Fate of Mercury in Cement Kilns

Transcription

1 Fate of Mercury in Cement Kilns Paper #1203 C.L. Senior, A.F. Sarofim Reaction Engineering International, Salt Lake City, UT E. Eddings Chemical and Fuels Engineering, University of Utah, Salt Lake City, UT ABSTRACT Mercury is one of a number of pollutants (like dioxins) that persist in the environment and bioaccumulate in the food chain. Because of its toxicity and the potential for bioaccumulation, mercury emissions to the environment are the subject of environmental regulation. U.S. EPA estimates that 87% of the man-made emissions of mercury come from point sources of combustion. There are currently emission limits on mercury from certain categories of combustion sources, including cement kilns and incinerators burning hazardous waste. Cement kilns that do not burn hazardous waste are not subject to these emission standards. However, EPA is currently reviewing the need for emission standards for mercury and other pollutants from cement kilns. In this paper, we review the chemistry of mercury in the cement-making process and develop a simple model for understanding the distribution of mercury among various streams in different types of cement kilns. INTRODUCTION EPA, in its Mercury Study Report to Congress 1 in 1997, put the amount of mercury released into the atmosphere from human activities between 50 and 75 percent of the total yearly release from all sources. According to this estimate, 158 tons of mercury was emitted annually from all human activities in the U.S., most of this (87%) from combustion point sources. Hazardous waste combustors, including Portland Cement facilities that burn wastes, contributed 7.1 tons per year or 4.4% of the total. Manufacturing sources contributed 10% to this estimate. The Portland Cement manufacturing industry (excluding facilities burning hazardous waste) was estimated to contribute 4.8 tons per year, or 3.1% of the total. Only a fraction of Portland Cement kilns in the U.S. burn hazardous waste. In a survey of cement plants in 20002, only 8 plants were identified as burning waste as a primary fuel (Table 1). The capacity of these plants corresponds to about 5% of the total manufacturing capacity in the U.S. However, a larger number of plants identified waste as an alternate fuel, corresponding to about 50% of the total manufacturing capacity. There are currently emission limits on mercury from certain categories of combustion sources, including facilities burning hazardous waste. Table 2 summarizes the current mercury emission standards for existing hazardous waste combustors in the U.S. and Europe, compared with those for municipal waste combustors in the U.S. The EPA uses a Maximum Achievable Control Technology (MACT) standard to specify control technologies for specific pollutants. For mercury emissions from hazardous waste 1

2 combustors, the MACT standard is feed rate control, based on input feed rates to the combustor. Table 1. Portland Cement Plants in the U.S. Using Waste as Fuel (Reference 2). Plants using waste as a primary fuel Capacity Process Plants tons/day 1000's tons/yr Dry 1 2, Wet 7 11,654 3,889 Total 8 13,784 4,569 Plants using waste as an alternate fuel Capacity Process Plants tons/day 1000's tons/yr Dry 7 13,498 4,295 Dry(Preheater) 8 13,444 4,334 Dry(Precalciner) 19 54,941 17,694 Wet 17 31,139 9,951 Total ,022 36,274 Table 2. Mercury emission standards for existing sources in µg per dry standard cubic meter in µg/dscm at 7% O 2 (Reference 3). Hazardous Waste Incinerators Hazardous Waste Cement Kilns Hazardous Waste Lightweight Aggregate Kilns Hazardous Waste Combustors (Europe) Large Municipal Waste Combustors Cement kilns that do not burn hazardous waste are not subject to these emission standards. However, EPA is currently reviewing the need for emission standards for mercury and other pollutants from cement kilns. MERCURY BEHAVIOR IN COMBUSTION SYSTEMS Coal is a common fuel used in cement kilns. Mercury is present in coal in low concentrations, on the order of 0.1 µg/g. Mercury concentrations in petcoke and tires, which are often burned in kilns, are lower. Mercury is also present in certain wastes burned in cement kilns, in various concentrations. In the high temperature combustion, all the mercury in the fuel is vaporized as elemental mercury. Coal and liquid wastes are injected through the primary 2

