGURE 2 Thermal processes in a flat plate collector.

concentrators, that is, lens or mirror-type collectors, whichcan only respond to the beam component).

Quse = mCρ(Tout − Tin),

where m is the fluid mass flow rate through the collector (inkilograms per second, Cρ the specific heat of fluid (joulesper kilogram Kelvin), Tout the output fluid temperature (indegrees Celsius), and Tin the inlet fluid temperature (indegrees Celsius).

An analytical expression for useful heat delivered is

Quse = Qin − Qloss , (2)

where the heat input is

Qin = GAc(τα)F (3)

where τ is the transmittance of the glass cover, α is theabsorptance of the surface, F the absorber heat removalfactor; the heat loss is

Qloss = AcUFT , (4)

where U is the global heat loss coefficient of collectorper unit aperture area (in watts per square meter Kelvin),and the difference in temperature is T = [(Tout + Tin)/2] − Ta .

Substituting Eqs, (3) and (4) into (2), and Eq. (2) into(1), we get

η = (τα)F − UFT/G.

This is the collector efficiency function (Fig. 3) of typey = abx , where the two collector constants a = (τα)F andb = UF are found by testing.

FIGURE 3 Efficiency functions of three collectors: (1) singleglazed, η = 0.73 − 8.7 (T/G); (2) double glazed, η = 0.68 − 4.9(T/G); and (3) selective surfaced, η = 0.63 − 3.3 (T/G).




If T = 0 (i.e., there is no loss), then ηo = (τα)F , whichis the no-loss efficiency, the y-axis intercept. The collectorstagnation temperature is reached when the efficiency isreduced to zero; thus

(τα)F = UF T /G ,

which is the x-axis intercept in Fig. 3.Figure 3 shows three typical efficiency functions: (1) a

single-glazed, (2) a double-glazed, and (3) a double-glazedselective surfaced collector.

For any surface, absorptance equals emittance for thesame wavelength of radiation. For ordinary black surfaces,the absorptance for high-temperature (∼6000◦K), short-wave solar radiation (α6000) is about the same as the emit-tance for long-wave radiation at normal operating temper-atures around 350◦K (e350).

Special selective surfaces can, however, be producedthat have a much greater α6000 value than e350; therefore,the losses can be substantially reduced. In some cases thea6000 /e350 ratio can be as high as 14, and reliable anddurable selective surfaces with ratio of about 8 are com-mercially available. Table I summarizes the properties ofsome selective surfaces.

If selective surfaces reduce the radiant losses, the con-vective losses can be reduced by a convection suppres-sion device, which is usually a transparent honeycombstructure (or lamellae) inserted between the glazing andthe absorber surface. The outer surface of the glass mayreceive an antireflective treatment (to increase transmit-tance), while the inside surface can have an infrared re-flective coating (to reduce outward radiation). Still furtherimprovements can be achieved by enclosing the absorberin a vacuum, which eliminates all convective losses.

The structurally strongest shape is the tube; this ledto the development of a variety of evacuated tubular col-lectors (ETCs), as shown in Fig. 4. These can producetemperatures in excess of 200◦C and can provide goodefficiencies up to about 150◦C.

Because improved collectors cost more, the choice mustdepend on a cost versus benefit evaluation. High efficien-

TABLE I Radiation Properties of Some Selective Surfacesa

Surface treatment α6000 e350 α6000/e350

Nickel black on galvanized steel 0.81 0.17 4.8

Copper black (NaOH + NaClO2 treatment of copper) 0.89 0.17 5.2

Copper oxide (CuO surface conversion) 0.93 0.11 8.5

Copper oxide on aluminium 0.93 0.11 8.5

Chromium black electrodeposited on nickel plated steel 0.95 0.09 10.5

Nickel black on nickel plated steel 0.94 0.07 13.4

aAfter Duffie, J. A., and Beckman, W. A. (1980). “Solar Engineering of Thermal Processes,” Wiley (Inter-science), New York.

FIGURE 4 Three evacuated tubular collectors (ETCs).

cies can be achieved by inexpensive collectors at low tem-perature operation. If higher temperature (>70◦C) heat isneeded, then improved insulation (small U value), doubleglazing, and a selective surface are warranted. Such im-provements would be more beneficial in a cold situation,where the operational T is large, but less so in a mildclimate.

For best results the collector should generally face duesouth (or north in the southern hemisphere), but a deviationof about 20◦ from this either way would not make muchdifference. To get the maximum annual total irradiation,the collector should face the annual mean position of thesun, which means that the angle of tilt (from the horizontal)should be the same as the geographical latitude. In mostplaces, however, less sun is available in winter; thus it isusual to give the tilt a winter bias, that is, to set the tilt asthe latitude +10◦ to reduce the difference between winterand summer performance.

II. HOT WATER SYSTEMS

The simplest thermosiphon domestic hot water (DHW)system is shown in Fig. 5. As the water in the collector isheated, it expands, becomes lighter, moves upward, dis-charges into the top part of the storage tank, and draws incooler, heavier water from the bottom of the tank. A nat-ural thermosiphon circulation is thus developed, whichgradually heats up the volume of water contained in thetank. It has been observed that this circulation exhibits




FIGURE 5 A simple thermosiphon domestic hot water system.(In terms of Table II: solar, direct, closed, filled, thermosiphon,recirculating system, with separate tank.)

self-regulating tendencies: the flow rate will adjust it-self so that the temperature increment from collector in-let to outlet (Tout–Tin) is approximately 10 K, at least inthe typical thermosiphon systems, and this holds true forwidely varying temperature and radiation conditions.

This system requires that the tank be at a level higherthan the top of the collector. In all cases, but especially ifit is exposed to outdoor conditions, the tank must be verywell insulated to prevent undue tank heat losses. If forsome reason the tank cannot be placed in such an elevatedposition, then a small pump must be introduced to drivethe fluid circulation. The pump must then be controlled,for example, by a differential thermostat. This will thenbe a forced circulation system, classified as active, whilethe thermosiphon system is accepted by most authors as apassive system.

The cold water feed can be provided by direct connec-tion to the mains (in which case all components must beable to withstand mains pressure) or from a feeder tank(which is in an elevated position and filled from the mainsthrough a float valve similar to the water closet-cisternfloat valve). For mains pressure systems most utilities orwater supply authorities require the installation of a non-return valve. Whereas low pressure (feeder tank) systemsare open to the atmosphere through a vent pipe, mainspressure systems must be protected against overpressure(due to thermal expansion of water as heated) by a pres-sure relief valve, as well as against collapse (implosion),which could be caused by a suction effect (vacuum for-mation) in case of an accidental water discharge, by an airintake (or breather) valve.

In areas where freezing can occur, the system mustbe protected from frost damage. In mild winter climates,where freezing occurs rarely, some heat may be sacrificedby a slow night circulation of water or by a small heatingelement built into the lower part of the collector. In colderclimates drain-back systems can be used (Fig. 6), which

FIGURE 6 A forced circulation drain-back system: when pump Astops, the content of the collector flows back into the tank.

automatically empty the collector when the temperaturedrops to near freezing. Alternatively, an indirect systemcan be used (Fig. 7) in which the primary (collector) cir-cuit is filled with an antifreeze solution (e.g., ethylene orpropylene glycol) or with a nonaqueous liquid. This is aclosed circuit and the heat is transmitted to the consum-able water via a heat exchanger. If the antifreeze agent istoxic, then a double-walled heat exchanger may be neces-sary as a safety measure. Such control and safety devicescan make an otherwise simple system quite complicated.

A solar DHW system could be designed to pro-vide 100% of the hot water requirement (i.e., a solar frac-tion of 1), but it would not be economical except in themost favorable situations. Figure 8 shows the solar frac-tion as a function of collector area for a typical DHWinstallation, and it is a clear illustration of the law of di-minishing returns. Most designers find the economic limitto be between a solar fraction of 0.6 and 0.8.

The remaining heating requirement must be supplied byan auxiliary heater, based on some form of conventional

FIGURE 7 An indirect domestic hot water system. (In terms ofTable II: solar, indirect, closed, filled, pumped, recirculating sys-tem, with separate tank.)




FIGURE 8 Solar fraction as function of collector area (for 180 liter/day hot water use, in Brisbane, using a double glazed, selective-surfaced collector), an illustration of the law of diminishing returns.

source of energy, unless the users are willing to changetheir life styles and put up with lukewarm water at timesor have a hot shower and do the washing only when solarhot water is available. The technically simplest form ofauxiliary heat is the use of an electric immersion heaterelement built into the hot water tank itself. It is recom-mended that this should be positioned at one half to twothirds of the way up the tank so that it heats only thetop third or half of the water volume. Should sunshinebecome available, the cooler water at the lower part ofthe tank remains ready for solar heating, so no heat iswasted.

Gas- or oil-fired heaters can also be used as auxiliaryheat sources, but these would be external to the tank, cir-culating its content through the heater or having a sep-arate heat transfer loop to supply heat to the hot watertank through a heat exchanger. From the point of viewof environmental sustainability gas auxiliary heaters aremuch preferred as they cause much less CO2 emissions.

An auxiliary heat source can also be installed as an after-heater. The solar part of the system would thus become apreheater. Only when draw-off occurs would the solar-heated water flow into the after-heater, at whatever thetemperature it is, and be brought up to the required temper-ature level. This after-heater may be a fIow-through-type(instantaneous) or a storage-type device. Under favorableconditions it may not be called on to operate, but it wouldserve as a standby unit. Table II shows a classification ofsolar DHW systems according to seven attributes.

The heating of swimming pools is an ideal applicationfor solar energy. Because the temperature requirement is

TABLE II Classification of Solar Domestic Hot WaterSystemsa

Descriptors

Attributes a b c

1. Energy source Solar only Solar preheat Solar + auxiliary

2. Coupling Direct Indirect —

3. Closure Open Vented Closed

4. Operation Filled Drain-back Drain-down

5. Circulation Thermosiphon Forced —(pumped)

6. Flow type Recirculating Series —connected

7. Storage Remote Close Integral(separate) coupled

a Any system should have one descriptor in each of the seven attributecategories.

very low (<30◦C), inexpensive often unglazed, collectorscan be used. Several collector systems that use plastic orrubber materials, which can be fitted to an existing roofsurface, are on the market. In almost all cases, these are ac-tive (pumped) systems, but generally the pool filter pumpis sufficient to drive the circulation through the collectors.

III. SPACE HEATING

Systems very similar to those just described can be usedfor the heating of buildings if scaled up by an order ofmagnitude. Whereas a DHW system for an average fam-ily house would have a collector of 3 to 5 m2 and a tankof 220 to 350 liters, a system for space heating wouldbe about ten times larger. With such a collector size, athermosiphon system is rarely practicable, so pumps arealmost always used; therefore, these are invariably clas-sified as active systems. Figure 9 shows the basic systemdiagram, including the three main functions: collection(C), storage (S), and distribution (D) of heat.

The distribution system can use a range of different heatemitters, such as fan convector units, panel radiators, andembedded pipe-coil fIoor-warming systems. All these areidentical to emitters used in conventional central heating

FIGURE 9 Principles of a water-based (hydronic) solar spaceheating system. C, collection; S, storage; D, distribution.




systems. The last type is particularly attractive for solaroperation because it requires only quite low temperatures(∼30◦C); therefore, higher collection efficiencies can beachieved.

Active solar space heating systems are rarely designedto give a solar fraction greater than about 0.7. Theremaining heating requirement can be provided by an in-dependent conventional heating system or by an auxiliaryheater coupled with the solar system. The heat can be pro-duced by electricity, by burning gas, oil, or coal, or evenby using firewood. The coupling (as shown in Fig. 10)may be in (1) the storage vessel, (2) the flow pipe to theemitters (in series with the solar), or (3) a storage bypassof the emitter loop (parallel to the solar source). If there isa solar heating system, then the domestic hot water sup-ply can also be coupled with this. The coupling must bean indirect one, that is, through a heat exchanger, becausethe heating system recirculates the water (or fluid) it isfilled with, whereas the content of the DHW system isconsumed and replaced by fresh cold water. There is analmost infinite range of possibilities for the combinationof space heating, DHW, and auxiliary heating systems.

Air can also be used as the heat transfer fIuid in collec-tors built for this purpose. The heat store in this case willmost often be a rock bin or a rock bed, that is, an insulatedcontainer filled with crushed rocks or pebbles (called a binif it is an upright container or a bed if it is a shallow layerunder the floor). The solar heated air will warm the rocks,and at a later time the air flow in the distribution circuitwill remove this heat and transport it to the spaces whereit is needed. Figure 11 shows a basic air heating systemdiagram. Instead of small pipes, more bulky air ducts mustbe used, and instead of pumps, fans, or blowers. A muchlarger quantity of air must be circulated than the volume of

FIGURE 10 Three ways of coupling an auxiliary heater with thesolar space heating system: (1) in the storage tank, (2) in the dis-tribution circuit flow pipe (series connection), and (3) in a storagebypass (parallel connection).

FIGURE 11 Principles of a solar air heating system. C, collection;S, storage; D, distribution.

water that would be necessary to transport a given quantityof heat because air has a low specific heat capacity and hasvery low density. Thus, the fan power requirement will begreater than the pump power required for an identical per-formance. Efficiencies of air heater collectors are lowerthan of water heater collectors. In spite of these factors,air systems are attractive for many reasons:

1. There is no need for frost protection.2. There is no need for the many safety devices.3. Air leaks are not as damaging as are water leaks.4. There is no need for heat exchangers and emitter de-

vices (warm air can be discharged directly into the rooms).5. The system is generally simple (and thus lower in

cost).

The use of phase-change materials (PCMs) for heat stor-age in both water and air systems is now on the increase.These PCMs are usually eutectic salt solutions, but paraf-fin wax and various other mineral oil products are alsoused, which go through a phase change (liquid to solidand vice versa) at a set temperature. The latent heat ofphase change can be used for heat storage without changein temperature, in addition to any sensible heat storage dueto temperature changes. For the storage of a given amountof heat, the volume required will be only about one sixthor one eighth the volume of a water tank having a similarheat capacity. Such phase-change materials are marketedencapsulated in plastic containers of various forms andshaped to provide a heat exchanger surface in contact withwater or air.

The last decade saw few radically new ideas but nu-merous refinements and improvements. Much attentionhas been directed at the development of manufacturingtechnologies to achieve better engineered and more cost-effective products.

In domestic hot water (DHW) systems the most popularmodel now is a close-coupled indirect system. The col-lector fluid is an aqueous glycol solution, with corrosioninhibitor added, which circulates through a heat exchangerjacket, surrounding the horizontal cylinder attached to thetop of the collector. This system allows the use of ordinarysteel sheet for the collector; thus, it is much less expensivethan copper or stainless steel collectors.




In larger, pumped systems the variable volume/variableflow rate method ensures optimum operating conditions:the collector flow rate is increased as the solar radiationinput increases; thus the output temperature remains nearconstant. If there is a hot water draw-off from the tankduring a no-sun period (e.g., at night), the tank water levelis allowed to drop, that is, no make-up water is admitted(which would reduce the tank temperature), until the nextday’s solar input period.

There are some doubts about the durability and cost-effectiveness of active solar space heating systems, butpassive systems are generally recognized as worth havingunder almost any circumstances. A perennial concern isthe optimization of solar window size. This provides thesolar input, but it is also the biggest loser of heat duringno-sun periods. Movable (night-) insulation was thoughtto be the only solution, but the systems so far developed arecumbersome to use or expensive or both. Great improve-ments have been achieved in recent years in transparentinsulation materials and constructions. These can act asquasi-diodes: admitting solar radiation but preventing con-ductive heat loss. Several types are already commerciallyavailable, but the development work continues.

Much work has been done in recent years, especially incold winter climates, on central (district-) space heatingsystems using seasonal storage. The potential of naturalformations (caves, aquifers) for thermal storage has beenproven.

IV. THERMAL BEHAVIOR OF BUILDINGS

Buildings are exposed to two sets of thermal influencesfrom the environment:

1. Air temperature, acting through heat conduction bythe enclosing elements or convection through air ex-change, that is, ventilation or air infiltration

2. Solar radiation, entering primarily through windows,but also warming the outside surfaces of solid enclosingelements and two sets of man-made influences

3. Incidental internal gains, that is, the heat producedby people, lights, and appliances

4. Deliberate heating or cooling

The designer’s aim is to create a controlled environ-ment in which the temperatures (as well as other environ-mental factors) are kept within narrow limits, as requiredby the occupants. Two sets of tools can be used to servethis aim:

1. Active controls, which are mechanical systems ofvarying complexity that provide heating or cooling and

rely on some form of energy supply (e.g., firewood, oil, orelectricity); and

2. Passive controls, which are the controls of heat flowby the building itself through its careful design; these mayinclude its orientation, shape, surface qualities, thermalproperties of its fabric, and the control of windows andopenings.

The terms active and passive controls differ from theterms active and passive solar systems. Passive controlsmay include passive solar systems, but also other thermalcharacteristics of the building. Active solar systems con-stitute only a small subset of active controls, a term thatincludes the whole range of conventional heating, venti-lating, and air conditioning (HVAC) systems.

Climatic design is the manipulation of passive controlsto create the best possible indoor conditions in a given cli-mate and thus to eliminate (or reduce) the need for activecontrols and consequent energy consumption. In climateshaving outdoor air temperature higher than that desiredindoors, both environmental influences tend to cause dis-comfort. Solar heat input should be eliminated or min-imized and ways found to dissipate the unwanted heat.When the outdoor temperature is below the comfort limitand thus causes an outward (negative) heat flow, solar ra-diation can be used to provide a positive heat input. Invery cold climates, all available means must be employedto minimize heat losses and maximize solar heat inputs.In moderate climates, the designer must carefully balancesolar heat inputs against heat losses. Solar overheatingcan occur even in the winter and even at latitudes greaterthan 50◦.

V. PASSIVE SYSTEMS

It is difficult to draw the line between a well-designedhouse and a passive solar one. A window oriented towardthe equator (south in the northern hemisphere and north inthe southern hemisphere) will admit a large amount of so-lar radiation. This will be absorbed by the floor and otherinternal surfaces, causing a temperature increase. Someof this heat will be transferred to the room air by con-vection or emitted as long-wave infrared radiation towardother (cooler) room surfaces. Window glass is practicallyopaque to such long-wave radiation, so it will be trappedin the room. (This is known as the greenhouse effect.)Another part of the absorbed heat may be conducted intothe body of material behind the absorbing surface, ele-vating its temperature. This material will in effect storethe energy as sensible heat. At a later time (e.g., at night)as the room cools down, the heat flow is reversed andthe stored heat is re-emitted into the room (Fig. 12). This




FIGURE 12 Principles of a direct gain (passive) system. G, globalirradiance; A, absorption; S, storage; E, emission.

mechanism is referred to as a direct-gain. system. A gooddirect-gain system will have large windows facing theequator and highly absorbent (dark) internal surfaces in ar-eas reached by direct solar radiation. These dark-surfacedelements should have a thermal capacity sufficient to storethe radiant heat input without an excessive rise in temper-ature. Dark colored quarry tiles or slates is a favored so-lution. [Thermal capacity is the product of the volume V ,density ρ, and specific heat capacity Cρ of the material:V × ρ × Cρ (m3 × kg/m3 × J/kgK = J/K).]

Excessive summer heat input can be avoided by sim-ple but carefully designed shading devices. The noonsolar altitude angle on equinox dates (March 21 andSeptember 23) is 90◦ minus the latitude (Fig. 13). If acanopy or awning above the equator-facing window is setto this angle, it will exclude all sun penetration for thesummer half of the year (the sun moves up 23.5◦ fromthe equinox position) and admit an amount of solar radi-ation increasing to a maximum at mid-winter, when thesun is 23.5◦ below its equinox position. This provides anautomatic seasonal adjustment, utilizing the sun’s appar-ent movement. In a cooler climate the cut-off date can bechanged to allow more heating, or in warmer climates toreduce the heat input. This mechanism works only for awindow with an equatorial orientation.

One disadvantage of direct-gain systems (especially ifthe thermal capacity is inadequate) is the large diurnal (dayand night) fluctuation of indoor temperatures. The reasonfor this is inherent in the system in that the absorber sur-face, which can reach quite high temperatures, is also theemitter surface. Therefore much heat is added to the room

FIGURE 13 Solar altitude angles at noon.

FIGURE 14 Principles of a mass wall (passive) system. Abbre-viations as in Fig. 12.

air simultaneously with the solar heat input and only the re-mainder is conducted into the storage mass. Delayed gaincan be provided by so-called mass-wall systems. These aretypically heavy masonry or concrete walls (Fig. 14) witha dark colored outer surface covered by glazing. The outersurface is the absorber and the inner surface is the emitter,so the heat flow is one directional and avoids the reversalof input/output heat flow of the direct-gain system. Theinside surface will reach its peak temperature with a timelag dependent on the thermal capacity and conductivity ofthe wall (typically 6–10 hr), so the maximum heat input tothe room can be provided at a time when it is most needed.

A variant of this is referred to as the water-wall system.Water in suitable containers is used as the thermal storagemass, but functionally it is the same as the masonry mass.There is a slight difference in behavior: whereas the ma-sonry wall can show a steep temperature gradient throughits thickness during the heating-up period, temperatures inthe water container will tend to be more homogeneous. Itis an efficient form of storage, but it gives a much shortertime lag than does solid masonry.

The Trombe wall (named after the French physicistFelix Trombe, who investigated the behavior of suchwalls) is a mass wall, with ventilator openings near boththe floor and ceiling (Fig. 15). If these vents are openduring the solar heating period, an upward air flow isgenerated in the space between the heated absorber andthe glazing, discharging the heated air into the roomthrough the top vent and drawing in cooler air from theroom through the bottom vent (a thermosiphon air loop).This provides a heating effect simultaneously with thesolar heat input. If the vents are closed, it behaves as doesthe mass wall. By regulating the opening and closing ofthe vents, the proportion of simultaneous and delayedgains can be adjusted at will.

FIGURE 15 Principles of a Trombe wall (passive) system. Ab-breviations as in Fig. 12.




FIGURE 16 Principles of an attached sun space (passive) sys-tem. Abbreviations as in Fig. 12.

The attached sunspace system utilizes a separate space(e.g., a greenhouse or sun porch) for heat collection duringthe day. It is sometimes referred to as the indirect-gainsystem. In principle, it is the same as the Trombe wall,except that the approximately 100-mm space between theabsorber and the glazing is extended to 2 m or more. Therewill be a large diurnal temperature fluctuation in this space;thus there should be a facility to isolate it from the room itserves (Fig. 16). During the day, the top and bottom ventsin the dividing wall are open, allowing a warm air flow intothe room. This can even be assisted by a small through-the-wall fan (this would be a hybrid system). At night thetemperature in the sun space may drop, but then the ventsare closed. The dividing wall can also act as the masswall described earlier, giving a delayed heat input into theroom.

