United States Air and Energy EngineeringEnvironmental Protection Research LaboratoryAgency Research Triangle Park, NC 27711

Research and Development EPA/600/SR-94/208 March 1995

E AP

NOX Control TechnologiesApplicable to Municipal WasteCombustion

Several technologies are available forreducing nitrogen oxide (NO X) emis-sions from municipal waste combus-tors (MWCs), including combustioncontrols, natural gas injection (NGI),selective non-catalytic reduction(SNCR), and selective catalytic reduc-tion (SCR). The full report documentsthe key design and operating param-eters, commercial status, demonstratedperformance, and cost of NGI, SNCR,and SCR, and identifies technology re-search and development needs associ-ated with them.

Two NGI processes have been devel-oped: (1) Methane de-NOX SM uses gasinjection to inhibit NO X formation andappears capable of reducing NO X emis-sions from MWCs by approximately60% and (2) reburning uses gas injec-tion to create reducing conditions thatconvert NO X formed in the primary com-bustion zone to molecular nitrogen.Because of the relatively high tempera-tures required for these NO X-reductionreactions, it may be difficult to suc-cessfully apply reburning to modernmass-burn waterwall MWCs. Long-termemission reductions are 45-65% forSNCR and 80-90% for SCR. Operationof SNCR processes near the upper endof their performance range can resultin unwanted emissions of ammonia orother by-product gases. An advancedversion of SNCR using furnace pyrom-etry and additional process controlsappears capable of achieving high NO Xreductions with less reagent than isneeded for conventional SNCR. Thecombination of NGI and SNCR (ad-

Project Summary

D. M. White, K. L. Nebel, M. Gundappa, and K. R. Ferry

vanced NGI) may be able to achieveoverall NO X reductions of 80-85%.

Comparing costs, SCR is the mostcapital intensive, followed by advancedSNCR and advanced NGI. Capital costsof NGI and conventional SNCR are com-parable. In terms of tipping fee impactand cost effectiveness, conventionalSNCR generally has the lowest costsof the evaluated technologies. For NGI,these costs depend on whether wasteis diverted and tipping fee revenuesare lost when applying this technol-ogy, along with the price of naturalgas. Depending on the selected NGIscenario, the resulting tipping fee im-pacts and cost effectiveness values canbe the highest of the evaluated tech-nologies. After specific NGI scenarios,the next highest tipping fee impactsand cost effectiveness values are forSCR. These high costs result from highcapital costs, as well as the cost ofcatalyst replacement and disposal.

This Project Summary was developedby EPA's Air and Energy EngineeringResearch Laboratory, Research Tri-angle Park, NC, to announce key find-ings of the research project that is fullydocumented in a separate report of thesame title (see Project Report orderinginformation at back).

IntroductionNitrogen oxides (NOX) are of environ-

mental significance because of their roleas a criteria pollutant, acid gas, and ozoneprecursor. The current New Source Per-formance Standards (NSPS) for municipalwaste combustors (MWCs) (40 CFR Part

2

60, Subpart Ea) limit NOX emissions to adaily average of 180 parts per million (ppm)at 7% oxygen (O2), dry basis.* By com-parison, typical NOX emissions from mod-ern mass-burn waterwall (MB/WW) MWCsrange from 220 to 320 ppm. To complywith the NSPS, most recently built MWCshave used a combination of combustioncontrols to limit NOX formation and selec-tive non-catalytic reduction (SNCR) to con-vert NOX to molecular nitrogen (N2). Be-cause of pressure to achieve even loweremission levels, questions have beenraised regarding the potential for advance-ments in NOX control technologies. To re-spond to these questions, the U.S. Envi-ronmental Protection Agency's (EPA's) Airand Energy Engineering Research Labo-ratory and the Department of Energy'sNational Renewable Energy Laboratory ini-tiated this assessment of three alternativeNO

X control technologies: natural gas in-jection (NGI), SNCR, and selective cata-lytic reduction (SCR). The objectives ofthe assessment were to (1) document thekey design and operating parameters,commercial status, demonstrated perfor-mance, and cost of each technology and(2) identify technology research and de-velopment needs.

