air cooler design sheiko

Effectively Design Air-Cooled Heat Exchangers

This primer discusses the

thermal design of ACHEs and the

optimization of the thermal design,

and offers guidance on selecting and

designing ACHEs for various appliclltions.

R. Mukherjee, Engmeers India. Ltd.

Based on bare-tube heat-transfer area. air-cooled heat exchangers (ACHEs) cost two to three times more than water-cooled heat ex-changers for the same heat duty (hardware costs only). There are two main reasons for this. First. the thermal conductivity of air is considerably lower than that of water. which results in a much lower heat-transfer coefficient. Second. since design ambient temperatures are always higher than design water temperatures. the mean temperature difference ( MTD J is always lower for an ACHE. especially at relative-ly low process-fluid outlet temperatures.

As a result of these two factors. the heat-transfer area of an ACHE is consid-erably larger than that of a water-cooled heat exchanger for the same duty. In addi-tion. the larger area requires an elaborate structural support system. which increases the cost further.

However. as all engineers know. the capital (or tixed) cost of e4uipment is only part of the story. What is important is the total cost - the sum of the tixed cost and the operating cost.

The operating costs for water cooling are much higher than those for air cooling. These include the costs of the initial raw water itself. makeup water. and treatment chemicals. the apportioned cost of the plant cooling tower. and the pumping cost. For

ACHE~. Lile uperaung cost IS just the cost of the power required to make the air flow across the tube bundles. As water becomes scarcer. the operating costs of water-cooled heat exchangers increase. thereby tilting the economics further in favor of air cooling.

26 FEBRUARY 1997 CHEMICAL ENGINEERING PROGRESS

This article outlines the advantages and disadvantages of ACHEs. explains and il-lustrates what ACHEs are. elaborates the various construction features available to accommodate different application require-ments. discusses the thermal design of ACHEs and the optimization of the ther-mal design. and examines several special applications.

Pros and cons of ACHEs Air-cooled heat exchangers offer several

important advantages over water-cooled exchangers.

Some of these are a direct result of water not being used as the cooling medium. The high costs of using water. including the costs for the raw water. makeup water. and treatment chemicals. are eliminated. The lo-cation of the cooler. and thus the plant itself. does not depend on being near a source of water (such as a river or lake l. Thermal and chemical pollution of the \\ater source are prevented. Maintenance costs are reduced. since frequent cleaning of the waterside of the cooler (necessitated by fouling such as scaling. biofouling. sedimentation. etc.) is not needed. And. installation is simpler be-cause water piping is eliminated.

Another advantage is that ACHEs can continue to operate. albeit at reduced ca-pacity. by natural convection when there is a power failure.

And hnally. control ot the process fluid's outlet temperature (and thereby the heat duty) is easily accomplished through vari-ous methods. such as switching fans on and off. use of two-speed or variablejipeed mo-tors. use of auto-variable fans (which allow

I

the blade angle to be adjusted even while the fan is in motion), and so on.

Limitations. Of course, ACHEs also have some limitations.

As noted earlier, the initial capital cost of an ACHE is considerably more than that of a water-cooled unit be-cause air has a much lower thermal conductivity and specific heat than water.

In cold climates, extensive winteri-zation arrangements have to be incor-porated to protect against freezing tem-peratures. This increases the initial capital costs even further.

An economical approach tempera-ture between the outlet process fluid and the ambient air is generally in the range of 10-l5C, whereas in water-cooled exchangers this approach tem-perature can be as low as 3-5C. This disadvantage is mitigated by having air cooling followed by trim cooling with water.

Because of the larger heat-transfer area, an ACHE requires a larger plot area than a water-cooled exchanger. However, this disadvantage can be overcome by locating an ACHE on a piperack so that no valuable plot area is wasted.

The low specific heat of air requires that large quantities of air be forced across the tube bundles. This is accom-plished by large-diameter fan blades rotating at high speeds, which pro-duces high noise levels.

The seasonal variation in air tem-perature can affect performance. and expensive control systems have to be incorporated to ensure stable operation.

ACHEs cannot be located near large obstructions, such as buildings, be-cause air recirculation can set in and reduce efficiency.

The design of ACHEs is relatively sophisticated. Because of this, there are fewer vendors of ACHEs than there are of water-cooled shell-and-tube heat exchangers.

For cooling viscous liquids, ACHEs become even more expensive due to the extremely low tubeside heat-trans-fer coefficient. (Such liquids yield con-siderably higher heat-transfer coeffi-

f_

I

cients when flowing on the outside of tubes in shell-and-tube heat exchang-ers, due to the much higher turbu-lence.) This situation can be remedied to a large extent by the use of tube in-serts. However, this technology has still not become very popular.

Optimizing air and water cooling

In many applications where the pro-cess outlet temperature is relatively low, air cooling alone may not be feasi-ble. For example, cooling a light hy-drocarbon liquid to 40-45C may not be feasible at a site where the design ambient temperature is 42C and the design cooling water temperature is 33C. In many such cases, a combina-tion of air cooling followed by trim cooling with water can be adopted.

For certain other services, air cool-ing may not be economically viable at all. For example, at the site just de-scribed (ambient= 42C, cooling water = 33C), air cooling may not be viable for a naphtha stabilizer condenser with inlet and outlet temperatures of 50C and 45C. respectively, due to an inor-dinately low temperature difference. Here, water cooling alone would be recommended.

Thus, there will be services where air cooling alone is suitable. others where a combination of air and water cooling can be used, and still others where only water cooling should be employed.

The optimum temperature break-point between air and water cooling (that is, the temperature at which the process fluid leaves the ACHE and en-ters the water-cooled heat exchanger) has to be established by the overall economics for the specific project. It will depend on the equipment costs for the air-cooled and water-cooled heat exchangers, the total cost of using water, and the cost of power. Generally speaking, this optimum temperature is around 15-20C more than the design ambient temperature.

The important point to note is that even for combination cooling (air plus water cooling), the ACHE will address

the major heat duty - 80% or more of the total - thereby considerably re-ducing the cooling water flow.

When using combination cooling, it is usually best to design the ACHE for a somewhat lower ambient temperature than would be used if there were no trim cooling and to design the trim cooler for that process fluid tempera-ture which would be delivered by the ACHE at the maximum (or near maxi-mum) ambient temperature. This is be-cause the increase in the trim cooler cost will be much less than the increase in the ACHE cost if it were to handle air at the higher ambient temperature. This concept is illustrated by the fol-lowing example.

Example 1. In a refinery, 71,848 kglh of kerosene was to be cooled from l83C to 43C. The maximum and minimum ambient temperatures at the site were 42C and l8C, respectively. Cooling water was available at 33C. The other relevant process parameters were: fouling resistance of kerosene = 0.0004 hm2C/kcal. fouling resis-tance of water = 0.0004 hm2C/kcal, allowable pressure drop of kerosene = 1.0 kg/cm2. allowable pressure drop of water= 0.7 kg/cm2. kerosene viscosity = 0.11 cP at 183C and 0.9 at 43C, and total heat duty = 5.87 MM kcal!h.

The optimum break-point tempera-ture between air and water cooling was calculated to be 58C. Thus, an ACHE was designed to cool the kerosene from 183C to 58C and a trim cooler (using water) to cool the kerosene from 58C to 43C. Because there was to be a trim cooler. 38C was used as the de-sign ambient temperature.

The ACHE was designed with one section having two tube bundles in par-allel and a total bare-tube area of 327 m2. Since the section had a high length-to-width ratio ( 12.5 m x 3.8 m), it uses three 9-ft-dia. fans.

A simulation of the ACHE's perfor-mance established that it the ambient temperature were 42C, the kerosene outlet temperature would ,be 60C. Therefore, the trim cooler was designed for this heat duty, that is, fi>r cooling the kerosene from 60C to 43C. A single

CHEMICAL ENGINEERING PROGRESS FEBRUARY 1997 27

AIR-COOLED HEAT EXCHANGERS

shell with a heat-transfer area of 157m2 was found to be adequate.

Now consider what would happen if the ACHE were to be designed for an ambient temperature of 42C and the trim cooler for cooling the kerosene from 58C to 43C. The ACHE bare-tube area would increase from 327 m2 to 369m2 (and the fan diameter from 9 ft to 10 ft). whereas the trim cooler heat-transfer area would decrease from 157 m2 to 137 m2. Thus. it would be more economical to design the ACHE for an ambient temperature of 38C and the trim cooler for a kerosene tem-perature equal to the outlet temperature from the ACHE (that is. not 58C but 60C) when the ambient temperature is the maximum ( 42C).

Construction features The American Petroleum Institute's

standard API 661. "Air-cooled Heat Ex-changers for General Refinery Ser-vices" (I). outlines the minimum re-quirements for design. materials. fabri-cation. inspection. testing. and prepara-tion for shipment of ACHEs. Although this standard is intended specitically for the petroleum retining industry. its use is widespread in the petrochemical. fer-tilizer. and general chemical industries as well. In fact. it is considered to be a standard for all ACHEs.