By placing some mass (e.g., concrete blocks or water-filled vessels) in the sun space, its diurnal temperaturefluctuation will be reduced, at the expense of heating theroom. If this space is a greenhouse or conservatory, thisis highly desirable, but it must be realized that there is aconflict of interest between the greenhouse use and thesun space as a heat source for the habitable room behindit. In many cases, greenhouses can become net losers ofenergy over the 24-hr cycle, but they are still popular aspleasant habitable spaces in their own right, albeit onlyduring sunshine periods.

In all four systems mentioned, double (or even triple)glazing can be used to reduce heat losses, but it shouldnot be forgotten that multiple glazing also reduces thetransmission of incoming solar radiation. Perhaps a moreeffective solution is the use of movable night insulation,which would substantially reduce the heat loss rate for thelarger part (up to 18 hr) of the winter day.

The last, and perhaps most attractive, generic system isthe water-roof (Skytherm) system patented by Harold Hay(of Los Angeles). The movable insulation is the essenceof this system, which can be used both for heating andcooling (Fig. 17). A metal roof deck becomes the ceiling,exposed internally, with water-filled plastic bags placed onit. Sliding insulating panels can cover the roof or can beremoved and “parked,” exposing the water bags. The sea-sonal operation is reversed: in winter, the water bags areexposed during the day and the water is heated by the sun;

FIGURE 17 Principles of a water roof (Skytherm) system.

they are covered during the night, and the heat is preservedwhile radiant ceiling heating is provided; in summer, theyare exposed overnight, and the water is cooled by con-vection and radiation to the sky; they are covered duringthe day, and solar heating is excluded while the cool ceil-ing provides a radiant heat sink and cools the room air byconvection.

A barrier to greater acceptance of the water-roof sys-tem is the complicated sliding cover mechanism and theunusual character of the resulting building. In contrast,the direct-gain system enjoys the greatest popularity be-cause it can be incorporated into any conventional buildingtype. Mass-wall and Trombe-wall systems are relativelyrare because they tend to dominate the architecture unlesshandled with some finesse. Probably the combination ofdirect gain (for instantaneous heat input) with some (non-ventilated) mass walls (for delayed gain) offers the mostbalanced performance as well as the greatest potential forarchitectural integration, for imaginative solutions.

VI. SOLAR COOLING

It is a contradiction in terms to talk about passive solarcooling. Passive cooling techniques exist, but these are (ifanything) anti-solar. Solar and other heat gains should beeliminated.

Night-time heat dissipation combined with some heatstorage capacity can provide passive cooling. (If the massis cooled down overnight, it will act as a heat sink dur-ing the day.) In dry climates evaporative cooling can beused to cool the ventilation air. Indirect evaporative coolers(Fig. 18) can extend the usefulness of this technique: theexhaust air may be cooled evaporatively and led througha heat exchanger, which will then cool the intake air with-out the addition of moisture. This could be considered asa hybrid system: the cooling is passive (evaporation) butit uses two electric fans and a small pump.

The stable (in hot periods, relatively cool) temperatureof the ground can be made use of, either by direct couplingof the building with the ground (e.g., earth-covered or




FIGURE 18 An indirect evaporative cooler: white arrows showthe passage of return air from the room which is evaporativelycooled by a spray in the vertical spaces of the plate heat ex-changer, then discharged; solid arrows show the path of freshair through the horizontal gaps of the heat exchanger, where it iscooled, then supplied to the room.

earth-integrated buildings) or by using underground pipesto cool the ventilation air. These are passive cooling meth-ods, but not solar ones.

The use of solar energy for the purposes of cooling isan attractive proposition because in the annual cycle, thepeak demand coincides with the maximum availability ofsolar radiation (in the summer). This is not so with solarheating, where the peak heat demand (in winter) occurswhen there is little solar radiation available.

Active solar cooling and air conditioning systems can beof two kinds: closed or open cycle. Closed-cycle systemsuse absorption refrigeration machines (chillers). These arebinary systems, using an absorbent and a refrigerant (asopposed to compression systems, which use a single fluidas refrigerant). Two systems are popular:

Absorbent Refrigerant

Water (H2O) Ammonia (NH3)

Lithium bromide (LiBr) Water (H2O)

These types of chillers are commonly used in gas orkerosene operated refrigerators. The difference is that heresolar-heated water would be used to provide heat input inthe generator (see Fig. 19). In the system shown the refrig-erant is expelled from the rich solution and the weakenedsolution is returned to the absorber. The refrigerant vapordissipates some of its heat to the environment (may bewater cooled) and condenses. This liquid enters the evap-orator through a throttle valve (or a high-resistance thinpipe), where, in the low-pressure chamber, it evaporatesand takes on some heat from its environment (e.g., froma water jacket), thus it produces chilled water. The vaporthen returns to the absorber and is absorbed by the ab-

FIGURE 19 Schematic diagram of an absorption chiller.

sorbent. This is an exothermic process, so heat is releasedto the environment.

Typically, the generator temperature needed is 80–90◦Cand chilled water of 8–10◦C can be produced. This chilledwater is then circulated through an air-to-water heat ex-changer (e.g., a fan-coil unit) and will cool the air suppliedto the rooms.

The coefficient of performance (CoP) of chiller systemsis defined as

CoP = rate of heat removal

rate of energy input to drive the system.

While compression chillers can achieve a CoP of 2 to4, absorption chillers rarely produce more than 0.5 to 0.6.Their use is justified only if inexpensive, low-grade (80–90◦C) heat is available in ample quantities.

Solar systems can produce hot water at such tempera-tures, but either very good quality flat plate or evacuatedtubular collectors (ETCs) must be employed; thus the col-lection system will be expensive. Numerous installationsof such systems have been completed, many of whichare technically successful, but very few (if any) would bejudged as economically viable.

In humid climates the effectiveness of evaporative cool-ing is limited by the already high moisture content of theair and the supply air would also become unpleasantly hu-mid. Moisture can be removed by either solid or liquiddesiccants, the former in a slowly rotating wheel enclosedin a wire mesh, the latter in the form of a fine spray. Theseare incorporated in “open cycle” cooling systems. Severalsuch systems are in the developmental stage, some com-ponents are already in use in conventional air conditioningsystems.

Their name comes from the fact that air from the atmo-sphere is taken through the system and discharged to therooms (so there is no closed circulation of some refriger-ant, as in cooling machines). All variants rely on evapora-tive cooling and the removal of moisture by some sorbentmaterial (desiccant), which will then be reactivated by




FIGURE 20 Two open cycle cooling systems.

solar heating. Figure 20 shows two such systems in dia-grammatic form: the first with a moisture transfer wheelcontaining silica gel, the second using a glycol spray.

VII. DESIGN METHODS

The thermal design of any building, as well as that ofdistinct solar systems, must be based on some predictivemodel. The first approach to an evaluation of expected heatflows can be based on steady-state assumptions by takingthe daily mean indoor-outdoor temperature difference andcalculating the 24-hr average solar gain. This is a relativelysimple calculation, but it only approximates reality. Theoutdoor temperatures vary between a maximum, shortlyafter midday and a minimum, normally just before sunrise.The solar heat input is usually restricted to less than onethird of the day, increases to noon, then decreases to sun-set; cloud conditions introduce a further variation. Indoortemperature requirements, as well as internal heat gains,may also vary with the time of day. Consequently, thetime-dependent thermal response of the building shouldalso be evaluated.

Many sophisticated mathematical models have beenconstructed and built into powerful computer programsfor the simulation of the thermal behavior of buildings.These tend to use very detailed climatic data and ana-lytically follow the heat flows through various buildingelements of differing exposure and construction. The timestep used is often 1 hr, but it can be as short as 10 min,and simulations can be run for a full year to predict indoortemperatures (for a free-running building) or the energyrequirements for systems to keep set indoor temperatures.These programs use hourly meteorological data, assem-bled to form a TMY (typical meteorological year) or TRY(typical reference year). The two major methods in useare based on finite difference and response factor meth-ods, respectively.

In the United States the most sophisticated and mostwidely used such program is DOE2, which is now avail-able also for PCs, in a Windows environment. In Europethe most refined one is ESPr (of the University of Strath-clyde, Glasgow), but this is rather a research tool than a de-sign tool. In Australia a comparable program is CheeNatof the CSIRO (Commonwealth Scientific and ResearchOrganisation), used also as the simulation engine in theNational House Energy Rating Scheme (NatHERS).

The previously cited are building thermal response,heating-, cooling load calculation, and energy use simu-lation programs; but as such they can also cope with mostpassive systems. For active solar systems the best simula-tion package is TRNSYS, a modular program developedby the University of Wisconsin.

For the practicing designer, the simplified methods aremore suitable. These are based on correlations of majorsystem parameters with the resulting performance, as pre-dicted by multiple runs of some sophisticated program.The most widely used of these is the solar-load ratio (SLR)method, developed by the Los Alamos Scientific Labora-tories and the unutilizability method of the University ofWisconsin. In Europe the Method 5000 is the most gen-erally accepted, adopted as the minimum requirement forCommission of European Communities (CEC) projects.

For active solar systems, the f-chart method is gener-ally used. All these can be used manually or with modestsized microcomputers. These methods predict the solarfraction, that is, the fraction of the total load (e.g., heatingrequirement) that is likely to be provided by the solar sys-tem. The simplicity of such methods allows the designerto examine many possible alternatives and select the mostappropriate ones.

Multiple simulation runs are also used for parametricstudies, in which one design variable at a time is sys-tematically changed and the performance of the system isexpressed as a function of that parameter. Sets of sensitiv-ity curves are thus produced, providing useful advice tothe designer.

With the dramatic increase in computing power of PCsand laptops a number of packages have been developedincorporating some sophisticated simulation system, butwith front ends and outputs to suit the designers’ or ar-chitects’ way of thinking. The program ENERGY10 ofthe NREL (National Renewable Energy Laboratory, inGolden, CO.) or the IEA (International Energy Agency)package ISOLDE are perhaps the best known examples.

VIII. USER INFLUENCE

It is relatively easy (and reliable) to predict the thermalbehavior of a building and its installations as physical




systems. Performance measurements in unoccupied build-ings have been used for the validation of many computer-based simulation programs. However, as soon as peopleoccupy these buildings, the predicted and measured resultsstart to diverge. This divergence is particularly pronouncedin the case of passive systems.

With active systems, the controls are mostly automatic,heat flows are well contained, and heat transfer processesare accurately predictable. With passive systems, heatflows are more diffuse and complex interchanges occur,which can be grossly influenced by user interference. Of-ten the appropriate adjustment of controls (e.g., of shadingdevices or of Trombe-wall vents) by the user at the righttime is required to achieve optimum performance of thesystem. Both the competence and the attitude of the usercan drastically influence the results.

A passive solar house, given to an unsympathetic (pos-sibly even hostile) user or a user lacking an understandingof the system, can use more energy than a conventionalhouse with an ordinary fuel-based heating system. Themost successful solar houses are the ones designed andbuilt by owner-occupiers. These owners are intimately in-volved with the system, are capable of adjustments, main-tenance, and even modification of the system, and are oftenemotionally committed to its success.

For tenanted houses, an instruction manual should beprovided, but preferably also an explanation and demon-stration should be given. Perhaps the financial incentive(saving on fuel bills) may be enough to obtain the tenant’scooperation.

IX. ECONOMICS

If an average conventional house is taken as the base caseand its annual energy use taken as 100%, it is suggestedthat about 30% of this can be saved by no-cost improve-ments and by the investment of a little thought and knowl-edge at the design stage. An increased expenditure at thedesign–construction stage, carrying the energy conserva-tion measures to their full potential, would increase thissaving to about 50%. If explicit passive solar systems arealso built into the design, the average saving could be 80%and, in favorable situations, even higher.

For the evaluation of such additional investments, theconcept of pay-back period is often used. The simple pay-back period (PPB) is

PPB = annual benefit

extra cos t= B

C

in terms of years. The annual benefit is taken as the so-lar contribution times the cost of conventional (auxiliary)energy, which would otherwise be used. A more accu-

rate definition would compare the present worth of futuresavings with the extra cost.

The length of an acceptable pay-back period dependson financing arrangements. With a passive system, it maybe the period of the mortgage loan (20–25 years). Withan active system, it should not be more than the life ex-pectancy of the collector array (18–20 years), but the aver-age householder would not consider a period greater than10 to 15 years. In a commercial or industrial situation,pay-back periods of 3 to 5 years are often demanded.

X. URBAN PROBLEMS

In most countries it is customary to build the front ofhouses facing the street. The front of a house generallymeans the location of the largest windows of the pri-mary habitable rooms. This may give good results forplots on the north side of east-west-running streets. Be-cause a southern or near-southern orientation is best inthe northern hemisphere, this should be considered at thestage of subdivision and street layout. So far, most solarbuildings have been erected on selected (therefore favor-able) sites in low density areas. If passive solar designis to become the norm (rather than the exception, as atpresent), then all new subdivisions should be designed topermit this.

Some changes in attitude may also be necessary, suchas the acceptance of buildings aligned other than paral-lel to the street, giving a stepped (sawtooth) streetscape,or the acceptance of major windows facing the back. Ifhigher densities are aimed at, the streets must run east–west or nearly so. Wider plots are necessary for north–south streets to allow sideways orientation of houses.With diagonally running streets (north–west–southeast ornortheast–southwest) wider plots are necessary to permita “twisted” positioning of the house on the site.

One of the major problems is that of avoiding over-shadowing of solar collectors or solar apertures by otherbuildings, that is, undisturbed solar access must be en-sured. The greater the building density, the more difficultthis problem will be. Much work has been done on thissubject from the legal and urban planning points of view,and attempts have been made in some areas to create leg-islation for the protection of solar access. Perhaps moresignificant (at least so far) are the few existing solar vil-lage projects produced by individual initiative and effort.Davis, California, is a notable example. The whole area isrun on a cooperative basis and it seems to work well bothphysically and socially.

Several authors have suggested that the best way of en-suring solar access is by use of a system of covenants ratherthan legislative measures. Whichever method is adopted,




FIGURE 21 An example of daily irradiation of a collector (day’s total = 4853 Wh/m2): between 10 a.m. and 2 p.m., 2681Wh/m2 = 55%. between 9 a.m. and 3 p.m., 3689 Wh/m2 = 76%; and between 8 a.m. and 4 p.m., 4375 Wh/m2 = 90%of total.

the technical questions are as follows: what are the criteria,what is the physical basis of measurement, and what de-gree of solar access is taken as reasonable without undulyrestricting development?

A roof-mounted solar collector is not as vul-nerable toovershadowing as a solar window of a passive system,which is near ground level, but ensuring total solar accessat all times, even for this, would be unreasonable. Withlow sun angles (just after sunrise and just before sunset),very long shadows are cast. However, the irradiance atsuch times is very low, therefore an obstruction would notcause a great loss.

Some authors suggest the criterion of no overshadowingbetween 9 a.m. and 3 p.m. (i.e., 3 hr on either side of solarnoon). Others are more demanding, suggesting 4 hr beforeand after solar noon (i.e., 8 a.m. to 4 p.m.); yet others aremore permissive, with limits of 10 a.m. and 2 p.m. (i.e.,±2 hr). All of these limits would apply to the mid-winterday. Figure 21 shows a typical irradiance curve for a north-facing collector (located in the Southern Hemisphere) witha 9 a.m. and 3 p.m. cut-off (other periods are shown bydashed lines). It is obvious that most of the radiation isreceived around solar noon, and the fraction lost (outsidethese time limits) is not excessive.

Two methods may be used for measurement as a basisfor enforcement: the solar envelope and the solar windowmethods. The former would define the maximum build-able volume possible on any site without casting undueshadows outside the site boundaries (Fig. 22). The latterwould define in three-dimensional terms the solid anglesubtended at an existing (established) solar device that

should not be penetrated by any obstruction (Fig. 23). Theformer is more generally applicable, giving protection notonly to existing solar users but also to any future buildings.

XI. RECENT DEVELOPMENTS

The earlier enthusiasm for solar cooling and air condition-ing systems has subsided: in spite of the improvement inCoP to 0.9 or even 1, they are often not only expensivebut also use so much parasitic energy that there would beno pay-back period even in energy terms. With the im-provement and cost reduction of solar photovoltaic cells,it may now be more effective to use such PV cells to gen-erate electricity, which then drives a compression chiller.With direct coupling of electrical output to the compressor

FIGURE 22 Solar envelope for a 20 × 40 m building site to ensuresolar access of adjoining properties between 10 a.m. and 2 p.m.on a mid-winter day.




FIGURE 23 Solar window (or unobstructed sky space) to ensuresolar access for an established device between 9 a.m. and 3 p.m.throughout the year.

motor, most cooling would be provided when the solar ra-diation is strongest, A system with refrigerant storage mayaltogether eliminate the need for battery storage of elec-tricity, which would otherwise be a large cost componentof solar–electric systems.

Passive systems, on the other hand, are gaining grounds.More and more architects, designers and builders under-stand and accept that every house is a solar house, everyhouse can make use of solar energy to a greater or lesserextent. Correct orientation is recognised as the most im-portant decision, which attracts no extra cost.

The design of the building envelope also shows an im-proving trend. Much more attention is paid to insulationand the prevention of thermal bridge effects. Perhaps thegreatest development occurred in windows. It used to beaccepted that the window is the weakest point of the build-ing envelope. In many climates the solar gain through awindow was less than the heat loss.

While a moderately good wall would have a U -value(thermal transmittance of 1 W/m2K, a single glazed alu-minium window would be over 6 W/m2K. With con-ventional double glazing units this may be reduced to3 W/m2K. Low emittance coating added gives a furtherreduction to 1.7 and if the cavity of the glazing unit is filledwith argon (instead of air) the result is about 1.6 W/m2K.Triple glazing, with two surfaces low-e coated and theunit filled with krypton: the U -value can be as low as0.85 W/m2K, if the frame is also improved with thermalbreaks or if the metal is replaced by vinyl. These are com-mercially available products. Experimentally, with a mul-tilayer “super”-window 0.63 W/m2K has been achieved(two outer glass layers and two internal low-e coatedplastic films). A double glazing unit, with low-e coat-ing and vacuum between the glass sheets (which necessi-tates the use spacer studs) reached so far the best results:0.56 W/m2K.

As the solar radiation transmittance of such multi-layerglazing is still 0.5–0.6, the designer can successfully uselarge (well oriented) windows without unduly increasingheat loss.

Another area of rapid development is building-integrated photovoltaics (BIPV). With stand-alone PVsystems the “balance of system” costs (mainly the sup-port structure and the storage batteries) is practically asmuch as the PV cell arrays. The support structure cost issaved if the arrays are mounted on a building and thereis no need for batteries if the system is grid-connected.Building elements are produced to incorporate PV cellsand are used as for example roofing tiles or spandrel pan-els in curtain walls. They are replacing nonsolar elements,which gives a cost saving.

Germany started the “thousand solar roofs” project in1990 and by 1995 some 2250 roof installations were com-pleted, each between 1 and 5 kW, giving a total installedcapacity of over 6 MW. Switzerland was not much behind.The European Commission installed over 11 MW PV ca-pacity since 1979 and the Netherlands alone (Shell) hasan annual production output of 20 MW. A new Germanplant has an annual output of 25 MW.

The United States launched a “million solar roofs”initiative in June 1997, which is to be completedby 2010 (this includes both PV and solar heaterpanels).

An increasing number of projects in the UK, Germany,Switzerland, the Netherlands, Denmark, Sweden, Japan,the United States and several other countries use PV panelsas wall cladding on large scale, often multistorey build-ings, referred to as “PV facades.”

Two directions of development represent two entirelydifferent philosophical attitudes to sustainable develop-ment. One is self-sufficiency, aiming for the “autonomoushouse,” which produces its own energy requirements (aswell as rainwater collection, waste water reuse, vegetablegardening, etc.). The other is acknowledging interdepen-dence, the benefits of connecting to the electrical grid,of water supply and other public services and attemptsto contribute to sustainability at the level of society as awhole. The utilization of solar energy in buildings is animportant element in both.

SEE ALSO THE FOLLOWING ARTICLE

ENERGY EFFICIENCY COMPARISONS AMONG COUNTRIES

• ENERGY RESOURCES AND RESERVES • PHOTOVOLTAIC

SYSTEM DESIGN • RADIATION PHYSICS • SIMULATION

AND MODELING • VENTILATION, INDUSTRIAL

BIBLIOGRAPHY

Anderson, B. (1977). “Solar Energy Fundamentals in Building Design,”McGraw-Hill, New York.




Department of Energy (1980–1983). “Passive Solar Design Hand-book,” Vols. I–III, U.S. Department of Energy, Washington,DC.

Duffie, J. A., and Beckman, W. A. (1980). “Solar Engineering of ThermalProcesses,” Wiley (Interscience), New York.

Givoni, B. (1994). “Passive and Low Energy Cooling of Buildings,” vanNostrand Reinhold, New York.

Hastings, S. R. et al. (1994). “Passive Solar Commercial and InstitutionalBuildings,” I. E. A.—John Wiley & Sons, Chichester.

Mazria, E. (1979). “The Passive Solar Energy Book,” Rodale Press,Emmaus, PA.

Szokolay, S. V. (1975/1977). “Solar Energy and Building,” ArchitecturalPress, London.

Szokolay, S. V. (1980). “World Solar Architecture,” Architectural Press,London, and Wiley (Halsted), New York.

Tabb, P. (1984). “Solar Energy Planning,” McGraw-Hill, New York.Watson, D., and Labs, K. (1983). “Climatic Design,” McGraw-Hill, New

York.

P1: GTM/GUR P2: GQT Final Pages Qu: 00, 00, 00, 00


Thermionic EnergyConversion

Fred HuffmanThermo Electron Corporation

I. IntroductionII. Fundamental ProcessesIII. Basic Principles of Thermionic ConversionIV. ElectrodesV. Experimental Results

VI. Applications

GLOSSARY

Arc drop Minimum electrostatic potential difference thatis required to maintain a plasma in an ignited discharge.

Barrier index Figure-of-merit parameter for character-izing the performance of a thermionic converter thatis given by the sum of the arc drop and the collectorwork function, thus accounting for the plasma plus thecollection losses. The lower this parameter, the higheris the converter performance.

Collector Cool electrode in a thermionic converter thatcollects the electron current.

Emitter Hot electrode in a thermionic converter fromwhich the electron current is emitted.

Fermi level Characteristic energy of electrons in a mate-rial under thermal equilibrium at which the probabilityof an allowed quantum state being occupied by an elec-tron is 1

2 .Plasma Ionized gas, containing approximately equal con-

centrations of positive ions and electrons, that is elec-trically conductive and relatively field free.

Space charge Negative electrostatic barrier formed as aconsequence of the finite transit time of the electronscrossing from the emitter to the collector.

Work function Heat of electron vaporization (and con-densation) from a material given by the difference inelectrostatic potential energy between its Fermi leveland a point just outside its surface.

A THERMIONIC energy converter transforms heat intoelectricity by evaporating electrons from a hot emitterand condensing them on a cooler collector. This deviceis characterized by extremely high operating tempera-tures, no moving parts, modularity, and the capacity tooperate inside the core of a nuclear reactor. Thermionicpower generators have been built utilizing combustion,solar, radioisotope, and reactor heat sources. Although

627

P1: GTM/GUR P2: GQT Final Pages


628 Thermionic Energy Conversion

commercial applications have yet to be realized, therm-ionic energy conversion has the potential to (1) producepower in space reliably using nuclear reactors, (2) co-generate electric power for a large fraction of industryunsuited to other energy converters, and (3) increasesubstantially the efficiency of utility power plants. Steadytechnical progress is bringing these possibilities closer toreality.