3 kiln burner and any mercury contained in them is exposed to flame temperatures, which will convert all the mercury to the gaseous, elemental form. Mercury is also found in the raw materials that enter the kiln. The range of mercury in limestone, the largest component of the raw material, is reported to be to 0.45 ug/g, but measurements are scanty. 4 In high temperature combustion systems, mercury exits the flame region in the elemental gaseous form (Hg 0 ). Subsequently, mercury can be oxidized homogeneously, oxidized heterogeneously or adsorbed by ash or activated carbon. Gas-phase thermodynamic equilibrium calculations suggest that HgCl 2 is the dominant oxidized species in flue gas at temperatures below F. Equilibrium predictions of mercury speciation do not agree with measured mercury speciation at the inlet to particulate control devices in coalfired power plants 5 nor with the speciation of mercury in medical waste incinerators utilizing a water quench after the combustion chamber, which suggests that mercury species are not in equilibrium as the flue gas cools. At moderate temperatures in a flue gas environment, mercury is thought to react with chlorine atoms, which are formed by the interaction of HCl with free radicals. Kinetic calculations have been carried out on simulated combustion flue gas containing chlorinated compounds. 5 These calculations showed that equilibrium was not achieved for chlorinated compounds in a rapidly cooling gas with cooling rates typical of the convection section of a coal-fired power plant waste-to-energy plant. Thus, it seems reasonable to conclude that the oxidation of mercury via chlorinated compounds does not reach equilibrium under conditions of rapid quenching. In incinerators that use a water quench, cooling is even more rapid than in combustion systems that generate steam. There is indirect evidence of the inability of wet scrubbers on incinerators with water quench to capture mercury, 6 presumably because in those systems, the mercury is still predominantly in the elemental form at the scrubber. Elemental mercury, as discussed below, is not removed by wet scrubbers, unlike oxidized forms of mercury. Experiments carried out on the speciation of mercury in a reactor that mimicked the flue gas composition and residence time of a typical incinerator 6 demonstrated that even at high (3000 ppmv) levels of HCl in the flue gas, mercury was not oxidized if the gas was quenched too rapidly. Mercury can be also oxidized heterogeneously or adsorbed by fly ash or activated carbon. Ash plays a role in both the adsorption of mercury and the oxidation of elemental mercury in flue gas at temperatures characteristic of particulate control device. Unburned carbon in ash has been suspected of adsorbing mercury in coal-fired power plants. In a dry process kiln, the gas exits the kiln at much high temperatures such that mercury will exit the kiln in the gaseous form. But in the preheater tower, mercury may be adsorbed on the raw meal, enriching both the kiln feed and the CKD. As with wet process kilns, concentrations of mercury in the kiln itself should be higher than predicted from the mercury content of the input streams. 3

4 Detailed measurements of mercury within Portand cement kilns have not been made. However, the behavior of thallium has been studied. Thallium is reported by Sprung 7 to have the highest volatility of the trace metals that were assayed in cement kilns; the other metals were Zn, Pb, As, Ni, Cr, Cd. Figure 1 shows the concentrations of thallium and other metals measured in solids collected from various stages in a preheater kiln. Volatile Figure 1. Distribution of selected metals in stages of preheater kiln. 7 Hourly quantity in feed material, g/h Stage: I II Thallium III IV Pb K 2 O x10-3 Cl x10-3 metals (e.g., thallium) condense at lower temperatures on the solids than other metals. Since the solids are being preheated before introduction into the kiln, the condensed metal will be recycled back into the kiln. The concentration of volatile metals will build up in kiln over time. Temperature, F The example of thallium suggests that mercury will have similar behavior; the boiling point of mercury is considerably lower than that of thallium. In a wet process kiln, the gases cool to 400 o F or less at the kiln exit. Mercury may be oxidized and/or it may condense on the raw material. In the latter case, the mercury will move along the kiln in the solid bed and then vaporize in mid-kiln after the solids dry and begin calcining. This will result in the gaseous mercury concentration within the kiln to build up to high levels, setting up a recycle loop for mercury within the kiln. As discussed below, some of the mercury will adsorb on the cement kiln dust (CKD) and be removed in the particulate control device and some will leave the kiln in gaseous form. When CKD is reinjected into the kiln, as is commonly done, levels of mercury in the kiln itself will be considerably higher than one would predict from looking at the content of mercury in the fuel and raw materials. MERCURY EMISSIONS FROM PORTLAND CEMENT MANUFACTURE Sources of mercury in cement kiln feed streams Coal is a primary fuel for many cement kilns, whether or not they burn hazardous waste. Coal contains very small amounts of mercury; typical values of mercury in bituminous coals of economic importance in the U.S. are 0.05 to 0.25 µg/g coal (dry basis). When translated into concentration in the flue gas, this corresponds to 4 to 20 µg/dscm at 7% O 2. 4