Figure 1. A tube bundle is an assembly of tubes, headers, tube supports, and frames.

A discussion of construction fea-tures should begin with a few basic definitions: Figure 2 . . \ bay (or section) comists of two or more tube bundles.

Tube /)[{nd/e - assembly of tubes. headers. tube ,upports. and frames (Figure I).

Ba\' or sectimz- the smallest inde-pendent part of an ACHE. complete with its tube bundles. fans. dri,es. mo-tors. supporting structure. and so on

--------- ------ -- ---

rate due to the loosening of the fins. they are generally not very popular. One place where they are common is in the corrosive marine atmospheres of offshore platforms. since they afford good protection against atmospheric corrosion of the base tubes and are su-perior to grooved fins.

Double, L-footed finned tubes. These offer better coverage of the base tube (Figure 6b). However, because they are more expensive (about I 0-15%) than single L-footed finned tubes. they are preferred only in ex-tremely corrosive atmospheres. The upper limit on process-fluid inlet tem-peratures for these finned tubes is 170C.

Figure 3. A bank consists of two or more units on the same structure. (Photo courtesy of ABB Lummus Heat Transfer.)

Grooved (or embedded) or G-type finned tubes. Here the fin is em-bedded in the tube by first plowing a groove in the tube wall and then stretching the fin material into the groove under sufficient tension to achieve a specified bond strength (Fig-ure 6c). These are the most commonly used tinned tubes in ACHEs.

surface). However. galvanized carbon steel finned tubes cannot be used for process fluid temperatures above 300C. The use of these tubes is not very prevalent. I See I 2 ). )

Since the coefticient of linear ex-pansion of aluminum is about twice that of carbon steel. a gap resistance between the tube and the tin material develops. As the operating temperature increases. the difference in the coefti-cients of expansion increases and so does the resulting gap resistance_ Thus. maximum operating temperatures. which depend on the type of bond be-tween the tube and the tin. have been established_ 1 These are specified for the various types of finned tubes in the following discussions.)

The standard tube O.D. is I in .. al-though 11 1~ in .. ! 1/2 in .. and even 2 in. diameters are employed in cases where the allowable tubeside pressure drop does not permit the use of 1-in.-O.D. tubes. The standard fin heights are Yx m., 112 m .. and 'Is in .. with the latter two being far more popular.

Single, L-footed finned tubes. This is a circular fin wrapped helically around the tube under tension (Figure 6a). Full coverage of the base tube by

the L-foot offers good protection against atmospheric corrosion.

These tins tend to become loose over time. resulting in significant deter-ioration in the airside performance due to the gap between the tube and the tins. Consequently. their usc is limitcu to applications where the process-tluid inlet temperature is less than 120"C. However. the majority of ACHE appli-cations have process inlet temperatures less than this.

Since the airside perfom1ancc of these tinned tubes is likely to dcterio-

/

G-finned tubes require a heavier tube wall than L-footed tubes. API 661 1 Clause -J..I.ll.3) specifies a minimum tube wall thickness of 2.1 mm for car-hon steel and low-alloy -,tee!. and 1.65 mm for stainless steel. For embedded-fin tubes. this thickness is measured at the bottom of the groove. Hence. em-bedded tin tubes have to be thicker than L-footed tin tubes by the groove ucpth 1 which is usually 0.3 mm ).

' '

Figure .J. ACHEs consist of bundles

Plan combined into hays. View hays combined into

units, and units combined into hanks.

Bundle--pq pq pq Front Elevation Fan~ ex::> ( ) ex::> ( ) ex::> (

1+---UnitA---! 1--unitB --1 1. Bank ~I


~--~~---------_;., ________ _


Figure 5. The ACHE's fans can be in a forced-draft (a) or induced-draft (b) configuration. a.

b.

These tinned tubes can withstand process-fluid inlet temperatures of up to 400oC due to the strong tin bond (the fin/bond resistance is considered negligible 1. They can also withstand cyclic operation without any loss of tin/tube contact. The disadvantage of G-tinned tubes is that the base tube material i, exposed to the atmosphere. so their tbe in aggressi\e atmospheres ( 'uch ~h marine 1 is not recommended.

Bimetallic finned tubes. Bimetallic tinned tubes have G-type tins embedded in an outer tube of aluminum that is stretched mer the ba~e tube (figure 6dl. These tubes ~ue not used \ery often. but they ;ue \\ell-suited for applications where the process fluid is at high pres-sure

-------

d.

, IIIII a. b.

c. d.

Figure 6. Finned tubes are available in a mriety of fin styles: Figure 7. ACHE~ can u~e plug (a}, cover-plate (b), manifold (c), or bonnet (d) headers. L-footed (a), double L-footed (b), grooved G-type (c). bimetallic

(d), and extruded (e).

Other tube bundle features Thbe supports. The tinned tubes

are supported by special aluminum support boxes or by zinc collars cast on the tubes themselves.

Side frames. TI1e ACHE\ side-frame serves two purposes. First. it supports the headers and the tube.., and makes the tube bundle a rigid . ..,elf-contained assembly that can be trans-ported and aected conveniently. Sec-ond. it serves as a seal and prevents the bypassing of air.

Thbe-to-tubesheet joint .. \t ltl\\ to medium pressures (]css than 70 bar!. the tubes are expanded into double grooves in the tubesheet. At high pres-sures I greater than 70 ban howcwr. the tubes arc strcngth-\\eldcd to the tubcshcet. (Strength-welding produces a proper tillet-weldcd joint \\Jth a much higher strength than a seal-weld-ed joint. which has no tiller. 1

Fans and drives Axial tans employed m ACHl:s dis-

place very large volumes of air against a low static pressure (typically 0.6-0.8 in. w.g.). These fans have characteristic per-formance curves that are specific to each fan manufacturer. The designer should

have access to fan curves that provide in-formation regarding volume lor mass I of air. static pressure. absorbed power. and noise. Some fan manubcturers e\en fur-nish computa sothvare to aid the de-signer in proper fan selection.

Fan diameter usuallv \aries be-t ween 6 ft and 18 ft. although fam with .smaller and larger diameters arc em-ployed in '>pecial circumstances.

A fan consists of two basic compo-nents - the huh and the blade'>.

The hub i.-. mounted nn the fan shaft and the blades are moumed on the hub. The hub may he made of cast 1ron. cast aluminum. or fabricated steel. \ lanufacturers usual!\ londuct static and d~ namic balancing of the hub in the shop.

Two t~pes of hubs ~ue a\ailable. With a manually adjustable hub. the blade angle can be altered nnly \\hen the fan is stationary. An auto-\ ariable hub includes a device (usually a pneu-matic controller) that can alter the blade angle even while the tan ISm monon. m order to control air flow. Control is usu-ally effected by means of a signal from a temperature indicator/controller !TIC) responding to the outlet temperature of the process fluid. 1 See ( 4 ). )

Blades can be of either metal (usual-! y aluminum) or fiberglass-reinforced plastic IFRP). Plastic blades are suitable only for temperatures of up to 70C.

Fan performance 1 air flow rate and -.tatic pressure 1 is determined by the number of blades. blade tip speed. blade angle. and blade width.

The effect of a change in tip speed on a fan's performance is dramatic. The volume of air tlow varies directly with the tip speed. the pressure varies as the square of tip speed. and the power consumption \aries as the cube of tip speed. Tip speed is normally lim-ited to 61 m/s. as noise becomes exces-sive at higher speeds.

Increasing the number of blades in-creases the fan's abi I i ty to work under pressure. Thus. one could use a 6-blade fan operating at a lower tip speed to deliver the same volume of air as a .+-blade fan. However. this can be carried too far - as the number of blades increases beyond six. multi-blade mterterence may actually reduce the efficiency of the fan. since each blade works in the disturbed wake of the preceding blade. Therefore, the number of blades has t


Plenum

+ / Induced Draft

''-.

/ 45" Max1mum Dispers1on Angle /

Centerline of Bundle

'-.. 45 Max1mum Dispers1on Angle {

Forced Draft

Figure if. The dispersion angle of the fan should be less than ./5".

Figure 9. An A.-frame configuration is well-suited to condensing senices.

All the blades of a fan should be ~et at the same angle for smooth operation. Usually. the blade angle is set between 12 and 2Y. This i~ because peii'or-mancc deteriorates at lo\\er angle~ and become~ unstable at higher angle~.

.-'\ fan \\ ith a \\ ider blade can he op-erated at a lower tip 'reed to ~h..:hiew the 'ame performance. Cun,equently. bn, with \\ ider blade, uperate less Illll.,il~. n1is leature 1' exploited b~ fan \Cilllnr, who offer 'JlcLial Ill\\ -noise fan ....

