To MACT or not to MACT: Mercury Emissions from Waste-to-Energy and Coal-fired Power Plants Nickolas J. Themelis 1 and Nada Assaf-Anid 2

Transcription

1 13th North American Waste to Energy Conference May 23-25, 2005, Orlando, Florida USA NAWTEC To MACT or not to MACT: Mercury Emissions from Waste-to-Energy and Coal-fired Power Plants Nickolas J. Themelis 1 and Nada Assaf-Anid 2 1 Department of Earth and Environmental Engineering, Earth Engineering Center, Columbia University New York City, NY (njt1@columbia.edu) 2 Department of Chemical Engineering, Manhattan College, Earth Engineering Center, Columbia University ABSTRACT During the combustion of fuel in Waste-to Energy (WTE) and coal-fired power plants, all of the mercury input in the feed is volatilized. The primary forms of mercury in stack gas are elemental mercury (Hg o ) and mercuric ions (Hg 2+ ) that are predominantly found as mercuric chloride. The most efficient way to remove mercury from the combustion gases is by means of dry scrubbing, followed by activated carbon injection and a fabric filter baghouse. Back in 1988, the U.S. WTE power plants emitted about 90 tons of mercury (Hg). By 2003, implementation of the EPA Maximum Achievable Control Technology (MACT) standards, at a cost of one billion dollars, reduced WTE mercury emissions to less than one ton of mercury. EPA now considers coalfired power plants to be the largest remaining anthropogenic source of mercury emissions. Approximately 800 million short tons of coal, containing nearly 80 short tons of Hg are combusted annually in the U.S. for electricity production. About 40% of this amount is presently captured in the gas control systems of coal-fired utilities. Since the concentration of mercury in U.S. coal is ten times lower than in the MSW feed and the volume of gas to be cleaned 55 times higher, the cost of implementing MACT by the U.S. coal-fired utilities is estimated to be about $25 billion. However, when this retrofit cost is compared to the total capital investment and revenues of the two industries, it is concluded that MACT should be affordable. Per kilogram of mercury to be captured, the cost of MACT implementation by the utilities will be twenty times higher than was for the WTE industry. However, implementation of MACT by the utilities will also reduce the emissions of other gaseous contaminants and of particulate matter. The authors: Nickolas Themelis is Stanley-Thompson Professor of Chemical Metallurgy, Department of Earth and Environmental Engineering, and Director of the Earth Engineering Center of Columbia University. Professor Nada Assaf-Anid is Chair of the Department of Chemical Engineering of Manhattan College in New York City and also Research Associate of the Earth Engineering Center of Columbia University. Dr. Assaf Anid specializes in the fate of organic and inorganic contaminants in water, sediments and soil. Mercury in WTE A study by Themelis and Gregory for the New York Academy of Sciences [1 ] described how the U.S. WTE industry reduced its mercury emissions from 89 tons, reported by NREL in 1988, to less than two tons by 2000 (Figure 1). This dramatic decrease was principally due to the implementation of MACT standards by the industry at a reported total cost of one billion dollars. A recent check by the authors of mercury emissions from the five WTE facilities in New Jersey [2] showed that, in 2003, the mercury emissions of these plants, which in 203

2 total process 2.05 million tons of MSW [3], had been reduced from 148 kilograms in 1999 to an estimated 44 kilograms in 2003 (Table 1). If the same trend applies throughout the U.S. (29 million tons of MSW processed in 2003 [3]) the total WTE industry emissions in 2003 would be in the order of 700 kilograms. It is believed that a large part of the reason for the further reduction of mercury emissions between 1999 and 2003 is that the concentration of mercury in the MSW has decreased as mercury has been phased out in consumer products. tons/year 9G NREL,l'IM ----, <.. _._.._.. WTE plants.._.. x Coal-fired plants Figure 1. Decrease in mercury emissions from WTE facilities, [6] Table 1. Change of mercury emissions at five NJ WTE facilities between 1999 and 2003 WTE % 2003 Facility Outlet Hg Outlet Hg reduct kg mercury conc. conc. ion per O2 from tons MSW Og/dscm