I. INTRODUCTION

A. Elementary Description

A thermionic converter is illustrated in Fig. 1. It consistsof a hot electrode, or emitter, separated from a cooler elec-trode or collector by an electrical insulator. The spacingbetween the emitter and collector is usually a fraction ofa millimeter. Electrons vaporized from the emitter crossthe gap, condense on the collector, and are returned to theemitter by way of the electrical load.

The converter enclosure is hermetically sealed so thatthe atmosphere between the electrodes can be controlled.In the conventional thermionic converter, the interelec-trode space is filled with cesium vapor from a liquid reser-voir at a pressure of ∼1 torr. The cesium performs twofunctions. First, it adsorbs on the electrodes to provide thedesired electron emission properties. Second, it providespositive ions to neutralize the electron space charge sothat practical current densities can be obtained from theconverter.

The thermionic converter is a heat engine utilizing anelectron gas as the working fluid. Thus, its efficiencycannot exceed that of the Carnot cycle. In effect, the

FIGURE 1 Thermionic energy converter.

temperature difference between the emitter and collectordrives the electrons through the load.

For a given set of electrodes, the output power of thethermionic converter is a function of the emitter temper-ature, collector temperature, interelectrode spacing, andcesium pressure. Characteristically, thermionic convert-ers operate at high electrode temperatures. Typical emittertemperatures range between 1600 and 2400 K and collec-tor temperatures vary from 800 to 1100 K to provide a highcurrent density (say, 5–10 A/cm2) at an output potentialof ∼0.5 V per converter. The efficiency of converting heatto electricity is usually between 10 and 15%.

B. Historical Background

The principle of thermionic conversion derives fromEdison’s discovery in 1885 that current could be made toflow between two electrodes at different temperatures ina vacuum. The analysis and experimental investigation ofthermionic emission from a hot electrode were performedby O. W. Richardson in 1912. W. Schlichter in 1915recognized this means of converting heat to electricity. Apatent was submitted on this topic in 1923. I. Langmuirand his associates characterized the electron and ionemission from cesium-adsorbed films on tungsten in the1920s.

There was some interest in thermionic conversion inRussia in the 1940s. In particular, A. F. Ioffe discussedthe thermionic converter as a “vacuum thermoelement”in 1949. Analytical studies of thermionic conversion thatneglected the effects of space charge and collector tem-perature were published in the early 1950s. The first ther-modynamic analysis of a thermionic converter was givenby G. N. Hatsopoulos in 1956.

In the late 1950s a remarkable metamorphosis tookplace when several groups in the United States and Russia(working, with few exceptions, independently and with-out knowledge of previous or concurrent efforts) achievedefficiencies of 5 to 10% and electrical power densities of3 to 10 W/cm2. V. C. Wilson pioneered investigations ofignited-mode converters.

During the 1960s and early 1970s, thermionic technol-ogy development in the United States and Russia was con-centrated on space reactor power systems. Remarkableprogress was made in both countries during this period. In1973, the United States terminated its space reactor pro-gram. However, Russia continued its efforts and by 1977had tested four TOPAZ thermionic reactors.

The demise of the U.S. space thermionic program co-incided with the OPEC oil action. These events providedthe motivation for transferring the thermionic technologythat had been developed for space to potential terrestrialapplications that could conserve premium fossil fuels.



Thermionic Energy Conversion 629

By the early 1980s, the wheel had made another cycle,and the U.S. combustion thermionic conversion programwas redirected to nuclear space activities. By 1985, onlya token terrestrial thermionic effort existed.

II. FUNDAMENTAL PROCESSES

A. Surface Phenomena

1. Fermi Level

To a first approximation, the long-range forces betweenelectrons in a metal are neutralized by the positive ioncores of the atoms. This implies that the metal lattice canbe thought of as a box in which electrons are free to move.The number of free electrons is taken as the number ofatoms times the valence of each atom. These valenceelectrons are responsible for electrical conduction by themetal; hence, they are termed conduction electrons, asdistinguished from the electrons of the filled shells of theion cores.

In the condensed state, the valence electrons are as-sociated with the quantum states of the entire crystalrather than states of individual atoms. By applying therules of statistical thermodynamics for spin 1

2 particles(i.e., Fermi–Dirac statistics) the electron energy distribu-tion f (E) in a metal at a temperature T is given by theexpression for the Fermi factor,

f (E) = 1

1 + exp[(E − EF)/kT ], (1)

where k is Boltzmann’s constant (k = 8.62 × 10−5 eV/K)and EF is an energy characteristic of the metal called theFermi energy. It is numerically equal to the energy forwhich the Fermi factor is 0.5. At T = 0, EF represents thehighest energy of the electrons in the metal.

FIGURE 2 Energy levels near a metal surface.

2. Work Function

A diagram of the energy levels in a conductor and the po-tential energy of an electron near the surface is shown inFig. 2. To remove an electron from the metal, it must havean energy above the Fermi level by the amount of φ eV;this parameter is defined as the work function. The workfunction can be thought of as the electron heat of vapor-ization. If an electron is introduced into the metal from thesurface, a quantity of heat, φ eV, will be given up. Thus,the work function is also the electron heat of condensation.It is convenient to think of the work function as a potentialbarrier between the inside of the metal and a point justoutside (say, 500 A) the metal surface that must be sur-mounted by an electron in order to escape from the metal.

Work function values vary between 4 and 5 eV for mostmaterials. However, they can range as widely as 1.8 eVfor cesium to 5.5 eV for platinum. Considerable varia-tion of work function values for a given material can befound in the literature. These variations are due to impuri-ties, surface films, nonuniform surfaces, and measurementtechniques.

3. Thermionic Emission

Thermionic emission from a metal can be explained interms of Fig. 2 and Eq. (1). At room temperature, fewof the quantum states above the Fermi level will be filled.However, as can be seen from the Fermi factor expression,there will always be some electrons occupying the high-energy states as long as T > 0. Due to their random motioninside the metal, many electrons will impinge on the sur-face. Those electrons with kinetic energies greater thanthe work function barrier may escape the metal. Thosethat cross the boundary at the surface will transform partof their kinetic energy (i.e., φ eV) into potential energy.




It can be shown, either from thermodynamic consid-erations or from the application of statistical mechanicsin connection with the quantum mechanics of electrons inmetals, that the current density J of electrons emitted froma uniform surface of a pure metal of absolute temperatureT can be expressed by the Richardson equation,

J = (1 − r )AT 2 exp(−φ/kT ), (2)

where A is a constant and r is the electron reflection coef-ficient of the surface (of the order of 0.05). The coefficientA is composed of a combination of fundamental physicalconstants,

A = 4πmk2e/h3 = 120 A/cm2 K2, (3)

where e is the electronic charge, m is the electronic mass,and h is Planck’s constant. If the reflection coefficient is as-sumed to be zero, it is convenient to express the Richardsonequation as

J = 120 T 2 exp(−11606 φ/T ), (4)

where J is the current density in amperes per(centimeter)2, φ is the work function in electronvolts, andT is the electrode temperature in kelvins. The value Jis called the saturation current density and correspondsto a zero electric field (horizontal potential distributionin Fig. 2). A strong applied electric field changes theelectron emission, because its effects are superimposedon those of the image force, and alters the shape of the po-tential distribution outside the electrode. The dotted-linedistributions in Fig. 2 correspond to electron-acceleratingand -decelerating fields. Equation (4) shows that the sat-uration current density of a surface increases rapidly withincreasing temperature and decreasing work function.

4. Cesiated Surfaces

Thus far, all practical thermionic converters have used ce-sium vapor because of the remarkably fortunate combina-tion of three major phenomena. First, Cesium atoms areadsorbed on the emitter and collector surfaces to reducethe work functions of these electrodes to values favor-able for energy conversion. No other electrodes have beenfound that provide long life at converter temperatures withwork functions suitable for energy conversion. Cesiatedelectrodes are quite stable since the adsorbed coveragesare due to the equilibrium between the evaporating andimpinging atoms. Second, some of the adsorbed cesiumatoms are evaporated as positive ions, which contributeto reducing the electron space charge between the emit-ter and collector. Third, electron collisions with cesiumatoms in the interelectrode space provide cesium ions toneutralize the space charge. Indeed, in most thermionicconverters, this is the primary ion source for this purpose.

B. Plasma Considerations

A plasma is a relatively field-free region of an ionizedgas that electrostatically shields itself in a distance thatis small compared with other lengths of physical interest.The space charge or strong field regions on the boundary ofthe plasma are called sheaths. The distance over which thepotential of a charge immersed in a plasma is reduced to anegligible amount can be shown, using Poisson’s equation,to be

λ =√

ε0kT/2Nee2, (5)

where Ne is the electron concentration and ε0 is the per-mittivity of free space. This distance is called the Debyeshielding distance. It gives the approximate thickness ofthe sheath that develops when the plasma contacts an elec-trode. Without such a sheath, the plasma would lose elec-trons much more rapidly than positive ions because of thegreater electron velocity.

A plasma is in thermal equilibrium when all species(electrons, ions, and atoms) have the same average en-ergy. At high particle densities, the collision frequency ishigh enough for this condition to exist. However, at thelow pressures (order of 1 torr) at which a conventionalthermionic converter operates, the collision frequency forelectrons is low enough that they can gain significant en-ergy between collisions. Under converter conditions, theelectron temperature Te may exceed the atom temperatureTa by a large factor. Typically, Te is of the order of 3000 K,whereas Ta � 0.5 (TE + TC). A plasma cannot greatly ex-ceed the density at which as many ions recombine locallyas are produced locally. In this condition, known as lo-cal thermodynamic equilibrium, the plasma properties areequivalent to those that would exist in equilibrium with ahypothetical surface at Te and emitting a neutral plasma.

III. BASIC PRINCIPLES OFTHERMIONIC CONVERSION

An idealized potential diagram of a thermionic energyconverter is shown in Fig. 3a. This diagram shows the spa-tial variation of the electrostatic potential perpendicular tothe electrodes. Since the potential energy of the electronsin the collector is greater than that in the emitter, the col-lected electrons can perform work as they flow back tothe emitter through the electrical load. The load voltage Vis given by the difference in the Fermi levels between theemitter and collector. The emitter and collector work func-tions are denoted by φE and φC, respectively. The currentdensity of the thermionically emitted electrons is givenby the Richardson equation. The output power density ofthe converter is the product of V times the load current




FIGURE 3 Ideal-mode converter. (a) Motive diagram; (b) curr-ent–density voltage characteristic.

density J . The J–V characteristic corresponding to the“ideal-mode” potential diagram given in Fig. 3a is shownin Fig. 3b.

An analogy can be drawn between thermionic conver-sion and the conversion of solar heat to hydropower. In thisanalogy, water plays the role of electrons, the sea surfacecorresponds to the emitter Fermi level, the lake surfacecorresponds to the collector Fermi level, and the gravita-tional potential corresponds to the electrical potential. Aswater is vaporized from the sea by heat, the water vapormigrates over the mountains to a cooler region, where itcondenses and falls as rain into a lake at a high altitude. Thewater returns through a turbine to the sea to complete thecycle by which solar energy is converted to hydropower.

A. Ideal Diode Model

The ideal diode thermionic converter model correspondsto a thermionic converter in which the emitter and col-lector are spaced so closely that no collisional or spacecharge effects take place. To reduce the complexity ofthe equations, ion emission effects will also be neglected.Although these assumptions do not strictly correspond toany thermionic converter, they do approach those of a veryclosely spaced diode operating in the vacuum mode. Theideal diode model defines the performance limit imposedby essential electron emission and heat transfer and pro-vides a basis for comparison with practical converters.

In Fig. 3a, the energy V + φC must be supplied to anelectron to remove it from the emitter and transport itacross the gap to the collector. The electron gives up heatin the amount of φC when it is collected. The collectedelectron has an electrostatic potential V relative to theemitter. The electron can then be run through an electricalload to provide useful work output on its route back to theemitter.

The current density J through the load consists of JEC,flowing from emitter to collector, minus JCE, flowing fromcollector to emitter. Thus,

J = AT 2E exp

(− V + φC

kTE

)− AT 2

C exp

(− φC

kTC

)(6)

for V + φC > φE. As the output voltage is reduced suchthat V + φC < φE,

J = AT 2E exp

(− φE

kTE

)− AT 2

C exp

(−φE − V

kTC

)(7)

In Fig. 3b, the saturation region of the J–V curve corre-sponds to negligible backemission. The open-circuit volt-age is given by setting J = 0 and the short-circuit cur-rent density for the case of V = 0. Equations (6) and(7) can be used to construct J–V curves for a range ofφE, φC, TE, and TC values. Usually, 1600 < TE < 2000 K,600 < TC < 1200 K, 2.4 < φE < 2.8 eV and φC � 1.6 eV.

The electrode power density is P = J V , and the thermalinput QIN to the emitter is given by

QIN = QE + QR + QL + QX, (8)

where QE is the heat to evaporate electrons, J (φE +2kTE)(this expression assumes negligible backemission fromthe collector), QR is the thermal radiation between elec-trodes, QL is the heat conducted down the emitter lead,and QX is the extraneous heat losses through structureand interelectrode vapor. Thus, the thermionic conversionefficiency is

η = JA(V − VL)/QIN, (9)

where A is the electrode area, VL is the voltage loss inemitter lead (= JARL), and RL is the resistance of emitterlead. For a given cross-sectional area of the lead, length-ening will increase VL and decrease QL; shortening willdecrease VL and increase QL. To obtain the maximumefficiency in Eq. (9), the lead dimensions must be op-timized. This optimization can be performed by settingdη/d RL = 0. Typically, VL � 0.1 V.

B. Operating Modes

In practice, the ideal motive and output characteristicillustrated in Fig. 3 is never achieved because of theproblems associated with the space charge barrier built upas the electrons transit from the emitter to the collector.Three of the basic approaches to circumventing thespace charge problem are (1) the vacuum-mode diode,in which the interelectrode spacing is quite small; (2)the unignited-mode converter, in which the positiveions for space charge neutralization are provided bysurface-contact ionization; and (3) the ignited mode, inwhich the positive ions are provided by thermal ionizationin the interelectrode space. The potential diagrams andJ–V characteristics of these three operating modes aregiven in Figs. 4a, b, and c, respectively.




FIGURE 4 Motive diagrams and J–V characteristics for (a) vac-uum, (b) unignited, and (c) ignited modes of operating a thermionicconverter.

In the vacuum mode (Fig. 4a) the space charge is sup-pressed by making the spacing quite small—of the orderof 0.01 mm. Although converters of this type have beenbuilt and operated with small-area electrodes, the practicaldifficulty of maintaining the necessary close spacing overlarge areas at high temperature differences has made thevacuum-mode diode mostly of academic interest.

Another approach to the space charge problem is tointroduce low-pressure cesium vapor between the emit-ter and collector. This vapor is ionized when it contactsthe hot emitter, provided that the emitter work function iscomparable to the 3.89-eV ionization potential of cesium.A thermionic converter of this type (Fig. 4b), in whichthe ions for neutralizing the space charge are provided bysurface-contact ionization, is said to be operating in theunignited mode. Unfortunately, surface-contact ionizationis effective only at emitter temperatures that are so high(say, 2000 K) as to preclude most heat sources. To operateat more moderate temperatures, the cesium pressure mustbe increased to the order of 1 torr. In this case (Fig. 4c) ce-sium adsorption on the emitter reduces its work function sothat high current densities can be achieved at significantlylower temperatures than in low-pressure unignited-modediodes. Most of the ions for space charge neutralizationare provided by electron-impact ionization in the interelec-trode spacing. This ignited mode is initiated when the out-put voltage is lowered sufficiently. This mode of operationis that of the “conventional” thermionic converters. Thisionization energy is provided at the expense of the out-put voltage via an “arc drop” Vd across the interelectrodeplasma. Because of the high cesium pressure, the electronsare scattered many times as they cross to the collector.

For a given electrode pair, emitter temperature, col-lector temperature, and spacing, the J–V characteristicis a function of cesium reservoir temperature. A typical

FIGURE 5 Typical output characteristics, parametric in cesiumreservoir temperature. Emitter, CVD W(110); collector, noibium;TE = 1800 K; TC = 973 K; d = 0.5 mm.

J–V family, parametric in cesium reservoir temperature,is given in Fig. 5.

C. Barrier Index

The barrier index, or VB, is a convenient parameter forcharacterizing thermionic converter development, com-paring experimental data, and evaluating converter con-cepts. It is an inverse figure of merit because the lower theVB, the higher is the converter performance. The barrierindex is defined as

VB = Vd + φC. (10)

Inspection of Fig. 4c shows that, the lower the φC and Vd,the higher is the output voltage V .

The barrier index can be defined operationally. For anygiven emitter temperature and output current, it is possibleto adjust cesium pressure, spacing, and collector tempera-ture to maximize the power output. The spacing envelopeof the optimized performance curves is shown in Fig. 6for a converter with an emitter temperature of 1800 K.This envelope is shifted by a constant potential differencefrom the Boltzmann line, which represents the idealcurrent–voltage characteristic. This potential differenceis defined as the barrier index. In Fig. 6, the barrier indexof 2.1 eV represents good performance for a thermionicconverter with bare metal electrodes, correspondingto Vd � 0.5 eV and φC � 1.6 eV. The equation for theBoltzmann line is

JB = AT 2E exp(−V/kTE). (11)

The Boltzmann line represents the idealized converteroutput (up to the emitter saturation current) for zero




FIGURE 6 Spacing envelope of the optimized performancecurves for a converter with a barrier index of 2.1 eV. Emitter,(110)W; collector, niobium; TE = 1800 K; TC and TR, optimum.

collector work function and zero collector temperature.Thus, the barrier index incorporates the sum of the collec-tor work function and the electron transport losses (due toionization, scattering, electrode reflection, and electrodepatchiness) into a single factor. For a real thermionic con-verter, the current density is

J = AT 2E exp[−(V + VB)/kTE]. (12)

Improvements in the barrier index can be translated intoeither higher efficiency at a given emitter temperature orthe same efficiency at a lower temperature.

The lead efficiency (i.e., the thermionic converter ef-ficiency, which takes into account the Joule and thermalheat losses down the electrical lead) is shown in Fig. 7 as

FIGURE 7 Lead efficiency as a function of emitter tempera-ture, parametric in barrier index (TC = 700–900 K; J = 10 A/cm2;φC = 1.3–1.7 eV).

a calculated function of emitter temperature for a rangeof barrier index values. These data emphasize that thethermionic converter is an extremely high temperaturedevice.

IV. ELECTRODES

To operate at a practical power density and efficiency, anelectrode should provide of the order of 10 A/cm2 overa long period at emitter temperatures of 1600 to 1800 Kwithout excessive evaporation. This requirement impliesan emitter work function of 2.4 to 2.7 eV.

A. Bare Electrodes

A bare electrode is defined as a surface that does not re-quire a vapor supply from the inter-electrode space tomaintain its emission properties. No bare electrode hasbeen found that meets the foregoing criterion. The conven-tional (Ba, Sr, Ca)O coating on a nickel substrate, whichis widely used (e.g., radio, oscilloscope, television, mi-crowave tubes), is not stable at the desired emitter tem-peratures. Lanthanum hexaboride showed promise, but itshigh evaporation rate and the extreme difficulty of bond-ing it to a substrate over a large area has eliminated itfrom thermionic application. For short-term experimen-tal devices, dispenser cathodes (barium compounds im-pregnated in a tungsten matrix) have given useful results.However, their current density and stability do not meetthe needs of a practical thermionic converter.

1. Cesium Pressure Relationship

In almost all vapor converters, the cesium pressure is ad-justed by controlling the temperature of a liquid cesiumreservoir. To a good approximation, the cesium vapor pres-sure pcs in torr can be calculated from

pcs = 2.45 × 108 T −0.5R exp(−8910/TR), (13)

where the cesium reservoir temperature TR is given inkelvins. It is also possible to control the pressure usingcesium-intercalated graphite compounds that shift the op-erating reservoir temperature to higher values, which maybe advantageous in some applications.

2. Rasor Plot

Experimental and theoretical results indicate that, for lessthan a monolayer coverage, the work function can be ex-pressed to a good approximation as

φ = f (T/TR) (14)




FIGURE 8 Cesiated work function of a surface versus the ratioof electrode temperature to cesium reservoir temperature, para-metric in the bare work function of the surface.

for a given substrate material. This relationship is shown inFig. 8, parametric in the bare work function φB of the sub-strate. Such a representation, which condenses a morassof electron emission data into a tractable form, is usuallycalled a Rasor plot.

3. Saha–Langmuir Equation

Cesiated surfaces emit positive ions as well as electrons.The fraction of the cesium impingement rate, g, that evap-orates as ions is given by the Saha–Langmuir equation,

g = 1

1 + 2 exp(I − φ/kT ), (15)

where I is the ionization potential of the impinging va-por atoms (3.89 eV for cesium) and φ and T refer to thesurface.

B. Additives

It can be advantageous to operate thermionic converterswith additives in combination with cesium. For some ap-plications, the improved performance can be substantial.

1. Electropositive Additives

The work function of a metal surface is reduced whenthe surface area is covered with a fraction of adsorbed

monolayer of a more electropositive metal. Such an addi-tive (e.g., barium) could reduce the emitter work functionwhile using cesium primarily for thermionic ion emission.Operating in this mode would require a very high emit-ter temperature compared with the conventional ignitedmode.

2. Electronegative Additives

The adsorption of a fractional layer of an electronegativeelement, such as oxygen, will increase the work function ofthe surface. However, Langmuir and his co-workers foundthat, when such a surface is operated in cesium vapor, thework function (for a given cesium pressure and substratetemperature) may be significantly lower than without theelectronegative additive. The thermionic converter perfor-mance can be improved by a lower collector work functionor a lower cesium pressure in the interelectrode space (orboth).

This behavior is a result of the interaction of adsorbedlayers with opposing dipole polarities. It should not betoo surprising in view of the Rasor plot (see Fig. 8), whichshows that, the higher the bare work function of a substrate,the lower is the cesiated work function for a given ratioT/TR.

V. EXPERIMENTAL RESULTS

This section summarizes the experimental results fromseveral classes of thermionic converters. It should providea background for choosing the most promising paths toimproved performance.

A. Vacuum-Mode Diodes

The J–V characteristic of a close-spaced diode, indicat-ing the space charge and retardingmode regions, is il-lustrated in Fig. 4a. Practically, it has not been possi-ble to operate thermionic diodes at spacings appreciablycloser than 0.01 mm. At such spacings, the output powerdensity is usually less than 1.0 W/cm2. Using dispensercathodes for the emitter (1538 K) and collector (810 K),G. N. Hatsopoulos and J. Kaye obtained a current densityof 1.0 A/cm2 at an output potential of 0.7 V. They en-countered output degradation due to evaporation of bar-ium from the emitter to the collector, which increased thecollector work function. Although the electrode stabilityproblem can be avoided by using cesium vapor to ad-just the electrode work functions, the practical problemof maintaining spacings close enough for vacuum-modeoperation over large areas at high temperatures has elimi-nated close-spaced converters from practical applications.