.-\PI 661 1 Clau~e -L2.J 1 'tipulate' the follm\ 1ng fur f~ub and fan hubs fur ACHL:

L There ... hould he at least I\\ o fans along thc tube lcngth. Hm\C\ cr. a 'in-glc-fan de,ign can he Lhcd in cxLep-tional circumstailLcs l'uch ~'' \cry

'm~ill unit'! il ~tgrLcJ to hy thc pur-cha,er ~llld the \ L'nuor. Thi, I\\ o-fan re-quin:ment h apparently based on

Lonsider~ilion' of reliahilit\ -- ,Jwuld

I '-

HTD may be used with motor drives rated 50 hp (37 kWl or less. For elec-tric motors rated above 50 hp (37 kWl. right-angle gear drives must be used. All steam turbine drivers must employ right-angle gear drives

Unlike flat and V-belts. HTD belts do not rely on friction for their pulling power. Rather. HTD belts utilize a new tooth design that substantially im-proves stress distribution and higher overall loading.

HTD belts do not stretch due to wear. are corrosion-resistant. and oper-ate at reduced noise levels. The belts are capable of transmitting higher torque at lower speed. thus improving the power capacity of toothed belts.

Maintenance i~ simple. ~o adjust-ments are required due to ~tretch or wear. HTD belts are ideal where main-tenance is difticult or 11here downtime could prove to be e.xtremely expensi1 e.

The plenum chamber i.s where the air delivered by a fan is distributed to 1 forced-draftl or ..:ollected fr11m 1 in-duced-draft 1 the whe bundle. It ,;on-sists (lf a rectangular h!l\ \lr a conical/rectangular transition l'tece. F!lr I' liced-draft units. the l'knum chamhcr C:lll he' 'lillare I ()J" i"L"L"l~111~UUJ"I t)J" contL~Li. ldll'rl'~ls l!lr imiuced-dLttl units. it ts imanahl~ cuntcal. .\ pant-lion is provided hemeen fans. ami the ~ap het11een the lUbe bundle ;md the pa111lion pLtte slwuld ll!ll exLeed 20 mnL

For f!lrced-draft units. the pknum chamber h~ts a c,)mcal llllct ~n the htll-l!llll l!l l"l'dUL"e inlet lthses. \\"hen l!l\\-lll H'-e L111' ;~re emplo) ed. th1' c' lllil~Li tnlet ts rLplacL'd h~ ;t hell-sh;~ped llHHilh.

Confiquration of ACHEs Horizontal nmtiguration .. \CHh ~tre usually conligured 111 the lwril!llltal position 1 Figure 51 because this make:. maintenance easier.

A-frame configuration. This de-sign (Figure l)) is employed almost e.x-clusively in power plants for condens-ing turbine exhaust sh~am. The tube bundles are mounted on a triangular frame \\ ith fans lncated bekm. The in-

clination from the horizontal is usually between -l5 and 60.

The A-frame configuration occupies 30--4()Cic less plot area than a horizontal configuration. Additionally. though no less imponantly. the A-frame is ideal for condensing. as it facilitates conden-sate drainage. The common header at the top of the unit allows uniform steam distribution with minimum pres-sure loss. which is imponant for the ef-ficient operation of vacuum steam con-densers. The A-frame configuration is. in fact. the basis of several patented freeze-proof designs.

Vertical configuration. \'enical orientation 1 Figure I 0) is generally em-ployed for packaged systems. such as compre~sor~ '' ith their intercoolers. These units are used where floor space is at a prermum.

A drawback is that thev are much more prone to performance deteriora-tion due to cross-\\ inds. Funhermore. multipass designs are not feasible for condensing sen ices.

Natural vs. mechanical draft \:atural drafttll\oiles nu tans- ~ur

thm i' h\ tuwral L,ln\ection LiliL' tu the 'tack L'ikLt .tcTds' the tuhe bundle .. \11 , \lettul 'tad. 1' su;netimes llbl~tlkd Ill tncrease the drart and thereh\ tilL' CtlPI-tng. The princtpal applicatiun ,,1 n~llu

r~tl draft is 111 dry Luulin~ r, ''' ers in power plants. \\here the lar~e L"ht11111L'Y of the L"!ltlling l!l\\ L'r esrahlishe' ~lll ap-preciable dr~1lt

~q] j .___

I

Most ACHEs are of the mechanical-draft design. Vast amounts of air are moved across tinned-tube bundles by a.xial fans driven hy electric motors. There are two main types of mechani-cal draft - forced draft and induced draft. In forced draft (Figure 5a). fans mounted below the tube bundles blow air across the tinned tubes. In induced draft (Figure 5b ). fans above the tube bundles suck air across the tinned tubes. Each type has advantages and disadvantages. and therefore preferred applications.

Forced-draft ACHEs. Forced-draft ACHEs have several advantages. Be-cause both fans and motors or drive transmissions are located below the tube bundles. accessibility for mainte-nance is much better. Because the fans are located below the tube bundles and handle the colder incoming air. the air pressure drop. and therefore the fan power consumption. 1.' somewhat lower. Fan blade life is longer. since e.xposure is to cold inlet air. Funher-more. e1en for relati1ely high air outlet temperatures 1 greater than 70~C). tiber-reinforced plastic blades can be used.

There are four principal Jisadlan-!ages !lf forced draft. First. di,trihution 'lf air across the tube hundk.' is poorer. ,ince the air lea\ es the tube bundles at ;~much l!l\\er \elocitl.

Second. hot ~1ir reurculation is more likclv to !lccur. due l!l the lll\\er dis-Lharge 1elocit~ ;md the ~thscncc of a stack. resultin~ 111 a lu~hcr ~ur inlet

:J Figure I 0. l'acka~ed ... ntcms lrpicallr employ rertical configurations.

;



temperature and consequently a lower MID. In low-MID applications. the deterioration in performance could be significant. Consequently. forced draft is not preferred where the cold-end temperature approach (the difference in temperature between the process outlet and the inlet air) is less than 5-8C.

Third. the ACHE is exposed to the elements (sun. rain. haiL and snow) un-less louvers or roofs are provided at the top of the tube bundles. This results in poorer stability and process control.

Finally. due to a very small stack ef-fect. natural draft capability in the event of fan failure is rather low.

Induced-draft ACHEs. The princi-pal advantages of induced draft are: better air distribution across the tube bundles: lower probability of hot air re-circulation (the air velocity at the dis-charge is usually more than twice that at the entrance): greater ability to oper-ate following fan failure (due to the much higher stack effect): better pro-cess control and protection from the ef-fects of rain. snow. haiL or sun: and no possibility of damage to fans or drives due to leaking products 1 if corrosive I.

Nomenclature A = heat transfer area D = fan diameter in m hrc, = airside heat-transfer coefficient hrc, = tubeside heat-transfer coeftlcient H P = absorbed shaft horsepower k = constant established by

perfonnance tests or furnished by the fan manufacturer for cakulating the overall sound power level

.HTD = l!lean temperature dilferencc f'WL = "mnd power level in dB

(reference level = I (}-12 watts) Q = heat-transfer duty r,., = tubesidc fouling rcSJstam:c rt." = airside fouling resistance r. = wall resistance

~PT = vel in dB (reference level = 0.0002 microbars)

TS = tip speed in m/s U = overall heat-transfer coefficient

However. induced draft does have several disadvantages. The air pressure drop and power consumption are high-er because the air being handled is hot-ter and lighter. In order to prevent dam-age to fan blades. V-belts. bearings. and other mechanical components. the exit air temperature has to be limited to about 90C oooc for FRP blades). The fans. being located above the plenum chamber. are less accessible for maintenance. Further, maintenance work may have to be carried out in the hot air caused by natural convection. Since the motor is usually located below the tube bundle. a long shaft is required to transmit the power to the fan located above the bundle. This rep-resents another potential problem area.

Thermal design An ACHE. like any other heat ex-

changer. must satisfy the following basic equation:

A= Q/(U)(MTDl (I) The overall heat-transfer coefficient.

U. is determined as follows: l/U = 11/ztc, + 1/lztc + rrr

+ r.,, + r ... (2)

Since the airside heat-transfer coef-ficient is generally much lower than the tubeside heat-transfer coefficient. it is necessary to use tinned tubes to make the airside heat-transfer coefficient more compatible with the tubeside heat-transfer coefficient.

Tubeside calculations. The tube-side heat-transfer coelticient and rres-sure drop calculations are rerformed just as they are in shell-and-tube heat exchangers. whether for single-phase cooling. condensing. or a combination of the two. This has been cmered ex-tensively in the literature (such as (o. 7)) and is rather common knowledge among chemical engineers. so it will not hi> di~cns-;f.'d bert".

Increased tubeside pressure drop. It sometimes happens that the allowable tubeside pressure drop is specified very low. This can penalize the ACHE de-sign considerably by requiring an inor-dinately high heat-transfer surface area.


For gases and condensers. the al-lowable pressure drop is generally be-tween 0.05 and 0.2 kg/cm2. depending upon the operating pressure - the lower the operating pressure. the lower the allowable pressure drop. For liq-uids, the allowable pressure drop is generally 0.5 to 0.7 kglcm2. However. if the liquid viscosity is high, a higher pressure drop is warranted for an opti-mum design.