Og/dscm (ref. 1) (ref. 8) 2003 A % 12 B * 78% 35.5 C % D % 5 E % Ave. 23 Mercury in coal Mercury exists in coal in minute quantities, ranging from 0.02 to 0.48 parts per million [4]. U.S. coals are at the low level of mercury concentration while some of the Chinese coals are at the high end. Since the U.S. power plants consume about 800 million tons of coal annually, at an assumed average of 0.1 ppm of mercury in coal, the amount of mercury input to U.S. coal-fired power plants (CFPP) is about 80 tons. A Department of Energy (DOE) estimate placed the amount of mercury emitted to the atmosphere by CFPP at 48 tons [5]. This is in agreement with EPA's estimate that about 64% of the estimated 75 tons mercury input to CFPP, i.e. 48 tons is emitted and the rest is captured in the present Air Pollution Control (APC) systems of these power plants [6]. Air Pollution Control Technologies The APCs of most coal-fired power plants resemble those of the WTE industry prior to MACT implementation: they consist of electrostatic precipitators (ESP) followed by wet scrubbers (WS). These systems may not be as effective in capturing very fine particles and volatile metals as the MACT APC systems that consist of dry scrubbers (OS) and fabric filter baghouses (FF). Also, the use of OS and FF allows the injection in the process gas, before it is filtered through the baghouse, of fine activated carbon particles (Activated Carbon Injection or ACI) to which both elemental and oxidized mercury vapors become attached [5]. Other volatile metal vapors, such as cadmium and dioxin molecules, also attach themselves to the fine carbon particles and are then captured in the fabric filter bags. Table 2 shows the existing and proposed WTE and CFPP Air Pollution Control systems. There are 89 WTE power plants [9]. EPA has reported [2] that there are 1,143 coal fired utility boilers of capacity greater than 25 megawatts of electricity [2]. An estimated 77% of these plants are equipped with either ESP or fabric filter controls. Of these, 83% are equipped with electrostatic precipitators [2]. MACT versus Cap and Trade EPA, having succeeded in reducing drastically the mercury emissions of WTE facilities is now addressing the current big emitter of mercury in the atmosphere. It has indicated that coal-fired power plants must also comply with the MACT 204

3 regulations that have proven effective in curtailing WTE emissions to one hundredth of the 1989 levels. In June 16, 2004, the Electric Power Research Institute (EPRI) submitted to EPA the "EPRI Comments on Proposed Emission Standards for Mercury Emissions of Electric Utility Steam Generating Units" (June 16, 2004, Docket 10 No. OAR ). This report suggests that "Cap and Trade" is a better way to reduce mercury emissions from coal-fired utilities than by imposing the MACT regulations. "Cap and trade" (C& T) is a policy approach for controlling large amounts of emissions from a group of sources at a cost that may be lower than if these sources were regulated individually. The "cap" is the allowable maximum amount of emissions over the set compliance period, in order to achieve the desired environmental effects. Each utility would be authorized to emit a certain amount and the total number of allowances would not exceed the industry cap. In addition, the sources under the "cap" can buy or sell allowances on the open market. Table 2. Existing and proposed WTE and CFPP C ontro IT ec hno i ogles ' WTE (89 plants) CFPP (1143 plants) Predominant Predominant Proposed existing APC existing APC APC systems systems systems under MACT Dry scrubbers Wet scrubbers Dry scrubbers Activated Carbon Injection (ACI) Activated Carbon Injection (ACI) Fabric filter bags ESPs Fabric filter bags Apparently, cap and trade worked well in the containment of S02 emissions and the EPRI report maintains that it would result in lower emissions of mercury than MACT. In the long run, EPRI showed that the mercury reduction plots for C& T and MACT converge. However, the mercury emission reduction projected by EPRI is not down to the 98% level that was achieved by MACT implementation in the WTE industry, but down to one third of the present level (i.e., from 48 tons down to 15) [7]. The likely reason is that as the concentration of contaminants in the process gas decreases, it becomes progressively