5 Figure 2 shows the cumulative distribution of mercury in solid fuels. These data come from EPA s Information Collection Request (ICR) and represent multiple fuel samples from every coal-fired power plant in the US taken during the fourth quarter of The figure demonstrates the range of mercury concentrations in solids fuels. Figure 2. Distribution of mercury concentrations in solid fuels from ICR, Part 2 data for fourth quarter, % Petcoke %Less than value 80% 60% 40% Tires Subbituminous Lignite Bituminous 20% 0% Hg content, ug/g Cement kilns in the United States have been co-firing hazardous waste with other fuels for 30 years. Most kilns that burn hazardous waste are wet process kilns. The reason for this has more to do with economics than process, however. Wet process kilns require more energy per ton of clinker than dry process kilns. Since fuel costs are an important part of the cost of cement production, burning a fuel that generates income helps wet process kilns reduce operating costs. Liquid hazardous wastes are most commonly fired in cement kilns through the primary burner. Liquid wastes include the following: Residues from industrial or commercial painting operations; Metal-cleaning fluids and lubricants; Electronic industry solvents; Solvents from automotive aftermarket operations. Some of these wastes contain trace metals, including mercury. Many of them also contain significant amounts of chlorine, which can affect the chemistry of mercury in the flue gas. The kiln feed materials also contain measurable amounts of mercury as illustrated in Table 3 using data from EPA s database of emissions from hazardous waste combustors (HWCs) 9 as part of the process of setting maximum achievable control technology (MACT) standards for HWCs. This database contains emissions data from 13 different 5

6 wet process cement kilns that burn hazardous waste. In some cases, metals and/or organics were spiked into the feed for testing purposes. In Table 3, a subset of the data is shown for samples taken without spiking of metals in the feed and presumably (from the low percentage of mercury in the raw material) without recycle of CKD into the raw material. All the mercury concentrations have been converted to the equivalent of µg/dscm at 7% O 2 in the flue gas. The amount of mercury entering the kiln in the raw material appears to be on the same order as that in the coal, for these particular kilns. The Table 3. Mercury in feed streams to select wet process kilns expressed as µg/dscm at 7% O 2 in the flue gas (Reference 9). Source ID Number Haz. Waste Raw Mat l Coal Total A hazardous waste typically has more mercury than either the coal or the raw material and there is a very wide range of mercury in the hazardous waste. Emissions of mercury from cement kilns In order to get an idea of the range of mercury emissions from cement kilns, we will look at two sets of data, one for kilns burning hazardous waste and one for kilns that do not burn hazardous waste. The Portland Cement Association compiled a database of cement kiln emissions, 10 including emissions of mercury for kilns that do not burn hazardous wastes. The mercury measurements are from 35 different sampling reports with a total of 50 measurements. Various types of kilns, fuels, and particulate control devices are represented in the database. Figure 3 shows a frequency distribution of the measured mercury emission. Mercury concentration is reported in µg per dry standard cubic meter (DSCM) at 7% O 2. Statistical analysis of the data did not reveal significant differences between the emissions as a function of type of process, but the mean mercury emission was higher for kilns with fabric filters as compared to those with electrostatic precipitators. One group of measurements falls into the expected range for coal-fired boilers; this might not be 6