To circumvent the problem of extremely small inter-electrode spacings in a vacuum converter, both magneticand electrostatic triodes have been investigated. However,neither device appears to be practical.

B. Bare Electrode Converters

1. Unignited-Mode Converters

Thermionic converters with bare electrodes have been op-erated in both the unignited and the ignited modes. Asdiscussed previously, unignited-mode operation requiresemitter temperatures above 2000 K. To approximate theideal diode J–V characteristics, electron–cesium scatter-ing must be minimized by interelectrode spacings of theorder of 0.1 mm. Under these conditions, the unignited-mode operation may be substantially more efficient thanthe ignited-mode operation, since the ions are supplied bythermal energy rather than by the arc drop, which subtractsfrom the output voltage.

Unfortunately, the high emitter temperatures requiredfor unignited operation eliminate most heat sources be-cause of materials limitations. For example, most nu-clear heat sources cannot be used on a long-term basisat these temperatures because of either fuel swelling orfuel–emitter incompatibility.

2. Ignited-Mode Converters

Thus far, essentially all practical thermionic convertershave operated in the ignited mode. This conventional con-verter has demonstrated power densities of 5 to 10 W/cm2

and efficiencies of 10 to 15% for emitter temperaturesbetween 1600 and 1800 K.

For a wide range of converter parameters, there exists anoptimum pressure–spacing product (or Pd) of ∼0.5-mmtorr. For higher values of this product, the plasma loss isincreased because of unnecessary electron–cesium scat-tering losses. From a practical viewpoint, it is difficult tofabricate large-area thermionic converters with spacingsof less than 0.25 mm. However, for long-lived space reac-tors, system studies indicate that interelectrode spacingsof up to 0.5 mm may be required to accommodate emitterdistortion due to fuel swelling.

The higher the bare work function, the lower is thecesium pressure required to maintain a given electronemission. For practical spacings, the reduced cesium pres-sure tends toward the optimum pressure–spacing productto reduce plasma losses and improve performance. Theoutput power density as a function of emitter tempera-ture, parametric in the bare work function of the emitter,is given in Fig. 9. It is evident that large increases in out-put power density can be achieved by the use of orientedsurfaces with bare work functions of greater than 4.5 eV.

FIGURE 9 Output power density as a function of emitter tem-perature, parametric in the bare work function of the emitter (J =10 A/cm2; d = 0.25 mm).

One avenue for improving thermionic converter perfor-mance is to fabricate emitters with preferred crystallineorientations with high bare work functions.

C. Additive Diodes

Controlled additions of oxygen into thermionic convert-ers can yield substantial improvement in power density,especially at spacings ≥0.5 mm. A comparison of theJ–V characteristics of two converters, identical exceptfor oxide collector coating, is shown in Fig. 10. The dra-matic increase in output is clear for the converter withthe tungsten oxide coating on the collector. In terms ofthe barrier index defined earlier, VB � 1.9 eV for the baretungsten collector diode. The improved performance isdue to a combination of reduced collector work functionand lower potential losses in the interelectrode plasma.The oxide collector supplies oxygen to the emitter so that

FIGURE 10 Comparison of the current density–voltage charac-teristics of two thermionic converters, identical except for oxidecoating on collector (—–, WOx , TC = 800 K; ---, W, TC = 975 K,TE = 1600 K; d = 1 mm).




a given current density can be obtained at a significantlylower cesium pressure. In effect, the added oxygen makesthe emitter act as if it has an extremely high bare workfunction. As a result of the lower cesium pressure, theoptimum interelectrode spacing increases.

VI. APPLICATIONS

Historically, the primary motivation for the developmentof thermionic converters has been their potential applica-tion to nuclear power systems in space, both reactor andradioisotope. In addition, thermionic converters are attrac-tive for terrestrial, solar, and combustion applications.

A. Nuclear Reactors

The capacity of a thermionic converter to operate effi-ciently at high emitter and collector temperatures insidethe core of a nuclear reactor makes it very suitable forspace power applications. This combination of character-istics results in a number of fundamental and developmen-tal advantages. For example, the necessity of radiatingaway all the reject heat from space conversion systemsputs a premium on the thermionic converter’s demon-strated capability of operating reliably and efficiently (typ-ically, 10–15%), even at high collector temperatures (up to1100 K). The capacity of the thermionic converter to op-erate inside the core of a reactor essentially eliminates thehigh-temperature heat transport system inherent in all out-of-core conversion systems. Thus, the thermionic reactorcoolant system is at radiator temperature so that, except forthe emitters, the balance of the reactor system (pumps, re-flector, controls, moderator, shield, etc.) operates near theradiator temperature. Even the high-temperature emittersare distributed inside the core in “bite-sized” pellets.

A cutaway drawing of a thermionic diode that has op-erated inside a reactor is shown in Fig. 11. The enricheduranium oxide fuel pellets are inside the tungsten emitter,which is made by chemical vapor deposition (CVD). Theoutside diameter of the emitter is ∼28 mm. The collec-tor is niobium, and the insulator seal is niobium–alumina.Thermionic diodes of this general design have operatedstably inside a reactor for more than a year, and an electri-cally heated converter has given constant output for morethan 5 years. Typically, emitter temperatures have rangedbetween 1700 and 2000 K.

A thermionic reactor is composed of an array ofthermionic fuel elements (TFEs) in which multiplethermionic converters are connected in series insidean electrically insulating tube, much like cells in aflashlight. In principle, it is possible to design reactorsystems with the thermionic converters out of core. This

FIGURE 11 Uranium oxide-fueled thermionic diode.

design concept presents a difficult problem of electricalinsulation at emitter temperature and vitiates many of thebasic advantages of the TFE reactor. It is also possibleto fuel the converter externally using a fuel element witha central coolant tube, the outer surface of which servesas the collector; such a design concept appears to bemore limited in system power than the flashlight TFEapproach.

The feasibility of the thermionic reactor system hasbeen demonstrated by in-core converter and TFE tests inthe United States, France, and Germany. As of the mid-1970s, four TOPAZ thermionic reactors had been built inRussia and tested at outputs of up to 10 kWe (kilowattselectrical).

B. Radioisotope Generators

A radioisotope thermionic generator uses the decay heatof a radioactive element. The design is usually modular. Amodule includes the radioisotope fuel, its encapsulation,a thermionic converter, heat transfer paths (for couplingheat from the fuel to the converter and from the converterto the radiator), hermetically sealed housing, and thermalinsulation. A variety of design concepts have been con-sidered (planar and cylindrical converters, heat pipe andconduction coupling, and several radioisotopes), and a fewfueled systems have been built. However, no report of anoperational use of a radioisotope thermionic generator hasyet been made.

The high operating temperature of the radioisotopethermionic converter poses difficult radiological safety




problems of fuel encapsulation integrity relative to launchpad accidents, launch aborts, and reentry. Since these prob-lems are more tractable at the lower operating tempera-tures of radioisotope thermoelectric generators, these unitshave been employed for remote and space missions requir-ing less than a kilowatt of electrical output.

A thermionic space reactor system does not suffer aradiological safety disadvantage compared with a ther-moelectric space reactor system since both systems arelaunched cold and neither reactor is started until a nuclearsafe orbit has been achieved. The radiological hazards as-sociated with the launch of a cold reactor system without afission product inventroy are significantly less than thoseof a launch of a radioisotope thermoelectric generator.

C. Combustion Converters

The demise of the U.S. space reactor effort in 1973, alongwith the accompanying OPEC oil action, provided the mo-tivation for utilizing on the ground the thermionic tech-nology that had been developed for space. There are anumber of potential terrestrial applications of thermionicconversion. These include power plant topping, cogener-ation with industrial processes, and solar systems.

Although refractory metals such as tungsten andmolybdenum operate stably in the vacuum of outer spaceat thermionic temperatures, converters operating in airor combustion atmosphere would oxidize to destructionwithin minutes without protection. Therefore, it is essen-tial to develop a protective coating, or “hot shell,” to isolatethe refractory metals from the terrestrial environment (seeFig. 1). The thermionic converter illustrated in Fig. 12is representative of the combustion-heated diodes thathave been constructed and tested. The dome is exposed tohigh-temperature combustion gases. Electrons evaporatedfrom the tungsten emitter are condensed on the mating

FIGURE 12 Combustion thermionic converter.

nickel collector, which is air-cooled. A ceramic sealprovides electrical insulation between the electrodes. Theelectrical power output is obtained between the flangeand the air tube, which also functions as the collectorlead.

The key component is the hot-shell–emitter (HS–EM)structure, which operates between the emitter andcollector temperatures. This trilayer composite structure(tungsten–graphite–silicon carbide) is fabricated byCVD. First, the graphite is machined to the desired con-figuration. Next, the inside of the graphite is coated withCVD tungsten to form the emitter and its electrical lead.Finally, the outside is coated with CVD silicon carbide toprotect the emitter from the combustion atmosphere.

The properties of the materials used in the HS–EMstructure complement one another. Tungsten is used be-cause of its low vapor pressure at emitter temperatures,high bare work function, low electrical resistivity, andcompatibility with cesium. Silicon carbide is chosen be-cause of its excellent high-temperature corrosion resis-tance, strength, thermal shock characteristics, and closematch to the thermal expansion of tungsten. The graphiteis selected, to match the thermal expansions of the tungstenand silicon carbide. The graphite is essential for providinggood thermal shock and thermal cycle characteristics tothe composite structure.

Arrays of such converters will be connected in series toprovide practical output voltage. Converters of this gen-eral design have operated stably for periods of as longas 12,500 h at emitter temperatures of up to 1730 K ina natural-gas burner. The HS–EM structure has survivedsevere thermal shock, thermal cycle, and pressure tests.

Although combustion thermionics has made substan-tial progress, it is not yet ready for commercialization.The first application would probably be a thermionicburner that could be retrofit onto industrial furnaces. A




thermionic burner is a combustor whose walls are linedwith thermionic converters. The emitters of the convertersreceive heat from the combustion gases and convert partof this heat into electricity while rejecting the balance ofthe heat from the collectors into the air for combustion.Operational experience with such units should provide thedatabase for subsequent power plant topping use.

The most recent cost estimate for an installedthermionic cogeneration system was approximately$1600/kW. Achievement of the corresponding convertercost of $540/kW would require additional investment inconverter and manufacturing development.

Relative to other advanced combustion conversion sys-tems, such as magnetohydrodynamics (MHD), ceramicblade turbine gas turbine, and potassium Rankine cy-cle, thermionic development costs should be substantiallylower. The cost effectiveness is a result of the modularityof thermionics, which makes it possible to perform mean-ingful experiments with small equipment. Thus, large in-vestments should not be required until there is a high prob-ability of success.

D. Solar Converters

Although the CVD silicon carbide converter was deve-loped for combustion applications, it is very suitable forsolar systems. Previous solar tests of terrestrial thermionicconverters had to utilize a window so that the high-tempe-rature refractory components could operate in an inertatmosphere. In addition to transmission losses, suchwindows are subject to problems of overheating andleakage.

A CVD converter has been solar-tested in a central re-ceiver heliostat array at the Georgia Institute of Technol-ogy. The test examined heat flux cycling, control of theoperating point, and mounting arrangements. The con-verter was mounted directly in the solar image with nocavity. The converter performance was comparable withcombustion measurements made on the same diode. Ther-mal cycling caused no problems, and the converter showedno degradation after testing.


ELECTRIC PROPULSION • NUCLEAR POWER REACTORS •PLASMA SCIENCE AND ENGINEERING

BIBLIOGRAPHY

Angrist, S. W. (1965). “Direct Energy Conversion,” Allyn and Bacon,Rockleigh, New Jersey.

Baksht, F. G., et al. (1978). “Thermionic Converters and Low Tem-perature Plasma” (Engl. transl.). Technical Information Center/U.S.Department of Energy, Washington, DC.

Hatsopoulos, G. N., and Gyftopoulos, E. P. (1974, 1979).“ThermionicEnergy Conversion,” Vols. 1 and 2. MIT Press, Cambridge, Mas-sachusetts.

Nottingham, W. B. (1956). Thermionic emission. In “Handbuch derPhysik,” Vol. 21. Springer-Verlag, New York.

Rasor, N. S. (1982). Thermionic energy conversion. In “Applied AtomicCollision Physics,” Ch. 5, Vol. 5. Academic Press, New York.

Ure, R. W., and Huffman, F. N. (1987). Thermoelectric and thermionicconversion. In “Standard Handbook for Electrical Engineers,” 12thed. McGraw-Hill, New York.

P1: GRB/GPA/GRD P2: GPA Final Pages/MBQ QC: GNB Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology EN016H-944 July 31, 2001 17:31

ThermoeconomicsGeorge TsatsaronisFrank CzieslaTechnical University of Berlin

I. Exergy AnalysisII. Review of an Economic Analysis

III. Thermoeconomic AnalysisIV. Thermoeconomic EvaluationV. Iterative Optimization of a Thermal System

VI. Recent Developments in Thermoeconomics

GLOSSARY

Exergy Useful energy or a measure for the quality ofenergy.

Exergy destruction Thermodynamic loss (inefficiency)due to irreversibilities within a system.

Exergetic efficiency Characterizes the performance of acomponent or a thermal system from the thermody-namic viewpoint.

Exergy loss Thermodynamic loss due to exergy transferto the environment.

Thermoeconomics Exergy-aided cost minimization.

THERMOECONOMICS is the branch of thermal sci-ences that combines a thermodynamic (exergy) analysiswith economic principles to provide the designer or op-erator of an energy-conversion system with informationwhich is not available through conventional thermody-namic analysis and economic evaluation but is crucial tothe design and operation of a cost-effective system. Ther-moeconomics rests on the notion that exergy (available

energy) is the only rational basis for assigning monetarycosts to the interactions of an energy conversion systemwith its surroundings and to the sources of thermodynamicinefficiencies within it. Since the thermodynamic consid-erations of thermoeconomics are based on the exergy con-cept, the terms exergoeconomics and thermoeconomicscan be used interchangeably.

A complete thermoeconomic analysis consists of (1)an exergy analysis, (2) an economic analysis, (3) exergycosting, and (4) a thermoeconomic evaluation. A ther-moeconomic analysis is usually conducted at the systemcomponent level and calculates the costs associated withall material and energy streams in the system and thethermodynamic inefficiencies (exergy destruction) withineach component. A comparison of the cost of exergydestruction with the investment cost for the same compo-nent provides useful information for improving the costeffectiveness of the component and the overall system bypinpointing the required changes in structure and param-eter values. An iterative thermoeconomic evaluation andoptimization is, among the currently available methods,the most effective approach for optimizing the design of a

659

P1: GRB/GPA/GRD P2: GPA Final Pages/MBQ QC: GNB


660 Thermoeconomics

system when a mathematical optimization procedure can-not be used due to the lack of information (e.g., cost func-tions) and our inability to appropriately consider importantfactors such as safety, availability, and maintainability ofthe system in the modeling process.

The objectives of thermoeconomics include

� Calculating the cost of each product stream generatedby a system having more than one product

� Optimizing the overall system or a specific component� Understanding the cost-formation process and the flow

of costs in a system

Thermoeconomics uses results from the synthesis, costanalysis, and simulation of thermal systems and providesuseful information for the evaluation and optimization ofthese systems as well as for the application of artificial in-telligence techniques to improve the design and operationof such systems.

I. EXERGY ANALYSIS

An exergy analysis identifies the location, the magnitude,and the sources of thermodynamic inefficiencies in a ther-mal system. This information, which cannot be providedby other means (e.g., an energy analysis), is very usefulfor improving the overall efficiency and cost effectivenessof a system or for comparing the performance of varioussystems (Bejan, Tsatsaronis, and Moran, 1996; Moran andShapiro, 1998).

An exergy analysis provides, among others, the exergyof each stream in a system as well as the real “energywaste,” i.e., the thermodynamic inefficiencies (exergy de-struction and exergy loss), and the exergetic efficiency foreach system component.

In the sign convention applied here, work done on thesystem and heat transferred to the system are positive.Accordingly, work done by the thermal system and heattransferred from the system are negative.

A. Exergy Components

Exergy is the maximum theoretical useful work (shaftwork or electrical work) obtainable from a thermal sys-tem as it is brought into thermodynamic equilibrium withthe environment while interacting with the environmentonly. Alternatively, exergy is the minimum theoreticalwork (shaft work or electrical work) required to form aquantity of matter from substances present in the environ-ment and to bring the matter to a specified state. Hence,exergy is a measure of the departure of the state of thesystem from the state of the environment.

The environment is a large equilibrium system in whichthe state variables (T0, p0) and the chemical potential ofthe chemical components contained in it remain constant,when, in a thermodynamic process, heat and materials areexchanged between another system and the environment.It is important that no chemical reactions can take placebetween the environmental chemical components. The en-vironment is free of irreversibilities, and the exergy of theenvironment is equal to zero. The environment is part ofthe surroundings of any thermal system.

In the absence of nuclear, magnetic, electrical, and sur-face tension effects, the total exergy of a system Esys canbe divided into four components: physical exergy E P H

sys ,kinetic exergy E K N , potential exergy E PT , and chemicalexergy EC H :

Esys = E P Hsys + E K N + E PT + EC H . (1)

The subscript sys distinguishes the total exergy andphysical exergy of a system from other exergy quantities,including transfers associated with streams of matter. Thetotal specific exergy on a mass basis esys is

esys = eP Hsys + eK N + ePT + eC H . (2)

The physical exergy associated with a thermodynamicsystem is given by

E P Hsys = (U − U0) + p0(V − V0) − T0(S − S0), (3)

where U, V, and S represent the internal energy, volume,and entropy of the system, respectively. The subscript 0denotes the state of the same system at temperature T0 andpressure p0 of the environment.

The rate of physical exergy E P Hms associated with a ma-

terial stream (subscript ms) is

E P Hms = (H − H 0) − T0(S − S0), (4)

where H and S denote the rates of enthalpy and entropy,respectively. The subscript 0 denotes property values attemperature T0 and pressure p0 of the environment.

Kinetic and potential exergy are equal to kinetic andpotential energy, respectively.

E K N = 1

2m �v2 (5)

E PT = mgz (6)

Here, �v and z denote velocity and elevation relative to coor-dinates in the environment (�v0 = 0, z0 = 0). Equations (5)and (6) can be used in conjunction with both systems andmaterial streams.

The chemical exergy is the maximum useful work ob-tainable as the system at temperature T0 and pressurep0 is brought into chemical equilibrium with the en-vironment. Thus, for calculating the chemical exergy,



Thermoeconomics 661

not only do the temperature and pressure have to bespecified, but the chemical composition of the environ-ment also has to be specified. Since our natural envi-ronment is not in equilibrium, there is a need to modelan exergy-reference environment (Ahrendts, 1980; Bejan,Tsatsaronis, and Moran, 1996; Szargut, Morris, andSteward, 1988). The use of tabulated standard chemi-cal exergies for substances contained in the environmentat standard conditions (Tref = 298.15 K, pref = 1.013 bar)facilitates the calculation of exergy values. Table I showsvalues of the standard chemical exergies of selected sub-stances in two alternative exergy-reference environments.The effect of small variations in the values of T0 and p0

TABLE I Standard Molar Chemical Exergy eCH of VariousSubstances at 298.15 K and pref

eC H (kJ/kmol)

Substance Formulaa Model Ib Model IIc

Ammonia NH3(g) 336,684 337,900

n-Butane C4H10(g) — 2,805,800

Calcium oxide CaO(s) 120,997 110,200

Calcium hydroxide Ca(OH)2(s) 63,710 53,700

Calcium carbonate CaCO3(s) 4,708 1,000

Calcium sulfate CaSO4 · 2H2O(s) 6,149 8,600(gypsum)

Carbon (graphite) C(s) 404,589 410,260

Carbon dioxide CO2(g) 14,176 19,870

Carbon monoxide CO(g) 269,412 275,100

Ethane C2H6(g) 1,482,033 1,495,840

Hydrogen H2(g) 235,249 236,100

Hydrogen peroxide H2O2(g) 133,587 —

Hydrogen sulfide H2S(g) 799,890 812,000

Methane CH4(g) 824,348 831,650

Methanol (g) CH3OH(g) 715,069 722,300

Methanol (l) CH3OH(l) 710,747 718,000

Nitrogen N2(g) 639 720

Nitrogen monoxide NO(g) 88,851 88,900

Nitrogen dioxide NO2(g) 55,565 55,600

Octane C8H18(l) — 5,413,100

Oxygen O2(g) 3,951 3,970

n-Pentane C5H12(g) — 3,463,300

Propane C3H8(g) — 2,154,000

Sulfur S(s) 598,158 609,600

Sulfur dioxide SO2(g) 301,939 313,400

Sulfur trioxide (g) SO3(g) 233,041 249,100

Sulfur trioxide (l) SO3(l) 235,743 —

Water (g) H2O(g) 8,636 9,500

Water (l) H2O (l) 45 900

a (g): gaseous, (l): liquid, (s): solid.b Ahrendts, 1980. In this model, pref = 1.019 atm.c Szargut, Morris, and Stewart, 1988. In this model, pref = 1.0 atm.

on the chemical exergy of reference substances might beneglected in practical applications.

The chemical exergy of an ideal mixture of N idealgases eCH

M,ig is given by

eCHM,ig =

N∑k=1

xk eC Hk + RT0

N∑k=1

xk ln xk, (7)

where T0 is the ambient temperature, eC Hk is the standard

molar chemical exergy of the kth substance, and xk is themole fraction of the kth substance in the system at T0.For solutions of liquids, the chemical exergy eC H

M,l can beobtained if the activity coefficients γk are known (Kotas,1985):

eCHM,l =

N∑k=1

xk eC Hk + RT0

N∑k=1

xk ln(γk xk). (8)

The standard chemical exergy of a substance not presentin the environment can be calculated by considering areversible reaction of the substance with other substancesfor which the standard chemical exergies are known. Forenergy-conversion processes, calculation of the exergy offossil fuels is particularly important.

Figure 1 shows a hypothetical reversible isotherm-isobaric reactor where a fuel reacts completely at steadystate with oxygen to form CO2, SO2, H2O, and N2. Allsubstances are assumed to enter and exit unmixed at T0,p0. The chemical exergy of a fossil fuel eCH

f on a molarbasis can be derived from exergy, energy, and entropy bal-ances for the reversible reaction (Bejan, Tsatsaronis, andMoran, 1996):

eCHf = −(�hR −T0�sR) + �eC H = −�gR + �eCH (9)

with

�hR =∑

i

νi hi = −hf +∑

k

νk hk = −HHV

�sR =∑

i

νi si = −sf +∑

k

νk sk

FIGURE 1 Device for evaluating the chemical exergy of a fuel.