Example 2. A vacuum-column bot-toms cooler was to be designed for the hydrocracker unit of an oil refinery. The principal process parameters were: heat duty = 5.764 x I ()6 kcallh. flow rate = 93.278 kg/h. inlet temperature = 180C. outlet temperature = 70C. al-lowable pressure drop = 0.5 kg/cm2. inlet viscosity = 4.2 cP. outlet viscosity = 15.0 cP. and fouling resistance = 0.0006 hm2 0 C/kcal. Carbon steel tubes of 25 mm 0.0. and 2.5 mm thickness were to be used. Since the ACHE would be located on a 12-m-wide piperack. the tube length was to be 12.5 m. The design ambient temper-ature was 42C.

Since the li4uid \ isco>ity was rather high. the allowable pressured drop of 0.5 kg/em~ was rather lmv. However. it was used to prepare a preliminary ther-mal design. as shown in Table I.

Due to the low allowable tubeside pressure drop. the tubeside heat-trans-fer COefticient \Vas l lll(y 08.6 kcal/hm' C. which represented X6.6r/r of the total heat-transfer resis-tance. In order to produce a more-eco-nomical design. a higher tubesiJe pres-'ure Jrop ot l.o kg/em: was permitted. The re\ ised Jesign i' aiso outlined in Table I. The merdesign was X.85?r. !he same as in the original design.

Note that in the re\ised Jesign. the tubeside heat-transfer coefticient was Y7.3 kcal/hm2 c (compared to o8.8 kcal/hm' 'C for the original design). ;md tht> owrall ht':-~t-tmnsfer coefficient was 80.07 kcal/hm2C (instead of 59.72 kcal/hm2 0 Cl. As a result. the number of sections could be reduced from three to two and the overall bare-tube heat-transfer area front 2.027 m 2 to 1.536 m2 The net result was a con-

l

l t

' I

siderable reduction in the cost and the plot area of the unit. Although the power consumption of each fan in the revised design was more (18.9 kW vs. 15.5 kW), the total power consumption of 75.6 kW was less than the 93.0 kW of the original design because there were fewer fans. The only negative ef-fect of the revised design was that the design pressure of the vacuum-column bottoms circuit increased marginally, resulting in a very minor increase in the cost of the other heat exchangers in the circuit upstream of the ACHE.

Airside calculations. The airside heat-transfer coefficient and pressure drop calculations are rather complex because they involve extended surface. For a detailed discussion on this sub-ject. see ( 8-10).

However. the designer should keep in mind that whereas the airside heat-transfer coefficient varies to the 0.5 power of air mas velocity. the pressure drop varies to the 1.75 power of the same.

Mean temperature difference. The MTD calculations for ACHEs are somewhat different from those of

~hell-and-tuhe heat exchangers. -.ince ACHEs employ pure cmsstlow. D1w,. the MTD curves furnished in the Tubu-lar Exchanger Manufacturers Associa-tion (TEMA l standards 1 II) are not ap-plicable to ACHEs. One source of in-formation on cross-flow MTD determi-nation is 1 12 ).

Interestingly. as the number of tube passes increases in an ACHE. the MTD also increases. For four or more tube passes. the MTD becomes equal to the true countercurrent l'vlTD detem1ined from the four terminal temperatures.

In \\ ater-cooled heat exchangers. the cooling water outlet temperature is usually limited to -1J-+5T. based upon considerations of scaling by re-verse-solubility -:.tits. In the case of ACHEs. however. there is no such lim-Itation. Consequently. the a1r tlow rate and its outlet temperature can vary to a greater extent.

When the air mass-flow rate is low-ered. the air outlet temperature in-creases. thereby reducing the MTD.

Table 1. Superior design with increased tubeside pressure drop (Example 2) .

Original Design, Revised Design, Lower lncntned

Pressure Drop Pressure Drop

Number of sections 3 2 Number of bundles per section 2 2 Number of tube rows 8 8 Number of tubes per row 44 50 Number of tube passes 8 10* Tubeside heat-transfer coefficient, kcal/hmz.oc 68.8 97.3 Tubeside pressure drop, kg/cm2 0.5 1.54 Overall heat-transfer coefficient, kcal/hmz.oc 59.72 80.07 Total air flow, kg/h 2.090.000 1,600,1100 Mean temperature difference. oc 68.19 67.14 Total bare-tube area, m2 2,027 1,536 Power per fan, kW 15.5 18.9 Total power, kW 93.0 75.6 Degree of overdesign, % 8.85 8.85

*In the revised design, two passes in each of the two upper rows and one pass in each of the six remaining rows.

Furthermore. the airside heat-transfer coefficient is reduced due to the hmer mass \ c'hlLity. thereby reducing the overall lk'at-tram.fer coctficicnr. Hoth these ctfech increase the required heat-transfa area. The ~ain is in the power consumption. as both the re-duced !low rate and the consequently lower pressure drop result in a lower power requirement.

The optimum thermal design will he the one that best balances these op-posing tendencies. This balance will depend upon the e.\.tent to which the airside heat-transfer coefticient is con-trolling. Optimitation of thermal de-sign is discussed in detail in the next -.ection.

Design ambient temperature. The selection of an appropriate design am-bient temperature is of the utmost im-portance. W1thout realiZing the conse-quences. many customers specify the highest temperature as the design air temperature. This is an extremely con-servative practice and will result in an unnecessarily high cost for the ACHE.

For example. if a plant site has a summer peak temperature of 45C and -1 winter lmv temperature of 2C. the design air temperature should not be -1-:'i"C. but rather somewhat lower. say 42"C. It is not prudent to penalize the ACHE design by basing it on a peak temperature that may occur only for a few hours during the entire year. The cost dift'erence between the two cases may he appreciable. especially if the \1TD is low.

The usual practice is to select that temperature \\ hich is not exceeded dur-ing 2-J'Ic of the total yearly hours of operation. Thus. a temperature varia-tion chart at sutticiently short intervals throughout the year i-. needed for a proper estimate of the design ambient temperature. Such data are a\ailable from the meteorological ottice.

J

--~$110:..


Table 2. Effect of design air temperature (Example 3). Design Ambient Mean Temperature Degree of Temperature, c Difference. c Overdesign, Ofo

40 41 42 43 44

0.2 kg/em". inlet temperature= 70.5C. outlet temperature = 55C. and fouling resistance = 0.0003 hm"C/kcal.

Using a design ambient temperature of 42C. the following thermal design was prepared: number of sections = 4. number of bundles per section = 2. tube dimensions = .~2 mm 0.0. x 2.5 mm thick x I 0.500 mm long. fin height = 64 mm. tin density = .+33 tins/m. total bare-tube area= 1.248 m2.

Varying the design ;unbient temper.l-ture from 40C to 44C n:sulted in a sharp reduction in the MTD. 11 ith a con-

~equent increounJ pPwer level. the tntal sound emitted by a -,ystem to the envi-ronment. is an ah-.olute LJUaillit\. The sound rre"ure k1 L'i is a relati1 e qu

-ing the usual two tube bundles is 6.4-7 mwide.

A 12-ft (3.657 m). 13-ft (3.962 m). or 14-ft (4.267 m) diameter fan fits in this width quite comfortably. Since there must be at least two fans along the tube length as noted earlier. a tube length of 8-10 m will satisfy the mini-mum 40o/c fan coverage criterion speci-fied in API 661 .

Sometimes. an ACHE section may be highly rectangular (that is. the length-to-width ratio is 2.5 or more). This may ~uise. for example. in the case of a large plant with a wide piper-ack (and thus a long tube length) but a small number of low-duty ACHEs. In such cases. three fans may be used along the tube length. Since the cost for such designs is higher. a 3-fan-per--,ection design should be a last resort.

Since ACHEs are usually located on top of a piperack and across it. the tube length is determined by the piperack width. The tube length is usually 0.5 m greater than the piperack width for convenience of locating the plenum chambers.

Sometimes. ACHEs arc grade-mounted. In -,uch cases. the ahmc lim-itation 11 ill not apply and tube length can he better optuni1.ed.

It i., c1idcnt that the number o! tube bundles 1\ iII han~ to be an even num-ber. ~~., two u! them arc combined to form one section.

fube u.D. Just a' in shell-and-tube heat c\-

dlangcrs. the smaller the tube diame-ter. the lcs-., expensive the ACHE 1\ ill he. Hui\CICL the -.,mailer the tube di-~tmctcr. the more difticult it 11ill he tu clean the tubes.

The -.,mallc-.,t tube diameter recom-mended hy API 661 .D. tuhes and 111 '' t uhe pa"c' ~ 1eld ~~ much more ccu-nomical dc-.,i!:!ll.