The assessment of achievable NOX

emissions presented by the report is basedon the average NOX reduction potential ofthese technologies applied to “typical”MWCs. The assessment does not exam-ine the potential severity or length of short-duration excursions in performance thatcan affect continuously achievable NOXemission rates associated with short aver-aging periods. The assessment also doesnot consider potential limitations on tech-nology performance that may result fromcombustor-specific design or operating re-strictions.

NOX Formation

The chemistry of NOX formation is di-rectly tied to reactions between nitrogenand O2. To understand NOX formation inan MWC, a basic understanding of com-bustor design and operation is useful.Combustion air systems in MB/WW MWCsinclude both undergrate (also called pri-mary) air and overgrate (also called sec-ondary or overfire) air. Undergrate air, sup-plied through plenums located under thefiring grate, is forced through the grate tosequentially dry (evolve water), devolatilize(evolve volatile hydrocarbons), and burnout (oxidize nonvolatile hydrocarbons) thewaste bed. The quantity of undergrate air

is adjusted to minimize excess air duringinitial combustion of the waste while maxi-mizing burnout of carbonaceous materialsin the waste bed. Overgrate air, injectedthrough air ports located above the grate,is used to provide turbulent mixing anddestruction of hydrocarbons evolved fromthe waste bed. Overall excess air levelsfor a typical MB/WW MWC are approxi-mately 80% (180% of stoichiometric [i.e.,theoretical] air requirements), withundergrate air accounting for 60-70% ofthe total air. In addition to destruction oforganics, one of the objectives of this“staged” combustion approach is to mini-mize NO

X formation.NOX is formed during combustion

through two primary mechanisms: fuel NOXformation and thermal NOX formation. FuelNOX results from oxidation of organicallybound N2 present in the municipal solidwaste (MSW) stream. Thermal NOX re-sults from oxidation of atmospheric N2.

Fuel NOX is formed within the flamezone through reaction of organically boundN2 in MSW materials and O2. Key vari-ables determining the rate of fuel NOXformation are the availability of O2 withinthe flame zone, the amount of fuel-boundN2, and the chemical structure of the N2-containing material. Fuel NOX reactionscan occur at relatively low temperatures[<1,100°C (<2,000°F)]. Depending on theavailability of O2 in the flame, the N2 com-pounds will react to form either N2 or NOX.When the availability of O2 is low, N2 isthe predominant reaction product. If sub-stantial O2 is available, an increased frac-tion of the fuel-bound N2 is converted toNOX. Testing conducted in the 1970s and1980s using coal showed that in O2-rich,highly mixed systems approximately 50%of the fuel-bound N2 can convert to NOX;in O2-starved staged-combustion systems,however, the rate of conversion decreasesto near 5%. Other testing has shown thatN2 associated with volatile compounds ismore readily converted to fuel NOX thanN2 associated with nonvolatile materials.Still other research involving coal and oilcombustion indicates that the extent ofconversion is related to the amount of N2available, with the degree of conversiondecreasing as the amount of fuel-boundN2 increases.

Thermal NOX is formed in high-tempera-ture flame zones through reactions be-tween N2 and O2 radicals. The key vari-ables determining the rate of thermal NOxformation are temperature, the availabilityof O2 and N2, and residence time. The keyreactions resulting in thermal NOX forma-tion are

N2 + O →← NO + N (1)

followed by

N + O2 →← NO + O (2)

N + OH →← NO + H (3)Because of the high activation energy re-quired for Reaction (1), thermal NOx for-mation does not become significant untilflame temperatures reach 1,100°C(2,000°F). Kinetic calculations (assuming30% excess air, average MSW proper-ties, and residence time of 0.5 second)predict thermal NO

X concentrations of <10ppm in MWCs. However, local flame tem-peratures may exceed 1,100°C and ther-mal NOX concentrations may be greaterthan these calculated model results.

Examination of MWC operating condi-tions suggests that most of the NOX emit-ted from MWCs (>80%) is attributable tofuel-bound N2. Based on typical MSW N2contents of 0.3-0.7%, the expected NOXemissions—assuming all of the fuel-boundN2 is converted to NOX—would be 1,000-2,500 ppm at 7% O2. As noted earlier, how-ever, actual emissions are generally be-tween 220 and 320 ppm at 7% O2, indicat-ing that perhaps 10-30% of the fuel N2 isconverted to NOX, with most of the remain-der forming N2.