more difficult to capture; the authors estimated the concentration of mercury in the CFPP combustion gas to be less than one fifth of that in the WTE combustion gas. It should be noted that the EPRI report to EPA (Table B.1-1, p.45) showed the utility industry in a better light by inflating the mercury tonnages emitted by other sources, especially the WTE industry. The emissions of the WTE industry were shown as 28.8 Mg/yr (1 Mg=1.1 short ton); the number of 8.2 Mg/yr was shown in parentheses with an explanatory note that this was "for 1999 emissions"[4]. In fact, the yearly tonnage of 28.8 Mg applied to the WTE emissions, as quoted in EPA's 1997 Report to Congress [6]; by 1999, the WTE emissions were only 2.1 Mg/y [1]. Also, the tons, shown by EPRI as being the total U.S. mercury emissions, is the old figure reported by EPA to Congress for [6]. Speciation of Mercury from Coal Fired Utilities Trace amounts of mercury are found in various types of waste and fuels (i.e. coal and oil). Through combustion of solid waste or fuels, mercury vapor is released in flue gas emissions as two major species: elemental mercury, Hg o, and oxidized (or ionic) mercury, Hg 2 +. Both can leave the stack in association with particulate matter [Hg--p]. Specifically, elemental mercury Hg O vapor is formed in the high temperature regions of coalfired boilers during combustion. Subsequent cooling of the flue gas leads to a series of reactions in which Hg O converts to ionic, Hg 2 +, and/or particulate mercury, Hg--p[6]. Figure 2 shows the predominant oxidation reactions that occur upon reaction of Hg o ( Q ) with chlorine radicals to form gaseous Hg(II)CI2 as the ultimate product, and the relative amounts of the particulate and gaseous species in the flue 205

4 gas. The estimated amounts present in the flue gas of WTE plants are shown in parentheses for comparison purposes. The speciation of mercury emissions is a function of rank of coal (bituminous, subbituminous, and lignite) as well as the control technologies (ESPs, wet scrubbers, etc.) used. Therefore, it can vary significantly from plant to plant. Most of the emitted mercury in subbituminous and lignite fired boilers is Hg o, while Hg 2 + is the predominant species emitted in bituminous fired boilers is [6]. This is probably caused by the higher sulfur and/or chlorine content in bituminous coal [5]. HgCl(g) + Cl Hg(II)CI2(g) (Eql) [10] Elemental Particulate Ionic \ CFPP: 54 % (WTE: 15 % 43 % [2] 80 % [11]) Figure 2: CFPP and WTE Hg speciation after combustion Fate of Mercury from Coal Fired Utilities Speciation is critical in determining the fate of mercury emissions, part of which remain in the global atmosphere for a long period of time or are deposited, through wet or dry deposition, to water surfaces, land, and vegetation. Oxidized mercury, Hg 2+ has a residence time of just hours in the atmosphere [1 3] because of its relatively high water solubility. After deposition on land, it is eventually washed into local rivers, lakes, and streams by rain and melting snow. In contrast, elemental mercury, Hg o, has a high vapor pressure and low water solubility and consequently a longer residence time in the atmosphere (estimated at 6-24 months) [1, 13] making it the dominant form of atmospheric mercury [1 4]. Therefore, once emitted, elemental mercury is more likely to be carried away by wind and enter the global mercury cycle, rather than deposit in the region around the point source [9]. This also explains why atmospheric Hg O concentrations represent the global atmospheric mercury pool and are not as good indicators of local deposition as Hg 2+ concentrations, which are highly responsive to changes in Hg emissions [1 4]. Figure 3 below depicts the fate of mercury once emitted to the atmosphere. Of the various reactions that occur in anoxic water zones, an imrortant one is the microbial methylation of Hg 2 + by sulfate reducing bacteria, yielding the toxic and bioaccumulative compound methylmercury, HgCH3. Wet Deposition Most of the total atmospheric Hg deposition to water and land surfaces occurs through wet 206