662 Thermoeconomics

�gR = �hR − T0�sR

�eCH =∑

k

νk eCHk

i = f, O2, CO2, H2O, SO2, N2

k = O2, CO2, H2O, SO2, N2

νi and νk ≥ 0 : CO2, H2O, SO2, N2

νi and νk < 0 : f, O2

Here, �hR, �sR, and �gR denote the molar enthalpy,entropy, and Gibbs function, respectively, of the reversiblecombustion reaction of the fuel with oxygen. HHV is themolar higher heating value of the fuel, and νk(νi ) is thestoichiometric coefficient of the kth (i th) substance in thisreaction.

The higher heating value is the primary contributorto the chemical exergy of a fossil fuel. For back-of-the-envelope calculations, the molar chemical exergy of a fos-sil fuel eCH

f may be estimated with the aid of its molarhigher heating value HHV:

eC Hf

HHV≈

0.95 − 0.985 for gaseous fuels (exceptH2 and CH4)

0.98 − 1.00 for liquid fuels

1.00 − 1.04 for solid fuels (10)

For hydrogen and methane, this ratio is 0.83 and 0.94,respectively.

The standard molar chemical exergy eCHs of any sub-

stance not present in the environment can be determinedusing the change in the specific Gibbs function �g forthe reaction of this substance with substances present inthe environment (Bejan, Tsatsaronis, and Moran, 1996;Moran and Shapiro 1998):

eC Hs = −�g + �eCH = −

∑i

νi gi +∑i =s

νi eCHi , (11)

where gi , νi , and eCHi denote, for the i th substance, the

Gibbs function at T0 and p0, the stoichiometric coeffi-cient in the reaction, and the standard chemical exergy,respectively.

In solar energy applications, the ratio of the exergy ofsolar radiation to the total energy flux at the highest layerof the earth’s atmosphere is 0.933 (Szargut, Morris, andSteward, 1988).

B. Exergy Balance and Exergy Destruction

Thermodynamic processes are governed by the laws ofconservation of mass and energy. These conservation lawsstate that the total mass and total energy can neither becreated nor destroyed in a process. However, exergy is

not generally conserved but is destroyed by irreversibili-ties within a system. Furthermore, exergy is lost, in gen-eral, when the energy associated with a material or energystream is rejected to the environment.

1. Closed System Exergy Balance

The change in total exergy (Esys,2 − Esys,1) of a closedsystem caused through transfers of energy by work andheat between the system and its surroundings is given by

Esys,2 − Esys,1 = Eq + Ew − ED. (12)

The exergy transfer Eq associated with heat transfer Q is

Eq =∫ 2

1

(1 − T0

Tb

)δQ, (13)

where Tb is the temperature at the system boundary atwhich the heat transfer occurs.

The exergy transfer Ew associated with the transfer ofenergy by work W is given by

Ew = W + p0(V2 − V1). (14)

In a process in which the system volume increases(V2 > V1), the work p0(V2 − V1) being done on the sur-roundings is unavailable for use, but it can be recoveredwhen the system returns to its original volume V1.

A part of the exergy supplied to a real thermal systemis destroyed due to irreversibilities within the system. Theexergy destruction ED is equal to the product of entropygeneration Sgen within the system and the temperature ofthe environment T0.

ED = T0Sgen ≥ 0. (15)

Hence, exergy destruction can be calculated either fromthe entropy generation using an entropy balance or directlyfrom an exergy balance. ED is equal to zero only in idealprocesses.

2. Control Volume Exergy Balance

Exergy transfer across the boundary of a control volumesystem can be associated with material streams and withenergy transfers by work or heat. The general form of theexergy balance for a control volume involving multipleinlet and outlet streams is

d Ecv,sys

dt=

∑j

(1 − T0

Tj

)Q j︸︷︷︸

Eq, j

+(

Wcv + p0dVcv

dt

)︸︷︷︸

Ew

+∑

i

E i −∑

e

Ee − E D, (16)

where E i and Ee are total exergy transfer rates at inletsand outlets [see Eq. (4) for the physical exergy associated



Thermoeconomics 663

with these transfers]. The term Q j represents the time rateof heat transfer at the location on the boundary where thetemperature is Tj . The associated rate of exergy transferEq, j is given by

Eq, j =(

1 − T0

Tj

)Q j . (17)

The exergy transfer Ew associated with the time rate ofenergy transfer by work Wcv other than flow work is

Ew = Wcv + p0dVcv

dt. (18)

Finally, E D accounts for the time rate of exergy destructiondue to irreversibilities within the control volume. Eitherthe exergy balance [Eq. (16)] or E D = T0 Sgen can be usedto calculate the exergy destruction in a control volumesystem.

Under steady state conditions, Eq. (16) becomes

0 =∑

j

Eq, j + Wcv +∑

i

E i −∑

e

Ee − E D. (19)

3. Exergy Destruction

The real thermodynamic inefficiencies in a thermal systemare related to exergy destruction and exergy loss. All realprocesses are irreversible due to effects such as chemicalreaction, heat transfer through a finite temperature differ-ence, mixing of matter at different compositions or states,unrestrained expansion, and friction. An exergy analysisidentifies the system components with the highest thermo-dynamic inefficiencies and the processes that cause them.

In general, inefficiencies in a component should beeliminated or reduced if they do not contribute to a re-duction in capital investment for the overall system or areduction of fuel costs in another system component. Ef-forts for reducing the use of energy resources should becentered on components with the greatest potential forimprovement. Owing to the present state of technologi-cal development, some exergy destructions and losses in asystem component are unavoidable (Tsatsaronis and Park,1999). For example, the major part of exergy destructionin a combustion processes cannot be eliminated. Only asmall part can be reduced by preheating the reactants andreducing the amount of excess air.

The objective of a thermodynamic optimization is tominimize the inefficiencies, whereas the objective of athermoeconomic optimization is to estimate the cost-optimal values of the thermodynamic inefficiencies.

Heat transfer through a finite temperature difference isirreversible. Figure 2 shows the temperature profiles fortwo streams passing through an adiabatic heat exchanger.The following expression for the exergy destruction E D ,q

due to heat transfer from the hot stream 3 to the cold

FIGURE 2 Temperature profiles and thermodynamic averagetemperatures for two streams passing through an adiabatic heatexchanger at constant pressure.

stream 1 can be derived (Bejan, Tsatsaronis, and Moran,1996):

E D ,q = T0 Q Tha − Tca

Tha Tca, (20)

where the thermodynamic average temperatures Tha andTca of the hot stream and the cold stream are given by

Ta =∫ e

i T ds

se − si= he − hi −

∫ ei v dp

se − si. (21)

Here, the subscripts i and e denote inlet and exit, respec-tively. For applications where the heat transfer occurs atconstant pressure, Eq. (21) reduces to

Ta = he − hi

se − si(22)

Equation (20) shows that the difference in the averagethermodynamic temperatures (Tha − Tca) is a measure forthe exergy destruction. Mismatched heat capacity ratesof the two streams, i.e., (mcp)h/(mcp)c = 1, and a finiteminimum temperature difference (�Tmin = (T3 − T2) inFig. 2) are the causes of the thermodynamic inefficien-cies. Matching streams of significantly different heat ca-pacity rates mcp in a heat exchanger should be avoided.Furthermore, the lower the temperature levels Tha and Tca ,the greater the exergy destruction at the same temperaturedifference (Tha − Tca).

The rate of exergy destruction associated with frictionE D, f r can be expressed as

E D, f r = − T0

Ta· m ·

∫ e

iv dp, (23)

where Ta is the thermodynamic average temperature ofthe working fluid and

∫ ei vdp is the head loss (Bejan,

Tsatsaronis, and Moran, 1996). The effect of friction ismore significant at higher mass flow rates and lower tem-perature levels. Although exergy destruction related tofriction is usually of secondary importance in thermalsystems, the costs associated with the exergy destructionmight be significant. Since the unit cost of electrical or me-chanical power required to feed a pump, compressor, or



664 Thermoeconomics

fan is significantly higher than the unit cost of a fossil fuel,each unit of exergy destruction by frictional dissipation isrelatively expensive.

Examples for the exergy loss include heat transfer to theenvironment (“heat losses”), dissipation of the kinetic en-ergy of an exhaust gas, and irreversibilities due to mixingof an exhaust gas with the atmospheric air.

Guidelines for improving the use of energy resourcesin thermal systems by reducing the sources of thermody-namic inefficiencies are discussed by Bejan, Tsatsaronis,and Moran (1996) and by Sama (1995). However, the ma-jor contribution of an exergy analysis to the evaluationof a system comes through a thermoeconomic evaluationthat considers not only the inefficiencies, but also the costsassociated with these inefficiencies and the investment ex-penditures required to reduce the inefficiencies.

C. Exergetic Variables

The performance evaluation and the design optimizationof thermal systems require a proper definition of the exer-getic efficiency and a proper costing approach for eachcomponent of the system (Lazzaretto and Tsatsaronis,November 1999). The exergetic efficiency of a compo-nent is defined as the ratio between product and fuel. Theproduct and fuel are defined by considering the desiredresult produced by the component and the resources ex-pended to generate the result.

εk ≡ EP,k

E F,k= 1 − E D,k + E L ,k

E F,k(24)

An appropriately defined exergetic efficiency is the onlyvariable that unambiguously characterizes the perfor-mance of a component from the thermodynamic view-point (Tsatsaronis, 1999). Table II shows the definition offuel and product for selected system components at steadystate operation.

The rate of exergy destruction in the kth component isgiven by

E D,k = E F,k − E P,k − E L ,k . (25)

Here, E L ,k represents the exergy loss in the kth compo-nent, which is usually zero when the component bound-aries are at T0. For the overall system, EL includes theexergy flow rates of all streams leaving the system (see,for example, the different definitions of EF for a fuel cellin Table II).

In addition to εk and ED,K , the thermodynamic evalua-tion of a system component is based on the exergy destruc-tion ratio yD,k , which compares the exergy destruction inthe kth component with the fuel supplied to the overallsystem EF,tot:

yD,k ≡ ED,k

EF,tot. (26)

This ratio expresses the percentage of the decrease in theoverall system exergetic efficiency due to the exergy de-struction in the kth system component:

εtot = EP,tot

EF,tot= 1 −

∑k

yD,k − EL ,tot

EF,tot. (27)

ED,k is an absolute measure of the inefficiencies in the kthcomponent, whereas εk and yD,k are relative measures ofthe same inefficiencies. In εk the exergy destruction withina component is related to the fuel for the same component,whereas in yD,k the exergy destruction in a component isrelated to the fuel for the overall system.

II. REVIEW OF AN ECONOMIC ANALYSIS

In the evaluation and cost optimization of an energy-conversion system, we need to compare the annual val-ues of capital-related charges (carrying charges), fuelcosts, and operating and maintenance expenses. Thesecost components may vary significantly within the planteconomic life. Therefore, levelized annual values for allcost components should be used in the evaluation and costoptimization.

The following sections illustrate the total revenue re-quirement method (TRR method) which is based on pro-cedures adopted by the Electric Power Research Institute(EPRI; 1993). This method calculates all the costs associ-ated with a project, including a minimum required returnon investment. Based on the estimated total capital invest-ment and assumptions for economic, financial, operating,and market input parameters, the total revenue require-ment is calculated on a year-by-year basis. Finally, thenonuniform annual monetary values associated with theinvestment, operating (excluding fuel), maintenance, andfuel costs of the system being analyzed are levelized; thatis, they are converted to an equivalent series of constantpayments (annuities).

An economic analysis can be conducted in current dol-lars by including the effect of inflation or in constant dol-lars by not including inflation. In general, studies involvingthe near term (the next 5–10 years) values are best pre-sented in current dollars. The results of longer term studies(20–40 years) may be best presented in constant dollars.

A. Estimation of Total Capital Investment

The best cost estimates for purchased equipment can beobtained directly through vendors’ quotations. The next



Thermoeconomics 665

TABLE II Exergy Rates Associated with Fuel and Product for Defining Exergetic Efficiencies of Selected Components atSteady State Operation

Component Schematic Exergy rate of product EP Exergy rate of fuel EF

Compressor, pump, E2 − E1 E3

quad or fan

Turbine or expander E4 E1 − E2 − E3

Heat exchangera E2 − E1 E3 − E4

Mixing unit m1(e3 − e1) m2(e2 − e3)

Combustion chamber E3 − E2 E1 − E4

Steam generator (E6 − E5) + (E8 − E7) (E1 + E2) − (E3 + E4)

Deaerator m2(e3 − e2) m1e1 − (m3 − m2)e1 − m4e4

Compressor with cooling E3 + E4 + E5 − E2 E1

air extractions

continues



666 Thermoeconomics

TABLE II (Continued )

Component Schematic Exergy rate of product EP Exergy rate of fuel EF

Expander with cooling E3 E1 + E4 + E5 − E2

air supply

Fuel cell W + (E6 − E5) Fuel cell as part of a system:

(E1 + E2) − (E3 + E4)

Stand-alone fuel cell:

E1 + E2

Ejector m1(e3 − e1) m2(e2 − e3)

Distillation columnb EC H2 + EC H

3 + EC H6 + EC H

7 (E4 − E5) + m7(eP H1 − eP H

7 )

− EC H1 + m6(eP H

6 − eP H1 ) + m3(eP H

1 − eP H3 )

+ m2(eP H2 − eP H

1 )

Gasifier EC H3 + (E P H

3 + E P H4 EC H

1 + EC H2 − EC H

4

− E P H2 − E P H

1 )

Steam reformer E3 − E1 − E2 E4 + E5 − E6

Evaporator including E2 + E5 − E1 E3 − E4

steam drum

a These definitions assume that the purpose of the heat exchanger is to heat the cold steam (T1 > T0). If the purpose of the heat exchanger isto provide cooling (T3 < T0), then the following relations should be used: E P = E4 − E3 and E F = E1 − E2.

b Here, it is assumed that eC Hj > eC H

1 ( j = 2, 3, 6, 7), eP H2 > eP H

1 , eP H6 > eP H

1 , eP H3 < eP H

1 , eP H7 < eP H

1 .



Thermoeconomics 667

best source of cost estimates are cost values from past pur-chase orders, quotations from experienced professionalcost estimators, or calculations using the extensive costdatabases often maintained by engineering companiesor company engineering departments. In addition, somecommercially available software packages can assist withcost estimation.

When vendor quotations are lacking, or the cost or timerequirements to prepare cost estimates are unacceptablyhigh, the purchase costs of various equipment items can beobtained from the literature where they are usually givenin the form of estimating charts. These charts have beenobtained through a correlation of a large number of costand design data. In a typical cost-estimating chart, whenall available cost data are plotted versus the equipmentsize on a double logarithmic plot, the data correlationresults in a straight line within a given capacity range.The slope of this line α represents an important cost-estimating parameter (scaling exponent) as shown by therelation

CP,Y = CP,W

(XY

XW

)α

, (28)

where CP,Y is the purchase cost of the equipment inquestion, which has a size or capacity XY ; and CP,W isthe purchase cost of the same type of equipment in thesame year but of capacity or size XW .

In practice, the scaling exponent α does not remain con-stant over a large size range. For very small equipmentitems, size has almost no effect on cost, and the scal-ing exponent α is close to zero. For very large capacities,the difficulties of building and transporting the equipmentmay increase the cost significantly, and the scaling ex-ponent approaches unity. Examples of scaling exponentsfor different plant components are provided by Bejan,Tsatsaronis, and Moran (1996). In the absence of othercost information, an exponent value of 0.6 may be used(six-tenths rule).

Predesign capital cost estimates are very often assem-bled from old cost data. These reference data Cref canbe updated to represent today’s cost Cnew by using anappropriate cost index.

Cnew = Cref

(Inew

Iref

)(29)

Here, Inew and Ire f are the values of the cost index todayand in the reference year, respectively. Cost indices forspecial types of equipment and processes are publishedfrequently in several journals (e.g., Chemical Engineering,The Oil and Gas Journal, and Engineering News Record ).

B. Calculation of Revenue Requirements

The annual total revenue requirement (TRR, total productcost) for a system is the revenue that must be collected ina given year through the sale of all products to compen-sate the system operating company for all expendituresincurred in the same year and to ensure sound economicplant operation. It consists of two parts: carrying chargesand expenses. Carrying charges are a general designationfor charges that are related to capital investment, whereasexpenses are used to define costs associated with the oper-ation of a plant. Carrying charges (CC ) include the follow-ing: total capital recovery; return on investment for debt,preferred stock, and common equity; income taxes; andother taxes and insurance. Examples for expenses are fuelcost (FC ) and operating and maintenance costs (OMC ).All annual carrying charges and expenses have to be es-timated for each year over the entire economic life of aplant. Detailed information about the TRR method canbe found in Bejan, Tsatsaronis, and Moran (1996) and inEPRI (1993).

C. Levelized Costs

The series of annual costs associated with carrying chargesCC j and expenses (FC j and OMC j ) for the j th year ofplant operation is not uniform. In general, carrying chargesdecrease while fuel costs increase with increasing yearsof operation (Bejan, Tsatsaronis, and Moran, 1996). Alevelized value TRRL for the total annual revenue require-ment can be computed by applying a discounting factorand the capital-recovery factor CRF:

TRRL = CRFn∑1

TRR j

(1 + ieff) j, (30)

where TRR j is the total revenue requirement in the j thyear of plant operation, ieff is the average annual effectivediscount rate (cost of money), and n denotes the plant eco-nomic life expressed in years. For Eq. (30), it is assumedthat each money transaction occurs at the end of each year.The capital-recovery factor CRF is given by

CRF = ieff (1 + ieff)n

(1 + ieff)n − 1. (31)

If the series of payments for the annual fuel cost FC j isuniform over time except for a constant escalation rFC

(i.e., FC j = FC0(1 + rFC) j ), then the levelized value FCL

of the series can be calculated by multiplying the fuelexpenditure FC0 at the beginning of the first year by theconstant-escalation levelization factor CELF.

FCL = FC0CELF = FC0kFC

(1 − kn

FC

)(1 − kFC)

CRF (32)



668 Thermoeconomics

with

kFC = 1 + rFC

1 + ieffand rFC = constant.

The terms rFC and CRF denote the annual escalation ratefor the fuel cost and the capital-recovery factor [Eq. (31)],respectively.

Accordingly, the levelized annual operating and main-tenance costs OMCL are given by

OMCL = OMC0CELF

= OMC0kOMC

(1 − kn

OMC

)(1 − kOMC)

CRF (33)

with

kOMC = 1 + rOMC

1 + ieffand rOMC = constant.

The term rOMC is the nominal escalation rate for the oper-ating and maintenance costs.

Finally, the levelized carrying charges CCL are obtainedfrom

CCL = TRRL − FCL − OMCL . (34)

The major difference between a conventional economicanalysis and an economic analysis conducted as part ofa thermoeconomic analysis is that the latter is done atthe plant component level. The annual carrying charges(superscript C I = capital investment) and operating andmaintenance costs (superscript OM) of the total plant canbe apportioned among the system components accordingto the contribution of the kth component to the purchased-equipment cost PECtot =

∑k PECk for the overall

system:

ZCIk = CCL

τ

PECk∑k PECk

; (35)

ZOMk = OMCL

τ

PECk∑k PECk

. (36)

Here, PECk and τ denote the purchased-equipment costof the kth plant component and the total annual time (inhours) of system operation at full load, respectively. Theterm Zk represents the cost rate associated with capitalinvestment and operating and maintenance expenses:

Z k = ZCIk + ZOM

k . (37)

The levelized cost rate of the expenditures for fuel C F

supplied to the overall system is given by

C F = FCL

τ. (38)

Levelized costs, such as ZCIk , ZOM

k , and C F , are used asinput data for the thermoeconomic analysis.

D. Sensitivity Analysis

An economic analysis generally involves more uncertain-ties than a thermodynamic analysis. In the above discus-sion, it has been assumed that each variable in the eco-nomic analysis is known with certainty. However, manyvalues used in the calculation are uncertain. A sensitiv-ity analysis determines by how much a reasonable rangeof uncertainty assumed for each uncertain variable affectsthe final decision. Sensitivity studies are recommended toinvestigate the effect of major assumptions about valuesreferring to future years (e.g., cost of money, inflation rate,and escalation rate of fuels) on the results of an economicanalysis.

III. THERMOECONOMIC ANALYSIS

The exergy analysis yields the desired information for acomplete evaluation of the design and performance of anenergy system from the thermodynamic viewpoint. How-ever, we still need to know how much the exergy destruc-tion in a system component costs the system operator.Knowledge of this cost is very useful in improving thecost effectiveness of the system.

In a thermoeconomic analysis, evaluation and optimiza-tion, the cost rates associated with each material and en-ergy stream are used to calculate component-related ther-moeconomic variables. Thermoeconomic variables dealwith investment costs and the corresponding costs associ-ated with thermodynamic inefficiencies. Based on ratio-nal decision criteria, the required changes in structure andparameter values are identified. Thus, the cost minimiza-tion of a thermal system involves finding the optimumtrade-offs between the cost rates associated with capitalinvestment and exergy destruction.

A. Exergy Costing

In exergy costing a cost is assigned to each exergy stream.The cost rate associated with the j th material stream isexpressed as the product of the stream exergy rate E j andthe average cost per exergy unit c j :

C j = E j · c j = m j · e j · c j , (39)

where e j is the specific exergy on a mass basis. A cost isalso assigned to the exergy transfers associated with heator work:

Cq = cq · Q; (40)

Cw = cw · W . (41)

Here, c j , cq , and cw denote average costs per unit of ex-ergy in dollars per gigajoule exergy. The costs associated



Thermoeconomics 669

with each material and energy stream in a system are cal-culated with the aid of cost balances and auxiliary costequations.

Sometimes it is appropriate to consider the costs ofphysical and chemical exergy associated with a materialstream separately. By denoting the average costs per unitof physical and chemical exergy by cPH

j and cCHj , respec-

tively, the cost rate associated with stream j becomes

C j = c j E j = cPHj EPH

j + cCHj ECH

j

= m j(cPH

j ePHj + cCH

j eCHj

). (42)

B. Cost Balance

Exergy costing involves cost balances usually formulatedfor each system component separately. A cost balance ap-plied to the kth system component expresses that the totalcost of the exiting streams equals the total cost of the en-tering streams plus the appropriate charges due to capitalinvestment and operating and maintenance expenses Z(Fig. 3).

n∑j=1

C j,k,in + ZCIk + ZOM

k︸︷︷︸Zk

=m∑

j=1

C j,k,out (43)

Cost balances are generally written so that all termsare positive. The rates ZC I

k and Z O Mk are calculated from

Eqs. (35) and (36).Introducing the cost rate expressions from Eqs. (39)–

(41), we obtainn∑

j=1

(c j E j )k,in + ZCIk + ZOM

k =m∑

j=1

(c j E j )k,out. (44)

The exergy rates E j entering and exiting the kth sys-tem component are calculated in an exergy analysis. Ina thermoeconomic analysis of a component, we may as-sume that the costs per exergy unit c j are known for allentering streams. These costs are known either from thecomponents they exit or, if they are streams entering theoverall system, from their purchase costs. Consequently,the unknown variables that need to be calculated with the

FIGURE 3 Schematic of a system component to illustrate a costbalance.

aid of the cost balance are the costs per exergy unit ofthe exiting streams. Usually, some auxiliary relations arerequired to calculate these costs.