Example ~- .\n m crhead umdenser '111 a eTude ct 1lumn 1\ as to be de;,igned j,,r a rclincr~ 11 ith the follm\ing pro-cess parameter-.,: heat duty = 19.5S4 X I 0'' kLal/h. !low rate = I.N.SOO h.g/h. inlet 1apor 11 eight fraction = 1.0. outlet vapor wetglll tracuon = u.(H. operaung pressure = 2.3 kg/em~ abs .. inlet tem-perature = 123''C. outlet temperature= 6Y'C. allowable pressure drop = 0.25 kgkm 2 fouling resistance = 0.0002 hm2 C!kcal. and de-,ign ambient tem-

Design #1 Design#2 Design#3 25-mm 0.0., 25-mm 0.0., 32-mmO.D .

2 Passes 1 Pass 2 Passes

3 3 3 2 2 2 5 6 5

48 46 38 16 16 16

433 433 433 67 67 73

1,384 1,591 1,401 33,050 37,988 30,691

2 2 2 4.572 4.572 4.267 0.573 0.072 0.214

2,112.8 1,027.2 1,381.5 2,130,000 2.750,000 2.420,000

5.8 10.97 11.95

713.3 826.04 747.32 433.4 I 364.8 400.9 9.72 i 23.12 21.87

I 3.78 i 3.61 3.61 I I i I

perature = _-;s.o C. Tubes were to he carbon 'tecl and 12.500 mm long.

A design \\as prepared \\ ith 25-mm-O.D .. 2.5-mm-thick tubes. 11 ith t\\O tube hundles in each of three -.,cc-tions: the other c"(lllstruction detail-, arc listed in Tahle 3. \Vith two tuhe pa-,se-.,. the tuheside pres-.,ure drop of O.'in J...g/cm: f~u c'\CCedcd the aJiowahic limit of 0.25 kg/em:. Due to the rather high tubeside heat-transfer coe!licient.

~~ low airside heat-transfer coclticient could he wlcrated and a low airs ide 1 c-locity could he maintained. leading t(J an unusual!~ low power consumption. However. the de;,ign was not accept-aote oecause ot me mgn tubeside pres--.,ure drop.

A single-pass design wa~ prepared next. Since the tubeside heat-transfer coefticient was signilican~y lower. the heat-transfer area required was much



larger. The number of tube rows in-creased from five to six. Note that the penalty for the single-pass design (as will be discussed later) has not been in-cluded. so the actual heat-transfer area will be even larger.

Since the design with 25-mm 0.0. tubes and one pass was uneconomical. the next design considered used 32-mm O.D .. 2.5-mm thick tubes. and two tube passes. This design was much more economical. as the heat-transfer area was appreciably less.

Fin height The usual fin heights are % in .. 1/~

in.. and 'IX in. The selection of tin height depends on the relative values of the airside and the tubeside heat-trans-fer coefficients. Where the airside heat-transfer coefficient is controlling t that is. it is the major resistance to heat transfer). a larger fin height ( 15.875 mm) will usually result in a better de-sign. If. however. the tubeside heat-transfer coefficient is controlling. it will be prudent to use a smaller tin height t 12.7 mm).

A higher tin height leads to a greater efticiency 1!1 converting pres--,ure drop to heat transfer on the air-,ide. However. \\ ith higher tins. fe\\ er tube' can he accommodated per row for the same bundle width.

In cases where the airside heat-trans-fer coefficient is controlling.

mized. Any increase in the number of tube passes results in the tubeside pres-sure drop exceeding the allowable limit. For example. by switching from 1/c-in.-high to %-in.-high fins. the num-ber of tubes c:m be reduced by about I 0%. thereby leading to roughly an 8'7c increase in the tubeside heat-transfer coefficient.

Example 6 (tubeside heat-trans-fer coefficient controlling). A \acu-um-column bottoms cooler had to be designed for a refinery hydrocracker unit. The principal process parameters were: flow rate= 93.278 kg/h (84.798 kg/h x 1.1 ). inlet temperature = 180C. outlet temperature = 70C. inlet viscos-ity = 4.2 cP. outlet viscosity = 15 cP. heat duty 5.764.000 kcal/h ( 5.240.()()() x 1.1 ). allowable pressure drop= 1.6 kg/cm2 fouling resistance = 0.0006 hm2'C/kcal. and design ambi-ent temperature = 42C. The specified tube size was 25 mm O.D .. 2.5 mm thick. and 12.500 mm long: the tube matetial was carbon steel.

Because the tubeside \ i;;co-,it\ ''a' rather high. the tubeside heat-transfer coefficient was controlling. ThereiPre.

~~ design \\a-, made '' ith tube, ha\ mg -in.-high tins and 27(1 finsim 1 the !t ,].

!11\\ 1ng sed ion on tin ;,pacing cxpLur1s \\ hy the maximum tin density 11I -LI.~ fins/m \\as not used 1. as detailed 111 Table 5.

The airside heat-transfer n.?'-i'tallL"L' \\as only 12.2'{ uf the tutal resJst~ulLe. ''here as the tubes ide re,istanLe \\ ~~' ~2.-lWi.

:'-Jext an altemati\'e design '' ~~' !'rL'-parcd \\ith tubes ha\ing --1n.-h1gh tins. The number uf tubes per n ''' '' ~~s reduced from 50 tu -t'i so that the bun-dle '' idth would be thl' same ~ts 111 the

\lrigin:.~l !the tube pitch in the 'L'L"l>!ld design \\as 6 7 mm. \ ' 60 mm ll 1r tllL' design \\ith 1 :-in.-high tins!. The ~m;,ide flow rate was ~tdjusted to J {"'lh() (\()() t:1/h t(' !P:~int_:~i!"' th~-. '-:IT""~"'~"' power consumption. The tuheside pre;,--,ure drop. 1.~5 kg/cm 2 was excessi\e.

Reducing the number of tube passes to nine resulted in the design becoming undersurfaced by 3.~4'/r. L"nfonunate-ly. the number of tubes per fO\\ could

not be increased. as the bundle width was already at the permitted maximum of 3.2 m. As expected (since the airs ide was not controlling). increasing the air flow rate made no significant improve-ment in performance.

Therefore. the only alternative was to increase the number of tube rows from eight to nine. The number of tube passes was kept at ten (two in the first row and one in each of the subsequent seven rows) and the air tlow rate was reduced from 1.660.000 to 1.600.000 k?fh to have the same airside power consumption.

This design was acceptable. as it was 5.7'7r oversurfaced and the tube-side pressure drop was !.39 kg/em:. However. while the hare-tube area \\a~ only marginally higher than that of the original design 1 I .55-1- m2 \ s. 1 . ."32 me). the finned area was considerabl;. higher 12-1-.037 m' '' 17.504 111:1. Consequently. the original design was superior.

!:in spacing The lugic described ~tho\ e for tin

height 1, ~tpplicahlc here. t


Table 6. Variation in airside heat-transfer coefficient and pressure drop with fin density (Example 7).

11 fins/in. 10 fins/in. 433fins/m 394fins/m

Airside pressure drop, mm H20 12.219 11.63 Step-wise reduction in air pressure drop, % - 4.8 Airside heat-transfer coefficient (based on

bare-tube area), kcal!hm2.C 961.0 883.9 Step-wise reduction in airside heat-transfer coefficient, % - 8.0 Power consumption, kW Step-wise reduction in power consumption,%

ble increase in the overall heat-transfer coefficient. so the increased power consumption will not be justified.

One might argue that instead of using a lower fin density in such cases. a lower air mass-flow rate may be used along 11 ith a higher tin density so that the power consumption is the same. Hm1 e1 cr. the lower air !low rate 11 ill result in an increase in the outlet air temperature and therefore a reduction in the \lTD. This will result in :Jn Jnle-rior design.

Numoer of tube rows The minimum number of tube

nm' recommended to establish a proper air !low pattern is four. al-though three tube rows can abo be used in exceptional circumstances 11hilc ma111taining a higher Olerdc-sign. \lost :\CHEs halT four to six tube nms. ;dthough eight or ewn ten m11s may he used occasionally.

The ad1antagc of using more tube rows 1s that more heat-transfer area can he aLclllllllHlllatcd in. the same bundle 11 idth. Thi.-. lcaJs to a reduction in the number of tube bunJlcs and scctiom .. thereby rcJucing the plot area re4uired fnr thL' .. \CHE and thus the cost.

The Jisadvantage is that in addition to increasing the fan horsepower (for the same air 1 elocity ). it also reduces the airside flow area for a given heat-transfer area and thereby the air flow rate ihelf. thus lowering the MTD. In

34.93 33.68 - 3.58

fact. this is one of the fascinating as-pects of ACHE design - even the coolant flow rate is variable and has to be optimized. For water-cooled heat exchangers. the cooling water flow rate is virtually fixed, depending upon its inlet temperature and maximum outlet temperature.

The airside pressure drop increases directly with the number of tube rows

~md so does the heat-transfer area. Ho}vewr. a.s the number of tube row.s increases. the airside 1elocity will have to be reduced for the same fan horse-power. and this will translate into a lower airside heat-transfer coefficient. This is why the number of tube rows has to be optimized carefully. For sim-plicity. this may he summarized by stating that the larger the number of tube rmvs. the lower will be the airside heat-transfer coctticient.