A number of evaluations of MWC NOXemissions data have attempted to definethe role of N2-containing materials in MSW(e.g., grass, leaves, wood, and foodwastes) on NOX emissions. The first ofthese evaluations was presented in 1987and suggested that fluctuations in mea-sured NOX levels were attributable to sea-sonal fluctuations in MSW composition.This evaluation was based on NOX com-pliance test data obtained from a numberof MWCs in the U.S. and overseas atdifferent times of the year, and concludedthat the seasonal variations in measuredNOX concentrations might be the result ofvariations in the amount of yard waste inthe MSW at different times of the year. Asimilar comparison of NOX emission con-centrations versus time of year for a num-ber of U.S. MWCs compiled by EPA in1989 to support the MWC NSPS did notshow any significant relationship betweenNOX concentration and time of year.

A more recent evaluation of NOX con-tinuous emission monitor data from 11MWCs located in the northeast U.S. foundno seasonal variations in NOX concentra-tions. This evaluation compared monthlyaverage NOX data from the individualMWCs (covering 12-36 consecutivemonths of operation for each unit) withthe estimated fraction of yard waste foundin northeastern MSW during each of thefour seasons. Monthly average NOX con-centrations from these units varied from

* Unless otherwise noted, all NOx concentrations used in

this summary are corrected to 7% O2 and are on a dry

basis.

3

140 to 310 ppm but did not show anyconsistent relationship between NOX con-centrations and the estimated percentageof yard waste.

Yet another evaluation compared NOXconcentrations measured during nine testruns conducted at the MB/WW MWC inBurnaby, British Columbia, to the amountof high-N2 organics (grass, leaves, brush,stumps, wood, food waste, textiles, rub-ber, and footwear) in the MSW fired dur-ing each run. During these tests, high-N2organics accounted for 25-47% and yardwaste accounted for 4-30% of the totalwaste stream. The estimated average N2content of the entire stream during eachrun ranged from 0.34 to 0.66%. NOX con-centrations in the flue gas during the runsvaried from 261 to 304 ppm. Statisticalanalysis of the MSW and flue gas datafrom each run showed no relationship (ata screening 80% confidence level) be-tween NO

X concentrations and MSW char-acteristics. The data suggest that, becauseof the staged-combustion design of theBurnaby MWC and other modern MB/WWMWCs, variations in NOX emissions appearto be attributable to differences in combus-tor design and operation, rather than wastecomposition.

NOX Control Technologies

NOX control technologies can be dividedinto two subgroups: combustion controlsand post-combustion controls. Combus-tion controls limit the formation of NOXduring the combustion process by reduc-ing the availability of O2 within the flameand lowering combustion zone tempera-tures. These technologies include stagedcombustion, low excess air, flue gas recir-culation (FGR), and NGI. Staged combus-tion and low excess air reduce the flow ofundergrate air in order to reduce O

2 avail-ability in the combustion zone. Anotheroption is FGR in which a portion of thecombustor exhaust is returned to the com-bustion air supply to both lower combus-tion zone O2 and suppress flame tem-peratures by reducing the ratio of O2 toinerts [N2 and carbon dioxide (CO2)] in thecombustion air system. One or more ofthese approaches are used by most mod-ern MWCs. Test data for these techniquesindicate that they can reduce NOX con-centrations by 10-30% compared tobaseline levels from the same units. ForNGI, two processes have been developed:(1) Methane de-NOX

SM uses gas injectionto inhibit NOx formation and (2) reburninguses gas injection to create reducing con-ditions that convert NOX formed in theprimary combustion zone to N2.

The most used post-combustion NOXcontrols for MWCs include SNCR andSCR. SNCR reduces NOX to N2 withoutthe use of catalysts. With SNCR, one ormore reducing agents are injected intothe upper furnace of the MWC to reactwith NOX and form N2. SNCR processesinclude Thermal DeNOX

TM, which is basedon ammonia (NH3) injection, NOXOUTTM,which uses urea injection, and the addi-tion of urea followed by methanol. Ther-mal DeNOX and NOXOUT have been usedpredominantly to date.