In the exergetic evaluation, a fuel and a product weredefined for each component of a system (see Table II). Thecost flow rates associated with the fuel C F and productC P of a component are calculated in a similar way tothe exergy flow rates E F and E P . Then, the cost balancebecomes

C P = C F + Z − C L . (45)

The term C L represents the monetary loss associatedwith the rejection of exergy to the environment. Table IIIshows the definitions of C P and C F for selected compo-nents at steady state.

C. Auxiliary Costing Equations

When the costs of entering streams are known, a cost bal-ance is generally not sufficient to determine the costs ofthe streams exiting a component because the number ofexiting streams is usually larger than one. Thus, in gen-eral, it is necessary to formulate some auxiliary equationsfor a component, with the number of these equations be-ing equal to the number of exiting streams minus one.The auxiliary equations are called either F equations orP equations depending on whether the exergy stream be-ing considered is used in the calculation of fuel or productfor the component (Lazzaretto and Tsatsaronis, November1999).

F Equations. The total cost associated with the removalof exergy from an exergy stream in a component mustbe equal to the cost at which the removed exergy wassupplied to the same stream in upstream components.The exergy difference of this stream between inlet andoutlet is considered in the definition of fuel for thecomponent.

P Equations. Each exergy unit is supplied to any streamassociated with the product of a component at the sameaverage cost cP,k . This cost is calculated from the costbalance and the F equations.

The following two examples illustrate the development ofthe F and P equations.

Example 1. Calculate the costs of the streams exiting theadiabatic turbine (subscript T ) or expander in Table III.The cost balance for this component is

C1 + ZT = C2 + C3 + C4. (46)

By grouping together the terms associated with fuel andproduct, we obtain



670 Thermoeconomics

TABLE III Cost Rates Associated with Fuel and Product for Selected Components at Steady State Operation

Component Schematic Cost rate of product C P Cost rate of fuel C F

Compressor pump, C2 − C1 C3

or fan

Turbine or expander C4 C1 − C2 − C3

Heat exchangera C2 − C1 C3 − C4

Mixing unit m1(c3,1e3 − c1e1) m2c2(e2 − e3)

with c3,1 = c3

+ m2

m1(c3 − c2)

Combustion C3 − C2 C1 − C4

chamber

Steam generator (C6 − C5) + (C8 − C7) (C1 + C2) − (C3 + C4)

Deaerator m2(c3,2e3 − c2e2) (m3 − m2)c1(e1 − e3)

with c3,2 = c3 + m4c1(e1 − e4)

+ m3 − m2

m2(c3 − c1)

Compressor with C3 + C4 + C5 − C2 C1

cooling airextractions

Expander with C3 C1 + C4 + C5 − C2

cooling airsupply

continues



Thermoeconomics 671

TABLE III (Continued )

Component Schematic Cost rate of product C P Cost rate of fuel C F

Fuel cell Fuel cell as part of a system: Fuel cell as part of a system:

C7 + (C6 − C5) + (C P H3 − C P H

1 ) (CC H1 + CC H

2 ) − (CC H3 + CC H

4 )+ (C P H

4 − C P H2 ) Stand-alone fuel cell:

Stand-alone fuel cell: C1 + C2

C7 + (C6 − C5)

Ejector m1(c3,1e3 − c1e1) m2c2(e2 − e3)wherec3,1 = c3 + m2

m1(c3 − c2)

Distillation columnb CC H2 + CC H

3 + CC H6 + CC H

7 − CC H1 (C4 − C5) + m7(cP H

1 eP H1

+ m6(cP H6 eP H

6 − cC H1 eP H

1 ) − cP H7 eP H

7 ) + m3(cP H1 eP H

1

+ m2(cP H2 eP H

2 − cC H1 eP H

1 ) − cP H3 eP H

3 )

Gassifier CC H3 + (C P H

3 + C P H4 CC H

1 + CC H2 − CC H

4

− C P H2 − C P H

1 )

Steam reformer C3 − C1 − C2 C4 + C5 − C6

Evaporator C2 + C5 − C1 C3 − C4

includingsteam drum

a These definitions assume that the purpose of the heat exchanger is to heat the cold steam (T1 > T0). If the purpose of the heat exchanger is toprovide cooling (T3 < T0), then the following relations should be used: C P = C4 − C3 and C F = C1 − C2.

b Here , it is assumed that eC Hj > eC H

1 ( j = 2, 3, 6, 7), eP H2 > eP H

1 , eP H6 > eP H

1 , eP H3 < eP H

1 , eP H7 < eP H

1 .

cp · E P = C4︸︷︷︸C P

= (C1 − C2 − C3)︸︷︷︸C F

+ Z T . (47)

The F equation for this component states that the costassociated with each removal of exergy from the enteringmaterial stream (stream 1) must be equal to the averagecost at which the removed exergy (E1 − E2 − E3) wassupplied to the same stream in upstream components.

C1 − C2 − C3 = c1 · (E1 − E2 − E3) (48)

Equation (48) is equivalent to

c2 = c3 = c1. (49)

The unknown cost rate C4 can be calculated with the aidof Eq. (48) or (49) and the cost balance [Eq. (47)].



672 Thermoeconomics

Example 2. Calculate the costs of the streams exitingthe adiabatic steam generator (subscript SG) shown inTable III, if streams 3 and 4 are (case a) rejected to theenvironment without additional expenses and (case b) dis-charged to the environment after using ash handling andgas cleaning equipment.

The cost balance for this component is

C1 + C2 + C5 + C7 + Z SG = C3 + C4 + C6 + C8. (50)

By grouping together the terms associated with fuel andproduct, we obtain

cp · E P = (C6 − C5) + (C8 − C7)︸︷︷︸C P

= (C1 + C2) − (C3 + C4)︸︷︷︸C F

+ Z SG . (51)

� Case a: The costs associated with streams 3 and 4 arecalculated from F equations: Each exergy unit in the ash(stream 3) and in the flue gas (stream 4) is supplied bycoal and oxygen at the same average cost.

C3

E3= C4

E4= C1 + C2

E1 + E2(52)

or

c3 = c4 = C1 + C2

E1 + E2(53)

The F equations [Eq. (52)] and the cost balance [Eq. (51)]may be used to calculate the cP value in Eq. (51). Then,the P equations supply the costs associated with streams 6and 8: each exergy unit is supplied to the main steamstream and to the reheat stream at the same average cost.

cP = C6 − C5

E6 − E5= C8 − C7

E8 − E7(54)

� Case b: All costs associated with the final disposalof streams 3 and 4 must be charged to the useful streamsexiting the boiler, that is, to the main steam and the hotreheat.

The following equations are obtained from the overallsystem (Fig. 4) and the cost balances for ash handling(subscript ah) and gas cleaning (subscript gc):

C9

E9= C10

E10= C1 + C2

E1 + E2, (55)

C3 = C9 − Z ah, (56)

C4 = C10 − Z gc. (57)

The cost rates C3 and C4 are usually negative in case b.The unknown cost rates C3, C4, C6, C8, C9, and C10 arecalculated from Eqs. (51) and (54)–(57).

FIGURE 4 Schematic of a steam generator to illustrate case b inExample 2.

For the system components having only one exitingstream, such as compressors without extractions, pumps,fans, mixing units, and ejectors, the cost associated withthis exiting stream can be calculated from the cost balanceEq. (43). When the costs of physical and chemical exergyare considered separately (e.g., for a distillation columnand a gasifier in Table III), and when more than one exergystream exits a system component, F or P equations arerequired, depending on whether the exiting stream beingconsidered belongs to the fuel or to the product.

The costing equations for the turbine and the steamgenerator are discussed above in Examples 1 and 2. Inthe following, the cost balances and the required F andP equations are given for the remaining devices shown inTable III.

� Heat exchanger (HX):

(C2 − C1) = Z H X + (C3 − C4) (58)

C4

E4= C3

E3(59)

� Combustion chamber (CC):

(C3 − C2) = ZCC + (C1 − C4) (60)

C4

E4= C1

E1(61)

� Deaerator (DA):

C4 + C3 = C1 + C2 + Z D A (62)

C4

E4= C1

E1(63)

� Compressor with cooling air extractions (C):

c3 E3 + c4 E4 + c5 E5 − c2 E2 = c1 E1 + ZC (64)

c3e3 − c2e2

e3 − e2= c4e4 − c2e2

e4 − e2= c5e5 − c2e2

e5 − e2(65)



Thermoeconomics 673

� Expander with cooling air supply (T ):

C3 = (C1 + C4 + C5 − C2) + Z T (66)

C2

E2= C1 + C4 + C5

E1 + E4 + E5(67)

� Fuel cell (FC ): The purpose of the fuel cell shownin Table III is to generate electricity and to supply usefulthermal energy.

C7 + (C6 − C5) = (C1 + C2) − (C3 + C4) + Z FC (68)

The F and P equations for a fuel cell as part (component)of a system are

CCH3

ECH3

= CCH1

ECH1

;CCH

4

ECH4

= CCH2

ECH2

(69)

CPH3 − CPH

1

EPH3 − EPH

1

= CPH4 − CPH

2

EPH4 − EPH

2

= C6 − C5

E6 − E5= C7

E7. (70)

For a stand-alone fuel cell (system), we obtain

C3

E3= C1

E1;

C4

E4= C2

E2(71)

and

C6 − C5

E6 − E5= C7

E7. (72)

For a stand-alone fuel cell, the exergies of streams 3 and 4are exergy losses .

� Distillation column (DC): The purpose of a distilla-tion column is to separate the chemical components inthe feed, i.e., to increase the chemical exergy of the sub-stances in the feed at the expense of the physical exergysupplied to the system with the feed and in the reboiler.For this component, it is appropriate to consider the costsof physical and chemical exergy for streams 1, 2, 3, 6, and7 separately.

The cost balance for the distillation column is given by

cC H2 EC H

2 + cC H3 EC H

3 + cC H6 EC H

6 + cC H7 EC H

7

− cC H1 EC H

1 + m6(cP H

6 eP H6 − cP H

1 eP H1

)+ m2

(cP H

2 ePH2 − cP H

1 eP H1

) + (c9 E9 − c8 E8)

= (c4 E4 − c5 E5) + m7(cP H

1 eP H1 − cP H

7 eP H7

)+ m3

(cP H

1 eP H1 − cP H

3 eP H3

) + Z DC . (73)

The F equations are

c5 = c4, (74)

cP H7 = cP H

3 = cP H1 . (75)

Each exergy unit is supplied to the product at the sameaverage cost (P equations):

m6(cP H

6 eP H6 − cP H

1 eP H1

)m6

(eP H

6 − eP H1

) = m2(cP H

2 eP H2 − cP H

1 eP H1

)m2

(eP H

2 − eP H1

)

= m2(cC H

2 eC H2 − cC H

1 eC H1

)m2

(eC H

2 − eC H1

) = m3(cC H

3 eC H3 − cC H

1 eC H1

)m3

(eC H

3 − eC H1

)

= m6(cC H

6 eC H6 − cC H

1 eC H1

)m6

(eC H

6 − eC H1

) = m7(cC H

7 eC H7 − cC H

1 eC H1

)m7

(eC H

7 − eC H1

) .

(76)

The exergy increase of the cooling water (stream 8) is anexergy loss.

c9 = c8 (77)

� Gasifier (GS):

The purpose of a gasifier is to convert the chemical ex-ergy of a solid fuel into the chemical exergy of a gaseousfuel. Due to the chemical reactions involved, the physicalexergy of the material streams increase between the inletand the outlet.

CC H3 + (

C P H3 + C P H

4 − C P H1 − C P H

2

)= CC H

1 + CC H2 − CC H

4 + Z GS (78)

When the chemical and physical exergies are consideredseparately, the following P equation is obtained:

CC H3

EC H3

= C P H3 + C P H

4 − C P H1 − C P H

2

E P H3 + E P H

4 − E P H1 − E P H

2

, (79)

CC H4

EC H4

= CC H1

EC H1

. (80)

� Steam reformer (SR):

C3 − C1 − C2 = C4 + C5 − C6 + Z S R, (81)

C6

E6= C4 + C5

E4 + E5. (82)

� Evaporator including steam drum (EV):

C2 + C5 − C1 = C3 − C4 + Z EV , (83)

c5e5 − c1e1

e5 − e1= c2e2 − c1e1

e2 − e1. (84)

IV. THERMOECONOMIC EVALUATION

In a thermoeconomic evaluation, the cost per unit of exergyand the cost rates associated with each material and exergystream are used to calculate thermoeconomic variables for



674 Thermoeconomics

each system component. From the exergy analysis we al-ready know the exergetic variables: rate of exergy destruc-tion E D,k ; exergetic efficiency εk ; and exergy destructionratio yD,k .

A. Thermoeconomic Variables

The definition of fuel and product for the exergetic effi-ciency calculation in a component leads to the cost flowrates associated with the fuel CF,k and product CP,k of thethe kth component. CF,k represents the cost flow rate atwhich the fuel exergy E F,k is provided to the kth compo-nent. C P,k is the cost flow rate associated with the productexergy E P,k for the same component.

The average cost of fuel for the kth system componentcF,k expresses the average cost at which each exergy unitof fuel (as defined in the exergetic efficiency) is suppliedto the kth system component.

cF,k = C F,k

E F,k. (85)

The value of cF,k depends on the relative position of thekth component in the system and on the interconnectionsbetween the kth component and the remaining compo-nents. As a general rule, the closer the kth component isto the product (fuel) stream of the overall system, usuallythe larger (smaller) the value of cF,k .

Similarly, the unit cost of product cP,k is the averagecost at which each exergy unit was supplied to the productfor the kth component:

cP,k = C P,k

E P,k. (86)

Using Eqs. (85) and (86), the cost balance for the kthsystem component can be written as

cP,k E P,k = cF,k E F,k + Z k − C L ,k . (87)

One of the most important aspects of exergy costing iscalculating the cost of exergy destruction in each compo-nent of the energy system being considered. The cost rateCD,k associated with exergy destruction E D,k in the kthsystem component is a “hidden” cost that can be revealedonly through a thermoeconomic analysis. It can be ap-proximated by the cost of the additional fuel that needs tobe supplied to this component to cover the exergy destruc-tion and to generate the same exergy flow rate of productE P,k :

C D,k = cF,k E D,k, when E P,k = const. (88)

For most well-designed components, as the exergy de-struction decreases or efficiency increases, the cost ofexergy destruction C D,k decreases while the capital in-vestment ZC I

k increases. The higher the C D,k value is,

the lower the Z k value. It is one of the most interestingfeatures of thermoeconomics that the exergy destructioncosts associated with a component are estimated and com-pared with the investment costs of the same component.This comparison facilitates the decision about the designchanges that might improve the cost effectiveness of theoverall system. In a thermoeconomic optimization we tryto find the appropriate trade-offs between C D,k and Z k .The C D,k values cannot be added to calculate the C D valuefor a group of components because each C D,k already con-tains information related to the interactions among the costof exergy destruction in upstream components. The C D

value for the group should be calculated using the averagecost per exergy unit of fuel cF for the group and the exergydestruction rate within the group.

Figure 5 shows the relation between investment cost perunit of product exergy ZC I

k /E P,k and exergy destructionper unit of product exergy E D,k/E P,k for the kth com-ponent. The shaded area illustrates the range of variationof the investment cost due to uncertainty and to multipletechnical design solutions that might be available. The in-vestment cost per unit of product exergy cZ

P,k increaseswith decreasing exergy destruction per unit of product ex-ergy or with increasing exergetic efficiency. This cost be-havior is exhibited by most components. The componentsthat exhibit a decrease of cZ

P,k with increasing efficiencydo not need to be considered in a thermoeconomic evalua-tion since for these components no optimization dilemmaexists: Among all available solutions we would use themost efficient component that has both the lowest specificfuel expenses and the lowest specific investment cost cZ

P,kand, thus, the minimum cP ,k value (Tsatsaronis and Park,1999).

The relative cost difference rk between the average costper exergy unit of product and average cost per exergy unitof fuel is given by

rk ≡ cP,k − cF,k

cF,k= cF,k(E D,k + E L ,k) + (

ZC Ik + Z O M

k

)cF,k E P,k

= 1 − εk

εk+ ZC I

k + Z O Mk

cF,k E P,k. (89)

This equation reveals the real cost sources in the kth com-ponent, which are the capital cost ZC I

k , the exergy destruc-tion within the component as expressed by C D,k , the oper-ating and maintenance costs, and the exergy loss. Amongthese sources the first two are the most significant onesand are used for calculating the exergoeconomic factor.

The exergoeconomic factor expresses the contributionof the capital cost to the sum of capital cost and cost ofexergy destruction:



Thermoeconomics 675

FIGURE 5 Expected relationship between investment cost and exergy destruction (or exergetic efficiency) for thek th component of a thermal system [Based on Fig. 1 in Tsatsaronis, G., and Park, M. H. (1999). “ECOS’99, EfficiencyCosts, Optimization, Simulation and Environmental Aspects of Energy Systems,” pp. 161–121, Tokyo, Japan, June8–10.]

fk ≡ ZC Ik

ZC Ik + C D,k

= ZC Ik

ZC Ik + cF,k E D,k

. (90)

The thermoeconomic variables ZC Ik and C D,k provide ab-

solut measures of the importance of the kth component,whereas the variables rk and fk provide relative measuresof the component cost effectiveness.

B. Design Evaluation

A detailed thermoeconomic evaluation of the design ofa thermal system is based on the following exergetic andthermoeconomic variables, each calculated for the kth sys-tem component:

E D,k rate of exergy destruction Eq. (25)

εk exergetic efficiency Eq. (24)

yD,k exergy destruction ratio Eq. (26)

ZC Ik cost rate associated with Eqs. (35) and (37)

capital investment

C D,k cost rate associated with Eq. (88)exergy destruction

rk relative cost difference Eq. (89)

fk exergoeconomic factor Eq. (90)

The values of all exergetic and thermoeconomic variablesdepend on the component type (e.g., turbine, heat ex-changer, chemical reactor). For instance, the value of theexergoeconomic factor is typically between 25 and 65%for compressors and turbines.

The following guidelines may be applied to the eval-uation of the kth system component to improve the costeffectiveness of the entire system:

1. Rank the components in descending order of costimportance using the sum ZC I

k + C D,k .2. Consider initially design changes for the components

for which the value of this sum is high.3. Pay particular attention to components with a high

relative cost difference rk , especially when the costrates ZC I

k and C D,k are high.4. Use the exergoeconomic factor fk to identify the

major cost source (capital investment or cost ofexergy destruction).

a. If the fk value is high, investigate whether it iscost effective to reduce the capital investment forthe kth component at the expense of thecomponent efficiency.

b. If the fk value is low, try to improve the componentefficiency by increasing the capital investment.

5. Eliminate any subprocesses that increase the exergydestruction or exergy loss without contributing to thereduction of capital investment or of fuel costs forother components.

6. Consider improving the exergetic efficiency of acomponent if it has a relatively low exergeticefficiency or a relatively large value for the rate ofexergy destruction, or the exergy destructionratio.



676 Thermoeconomics

V. ITERATIVE OPTIMIZATIONOF A THERMAL SYSTEM

Design optimization of a thermal system means the mod-ification of the structure and the design parameters of asystem to minimize the total levelized cost of the systemproducts under boundary conditions associated with avail-able materials, financial resources, protection of the envi-ronment, and government regulation, together with safety,reliability, operability, availability, and maintainability ofthe system (Bejan, Tsatsaronis, and Moran, 1996). A trulyoptimized system is one for which the magnitude of everysignificant thermodynamic inefficiency (exergy destruc-tion and exergy loss) is justified by considerations relatedto costs or is imposed by at least one of the above boundaryconditions.

The iterative thermoeconomic optimization techniquediscussed here is illustrated with the aid of a simplifiedcogeneration system as shown in Fig. 6. The purposeof the cogeneration system is to generate 30 MW netelectric power and to provide 14 kg/s saturated steam at20 bar (stream 7). A similar cogeneration system is usedin Penner and Tsatsaronis (1994) for comparing differ-ent thermoeconomic optimization techniques. It shouldbe emphasized that the cost minimization of a real systemaccording to Fig. 6 is significantly simpler compared withthe case considered here, since in a real system engineershave to select one of the few gas turbine systems availablein the market for a given capacity and then have to optimizethe design of only the heat-recovery steam generator. Inthis example, however, to demonstrate application of theiterative thermoeconomic optimization technique with theaid of a simple system containing different components,we assume that we can freely decide about the design ofeach component included in the gas turbine system.

The decision variables selected for the optimization arethe compressor pressure ratio p2 /p1, the isentropic com-

FIGURE 6 A simple cogeneration system.

pressor efficiency ηsc, the isentropic turbine efficiency ηst ,and the temperature T3 of the combustion products enter-ing the gas turbine. All other thermodynamic variables canbe calculated as a function of the decision variables. Allcosts associated with owning and operating this system arecharged to the product streams 7 and 10. This includes thecost rate associated with the exergy loss from the overallsystem (stream 5).

The total cost rate associated with the product for theoverall system C P ,tot is given by (Bejan, Tsatsaronis, andMoran, 1996)

C P,tot = (C7 − C6) + C10 + C5 + Z other

= C1 + C8 +∑

k

Z k + Z other!= min (91)

k = AC, CC, GT, H RSG

Here, Z other denotes the levelized cost associated with thecapital investment for other system components not shownin Fig. 6. The subscripts AC, CC, GT, and HRSG refer tothe components air compressor, combustion chamber, gasturbine, and heat-recovery steam generator, respectively.Equation (91) represents the objective function for the costminimization.

The iterative thermoeconomic optimization of the co-generation system is subject to the following constraints:6.0 ≤ p2/p1 ≤ 14.0; 0.75 ≤ ηsc ≤ 0.89; 0.8 ≤ ηst ≤ 0.91;and 900 K ≤ T3 ≤ 1600 K. The minimum temperature dif-ference �Tmin in the heat-recovery steam generator mustbe at least 15 K. A first workable design (base-case design)was developed using the following values of the decisionvariables:

p2/p1 = 14, T3 = 1260 K, ηsc = 0.86, ηst = 0.87.

Relevant thermodynamic and economic data as well as thevalues of the thermoeconomic variables for the base-casedesign are shown in Tables IV and V. The thermoeco-nomic variables are used to determine the changes in thedesign of each component to be implemented in the nextstep of the iterative optimization procedure. Knowledgeabout how the decision variables qualitatively affect theexergetic efficiency and the costs associated with each sys-tem component is required to determine the new values ofthe decision variables that improve the cost effectivenessof the overall system. Engineering judgment and criticalevaluations must be used when deciding on the changesto be made from one iteration step to the next.