It follows. then. that when the tube-side heat-transfer cocfticient is control-ling. a larger number of tube rows may be used because the lower airsidc heat-transfer coefticient 11 ill not adversely a/teet the Jesign hince it is not con-trolling). However. when the airsidc heat-transfer cocfticient is controlling. the number of tube rows wi II have to be limited to a lower value for an opti-mum design.

Thus. for services such as con-densers. water coolers. and light hydro-carbon litjuid coolers having a high tubeside heat -transfer coefficient. the


9 fins/in. 8 fins/in. 7 fins/in. 354fins/m 315 fins/m Z76fins/m

11.039 10.472 9.911 5.1 5.1 5.4

804.8 727.6 650.2 9.1 9.6 10.6

32.43 31.22 30.03 3.71 3.73 3.81

optimum number of tube passes is like-ly to be four or five. However. for gas coolers and viscous liquid coolers. the optimum number of tube rows is likely to be six. seven. or even higher. with the maximum generally being eight to ten. This is because with a larger number of tube rows. the airside heat-transfer coef-ficient will be inordinately low tdue to the low 1elocity sustainable for a nor-mal power consumption 1 or the airs ide pressure drop \\ill he inordinatelv hic-h 1 if a normal airs ide 1 clocity ts used 1.

Fan power consumption Fan power 1aries directly as the vol-

umetric air flow rate and the pressure drop; the 1olumetric air flow rate varies directly as the air mass velocity. while the pressure drop 1arics to the 1.75 power of air mass 1 elocity. Thus. air pressure drop 1arie" to the 1.75 power and fan horsepower to the 2. 75 power

I \

i I r I

air mass velocity will depend upon the extent to which the airside heat-trans-fer coefficient is controlling. In a situa-tion where the airside heat-transfer co-efficient is highly controlling, the opti-mum air mass velocity will be higher than in a situation where the airside heat-transfer coefficient is much less controlling.

Example 8. For convenience. the tube metal resistance (which is negligi-ble) will be ignored. and the tubeside fouling resistance and tubeside film re-sistance will be combined into a single term called the tubeside resistance.

Let us first look at what happens when the airside is controlling. The air-side heat-transfer coefficient (htc0 ) = 800 kcal/hm2C and the tubeside heat-transfer coefficient (/Jtc,) = 4.000 kcal/hm2 0 C: so. the overall heat-transfer coefficient ( U) 66 7 kcal/hm2'C. If the air flow rate is in-creased by 25S'T. htc,, will increase to 894.4 kcalfhm2"C. so that U will in-crease to 731 kcal/hm2 0 C. a differ-ence of 9.69C. This has been achie\'ed at the expense of a ( 1.25)1 7'. or 4X'7r.

increase in air pressure drop and a (1.25)2 75. or 85%. increase in fan power consumption.

Now let's consider the case where the tubeside is controlling. with htca = 800. htc, = 250. and U = 190.5. all in units of kcal/hm2 0 C. The same 25o/c increase in air flow rate will again cause htca to increase to 894.4 kcallhm2 0 C. Here. U will increase from 190.5 to 195.4 kcal/hm2'C. an increase of only 2.6%. As earlier. the airside pressure drop will increase by 48% and the fan power consumption by 859c.

Thus. for the same increase in pres-~ure drop and fan power consumption. the increase in the overall heat-transfer coefficient is 9.6% when the airside is controlling but only 2.69c \Vhen the tubeside IS controlling. Therefore. while this increase in air flow may be justifiable when the airside controls (depending upon the actual heat-trans-fer area. :1ir pressure drop. and fan power consumption). it is certainly un-likely to he justifiable when the tube-side controls.

lOf----------~~L---~-7+---r---~--~~ 9f----------+~----~F-+---r---~~_,~ 8~--------~L---~F-~--~~--~~~ 7~------~~----~---+--~~--~~~ sr-------~~----~---+--~~--~~~ 5 f-----i:-+--+

! i I

I 2 3 4 5 6 8 9 10

Air Mass Velocity

Figure 11. Variation of airside heat-transfer coefficient. preswre drop, and power consumption with air mass l'eiocity.

Tube pitch Although tubes can be laid out in

either a staggered or an in-line pattern. the former is almost invariably em-ployed because it results in much better performance in terms of conversion of pressure drop to heat transfer. Tube pitch has a \ery profound effect on air-side performance. The transverse pitch is more crucial and is what is implied by the term "tube pitch ... The longitu-dinal pitch has much less influence and is usually 80--90\'c of the transverse pitch. in order to minimize the height of the tube bundles :1nd thereby the cost.

Designers tend to usc the following standard combinations of bare-tube O.D .. finned-tube O.D .. and tube pitch. as they tend to he optimum:

l-in./2-in./2.375-in. (25-mm/50-rnm/60-mm)

l-in./2 . .:!5-in./2.625-in. (25-mm/57 -rnm/6 7 -mm)

However. in man: 'ituations. these may not be the opurnum. So. the tuhe pitch should he \ ~med ami the opti-mum c~tahli-,heJ lor c~tc:h application.

The nonnal ran!2e llf tube pitch for the 1-in./2-in. 1 hare-tube 0.0./tinned-tube O.D.I combination i~ 2.125 in. to 2.5 in. Similarly. for a l-in./2.25-in. combination. the normal range of tuhe pitch i~ 2.'375 in. to 2.75 in.

At a relatiwly ll)\\er tube pitch. the air pressure drop ~tnd therefore the power con-,umption tend to he high for the ~arne airside heat-transfer coefti-cicnt. In other \\llrd>. a, the tuhe pitch i., decreased. the ~ur,ide pressure drop and power consumption increase more rapidly than dues the airsidc: heat-trans-fer coefticient.

This i~ illwMated b\ the results of a study >ummari7ed in Table 7. The tubes were 1-in.-O.D. tubes with 1/c-in.-iugii tin,; t'uut tui>e f.Jildte' -2.125 in .. 2.25 in .. 2.'375 in .. and 2.5 in. - and four tin densities - 5 fins/in. (fpi). 7 fpi. 9 fpi. and 11 fpi - were evaluated. The combinC)tion of 2.625 in. tube pitch and I I fpi \\as consid-

CHEMICAl ENGINEERING PROGRESS FEBRUARY 1997 41


-

Table 7. Effect of tube pitch and fin density on heat-transfer coefficient (HTC) and pressure drop (PO).

5 fins/in. 197 fins/m

PO Tube Pitch= 2.125 in. 74.2 Tube Pitch = 2.25 in. 65.9 Tube Pitch= 2.375 in. 59.1 Tube Pitch= 2.5 in. 53.8

erect the base case (since this a very common configuration). and 100 units were assigned for both heat-transfer coefficient ( HTC) and pressure drop IPD). The performances of the other combinations. in terms of heat-transfer coefficient and pressure drop, were de-termined relative to the base case.

As expected. the reduction in pres-sure drop was much sharper than the reduction in heat-transfer coefticient. Typically. for each step change in tube pitch. the pressure drop decreased hy about I Wlr. whereas the heat-transfer coefticicnt decreased hy only about 2';. This panern \\a.'> esscntiall~ the 'ame for all the various tin densities.

The designer may. therefore. kel that the higher the tuhe pitch. the better the performance of an ACHE. \\'hile this is basically true. there is another extremely imponant factor that should not he ()\erlool\ed: As the tuhe ritch is increased t(Jr a giwn number of tubes. the tuhe bundle \\ idth \\ill increase (even the fan diameter \\ill increase at a cenain point!. thus pushing up the L'Ost of the ,\CHE. Consequently. the tuhe pitch should he optimized in the 11\'Cm// L'lllltext.

While the optimum tuhL' pitd1 may vary from situation to situation. design-ers typically prefer to w.e a 2 . .:n5-in. tube pitch for a 1-in./2-in. combination.

. . l '-'' llll.> eO:IIC:l.lli_)' 1Lf)l

Table 9. Process parameters, and construction and performance details, for Example 10,

a combined service.

ACHE#1 ACHE#2 ACHEI3 Stripped Light Stripped Heavy Circulating Heavy

Service Cycle Oil Cooler Naphtha Cooler Naphtha Cooler

Flow rate, kg/h 48,949 Heat duty, MM kcaVh 2.48 Inlet temperature, c 210 Outlet temperature, c 66 Operating pressure, kg/cm2 gabs. 9.8 Allowable pressure drop, kg/cm2 0.7 Fouling resistance, hm:z.ctkcal 0.0004 Inlet viscosity, c P 0.4 Outlet viscosity, cP 2.66 Number oftubes per row 48 Number of rows 6 Number of tube passes 6 Approximate bundle width, m 3.0 Air flow rate, kg/h 370,700 Air inlet temperature, c 42 Air outlet temperature, c 69.9 StatiC pressure, mm w.g. 13.0 Tubeside pressure drop, kg/cml 0.7 Mean temperature difference, c 52.8 Heat-transfer area I bare). m1

'ure drop in the various passes than would he possible in an even distribu-tion of 1.5 tube rows in all passes. This re.,ults 111 a higher overall pres-'ure-dnlpt(l-heat -transfer conversi\lll. which in turn results in a lower heat-transfer area.