5 deposition [1 3] via rain and snowfall containing soluble Hg 2+. \ Figure 3: The Mercury cycle [12] Evasion [RH/!II1Ild AntIIrcpogenc & Nallllall (II) ;::: Hg(P) PanlCUlale Removal The unusually stable monoatomic gas, Hg o (g), becomes more soluble in water through fast oxidation by 03 (g) to Hg 2 +. The latter builds up at a steady state concentration as a function of the concentrations of surrounding gases, including S02(g), as depicted in equation (2) and Figure 4 [13]. I / 6 /,' ' I //;/ // 0 / 1 " I Hg2+(aq), (\ / I I ' /6 / /6 I /! ' / I / 0', I / / Figure 4. Schematic of the heterogeneous environmental chemistry influencing wet deposition of emitted Hg [13] [Hg2\aq)] = k[hgo(g)][03(g) ] [H+(aQ)]2[S02(g)r1 (Eq2) J ' 0 Comparison of costs of implementing MACT Table 3, on the following page, shows that the process gas generated by utilities, per ton of fuel used, is twice that of WTE facilities. Because of the much lower concentration of mercury in coal process gas and larger volume of gas generated by the U.S. utilities (Table 3), the cost of MACT implementation in the utilities industry, per kilogram of mercury captured, will be much higher than it was for the WTE facilities. However, another way to look at this issue is as follows: What will be the relative cost of MACT implementation by the utilities, in terms of a) incremental capital cost (over and above initial capital investment, and b) revenues from the sale of electricity? Table 3 shows that if the U.S. utilities were to retrofit to the same standard as was required of the WTE industry, the retrofit cost would be 24% of the annual revenues of the utilities industry. In comparison, the cost of implementing MACT for the WTE industry was 48% of its annual revenues. Also, the one-time retrofit cost to the WTE industry was 6 cents per kwh generated in one year, while the one-time retrofit cost to the utilities would be only 1 cent per kwh generated. The Appendix to the EPRI Comments to EPA [7] showed that their estimated costs of MACT implementation to the utilities industry (page 102 Table VI-1 ) are $27.8 billion for MACT and $1 9.7 billion for Cap and Trade. The EPRI MACT cost number is fairly close to the one that was calculated in this study by extrapolating the MACT retrofit of the WTE industry (see Table 3, Note 10). Per kilogram of mercury to be captured, the cost of MACT implementation by the utilities will be twenty times higher than was for the WTE industry. However, implementation of MACT by the utilities would not only decrease mercury emissions but also other volatile metal emissions, such as cadmium, and also particulate matter. Thus, the general adoption of dry scrubbers, activated carbon injection, and fabric filters by the U.S. utilities would benefit the environment in several ways. 207

6 Table 3. Comparison of Costs of MACT Retrofit by U.S. WTE Industry to Projected Costs of MACT Retrofit by the U.S. Coal-fired Utilities US WTE Industry All US Coal-fired utilities Number of facilities L Fuel combusted, million short tons/year Volume of gas to be cleaned, Nm3/short ton fuel 5,000 10,000 Total process gas, million Nm3 (Normal cubic 145,000 8,000,000 meter at 1 atm, O C) Kilograms of mercury in combustion ases (at assumed 1 ppm of mercury in MSW1 and 0.10 ppm average in coal) 25,700 72,800 Kilograms mercury captured presently 25,700 29,100 Estimated kilograms of mercury emitted to the atmosphere presently < Kilograms to be captured by MACT at utilities, at assumed 75% capture of present Hg emissions Power plant nameplate capacity, MW 2, ,000L Electricity produced, million kwh/year 16,000 1,560,000" Total initial investment, $million 9, ,000 Total revenues, $million 2,000 93,600 H Estimated cost of MACT retrofit, $million 1,000" 24,700'u $ retrofit / $ initial investment $ retrofit / $ revenues $ retrofit / kwh generated $ of investment per annual kg Hg captured, or to be captured (utilities), by means of MACT implementation $39,000 $754,000 Notes: 1. lwsa, Directory of WTE Facilities, 2. ElAiDOE: ht1]l:llwww.eia.doe.gov/cnea{/electricitji/il!l!./htmlllt5[l.01.html 3. Estimated at net of 550 kwh per ton MSW 4. ElAiDOE: ht1]l:llwww.eia.doe.govlneic/guick[actslguickelectric.htm Estimated at $330 per annual ton of capacity 6. Estimated at $l,ooolkw of installed capacity 7. Estimated at average of $70lton (tipping fees +sale of electricity) 8. At assumed $0.06 per kwh 9. As reported by lwsa Scaled up from retrofit costss of WTE industry multiplied by ratio of volume of process gas to be treated to the power of 0.8, in order to reflect economies of scale II. Calculated from measured average mercury concentration in furnace gas of five NJ WTE facilities 5. Conclusions The primary forms of mercury in the stack gas are elemental mercury (Hg o ) and mercuric ions (Hg 2 + ) that are predominantly found as mercuric chloride [Hg(II)CI2]. The most efficient way to remove mercury from the process gas is by means of dry scrubbing, followed by activated carbon injection and fabric filter baghouses. By spending an estimated $1 billion to implement the superior air pollution control systems of MACT regulations, the U.S. 208