A. Evaluation of the First Design Case(Base-Case Design)

Table V summarizes the thermoeconomic variables andthe purchased equipment costs P EC calculated for each



Thermoeconomics 677

TABLE IV Mass Flow Rate, Temperature, Pressure, Exergy Flow Rates, Cost PerExergy Unit, and Cost Flow Rate for Each Stream in the Base-Case Design of theCogeneration System

m j T p EP Hj EC H

j E j c j C j

Nr. (kg/s) (K) (bar) (MW) (MW) (MW) ($/GJ) ($/h)

1 132.960 298.15 1.013 0.000 0.000 0.000 0.00 0

2 132.960 674.12 14.182 49.042 0.000 49.042 32.94 5815

3 134.943 1260.00 13.473 115.147 0.351 115.498 18.11 7532

4 134.943 754.35 1.066 27.518 0.351 27.869 18.11 1817

5 134.943 503.62 1.013 7.305 0.351 7.656 18.11 499

6 14.000 298.15 20.000 0.027 0.035 0.062 0.00 0

7 14.000 485.52 20.000 12.775 0.035 12.810 34.25 1579

8 1.983 298.15 20.000 0.904 101.896 102.800 4.56 16899 52.479 21.47 4057

10 30.000 21.47 2319

component of the cogeneration system. According tothe methodology described in Section IV.B, the compo-nents are listed in order of descending value of the sumZC I + C D . The compressor and the turbine have the high-est values of the sum ZC I +C D and are the most importantcomponents from the thermoeconomic viewpoint.

The value of the exergoeconomic factor f for the aircompressor (AC) is high compared with a target value ofbelow 65%. The f factor indicates that most of the costsassociated with the air compressor are due to capital in-vestment. Although the cost per exergy unit of fuel cF

supplied to the air compressor is the highest among allsystem components, the cost rate C D associated withexergy destruction E D in this component is relativelylow. Only 6.46% of the total exergy destruction occursin the air compressor. A decrease of the capital invest-ment costs for the air compressor might be cost effectivefor the entire system even if this would decrease the ex-ergetic efficiency εAC. These observations suggest a sig-nificant decrease in the values of p2/p1 and ηsc. Com-pared to the isentropic efficiency, the influence of the pres-sure ratio on the compressor exergetic efficiency is low.Here, a reduction of p2/p1 aims at a decrease of Z AC

only.

TABLE V First Design Case (Base-Case Design)a

P EC ε EF EP ED yD cF cP C D ZC I r f ZC I ++ C D

Name (103$) (%) (MW) (MW) (MW) (%) ($/GJ) ($/GJ) ($/h) ($/h) (%) (%) ($/h)

AC 8732 93.56 52.48 49.10 3.38 3.29 21.47 32.90 261 1112 53.2 81.0 1374

GT 3282 94.12 87.63 82.48 5.15 5.01 18.11 21.47 335 418 18.5 55.4 754

HRSG 1297 63.07 20.21 12.75 7.46 7.26 18.11 34.41 486 165 90.0 25.3 652

CC 139 64.65 102.80 66.46 36.34 35.35 4.56 7.18 597 17 57.3 2.9 615

a T3 = 1260 K, p2/p1 = 14, ηsc = 0.86, ηst = 0.87; overall system: εtot = 41.58%; C P,tot = $4587/h; C L ,tot = C5 = $499/h; HRSG: �Tmin = 88 K.

The value of the exergoeconomic factor f for thegas turbine expander (GT) is within the range of the targetvalues (between 25 and 65%, Section IV.B). Therefore,no recommendations can be derived from the thermoeco-nomic evaluation of the gas turbine expander.

The heat-recovery steam generator (HRSG) has thehighest value of the relative cost difference r . The ex-ergoeconomic factor f indicates that most of the costs forthis component are related to the cost rate associated withexergy destruction C D [Eq. (89)]. The difference in theaverage thermodynamic temperature between the hot andthe cold streams is a measure for the exergy destructionin the heat-recovery steam generator [Eq. (20)]. A reduc-tion in the value of the temperature T3 and/or an increasein the isentropic turbine efficiency ηst or the pressure ra-tio p2/p1 lead to a decrease in the temperature T4. Sincethe temperature profile of the cold stream and the heatingduty in the heat-recovery steam generator are fixed, boththe average temperature difference and the temperature T5

of the exhaust gas would decrease. These changes shouldresult in an increase in both the exergetic efficiency andthe capital investment in the heat-recovery steam gener-ator, as well as in a decrease in the exergy loss from theoverall system.



678 Thermoeconomics

TABLE VI Sample of a Qualitative Decision Matrix for the First Design Case (Base-CaseDesign)a

Decision variablesSystem Objective

component ZC Ik or εk T3 p2/p1 ηsc ηst ZC I

k ++ C D,k

1260 14 0.86 0.87 Initial values

AC ↓ — ↓ ↓ — 1374

GT — — — — — 753

HRSG ↑ ↓ ↑ — ↑ 652

CC ↑ ↑ ↑ ↓ — 615

— ↓ ↓ ↑ Suggestions

1260 12 0.84 0.88 New values

a The symbols ↓, ↑, and —mean a decrease, an increase, or no change, respectively, in the value of adecision variable.

The combustion chamber (CC) has the largest ex-ergy destruction rate. However, according to the sumZC I + C D , the economic importance of this componentis rather small. Each exergy unit of the fuel is supplied tothe combustion chamber at a relatively low cost. Hence,each unit of exergy destruction can be covered at the samelow cost. The low value of the exergoeconomic factor fshows that C D,CC is the dominant cost rate. However,only a part of the exergy destruction can be avoided bypreheating the reactants or by reducing the excess air. Ahigher value of p2/p1 and/or a lower value of ηsc lead toan increase in the temperature T2 of the air entering thecombustion chamber, whereas a lower temperature of thecombustion products T3 reduces the amount of excess air.

Table VI shows a sample of a qualitative decisionmatrix for the base-case design which summarizes thesuggestions from the thermoeconomic evaluation of eachcomponent. Decreasing the values of the pressure ratiop2/p1 and the isentropic compressor efficiency ηsc aswell as increasing the isentropic turbine efficiency ηst areexpected to improve the cost effectiveness of the cogen-eration system. Note, that the decrease in the p2/p1 valuecontradicts the corresponding suggestions from the heat-recovery steam generator and the combustion chamber.However, changes suggested by the evaluation of a com-ponent should only be considered if they do not contradict

TABLE VII Second Design Case (1. Iteration)a

P EC ε EF EP ED yD cF cP C D ZC I r f ZC I++ C D


AC 4633 92.37 48.70 44.98 3.72 3.55 18.00 25.31 241 601 40.63 71.4 841

GT 3802 94.77 83.04 78.70 4.34 4.14 14.47 18.00 226 493 24.40 68.5 719

CC 138 64.26 104.85 67.37 37.48 35.74 4.56 7.22 616 18 58.15 2.8 634

HRSG 1245 61.62 20.69 12.75 7.94 7.57 14.47 29.00 414 161 100.45 28.1 575

a T3 = 1260 K, p2/p1 = 12, ηsc = 0.84, ηst = 0.88; overall system: εtot = 40.77%; C P,tot = $3913/h; C L ,tot = C5 = $446/h; HRSG: �Tmin = 105 K.

changes suggested by components with a significantlyhigher value of the sum ZC I + C D . The temperature T3

remains unchanged, since contradictory indications areobtained from the evaluations of the heat-recovery steamgenerator and the combustion chamber. The values ofthe sum ZC I + C D for both components are in the samerange. After summarizing the results from the evaluationof the base-case design, the following new values areselected for the decision variables in the second designcase: T3 = 12604 K (unchanged), p2/p1 = 12, ηsc =0.84, and ηst = 0.88.

B. Evaluation of the Second DesignCase (1. Iteration)

Through the changes in the decision variables, the valueof the objective function C P,tot is reduced from $4587/hto $3913/h, and the cost rate associated with the exergyloss C5 decreased from $499/h to $446/h. The new val-ues of the thermoeconomic variables are summarized inTable VII. The sum ZC I + C D shows that the air com-pressor and the gas turbine expander are still the mostimportant components from the thermoeconomic view-point. The importance of both components is due to therelatively high investment cost rate ZC I and, to a lesserextent, to the high fuel cost cF for these components. The



Thermoeconomics 679

cost effectiveness of these components may be improvedby lowering the investment cost at the expense of the ex-ergetic efficiency. This can be achieved by a decrease inthe values of the isentropic efficiencies ηsc and ηst . Forthe third design, ηsc and ηst are reduced to 0.82 and 0.87,respectively.

Compared to the base-case design, the combustionchamber has now a higher relative cost importance thanthe heat-recovery steam generator. To reduce the cost ofexergy destruction in the combustion chamber, the valueof T3 is increased to 1320 K.

Although the value of the relative cost difference r in-creased for the heat-recovery steam generator from thebase-case design to the second design, the sum ZC I + C D

for the heat-recovery steam generator decreased. The mod-ifications in the upstream components (AC and CC) andthe interactions among the system components lead to adecrease in the cost per exergy unit of the fuel cF,H RSG .Therefore, the cost associated with exergy destruction isreduced even if the exergy destruction rate for the heat-recovery steam generator increased in the second design.The reduced pressure ratio p2/p1 outweighed the effect ofthe increased isentropic turbine efficiency ηst on the tem-perature T4. Hence, T4 increased in the second design in-stead of an anticipated decrease. These observations showthat the suggested changes based on the thermoeconomicevaluation of the other components may affect the costeffectiveness of the heat-recovery steam generator nega-tively. The pressure ratio p2/p1 remains unchanged in thethird design, since the suggested changes in ηst and T3

already lead to a higher temperature T4, which results in alarger exergy destruction rate in the heat-recovery steamgenerator.

C. Third Design Case (2. Iteration)

As a result of the changes in the decision variables,the value of the objective function C P,tot decreased to$3484/h. Table VIII shows the new values of the ther-moeconomic variables for each component. The costs as-sociated with the compressor and the turbine are reducedsignificantly, whereas the sum ZC I + C D is almost un-

TABLE VIII Third Design Case (2. Iteration)a

P EC ε EF EP ED yD cF cP C D ZC I r f ZC I++ C D


AC 3251 91.47 46.67 42.69 3.98 3.70 15.18 20.95 218 430 38.1 66.4 648

CC 135 65.30 107.68 70.32 37.36 34.70 4.56 7.10 614 18 55.5 2.8 632

GT 2852 94.59 81.06 76.67 4.38 4.07 12.34 15.18 195 377 23.0 66.0 572

HRSG 1134 58.78 21.69 12.75 8.94 8.30 12.34 26.09 397 150 111.4 27.4 547

a T3 = 1320 K, p2/p1 = 12, ηsc = 0.82, ηst = 0.87; overall system: εtot = 39.70%; C P,tot = $3484/h; C L ,tot = C5 = $453/h; HRSG: �Tmin =142 K.

changed for the combustion chamber which now has thesecond highest value. The values of the exergoeconomicfactor f AC and fGT are close to their target values. Furtherimprovements in the cost effectiveness may be achievedby slightly decreasing the isentropic efficiencies and sig-nificantly increasing the temperature T3, even if the exergydestruction rate in the heat-recovery steam generator andthe cost rate associated with the exergy loss of the overallsystem C5 increase.

VI. RECENT DEVELOPMENTSIN THERMOECONOMICS

Complex thermal systems cannot usually be optimizedusing mathematical optimization techniques. The reasonsinclude system complexity; opportunities for structuralchanges not identified during model development; in-complete cost models; and inability to consider in themodel additional important factors such as plant avail-ability, maintenability, and operability. Even if mathemat-ical techniques are applied, the process designer gains noinsight into the real thermodynamic losses, the cost for-mation process within the thermal system, or on how thesolution was obtained.

As an alternative, thermoeconomic techniques provideeffective assistance in identifying, evaluating, and reduc-ing the thermodynamic inefficiencies and the costs in athermal system. They improve the engineer’s understand-ing of the interactions among the system components andvariables and generally reveal opportunities for design im-provements that might not be detected by other methods.Therefore, the interest in applying thermoeconomics hassignificantly increased in the last few years. In the follow-ing, some recent developments in thermoeconomics arebriefly mentioned.

To evaluate the thermodynamic performance and costeffectiveness of thermal systems and to estimate thepotential for improvements it is always useful to knowfor the most important system components the avoid-able part of exergy destruction, the cost associated withthis avoidable part, and the avoidable investment cost



680 Thermoeconomics

associated with each system component. Improvement ef-forts should then focus only on these avoidable parts ofinefficiencies and costs. Figure 5 illustrates the assessmentof specific unavoidable exergy destruction EU N

D ,k / E P ,k andspecific unavoidable investment cost ZU N

k / E P ,k for thekth component. The avoidable parts are obtained as thedifference between total value and avoidable part. Moreinformation about this topic is provided by Tsatsaronisand Park (1999).

The design and improvement of a thermal system ofteninvolve application of heuristic rules. Due to the com-plexity of energy conversion systems, as well as to theuncertainty involved in some design decisions, computerprograms using principles from the field of artificial intel-ligence and soft computing are useful tools for the processdesigner in improving a given design and in developing anew cost-effective thermal system. The benefits of com-bining knowledge-based and fuzzy approaches with aniterative thermoeconomic optimization technique are dis-cussed together with some applications by Cziesla (2000).

Calculating the cost of each product stream generatedby a thermal plant having more than one product is an im-portant subject for which several approaches have been de-veloped in the past. Some of these approaches use exergy-based or thermoeconomic methods; however, the resultsobtained by different methods may vary within a widerange. Recently, a new exergy-based approach was de-veloped (Erlach, Tsatsaronis, and Cziesla, 2001) for (a)assigning the fuel(s) used in the overall plant to the prod-uct streams and (b) calculating the costs associated witheach product stream with the aid of a thermoeconomicevaluation. This new approach is general, more objectivethan previous approaches, and flexible, i.e., it allows de-signers and operators of the plant to actively participate inthe fuel and cost allocation process.


ENERGY EFFICIENCY COMPARISONS AMONG COUNTRIES

• ENERGY FLOWS IN ECOLOGY AND IN THE ECONOMY

• ENERGY RESOURCES AND RESERVES • GEOTHERMAL

POWER STATIONS • SOLAR THERMAL POWER STATIONS

• THERMODYNAMICS

BIBLIOGRAPHY

Ahrendts, J. (1980). “Reference states,” Energy—Int. J. 5, 667–677.Bejan, A., Tsatsaronis, G., and Moran, M. (1996). “Thermal Design and

Optimization,” Wiley, New York.Cziesla, F. (2000). “Produktkostenminimierung beim Entwurf kom-

plexer Energieumwandlungsanlagen mit Hilfe von wissensbasiertenMethoden,” No. 438, Fortschr.-Ber. VDI, Reihe 6, VDI Verlag,Dusseldorf.

Erlach, B., Tsatsaronis, G., and Cziesla, F., A new approach for as-signing costs and fuels to cogeneration products. In “ECOS’o1,Efficiency, Costs, Optimization, Simulation and EnvironmentalAspects of Energy Systems,” pp. 759–770, Istanbul, Turkey,July 4–6.

Kotas, T. J. (1985). “The Exergy Method of Thermal Plant Analysis,”Butterworths, London.

Lazzaretto, A., and Tsatsaronis, G. (November 1999). On the calcu-lation of efficiencies and costs in thermal systems. In “Proceedingsof the ASME Advanced Energy Systems Division” (S. M. Aceves,S. Garimella, and R. Peterson, eds.), AES-Vol. 39, pp. 421–430,ASME, New York.

Moran, M. J., and Shapiro, H. N. (1998). “Fundamentals of EngineeringThermodynamics,” Wiley, New York.

Penner, S. S., and Tsatsaronis, G., eds. (1994). “Invited papers on exer-goeconomics,” Energy—Int. J. 19(3), 279–318.

Sama, D. A. (September 1995). “The use of the second law of ther-modynamics in process design,” J. Eng. Gas Turbines Power 117,179–185.

Szargut, J., Morris, D. R., and Steward, F. R. (1988). “Exergy Analysis ofThermal, Chemical, and Metallurgical Processes,” Hemisphere, NewYork.

Electric Power Research Institute (1993). “Technical assessment guide(TAGTM),” Vol. 1, TR-102276-V1R7, Electric Power ResearchInstitute, Palo Alto, CA.

Tsatsaronis, G. (1999). Strengths and limitations of exergy analy-sis. In “Thermodynamic Optimization of Complex Energy Systems”(A. Bejan, and E. Mamut, eds.), Vol. 69, pp. 93–100, Nato ScienceSeries, Kluwer Academic, Dordrecht/Normell, MA.

Tsatsaronis, G., and Park, M.H. (1999). On avoidable and unavoid-able exergy destructions and investment costs in thermal systems.In “ECOS’99, Efficiency, Costs, Optimization, Simulation and Envi-ronmental Aspects of Energy Systems,” pp. 116–121, Tokyo, Japan,June 8–10.

P1: GNH Final Pages Qu: 00, 00, 00, 00

Encyclopedia of Physical Science and Technology En017B-817 August 2, 2001 18:38

Waste-to-Energy (WTE) SystemsS. S. PennerUniversity of California, San Diego

I. Historical OverviewII. Introductory Remarks on WTE PlantsIII. Initial Growth Rates for MWI Use in the United

StatesIV. Integrated Waste ManagementV. The 1999 Integrated Waste Services

Association WTE Directory of U.S. PlantsVI. Life Cycle Comparison of WTE

with Selected Fossil-Fuel Burners

for Electricity ProductionVII. Current and Future Applications of Waste

Incineration WorldwideVIII. Potential Electricity Production from WTE

Systems in the United StatesIX. Representative MWI SystemsX. Electricity and Tipping-Fee RevenuesXI. Environmental Issues

GLOSSARY

Combustor (or Burner or Incinerator) Type1

AFBC Atmospheric fluidized-bed combustor.AFBCRW Atmospheric fluidized-bed combustor with

refractory wall.CFBC Circulating fluidized-bed combustor.CFBCWW Circulating fluidized-bed combustor with a

water wall.MB Mass burner (a system that combusts all of the in-

coming material).MBRW Mass burner with refractory wall.MCU Modular combustion unit.MWC or MWI Municipal-waste combustor or munic-

ipal-waste incinerator.RBCCRW Rotary bed combustion chamber with refrac-

tory wall.

1The letter for combustor (C) may be replaced by B for burner or Ifor incinerator.

RDF Refuse-derived fuel.RDFC Refuse-derived-fuel combustor.RDFP Refuse-derived-fuel processor (other than a

combustor).RWWC Rotary water-wall combustor.SSCWW Spreader and stoker combustor with a water

wall.WTE Waste to energy.

APC (Air-Pollution Control) System

CI (Activated) carbon injection.CyS Cyclone separator.DSI Duct sorbent (dry) injection (downstream of the

furnace).ESP Electrostatic precipitator.FF Fabric filter.FSI (Dry) furnace sorbent injection.SDA Spray dryer absorber.SDS Spray-dryer scrubber (used interchangeably with

SDA).

631

P1: GNH Final Pages


632 Waste-to-Energy (WTE) Systems

SNCR Selective noncatalytic reduction (used for NOx

control).WS Wet scrubber.

Energy Output and Material Recovery

COG Cogenerators (produce both electricity and steam).MWe Megawatts of electrical power (gross).STM/hr Steam production per hour.TPY of FeX Tons per year of ferrous metals recovered.TPY of O Tons per year of materials other than ferrous

metals recovered.

I. HISTORICAL OVERVIEW

About 25 years ago, augmented construction of WTE sys-tems enjoyed wide support in the United States as a pre-ferred solution to two major problems by (i) providing forthe disposal of rapidly increasing sources of municipalwastes while (ii) supplying low-cost, mostly renewablefuels for electric-power and steam generation. The impor-tant contribution to waste management is the result of a 90to 95% reduction in the remaining waste volume, whichis accompanied by the production of stable, nonleachable,nonhazardous ash that is usable in light aggregates forstructural fill, road base, and subbase and as an additive toconcrete or asphalt.

With the recognition of possible health effects thatmay be caused by emissions of even minute amounts ofmetals such as mercury or cadmium and of toxic com-pounds of the type belonging to the 210 congeners ofchlorinated dioxins and furans, opposition to waste in-cineration became widespread, especially in the UnitedStates and Germany. Because of the availability of largeareas of open and unused land in the United States, ex-cept in the vicinity of many of our urban populationcenters, there was renewed development of now “san-itary” landfills (i.e., landfills with impoundments, sealsagainst leakage to the local water table, etc.). The pre-ferred environmental agenda may be summarized by thegoal of the “3 Rs” standing for “Reduce, Reuse, Recy-cle.” Complete success of this agenda would obviouslyeliminate wastes and remove the necessity for waste man-agement altogether. Although considerable reductions inwaste generation were achieved for a large number ofmanufactured products, these reductions were accompa-nied by the creation of new sources of waste associatedwith the development of new technologies. Unfortunately,the net result has not been a significant reduction in percapita waste generation. EPA estimates have remainedat about 1500 lbs of waste per person per year in theUnited States.

It should be noted that WTE has one important and en-vironmentally attractive feature. Since a large percentageof combustible municipal waste is derived from biomass,WTE represents in effect one of a very small numberof economically viable technologies for plant-derived-product utilization in combustors. These have the dis-tinguishing feature of allowing energy generation with-out producing a net addition of carbon dioxide to theatmosphere.

New WTE plants are now in the planning stage or un-der construction in Southeast Asia, Eastern Europe, LatinAmerica, and elsewhere. In the developed countries, es-pecially the United States, Japan, and Western Europe,stringent emission controls are mandated by regulatoryagencies. The complex assessments of possible health ef-fects fall beyond the scope of our deliberations and shouldproperly be provided by toxicologists and epidemiolo-gists. Here, we content ourselves with descriptive materialon system design, performance, and EPA-imposed emis-sion regulations, as well as brief remarks on the economicaspects of technology utilization. WTE systems have prob-ably become the most severely regulated of all combustiontechnologies with respect to allowed emissions of poten-tially hazardous compounds and elements.

The primary competition for waste incineration is rep-resented worldwide by waste disposal in landfills. San-itary landfills generally have disposal costs that are notmuch less than those for incineration, when allowance ismade for energy recovery. Old-fashioned landfills can beconstructed at much lower cost than either modern in-cinerators or sanitary landfills. For this reason, they haveremained the preferred option in countries with large openspaces and low wast-disposal budgets. The use of WTEsmade strong gains in the United States while methodsfor land disposal were improved and sanitary landfills be-came the norm. At this point, the use of WTEs became aless compelling option. In the EPA hierarchy of preferredmethods for municipal waste disposal, the use of land-fills and of municipal waste incinerators occupy the bot-tom two positions because of presumed or demonstratedundesirable health effects. Of these least desirable twooptions, which is the worst? The author of at least onestudy (Jones, 1997) awards the bottom prize to landfillsafter considering a wide range of estimable environmentalimpacts for the durations of the service lives of the twosystems.

II. INTRODUCTORY REMARKSON WTE PLANTS

WTE plants are constructed and sold by a number ofvendors and have some common design features (Penner

P1: GNH Final Pages


Waste-to-Energy (WTE) Systems 633

et al., 1987). The operational sequence involves trashdelivery and storage in a large pit, often followed byshredding, removal of magnetic materials, and dryingbefore loading onto a stoker-grate bed. An average heatcontent for the incoming dry waste in the United Statesis around 11,600 kJ/kg (5000 Btu/lb) with considerablevariability depending on the location (e.g., urban or ruralsources) and time of year. The worldwide variability ofheat content is even greater because waste generation re-flects ethnic materials-use patterns, as well as preincinera-tion separations such as removals of plant wastes, papers,plastics, glass, metals (aluminum and iron), etc., whichmay contribute significantly to heat-content reductions orincreases.