235

This can he restated as follm\;: If the number ot tubes per pass \\ere to he identical in all the passes. the initial passes 1where there is more \apor and consequently a lower density) will have a higher pressure drop \Vithout a corresponding!~ higher heat-transfer coetticient. The subsequent tube passes will hme much lower pressure drops. The design will. therefore. not be opti-mum with respect to the best utiliza-tion of available pressure drop.

Example 9. An atmospheric-col-umn overhead condenser using air as the cooling medium was to be de-

19,216 74,114 1.02 1.64 165 177 66 138 9.4 8.8 0.5 0.7

0.0003 0.0003 0.3 0.28 0.7 0.36 15 12 6 6 4 2

1.0 0.8 115,800 88,700

42 42 72.6 119.0 13.0 13.0 0.2 0.6 50.9 73.9 73.5 I 58.8 I

'igned for the crude distillation unit of a refinery based on the following pa-rarnetep,: heat duty= A7.98 MM kcal/h 156.A5 x a safetv factor of 1.2 l. over-head \apor t1nw rate = 501.7-W kg/h 1418.120 kg/h x 1.2) !hydrocarbon = 95.8 'r\\t. and steam= 4.2


to be used. as the ACHEs were to be mounted on a 10-m-wide piperack. Since the heat duties were rather small. a combined design wa~ prepared wherein three tube bundles (one for each service) were combined in a sin-gle section. using G-llnned tubes with 12.5-mm-high aluminum tins. ~paced at 43 tins/m. at a transverse pitch of 60 mm, as elaborated in Table 9. Two fans. 3.657 m in diameter. were select-ed to force the total air flow rate of 575.200 kg/h against a static pressure of 13 mm w.g .. consuming 16...+ kW each: two motors of 18.75 kW each were selected to drive the fans. Split headers were used to handle all three bundles. since all of them had design temperatures in excess of 200-c and the first two had a high temperature difference (between the inlet and the outlet temperatures of the process fluid) as well.

It may be noted that the air outlet temperature from ACHE #3 was rather high. at li9C. This \\a-, a di-rect consequence of the high \lTD. which required a relatively 'mall heat-transfer area and therefore :1 rather 'mall face area. through \\hich '1nl;. :1 rciati\cl\ low tlow rate of air Lould be

!'J.\~etJ. TI1e di-,ad\antage in combining dif-

ferent scniccs in one \ection i-., the loss of individual control. Consequently. this practice is nnt recommended for condcn .... crs. but is limited lt1 product coolers.

Recucuiatton ACHEs There an: situations where :1 1111111-

murn tubcwall temperature !llU't he maintained . .-\n example i' the O\ cr-head Londenscr for a -,our \\ ~llcr 'trip-per. where if the tuhewall temperature falls below 70 C. solidification fouling takes place due to the deposition of ammonium -,alts. Since the ambient temperature varies from dav to night and from season to season. a special arrangement is needed to maintain a constant air temperature (the design ambient temperature) at all times to the fans delivering air to the condenser tube bundles (forced draft).

I I

h~l.,..-r-n--=-:--:::'-~-r--r-rJ,.-~~~ Bypass Louver

{

Figure I 2. A.utomatic louvres on recirculation ACHEs maimain a constant air temperature to the fans.

Such a situation can employ a recir-culation ACHE. wherein automatic lou-vers at the top and side( s) of the hous-ing control the extent of recirculation. As shown in Figure 12. a pan of the air leaving from the top of the tube bundles is recirculated and mi.xcs \\ ith fre'h :ur entering from the 'ide( s 1. -,o that the combined temperature j.., prcn,ely the design ambient temperature. Obviously. the lower the ambient temperature. the greater will be the extent of recircula-tion. Automatically controlled lou\ ers (sensing the atr temperature just below the fans 1 at both the top and the 'ide( s I ensure the desired air temperature ~tt all times. TI1is ammgement can ~li"l he used for cooling hean c-rude 'tock.., with high pour poinh.

One difficulty 1\ith recirculation ACHEs \1 ill he startup during 11 inter. or even summer c\enings if the ambi-ent temperature fall!, below the design ambient temperature. This problem Lan be overcome by placing 'team Loils below the tube bundles - the cold am-bient air is warmed up to the design ambient temperature by passing low-pressure steam through the coils and having total air recirculation until the air temperature reaches the design am-bient temperature. The steam supply

can then be stopped. as the ACHE will be able to take care of it-;elf.

'-iumtdified :~CHEs Normal air-cooled hcut exchangers

Lan cool a pnlccs' fluid unly to a tem-perature that i, htghcr than the dry-bulb tt:mperature ol the ambient .ur. Therefore. in hot and area'>. the utility ,1t' air cooling is _,e1erely limited.

However. hy humidifying the hot dry air. the temperature can he brought down considerably. thereby extending the capabilit;. of air cooling 'igniticantly.

Humidified ACHE, arc nonnallv of the induced-draft ty pc. \Vater IS 'prayed into the air 'trcam before it L'tHnes in contact \\ ith the heat ex-changer sllli'ace. thereby lo\\ering the air temperature due to e\aporation. \1ist eliminators arc needed tn prevent droplets of water from entering the tube hundle. The \\:ncr that does not evaporate is collected in a basin be-neath the cooler and recirculated.

The installed cost of humidified ACHEs is rather high. And. the cost of makeup water represents an additional expense. Another limitation of humidi-fied air cooling is that soft ~ater must be used. otherwise there will be scaling


of the tinned tubes with consequent de-terioration in cooler performance.

However. since humidification en-hances the cooling capability consid-erably. a proper economic study will have to be carried out to establish the viability of this type of exchanger. It is important to remember that in arid areas. the scarcity of water may ne-cessitate this mode of cooling for ser-vices with streams that must be cooled to temperatures below summer day temperatures.

An important factor to consider is that humidification of the coolant air will be required only during the day-time in the summer. which may repre-sent only 4--5% of the total operation of the equipment. Certainly. the selec-tion of the design ambient temperature will have to be done very carefully.

Tube inserts Tube inserts considerably enhance

the tubeside heat-transfer coefficient under laminar flow conditions. The tirst-generation tube inserts were twist-ed tapes that imparted a swirling mo-tion to the tubesidc fluid. thereby aug-menting heat tri1311 ust>d carhnn steel tubes that measured I in. 0.0. x 12 BWG (minimum. under groove). Due to the rather high viscosity. laminar flow in all three coolers resulted in extremely low tubeside heat-transfer coefficients. Consequently. the required plot area was

Table 10. Process parameters and thermal design for Example 11, designing three coolers for an offshore platfor~

'~

ACHEI1 ACHEI2 ACHE13 Auid handled Synthetic oil Turbine oil Turbine oil

Flow rate, kg/h 2,074 Inlet temperature, c 74 Outlet temperature, c 54 Operating pressure, kg/cm2 abs. 6.2 Allowable pressure drop, kg/cm2 0.7 Fouling resistance, hm:z.C/kcal 0.0004 Heat duty, kcal/h 19,030 Inlet viscosity, cP 8.3 Outlet viscosity, cP 14 Number of tube bundles 1 Number of tubes per row 2 Number of tube rows 7 Bundle width, m 0.2 Number of tube passes 2 Tubeside pressure drop, kg/cm2 0.7 Heat-transfer area (bare). m2 7.2 Air flow rate, kg/h 13,000 Inlet air temperature, oc 40 Outlet air temperature, c 46.1 Mean temperature difference, c 19.9 Overall heat-transfer coefficient.

kcal/hmzoc !

considerably higher than the allowable \ alue. and th~ tubes ide pressure drops exceeded th~ allowable limits.

Theret\1re. rL'\i.sed designo, \\L'r~ :il-temptcd incorp.H:tting wir~-tin insens in tubes that were 6.:\ Ill long and had 1 :-in.-high aluminum tins spaced at .fJ.~ lin-Jm. .\ combined section using one tube bun-dle for each sef\icc could be ~:.t~ilv ac-commlxlated \\ ithin the spccitied plot :trea 1 the actual plot area \\ :L~ 2.0 I m x 6.5 mi. ll1c katures l 1f the thermal design are detailed in Table I 0. Three fans. I j2-l rn in dinmt>ter. were emploved to drive the total air flow of 188.530 kg!h across the three tube bundles.

It should be highlighted here that the overall heat-transfer coefficients of LB. 207. and 261 kcal/hmcC ob-tained with wire-tin tube inserts are

133

7,438 35,449 66 75 54 54

46.0 7.0 1.0 1.0

0.0004 0.0004 44,440 373,670

10 9.2 21 21 1 1 4 23 7 7

0.34 1.47 1 1

0.91 1.0 14.3 82.3

26,000 149,530 40 40

47.1 50.4 15.0 17.4

207 261 I

llrder-of-magnitudc times higher than what can he achie\ed with bare tubes.