SCR is an add-on control technologythat catalytically promotes the reactionbetween NH3 and NOX. SCR systems canuse aqueous or anhydrous NH3, with theprimary differences being the size of theNH3 vaporization system and the safetyrequirements.

Technical ApproachThis assessment was conducted using

information obtained from published lit-erature and contacts with technology ven-dor and MWC industry personnel. For eachtechnology, information was collected onthe key process variables; commercial ap-plications in Europe, Japan, and theU.S.; recent research and developmentactivities; and costs. In addition to thecurrent versions of NGI and SNCR, ad-vanced concepts for the two technologieswere examined, including advanced NGI(which combines conventional NGI andSNCR) and advanced SNCR (which em-ploys additional process control equipmentto enhance conventional SNCR perfor-mance). This information was then usedto develop a series of computer-basedspreadsheets designed to maintain basicmaterial and energy balances for the tech-nology and to calculate technology costs.

The output from these spreadsheets ispresented in two formats. The first formatis a table presenting capital costs, tippingfee impacts, and cost effectiveness lev-els** for each NOX control technology asa function of MWC size and a second keytechnology variable. An example of this

output, based on conventional SNCR, isshown in Table 1. For example, the esti-mated capital cost for SNCR operating at60% NO

X reduction on a 400 ton per day(tpd) MWC is $1,980/tpd of capacity. Forthe same NOX reduction level and MWCsize, the tipping fee impact is estimated atapproximately $1.50 per ton of MSW pro-cessed, and the cost effectiveness is ap-proximately $1,270 per ton of NOX re-moved from the flue gas.

The second output format is a graphshowing the sensitivity of tipping fee im-pact and cost effectiveness to key pro-cess variables. An example of this output,also based on conventional SNCR, is pre-sented in Figure 1. For example, the pro-cess variable having the greatest effecton cost is MWC size. Compared to the400 tpd “reference” MWC, reducing plantsize to 100 tpd increases the tipping feeimpact from roughly $1.50 per ton of MSWto almost $4.00 per ton (shown on the leftY-axis), and increases the cost effective-ness value from $1,270 per ton of NO

Xremoved to $3,380 per ton (shown on theright Y-axis).

Technical Status of EvaluatedControl Technologies

The technical status of each evaluatedcontrol technology is summarized in Table2. As noted, two NGI processes have beendeveloped. Methane de-NOX uses gas in-jection to inhibit NOX formation and ap-pears capable of reducing NOX emissionsfrom MWCs by approximately 60%. Thesecond approach, reburning, uses gas in-jection to create reducing conditions thatconvert NOX formed in the primary com-bustion zone to N2. Because of the rela-tively high temperatures required for theseNOX-reduction reactions, it may be difficultto successfully apply reburning to modernMB/WW MWCs. Short-duration tests ofthese processes have been conducted atMWCs in the U.S. and Europe. However,neither process has been adequately de-veloped and demonstrated to be consid-ered ready for commercial applications.

Both SNCR processes (Thermal DeNOX

and NOXOUT) and SCR are considered tobe commercially available. Long-term NOXreductions are 45-65% for SNCR and 80-90% for SCR. Operation of these pro-cesses near the upper end of their perfor-mance range can result in unwanted emis-sions of unreacted NH3 or other by-prod-uct gases. An advanced version of SNCRusing furnace pyrometry and additionalprocess controls has been tested on atleast two MWCs and appears capable ofachieving high NOX reductions with lessreagent than is needed for conventional

**Capital costs include purchased equipment costs,installation, engineering and home office expenses,and process and project contingencies. Tipping feeimpact is calculated by dividing the technology's totalannualized cost by the annual tonnage of MSW pro-cessed. Tipping fee impact is an incremental cost thatindicates the potential cost of the technology on theMSW generator. However, it does not necessarilyreflect the amount by which the plant's tip fee willincrease as a result of applying the control technology.Cost effectiveness is calculated by dividing thetechnology's total annualized cost by the tonnage ofreduced NOx emissions and indicates the cost of thecontrol relative to its environmental benefit.