Although different MWIs have different designs, thereare some common performance features. Grate systemstransport trash through the combustion zones with sequen-tial drying, devolatilization, and burning. Primary (under-fire) air is forced upward through the grate. The combinedaction of forced air and grate movement facilitates mixingand combustion. Grate residence times vary but fall gener-ally in the range of 20 to 40 min. The ash-handling systemcontrols ash movement. Hot gases rise from the grate intothe furnace where secondary air injection causes completeturbulent burnout of combustibles. Overall air-to-fuel ra-tios are rich in oxidizer and are generally about 2 timesstoichiometric.

MWIs have water-walled furnaces. The boiler feedwa-ter is preheated by radiative and convective heat trans-fer to the walls before superheating by flue gases down-stream. If desired, an underfire preheater may be used orthe furnace feedwater may be preheated in an economizerdownstream of the superheater. Removal of particulatesfrom the gaseous products is now generally accomplishedin baghouses. These have largely replaced electrostaticprecipitators, which have been found to be less effectivethan baghouses for the removal of trace metals, diox-ins, and furans. Furthermore, wet and dry scrubbers forflue gas are employed in specially designed systems tolower the effluent-gas concentrations of toxic materialssuch as mercury, cadmium, and congeners of dioxins andfurans.

A summary of operational WTE plants for 1996 in theUnited States is given in Table I.

III. INITIAL GROWTH RATES FOR MWIUSE IN THE UNITED STATES

Before the normal growth rate of MWIs was seriously in-terrupted in the United States around 1988–89 becauseof environmental concerns involving especially dioxinand furan emissions, a logistics curve was used to esti-

TABLE I Operational WTE Plants in the United States in May1996a

Number of Design AnnualTechnology operating capacity capacity

type plants (mt/day) (106 mt)

Mass-burn plants 70 72,800 22.6

Modular plants 21 2,530 0.8

Incinerators 21 2,880 0.9

RDF production and 15 22,600 7.0combustion

Processing only 11 4,530 1.4

Combustion only 8 3,000 0.9

Totals 146 108,000 33.6

MWCs without RDFs 135 104,000 32.2

WTE plants 114 101,000 31.3

[From Taylor, A., and Zannes, M. (1996). Integrated Waste ServicesAssociation, Washington, D.C.]

a See Glossary for brief definitions of technology types.

mate the rate of market penetration as well as the ultimatevalue for market saturation (Penner and Richards, 1989).The latter was based on an assumed ultimate per capitamunicipal waste generation rate of 1200 lbs/person (p)-yrin view of stringent conservation efforts at the input end.With the optimistic value of 30% of total waste recy-cled, ultimate disposal would involve 840 lbs per per-son per year dedicated for incineration assuming thatlandfilling would become unacceptable. With this scena-rio and a stable U.S. population of 300 × 106, technologysaturation would correspond to 252 × 109 lbs/yr of wasteincineration. At $105/(t/d) as the required capital invest-ment, the total potential capital investment in incine-ration facilities becomes [$105/(t/d)] × (252 × 109 lbs/yr)× (1 t/2000 lbs) × (1 yr/365d) � $36 × 109. Because newconstruction of MWIs in the United States was essentiallyterminated in 1993, only about one-fifth of this investmenthad actually been made by the end of 1998.

Technology growth rates according to the logisticscurve may be estimated from the Verhulst equation,

F(t)/[1 − F(t)] = {F(t1)/G − (t1)} exp k(t − t1), (1)

where F(t) is the ratio of technology utilization at time tto technology utilization at market saturation (t → ∞). Ifwe assume that F(t1) = 0.01 for t1 = 1978 and determineF(t2 = 1988) from the known 1988 U.S. incineration ca-pacity of 5.09 × 104 t/d, then k = 0.297 yr−1 and

F(t)/[1 − F(t)] = 0.196 exp(0.297�t), �t = t − 1988(2)

Although Eq. (2) was followed reasonably well dur-ing the early and middle 1980s, investments in MWIs

P1: GNH Final Pages



TABLE II Management of Municipal Solid Wastesa in Selected Countries (1996)

Percentage Waste-to-energy Waste to Percentage ofCountry recycled (%) landfill (%) WTE Ash Usedb

Canada 21 5 74 Very small

Denmark 23 48 29 60–90 (bottom ash)

France 13 42 42 45 (bottom ash)

Germany 18 36 46 60 (bottom ash)

Japan 5 72 23 10 (bottom ash)

Netherlands 20 35 45 40 (fly ash), 90 + bottom ash

Sweden 19 47 34 Under consideration

United States 23 15 62 6.4 (fly + bottom ash)

a Dogett et al. (1980).b Bottom ash is discharged from the furnace and used for pavement construction, fly ash is

discharged from the pollution-control equipment and may be used in asphaltic materials. Unusedash is landfilled.

decreased greatly around the end of the decade andfurther increases in F(t) ceased altogether around1993.

IV. INTEGRATED WASTE MANAGEMENT

Integrated Waste Management (IWM) refers to the co-herent composite approach used in waste managementwith first preference generally accorded in decreasing or-der to the 3 Rs: reduce, reuse, recycle. Reducing wastehas proved to be an elusive goal because populationand per capita income growths have tended to counter-act more efficient packaging and manufacturing. In 1988,per capita solid-waste production was estimated at 1000to 1500 lbs/yr. An EPA estimate for the year 2000 is215 × 106 mt/yr or, for a population of 300 × 106 peo-ple, per capita solid waste production of 1600 lbs/yr perperson, which is uncomfortably close to the high estimatemade 12 years earlier (Dogett et al., 1980). The EPA ex-pects that one-third of this waste will be recycled, which isa high estimate and about 50% above the levels achieved(Dogett et al., 1980) during 1996 in any developed country(see Table II).

It is evident from Table II that the ratio of waste-to-It is evident from Table II that the ratio of waste-to-energy (WTE) to waste-to-landfill (WTL) varies widelyfor the listed sample of developed countries from a lowof 0.07 in Canada to 0.24 in the United States to 0.78 inGermany and 3.1 in Japan. Large ratios of unpopulatedto populated land areas are important factors in allowingwidespread use of landfills. As probably the most severelyregulated industry in terms of pollutant emissions with re-gard to health and environmental impacts, MWIs haveserved to teach us a great deal about monitoring, con-trolling, and coping with minuscule emissions of toxicmaterials.

V. THE 1999 INTEGRATED WASTESERVICES ASSOCIATION WTEDIRECTORY OF U.S. PLANTS

There were 111 operating plants in the United States in1999, of which 103 (64 operated by IWSA members2)were WTE plants often consisting of a number of furnaceunits with capacities ranging from about 1120 for modules(individual combustors) to 73,920 tpd (tons per day) forcomplete mass burn units and an aggregate annual capac-ity of 32.7 × 106 t of which 30.7 × 106 t (26 × 106 t byIWSA members) were used for energy production (Kiserand Zannes, 1999). This waste processed in WTE facilitiesrepresents about 15% (∼13% by IWSA members) of thetotal 1999 U.S. waste disposal, serves 39 (33 for IWSAmembers) million people, is used in 31 (23 for IWSAmembers) states, involves the recovery of 773,000 tpyof ferrous metals, on-site recycling of paper, ash, non-ferrous metals, etc., for a total of 462,000 tpy, generates2770 MWe of which 2,570 (2240 for ISWA members) areMWe and 200 MWe equivalent are generated as steam.

Air-pollution control devices are widely but not uni-versally used as follows: Of 69 mass-burn units, 51 haveSDAs (spray dryer absorbers or scrubbers), 1 has a WS(wet scrubber), 0 have a C (cyclone), 9 have LIs (limeinjection), 49 FFs (fabric filters), 22 ESPs (electrostaticprecipitators), 25 SNCRs (selective noncatalytic reductionfor NOx control), and 28 CIs (activated carbon injections);of 13 modular systems, 2 have SDAs, 3 WSs, 1 has aC, 2 have LIs, 3 FFs, 7 ESPs, none has an SNCR, and2 have CIs; of 13 RDFs (refuse-derived-fuel) facilities

2The IWSA included in 1999 the following companies: AmericanRef-Fuel Co., Constellation Power Inc., Energy Answers Corp., FosterWheeler Power Systems Inc., Katy-Seghes Inc., Montenay Power Corp.,Ogden Energy Group Inc., Westinghouse Electric Corp., and Wheelabra-tor Technologies Inc.

P1: GNH Final Pages



processing and burning the fuel, 9 have SDAs, 2 WSs,1 has a C, 1 an LI, 8 have FFs, 4 ESPs, 3 SNCRs, and2 CIs; of 8 RDF-Cs (fuel-burning only), 2 have SDAs,none has a WS, 1 has a C, 2 have LIs, 5 FFs, 2 ESPs,none has an SNCR or CI. According to EPA standardsfor the end of December 2000, units processing morethan 250 tpd must meet R-MACT (revised maximumachievable control technology) standards for acid-gas con-trol [by using SDA and DSI (dry duct sorbent injectiondownstream of the furnace), FSI (dry furnace sorbentinjection), CYC (cyclone separation), or any combina-tion of these], NOx control (e.g., with SNCR), controlsfor Hg, total tetra-to-octachloro dioxins and furans, andother specified pollutants (e.g., by using CI). Meeting re-vised MACT limits will require pollution-control retrofitsat many plants, although the R-MACT limits are alreadyexceeded at some of the large modern plants as is illus-trated by the following data: for tetra-to-octachloro diox-ins and furans, optimally achieved performance is 1 to12 nanogram (ng)/dscm (dry standard cubic meter) vsremaining R-MACT limits of 30 ng/dscm for systemsusing SD/FF and 60 ng/dscm for SD/ESP systems; forHg, the R-MACT limit is 0.080 mg/dscm vs 0.008 to0.035 mg/dscm achieved; for Pb, the R-MACT limit is0.44 mg/dscm vs 0.001 to 0.027 mg/dscm achieved; forCd, the R-MACT limit is 0.040 mg/dscm vs 0.001 to0.004 mg/dscm achieved; for particulate matter (PM), theR-MACT limit remains 27 mg/dscm3 vs 1 to 14 mg/dscm3

achieved; for HCl, the R-MACT limit is 31 ppmv (parts permillion by volume) vs 2 to 27 ppmv achieved; for SO2, theR-MACT limit is 29 ppmv vs 1 to 21 achieved; for NOx, thenew R-MACT limit is 205 ppmv for MB (mass burn)/WW(water wall), remains at 250 ppmv for MB/RWW (rotarywater-wall combustor), remains at 250 ppmv for RDF,is reduced to 180 ppmv for FBC (fluidized-bed com-

fractory walls) vs 36 to 166 ppmv achieved in the bestunits.

It should be noted that the R-MACT limits represent theaverage of the best 12% of permitted and operating sys-tems rather than the best achieved values for any system.These perform at the lower limits in the

A life cycle (LC) comparison was prepared (Ecobalance,

overview given inthe preceding paragraph.

VI. LIFE CYCLE COMPARISONOF WTE WITH SELECTEDFOSSIL-FUEL BURNERS FORELECTRICITY PRODUCTION

Inc., 1997) for IWSA in 1997 by Ecobalance Inc. Theprimary results in relative units per kWeh produced are

the following: acidification potential in terms of g of H+

equivalent for WTE = 0.056 vs 0.37 for coal, 0.21 forfuel oil, and 0.061 for NG; g of CO2 produced = 554 forWTE vs 1333 for coal, 871 for fuel oil, and 665 for NG;natural resources depletion index = −0.69 for WTE vs1.0 for coal, 14.3 for fuel oil, and 28.6 for NG. Theseestimates depend among many other factors on the heats ofcombustion and compositions of MSW, emission speciesand concentrations for the three fuels, source compositionsand production methods for the fossil fuels, designs andconstruction of plants and auxiliaries, ash composition andmanagement, etc.

VII. CURRENT AND FUTUREAPPLICATIONS OF WASTEINCINERATION WORLDWIDE

An extensive assessment (H. Kaiser Consultancy, 1996)of current and future applications of municipal waste in-cinerations contains the following estimates as of 1996:In spite of recent selective shutdowns (e.g., 2 out of 3operating units in Hong Kong), 2400 large-scale combus-tors remain in operation, 150 units are under construc-tion, and 250 additional units are expected to be builtby 2005, i.e., about 2800 units should be in service by2005. The dominant technology will remain grate incin-eration with growth especially for fluidized-bed combus-tion and, to a lesser extent, for pyrolysis units. Small- andmedium-size plants (up to 300 mt/day) will be preferred.WTE plants will represent the systems of choice be-cause of the economic values of electricity generation andsteam production. System utilization will grow preferen-tially in Asia with rapid urbanization and also in EasternEurope.

New facilities will ultimately have to meet rigorous en-vironmental controls everywhere and will therefore be-come more costly with time as R-MACT cleanup andpossibly later tightening of these standards become uni-versal requirements. A rough estimate of a typical plantcost for a 300 mt/day plant with R-MACT cleanup sys-tem is about $100 to $150 ×106 in 1999 U.S. dollars,corresponding to $12.5 ×109 for 250 new units, i.e., amarket for new systems of about $3 ×109/year to the year2005.

VIII. POTENTIAL ELECTRICITYPRODUCTION FROM WTE SYSTEMSIN THE UNITED STATES

The potential for electricity production may be estimatedfor the United States by using a procedure described in

P1: GNH Final Pages



1988 (Penner et al., 1998). The method is applicable tocountries other than the United States.

With stringent implementation of energy-conservationmeasures, steady-state generation of municipal wastesin the United States may possibly be restricted to1200 lb/yr-p. This waste has roughly (Penner et al., 1998)an energy content of 104 kJ/kg. By the year 2050, a steady-state population of about 325 ×106 will then produce wasteleading to an electric power output from incineration of

PMSW1 � [(1200 lb/yr p) × (0.7)] × (325 × 106 p)

× (104 kJ/kg) × (1 kg/2.2 lb)

× (2.78 × 10−4 kWh/kJ) × (1 yr/8.76 × 103 h)

× (0.2 kWe/kW) × (1 MWe/103 kWe)

= 7.9 × 103 MWe

if 70% of the waste is incinerated while 30% is recycled orreused and 20% conversion efficiency of the heat releaseto electricity is achieved. The use of similar procedures for

FIGURE 1 Schematic of the MARTIN-System Technology (A) and furnace details (B), as implemented at the Bristol,CT (USA) resource-recovery facility. After weighing in the tipping hall (1), collection vehicles (2) dump unsorted wasteinto the storage pit (3). Overhead cranes (4) lift the waste to a bin (5) from which it enters the feed hopper (6) leadingto the feed chute. Hydraulic ram feeders (7) provide controlled charging of the proprietary Martin reverse-acting stokergrate (8). A forced-draft fan (9) supplies primary underfire air below the grate (10) to the refuse burning on the stokergrate. Secondary above-fire air is injected at the front and rear walls above the grate to induce turbulence and causeburnout of the primary combustion products formed at and above the grate and in the furnace (12). The furnace wallsand dividers between the boiler sections (13) are welded solid membranes. In the third pass (14) of the multipassboiler, the flue gases are cooled as they enter the horizontal superheater, which contains vertical superheater tubes(15). Economizer tubes (16) are located in the fifth pass of the boiler. During operation, the boiler surfaces are cleanedby sootblowers in the third, fourth, and fifth passes. A lime slurry is injected in the dry scrubber (17) to neutralize andcapture acidic components of the flue gases. Spent lime is collected in the fabric-filter baghouse (18) together withfly ash. An induced-draft fan (19) moves the cleaned flue gases to the exhaust stack (20). Particulates collected inthe dry scrubber (17) and baghouse (18) are moved via fly-ash conveyers (21) to the proprietary discharger (22) andthen transferred by conveyors (23) to residue storage before disposal in a landfill (or, at other locations, use in themanufacture of aggregate construction materials). The high-temperature steam produced in the boiler may be usedfor power generation on passage through a turbine generator or as input to a district-heating system.

hazardous and biomedical wastes led (Penner et al., 1998)to estimates of 3.9 × 103 MWe and 0.46 × 103 MWe,respectively.

IX. REPRESENTATIVE MWI SYSTEMS

A widely used system was developed by the Martin Com-pany of Munich, Germany. The U.S. and Japanese li-censees of this technology are, respectively, Ogden Corp.and Mitsubishi, both of which have important marketshares in their home countries and elsewhere. A schematicof the Martin operating system is shown in Fig. 1. Thenovel feature of the process is the reverse-acting stokergrate, which is inclined at an angle of about 26◦ withrespect to the horizontal to allow the trash to move for-ward by gravity while it is not pushed. The grate has se-quential fixed and moving grate steps, with the latter stir-ring the waste against the downward direction. As theresult, the burning refuse layer is continuously rotated

P1: GNH Final Pages



FIGURE 1 (continued )

and mixed at a constant bed depth while red-hot mate-rial continues to move back in the direction of the feederend. The net result is that vigorous refuse drying, igni-tion, and combustion occur along the entire grate length.These processes are properly controlled by dividing theentire grate into distinct sequential segments that are sep-arately supplied with computer-monitored and controlledair for primary combustion. Partially burned combustionproducts rise above the grate and flow upward parallelto water-cooled furnace walls. Secondary air injectorsprovide the needed oxygen for complete combustionof fuel molecules. The total injected air-to-fuel ratiois about 180% of stoichiometric to ensure completeoxidation.

A forward-acting grate has been used by a number ofvendors including Deutsche Babcock Anlagen (DBA), itsAmerican licensee American Ref-Fuel, Combustion En-gineering, Vølund, von Roll, and others. The grate anglewith respect to the horizontal is now greatly reduced below26◦ since the primary force acting in the forward direction

is provided by the moving grate. Underfire and secondaryair are injected in a manner analogous to that used in theMartin system and the furnace is again constructed withwater-containing walls.

DBA developed the Dusseldorf grate with large rotatingdrums for stirring the incoming trash flow and facilitatingintimate contact between primary underfire air and theflowing trash. Secondary air injection and burnout areachieved in a manner resembling the procedure sketchedin Fig. 1.

Vølund has designed a system in which a rotary kiln islocated downstream of a forward-acting grate. Gas-phaseconversions have been implemented in furnaces with par-allel flows of the primary combustion products and sec-ondary air flows, as well as with counterflows and mixedflows of these gases.

We refer to the literature for such topics as incinera-tor modeling, temperature-profile estimations and verifi-cation, fouling of chamber walls, and formations of toxiccompounds such as furans and dioxins in the gas phase

P1: GNH Final Pages



and on fly ash (which is probably the dominant process),as well as designs of cleanup systems that have led to theR-MACT standards described previously.

X. ELECTRICITY ANDTIPPING-FEE REVENUES

As an example, we consider a city with 1 million peo-ple generating 1.5 kg/day-p of municipal waste. The to-tal waste production is then (1.5 kg/day-p) × (106 p) ×(1 mt/103 kg) = 1.5 × 103 mt/d. For an assumed heatof combustion of 10 MJ/kg and a fossil-energy-to-electricity conversion efficiency of 0.2, electricity gener-ation per mt of refuse is then (104 kJ/kg) × (103 kg/mt) ×(1 kWeh/3600 kJ) × 0.2 = 560 kWeh/mt. With refusegeneration of 1.5 × 103 mt/d, the total electricity pro-duction is then (560 kWeh/mt) × (1.5 × 103 mt/d) ×(1 d/24 h) = 35 MWe. A competitive price for electric-ity sales in 1995 was about $20 × 10−3/kWeh so that thetotal value of the electricity generated per year becomes35 MWe × ($20 × 10−3/kWeh) × (103 kWe/MWe) ×(24h/d) = $16,800/d. If the refuse tipping fees are $20/mt,then the return from tipping fees is ($20/mt) × (1.5 ×103 mt/d) = $30 × 103/d, i.e. of the total return of$46,800/d, about 36% is derived from electricity gener-ation for the specified costs.

XI. ENVIRONMENTAL ISSUES

The stringent R-MACT control specifications have beendeveloped over a long period of time (more than 15 years)and reflect a compromise between what is possible andwhat is desirable in order to protect public health while al-lowing improvements of valuable technologies. The emis-sion specifications for each component and especially forthe 210 congeners of chlorinated dioxins and furans werearrived at after completion of a great deal of research andmany medical and epidemiological assessments of possi-ble harm done by emissions to the air and/or water by thesecompounds. These studies have not yet been concluded.

Municipal waste incinerators have figured prominentlyin the debate about environmental equity. However, a 1990census showed that modern WTE facilities are predomi-nantly located in white, middle-class neighborhoods with6% more whites than corresponds to the national averageand a median household income 7% abovaverage (Carr, 1996).

e the national


ENERGY FLOWS IN ECOLOGY AND IN THE ECONOMY

• ENERGY RESOURCES AND RESERVES • HAZARDOUS

WASTE INCINERATION • RADIOACTIVE WASTE DISPOSAL

• RENEWABLE ENERGY FROM BIOMASS • WASTEWATER

TREATMENT AND WATER RECLAMATION

BIBLIOGRAPHY

Carr, G. L. (1996). “Environmental equity: Does it play a role in WTEsiting?” J. Hazardous Mat. 47, 303–312.

Dogett, R. M., O’Farrell, M. K., and Watson, A. L. (1980). “Forecasts ofthe Quantity and Composition of Solid Waste,” 171 pp., EPA-600/5-80-001, EPA, Cincinnati, OH.

Ecobalance, Inc. (May 1997). “Life Cycle Comparison of Energy Pro-duction of Waste-to-Energy Facility to Other Major Fuel Sources.”

Jones, K. H. (March–April 1997). “Comparing air emissions from land-fills and WTE plants,” Solid Waste Technol., 28–39.

H. Kaiser Consultancy (1996). Philosophenweg 2, D-72076, Tubingen,Germany.

Kiser, J. V. L., and Zannes, M. (February 1999). IWSA, 1401 H StreetNW, Washington, D.C.

NREL/BR-430-21437 (December 1996). Published by the National Re-newable Energy Laboratory (NREL), 1617 Cole Boulevard, Golden,CO 80401-3393.

Penner, S. S., and Richards, M. B. (1989). “Estimates of growth ratesfor municipal waste incineration and environmental control costs forcoal utilization in the U.S.,” Energy—Int. J. 14, 961–963.

Penner, S. S., Wiesenhahn, D. F., and Li, C. P. (1987). “Mass burning ofmunicipal wastes,” Ann. Rev. Energy 12, 415–444.

Penner, S. S., Chang, D. P. Y., Goulard, R., and Lester, T. (1988). “Wasteincineration and energy recovery,” Energy—Int. J. 13, 845–851.

Taylor, A., and Zannes, M. (May 1996). Integrated Waste Services As-sociation (IWSA), 1401 H Street NW, Washington, D.C.

encyclopedia of physical science and technology - energy

Documents