.'an.JiJie rinnmg l1ensity Variable tinning density can be usdul

for the cooling of liquids with \'iscosity that L'hanges considerably from inlet to tlUtlet. With the incrc:.tsc in viscosity. th~ tubeside heat -tr


Consequently. when cooling a viscous liquid. it is often advanta-geous to have a higher tinning densi-ty in the upper rows and a lower tinning density in the lower rows. The degree of overdesign may be slightly less. but there will be an ap-preciable savings in power consump-tion. Furthermore. the greater the variation in viscosity of the tubeside liquid. the greater is the potential for this type of variable tinning.

The vacuum-column bottoms cooler discussed in Example 6 earlier had two sections. each having two tube bun-dles. and each bundle having 50 tubes per row. eight rows. and ten tube pass-es. The fin density was 276 fins/m. The effect of alternative tinning can be evaluated by keeping the fin density at 276 tins/m in the top six tube rows and changing it to 197 fins/m in the lower two tube rows. The fan power require-ment drops from 18.92 kW to 18.4 kW.

Literature Cited 1. American Petroleum Institute, "Air-

cooled Heat Exchangers for General Refin-ery Services," API Standard 661, 3rd ed., American Petroleum Institute. Wa~hington. DC (Apr. 1992).

2. Rhoades, M. E-, and L. D. Paul, corro-sion Resistance of Aluminum and Galva-nized Fins on Air-Cooled Exchanger Tub-ing," in "Heat Transfer - Atlanta 1993," AIChE Symposium Series, 89 (295), pp. 268-273 (1993).

3. Taborek, J., "Bond Resistance and Design Temperatures for High-Finned Tubes - A Reappraisal.'' Heal Transfer Em;ineering, 8 i2l. pp. 2&-3..\ 11987\.

..). :\lonroe. R. C., c,msider Variable Pitch Fans: Hrdrocarhmz Processing. 59 I 121. pp. 122-128iDec. 1980).

5. American Petroleum Institute, "General Purpose Steam Turbines for Refinery Ser-vice." API Standard 611. 3rd ed .. American Petroleum Institute. Washington. DC (Au-gust 1988. reartlrrned May 1991 ).

6. McKetta. J. J .. ed .. "Heat Transfer Design Methods." ;\larcel Dekker. :'-lew York (1992).

7. Zukauskas, A.. "Air-cooled Heat Ex-changers: in "Heat Exchangers: Therrnal -Hydraulic Fundamentals and Design," S. Kakac. eta/ .. eds .. Hemisphere Publishing. New York. pp . ..\9-83 ( 1981 ).

8. Briggs, D. E .. and E. H. Young, "Convec-tion Heat Transfer and Pressure Drop of Air Flowing Across Triangular Pitch Banks or t'mnea i uoes. m "tleat 1 ranster -Houston." .4./CizE Symposium Series. 59 (41), pp. 1-10 (!963).

9. Robinson, K. K., and D. E. Briggs, "Pres-sure Drop of Air Flowing Across Triangu-

Jar Pitch Banks of Finned Tubes." in "Heat Transfer - Los Angeles," .4./ChE Sympo-sium Series, 62 (64). pp. 177-184 (1966).

10. Webb, R. L., "Airside Heat Transfer in Finned Tube Exchangers." Heal Exchange Engineering. l (3). pp. 34-49 (Jan.-Mar. 1980).

11. Thbular Exchanger Manufacturers As-sociation, "Standards of the Tubular Ex-changer Manufacturers Association." 7th ed., TEMA. New York f 1988).

12. Kern, D. Q., and A. D. Kraus, "Extended Surface Heat Transfer." ;\1cGraw-Hill.

NewYorki197~l. 13. Farrant. P. E., ";\Ioise and its Influence on

Air Cooled Heat Exchanger Design: ;n "Heat Exchanger., - Theory and Prac-tice ... J. Taborek. et a/ .. eds .. Hemisphere Publishing, 'lew York. pp. 383-390 (! 983). 1~. Paikert. P . and K. Ruff. '"State of the An

for Design of Air Cooled Heat Exchangers with Noise Level Limitation: in "Heat Ex-changers - Theory ;1nd Practice: J. Ta-borek. cl a/ .. eds .. Hemisphere Publishing. New York. pp. 383-396 I 1983).

15. Sallenbach. H. G . "Save Energy - L\e Natural Convecrzon Air-Cooled Exchang-ers." Hrdmcarhon Pmcessmg. pp. 79-86 (July 19831

16. Rozenman, .1 .. et at .. "The Effect of l'n-equal Heat Loads on rhe Perfonnance of A.:-cooled Conden'>ers." in Heat Transfer - Research and Design." .4./ChE Sympo szum ::.erzes. ;u 1 IJil). pp. 1 ill-1M (1':1/3).

17. LarinoiT, l\1. W., et aL, "Design and Spec ification of Air-cooled Steam Condensers." Chern. Eng., 85 (!2). pp. 86-94 (May 22,1978).


whereas the degree of overdesign falls from 8.85% to 8.25%. Switching to 276 fins/m in the upper four rows and 197 fins/m in the lower four rows re-duces the fan power requirement fur-ther to 17.88 k W and reduces the overdesign to 7.56%. Thus. for a 1.2% reduction in overdesign. the power consumption is cut by 5.817c.

Natural convection In cases where the heat duty is very

small and the tubeside heat-transfer co-efficient is rather low. no fans need be used. so that power consumption is saved ( 15 ). This is illustrated by the following example.

Example 12. A vacuum residue cooler for a refinery was to be designed based on the following data: heat duty = 476.400 kcaUh. flow rate = 17,604 kg/h. inlet temperature= 267C. outlet temperature = 220C. inlet viscosity = 18 cP. outlet viscosity = 25.3 cP. foul-ing resistance = 0.002 hm="Cikcal. and allowable pressure drop = 1.0 kgJcm2. C;rrbon \lee! tubes. 25 mm O.D. x 2.5 mm thick x o.O m long. were to be used.

.-\. natural-draft ACHE ''as de-'igned a\ follo11-.. It had ,,ne section having one wbe bundle. -to tubes/row. four tube nm s. and eight tube passes. Fin height ''as 12.5 mm. transverse tube pitch 1\as 60 mm. and tin density was -BJ tins/m. A 3-m-high \tack was incorporated.

The tuheside and airside heat-transfer Loeft!cients 1wre 61.53 1--cal/hm: C and 239.8 !.-cal/11m: C representing 70.-t

Figure 13. Wire-fin tube inserts improve heat transfer by increasing turbulence. (Photo courtesy of Cal Gavin, Ltd.)

have resulted in a negligible increase in the overall heat-transfer coefficient. However. the cost would have been sig-nificantly higher because of the costs for the fans, drives. plenums. and so on.

Air-cooled vacuum steam condensers

Steam turbines are very widely used in the chemical process industries for driving not only electric generators. but also various types of pumps. fans. compressors. and other equipment. Steam condensers are required to con-dense the exhaust of these turbines :.~nd return the condensate to the boiler. This condenser can enher be \\ater-cooled or air-cooled. (The advantage~ and disadvantages of each are the same as elaborated at the beginning of the article.) A typical A-frame air-cooled vacuum steam condenser ( ACVSC) i~ shown in Figure 9.

The main problem with the ACVSC is not the condensation of the steam. but the evacuation of the noncondens-ables. (The noncondensables are the gases that enter the vacuum section of the power cycle from the atmosphere as well as from the chemicals used for the treatment of boiler feedwater.l Fail-

Acknowledgment The author is grateful to the management of Engineers India Ltd. for permission to pub-lish this article and acknowledges the use of Heat Transfer Research, Inc.,'s ACE com-puter program for the worked-out examples.

ure to eliminate the noncondensables can cause freezing of condensate in winter. loss of performance due to blanketing of the heat-transfer surface. and absorption of noncondensables by condensate and subsequent corrosion of tube metal. Thus. a successful ACVSC must continuously and totally collect and eliminate all noncondens-ables from the system.

The trapping of noncondensables inside the condenser tubes is a direct consequence of the variation of coolant air temperature across the tube bundle. Consider a single-pass condenser hav-ing four tube rows. The tubes of the lowest row are expm;ed to the coldest air while the tubes of the upper rows are exposed to progressively hotter air. Therefore. the tubes in the lowest row condense more steam (due to the high-er MTDl. while those in the upper rows condense less and less steam. Consequently. the pressure drop \\ill be the highest in the tubes in the lowest row and progressively lower in the tubes of the upper rows. ll1is \\ill cause a back flow 1 lf noncondenqbles from the tubes of the upper rows to the tubes of the lowest row. which can eventually lead to gas blanketing of a substantial fraction of the heat-transfer surface. This problem \viii be less acute in. but not absent from. a 2-passt

air cooler design sheiko

Documents