4

Table 1 . Model Plant Cost Estimates for Conventional SNCR a

Total Capital Cost Tipping Fee Impact Cost Effectiveness($1000/TPD Capacity) ($/ton MSW) ($/ton NOx)

NOx Reduction (%) 45 60 65 45 60 65 45 60 65

100 TPD Mass Burn MWC 5.05 5.06 5.07 3.74 3.91 4.06 4,308 3,378 3,235

400 TPD Mass Burn MWC 1.97 1.98 2.00 1.30 1.47 1.62 1,496 1,268 1,289

750 TPD Mass Burn MWC 1.49 1.50 1.52 0.92 1.09 1.24 1,058 940 986

a$/ton can be converted to $/Mg by multiplying by 1.1.

SNCR. Combining NGI and SNCR toachieve an overall NOX reduction of 80 -85% may be feasible but will require test-ing to evaluate the interactions betweenthe temperature and residence time re-quirements of each technology.

Comparative Costs ofEvaluated ControlTechnologies

As part of this study, cost evaluationswere conducted for several variations ofNGI, SNCR, and SCR. For the evaluationof NGI, two scenarios were examined.The first scenario, referred to as NGI-100,

assumes the MWC is firing MSW at 100%of its design heat input capacity. There-fore, under this scenario, the rate of wastefired must be reduced by an amount com-parable to the heat input from natural gas.This results in a reduction of tipping feerevenues. The second scenario, referredto as NGI-85, assumes that the MWC isfiring MSW at 85% of its design heat inputcapacity because of insufficient MSW flowor because the unit was designed withexcess heat input capacity. In this case,natural gas can be used to reduce NO

Xemissions without displacing any wasteand, therefore, without a loss of tipping

fee revenues. The average NOX reductionassumed in both scenarios is 60%.

Four variations of SNCR technologywere examined: (1) conventional SNCR,with an average NOX reduction of 60%,(2) advanced SNCR, with an average NOXreduction of 60% with lower reagent usecompared to conventional SNCR, (3) asecond advanced SNCR option with a NOXreduction of 70% at a higher reagent feedrate, and (4) advanced NGI, which com-bines conventional NGI with conventionalSNCR to achieve a NOX reduction of 80%.The SCR evaluation focused on a cold-side system with a NOX reduction level of

Figure 1 . Effect of unit size, reagent cost, and annualized capital on tipping fee impact and cost effectiveness for conventional SNCR.

4

3.5

3

2.5

2

1.5

1

Reference MWC ParametersUncontrolled NOx = 250 ppmNOx Reduction = 60%

3,455

3,023

2,591

2,160

1,728

1,296

864

Cos

t Effe

ctiv

enes

s ($

/ton

of N

Ox)

Tip

ping

Fee

Impa

ct (

$/to

n of

MS

W)

Unit Size (tpd) 100 200 300 400 500 600 700

Reagent Cost ($/ton) 210 240 270 300 330 360 390Annualized Capital ($1000/yr) 52 60 67 75 82 90 97

Unit Size Reagent Cost Annualized Capital

5

Table 2 . Technical Overview of Evaluated NOX Control Technologies

Technology Commercial Status NOX Control Performance Technical Issues

NGI: Tested on MWC in Olmsted Testing achieved up to 60% Scaleup of technology to largerMethane County, Minnesota. NOx reduction without furnaces (> 100 tpd).deNOX

SM increasing CO emissions.

NGI: Applied to fossil fuel boilers; MWC testing encountered high Reburn zone temperatures inReburning use on MWCs limited to test CO levels when NOX reductions MWCs may be too low for NOX

program in Malmo, Sweden. exceeded 30 - 40%. reduction reactions.

SNCR: Applied to six MWCs in U.S. Achieved short-term reductions Impact of furnace temperatureThermal plus others overseas. of 45 - 75%, depending on NH3 swings on NOX and NH3DeNOX

TM injection rate. Plume visible at emissions. Control of NH3 sliphigher reduction levels. and visible plume.

SNCR: Applied to two MWCs in U.S. Comparable to NH3 injection. Similar to NH3 injection. N2ONOXOUTTM plus others overseas. emissions also of concern.

Advanced Tested at MWCs in Lancaster, Achieved 60 - 75% NOX Demonstration of long-termSNCR Pennsylvania, and Munich, reduction with less reagent than performance capability.(improved Germany. is needed with conventional Additional process controlsprocess SNCR. Plume visible at higher may benefit other combustorcontrol) reduction levels. operations.