7 WTE industry succeeded in reducing mercury emissions from about 90 tons back in 1988 to less than one ton in Due to the fact that the concentration of mercury in U.S. coal is much lower than in the present MSW feed, and the total volume of gas to be cleaned 55 times greater, the cost of implementing MACT to the U.S. coal-fired utilities was estimated to be about $25 billion, vs the $1 billion spent by the WTE industry. However, when the cost of implementation is compared to the capital investment and the revenues of the two industries, it is concluded that MACT should be affordable for the utilities. The cost of MACT implementation to the utilities, per kilogram of mercury to be captured, will be twenty times greater than in the WTE case. However, implementation of MACT by the utilities will also reduce the mission of other gaseous contaminants and also of particulate matter, a major remaining cause of adverse health effects. REFERENCES [1] Themelis, N.J., Gregory, A.F., "Sources and Material Balance of Mercury in the New York-New Jersey Harbor", Report to the New York Academy of Sciences, October 3,2001, pp , also: N.J. Themelis and A. Gregory, "Mercury Emissions from High Temperature Sources in NY/NJ Hudson Raritan Basin", in North American Waste to Energy Conference (NAWTEC 10) Proceedings, ASME International, Philadelphia, May Division, U.S. Environmental Protection Agency, EPA 600/R-01/109 NRMRL ORD, Dec [7] Yager, J., "EPRI Comments on EPA Proposed Emission Standards/Proposed Standards of Performance, Electric Utility Steam Generating Units: Mercury Emissions", Palo Alto, California, pp. 45; , June [8] Report to Congress, U.S. Environmental Protection Agency "Mercury Study Report to Congress ", Volume II. Inventory of Anthropogenic Emissions in the U.S., EPA-452/R , pp A-15, December [9] Sullivan, T.M, Lipfert, T.D., Morris, S.M., "The Local Impacts of Mercury Emissions from Coal Fired Power Plants on Human Health Risk", Brookhaven National Laboratory, Upton, New York, May 2003, pp. 7 [10] Edwards, J. R, Srivastava, R V., Kilgroe, J.D., "A Study of Gas-Phase Mercury Speciation Using Detailed Kinetics", Journal of the Air & Waste Management Association, pp 1, [11] Licata, A., Balles, E., SchOttenhelm, W., "Mercury Control Alternatives for Coal-Fired Power Plants", Presented at Power Gen 2002 Orlando Flo, p.10, December [12] EPA Report to Congress, Volume III: "Fate and Transport of Mercury in the Environment",, p.2 et seq., December [13] McCormac, B. M., "Mercury in the Swedish Environment: Recent Research on Causes, Consequences and Corrective Methods", Water, Air & Soil Pollution: International Journal of Environmental Pollution, vol. 55, Kluwer Academic Publishers, Boston, pp , [14] Mason, R, Abbott M., Bodaly, A., et al. Monitoring the Response to Changing Mercury Deposition, Environmental Science &Technology, 15A-22A, [2] Sikdar, H., New Jersey Department of Environmental Protection, private communication, Oct [3] Zannes, M. and J.v.L. Kiser, "2004 Directory of U.S. WTE Facilities", [4] Quick, J.S., Tabet, D.E., S. Wakefield, S., R Bon, R L. "Optimizing Technology to reduce Mercury and Acid gas emissions from Electric Power Plants", Prepared by The Utah Geological Survey for the United States Department of Energy, Salt Lake City, Utah, pp , February 2004 [5] Feeley, T.J., Murphy, J., Hoffman, J., Renninger, SA, "A Review of DOE/NETL's Mercury Control Technology R&D Program for Coal-Fired Power Plants", Prepared by National Energy Laboratory, United States Department of Energy, pp 1; 6, April 2003 [6] "Control of Mercury Emissions from Coal-fired Electric Utility Boilers", Office of Research and Development, Air Pollution Prevention and Control 209