Advanced NGI Concept only. Potential for 80% reduction. Interaction of temperature and(conventional residence time needs for eachNGI plus technology.SNCR)

SCR Installed at 20 MWC plants in 80 - 90% NOX reduction. Catalyst life in hot-side systems.Europe and Japan.

80%. A limited evaluation of a hot-sideSCR system was also conducted.

Costs for the different control technolo-gies are compared in Figures 2, 3, and 4.Figure 2 presents the capital costs foreach of the technologies. The advancedSNCR system with the 70% NO

X reduc-tion and the hot-side SCR system are notrepresented, since only a limited analysisof these scenarios was conducted. Asshown by the figure, the capital costs foreach of the technologies increase withincreasing unit size. SCR is the most capi-tal intensive of the technologies, costing 4to 5 times more than the next highesttechnology. Advanced SNCR and ad-vanced NGI have the next highest capitalcosts, with both technologies estimated tocost $1 to 2 million per combustor. Thecapital costs associated with NGI and con-

ventional SNCR are comparable, at lessthan $1 million per combustor.

The tipping fee impacts and cost effec-tiveness values presented in Figures 3 and4, respectively, include both annualizedcapital costs and operating and mainte-nance costs. Conventional SNCR gener-ally has the lowest costs of the technolo-gies. NGI-100 has the highest tipping feeimpacts for all but the 100 tpd MWC, forwhich SCR has slightly higher costs. Costeffectiveness values for NGI-100 are thehighest for all combustor sizes. Both ofthe NGI-100 and NGI-85 scenarios resultin an incremental cost increase; however,the revenue loss associated with divertingMSW when firing natural gas under NGI-100 results in much higher costs for thiscontrol technique. These revenues are notlost with the NGI-85 scenario, plus rev-

enues are received under this scenariofrom sale of additional electrical produc-tion. The other key variable affecting bothNGI scenarios is the price of natural gas.The average gas price used in these fig-ures is $3.50/106Btu.

The high tipping fee impacts and costeffectiveness values with SCR result fromthe high capital costs for this technology,as well as from the cost of catalyst re-placement and disposal. Advanced SNCRand advanced NGI have higher costs thanNGI-85 applied to the small combustorsize but are fairly similar when applied tothe medium and large combustor sizes.

6

Figure 3. Comparison of tipping fee impact.

12

10

8

6

4

0

2

100 400 750Unit Size (tpd)

Cap

ital C

ost (

$)(M

illio

ns)

NGI-100 60%

Advanced SNCR 60% Advanced NGI 80%

NGI-85 60% Conventional SNCR 60%

SCR 80%

Figure 2 . Comparison of capital cost.

25

20

15

10

5

0100 400 750

Tip

ping

Fee

Impa

ct (

$/to

n M

SW

)

Unit Size (tpd)

NGI-100 60% NGI-85 60% Conventional SNCR 60%

SCR 80%Advanced NGI 80%Advanced SNCR 60%

7

Figure 4 . Comparison of cost effectiveness.

Cos

t Effe

ctiv

enes

s ($

/ton

NO

x)

Conventional SNCR 60%

18

0100 400 750

Unit Size (tpd)

NGI-85 60%

SCR 80%Advanced NGI 80%Advanced SNCR 60%

NGI-100 60%

(Tho

usan

ds)

16

14

12

10

8

6

2

4

8

D. M. White, K. L. Nebel, M. Gundappa, and K. R. Ferry are with Radian Corp.,Research Triangle Park, NC 27709.

James D. Kilgroe is the EPA Project Officer (see below).The complete report, entitled "NOx Control Technologies Applicable to Municipal

Waste Combustion," (Order No. PB95-144358; Cost: $27.00, subject to change)will be available only from:

National Technical Information Service5285 Port Royal RoadSpringfield, VA 22161Telephone: 703-487-4650

The EPA Project Officer can be contacted at:Air and Energy Engineering Research LaboratoryU.S. Environmental Protection AgencyResearch Triangle Park, NC 27711

United StatesEnvironmental Protection AgencyCenter for Environmental Research InformationCincinnati, OH 45268

•

• •

•

Official BusinessPenalty for Private Use $300

EPA/600/SR-94/208

BULK RATEPOSTAGE & FEES PAID

EPAPERMIT NO. G-35