7 surprising since many of these kilns burn coal as the primary fuel. What is surprising, however, is the number of measurements above 50 µg/dscm. Figure 3. Distribution of stack concentrations of mercury (in µg/dscm at 7% O 2 ) for selected kilns that do not burn hazardous waste (Reference 10). Range of Coal-fired boilers Range of MWCs Number of Tests in Range >500 Mercury, ug/dscm Figure 4. Distribution of stack concentrations of mercury (in µg/dscm at 7% O 2 ) for selected kilns that burn hazardous waste (Reference 9). Number of Tests in Range Range of Coal-fired boilers Mercury, ug/dscm Range of MWCs >500 The EPA database of emissions from hazardous waste combustors (HWCs) 10 previously mentioned contains emissions data from 13 different wet process cement kilns that burn hazardous waste. In some cases, metals and/or organics were spiked into the feed for testing purposes. Figure 4 summarizes the distribution of mercury emissions based on 60 different samples. The mercury emission for each sample represents an average of three measurements. 7

8 Table 4 compares the two datasets. The non-hazardous waste dataset represents all types of kilns and the hazardous waste dataset, only wet process kilns. However, there was no statistically significant difference noted in the former dataset among various kiln types. The dataset of hazardous-waste burning kilns has a mean and median that are about twice that of the dataset from the kilns that don t burn hazardous waste. Since the compositions of wastes vary, no absolute comparison can be made. However, we can get an idea of the range of mercury emissions from cement kilns, whether or not they fire hazardous wastes. Table 4. Mercury emissions from cement kilns burning hazardous waste 9 and not burning hazardous waste 10 (in µg/dscm at 7% O 2 ). Kiln type Wet All HW non-hw Median Mean Standard deviation Minimum Maximum There is an inherent recycle of a volatile metal such as mercury within the kiln, both due to recycle of mercury-containing cement kiln dust and to large temperature gradients (in wet process kilns), it may take a long time for mercury to reach steady state in a cement kiln. This makes it difficult to make accurate mass balance measurements of mercury, particularly when mercury is only spiked in the feed for short periods of time or where mercury concentrations are changing rapidly in the feed stream. To illustrate the difficulties in closing the mercury mass balance, we compare the mercury in the feed streams to the mercury in the stack for selected samples from the EPA HWC database. 9 We have excluded any measurements in which mercury was spiked into the feed, with the understanding that it can take a long time to come to steady state when metals are spiked into the feed. The data on mercury in the raw material appear to include both CKD and raw meal in some cases, which suggests that CKD was being recycled for those cases. We have adjusted the concentration of mercury in the raw material in those cases either to reflect the mercury level in the absence of recycle (if those data are available for that kiln) or the mercury level in typical raw material. Figure 5 shows the ratio of mercury in the stack emission to mercury in the feed. There was generally a significant amount of mercury in the stack, relative to the feed concentration. In many cases, the amount of mercury in the stack was greater than the amount of mercury in the feed. As discussed above, this illustrates the inherent difficulty of obtaining a steady state measurement of mercury in cement kilns. 8

9 Figure 5. Ratio of mercury in stack emission to mercury in feed stream for selected wet process kilns with and without recycle of cement kiln dust (Reference 9). Number of Tests in Range No Recycle Recycle Hg in stack emission/hg in Feed The distribution of mercury in a dry process, preheater kiln is illustrated in Figure 6. These data are taken from a large study of trace metals in many different kilns. 11 In this particular kiln, mercury was also spiked into the feed, but the amount of the spike has been removed from the mass balance because of the uncertainty of the time needed to Figure 6. Mercury flow rates in a dry process, preheater kiln; all flows in lb/hr (Reference 11). Main Stack: Feed: Preheater Bypass Stack: Cooler Stack: Bypass Dust: Kiln Cooler Clinker: Inputs: lb/hr Outputs: lb/hr Mass Balance Closure: 43% Tires: Liquid Waste: Coal:

10 reach steady state. The mass balance closure is 43%. MODELING MERCURY IN CEMENT KILNS Models have been developed 12 that can be used to predict the distribution of trace metals including mercury in cement kilns. In this work, a model previously developed by Reaction Engineering International was used to examine the effect of recycle on mercury emissions from cement kilns. The model for the long, wet kilns is illustrated in Figure 7. Table 5 gives the assumptions used in the model. As shown previously, recycle of kiln dust back into the kiln has an impact on mercury emissions. Figure 8 illustrates the predicted stack emissions with and without recycle of cement kiln dust. Stack emissions are higher with recycle, which is the behavior that has been observed (Figure 5). Figure 7. Long, wet kiln process model for mercury. Main Stack Coo Vent M18 M19 A M9 Long Kiln M1 M2 M3 M4 Raw Meal M21 Kiln fuel M6 M5 M13 M15 M14 M10 M11 M12 C Clinker Table 5. Assumptions in long kiln process model. Dust entrainment in kiln, g CKD/g clinker 8.0% Hg vaporization/recycle in kiln 90.0% Dust entrainment in clinker cooler 10.0% Collection efficiency, main ESP 95.0% Collection efficiency, cooler ESP 30.0% CKD recycle 0.0% Hg concentration in fuel, mg/kg 0.10 Hg concentration in limestone, mg/kg 0.10 Fuel usage, kg fuel/kg clinker 0.20 Limestone usage, kg/kg clinker

11 The model for the long, wet kilns is illustrated in Figure 8. Table 6 gives the assumptions used in the model. As shown previously, recycle of kiln dust back into the kiln has an impact on mercury emissions. In precalciner kilns, the preheater/precalciner can be bypassed. A portion of the flue gas is often recycled through a separate stack. The bypass stream does not pass through the preheater, where mercury can condense on the raw material and be recycled into the kiln. Figure 9 illustrates the predicted emissions from the main stack and bypass stacks. The percentage of flue gas that exits through the bypass stack has an influence on stack concentrations of mercury. Bypass shifts mercury emissions from the main stack to the bypass stack. Figure 8. Precalciner kiln process model for mercury. Bypass Stack M7 B M17 Main Stack Preheater/Precalciner M16 Cooler Vent M18 A M9 M8 M19 M20 M1 M2 M3 M4 Inputs M22 M21 Precalciner fuel Kiln fuel Kiln M6 M5 M13 M15 M14 M10 M11 M12 C Clinker Table 6. Assumptions in precalciner process model. Dust entrainment in kiln, g CKD/g clinker 14.0% Hg vaporization/recycle in kiln 70.0% Dust entrainment in clinker cooler 10.0% Kiln bypass 30.0% Carryover from preheater to APCD 60.0% Collection efficiency, main ESP 30.0% Collection efficiency, bypass ESP 30.0% Collection efficiency, cooler ESP 30.0% Hg concentration in fuel, mg/kg 20.0% Hg concentration in limestone, mg/kg 0.26 Fuel usage, kg fuel/kg clinker 0.11 Fraction of fuel burned in kiln 1 Limestone usage, kg/kg clinker

12 Figure 9. Effect of bypass ratio on the emissions of mercury from main and bypass stack on precalciner kiln Main Stack Emissions Hg/Hg Input Bypass Stack Emissions Bypass ESP Dust % 5% 10% 15% 20% 25% 30% 35% Bypass CONCLUSIONS There are currently emission limits on mercury from certain categories of combustion sources, including cement kilns that burn hazardous waste. Cement kilns that do not burn hazardous waste are not subject to these emission standards. However, EPA is currently reviewing the need for emission standards for mercury and other pollutants from cement kilns. Mercury enters a cement kiln in the coal, in the raw material (kiln feed) and in hazardous wastes (if they are burned). Mercury leaves the kiln in the clinker, CKD or via stack emissions. The distribution of mercury within a cement kiln is difficult to measure quantitatively because of the difficulty in reaching a steady state. There is an inherent recycle of mercury between the hot and cold end of a kiln, and also from the reinjection of mercury-containing CKD into the kiln. Models can be used to predict the distribution of mercury in a cement kiln. In the future, such models might be used in conjunction with control technology to minimize emissions of mercury from cement kilns. REFERENCES 1 Keating, M.H., et al. Mercury Study Report to Congress, Volume I: Executive Summary, EPA-452/R , December Portland Cement Association, U.S. and Canadian Portland Cement Industry: Plant Information Summary, Data as of December 31, 2000 (Portland Cement Association, 2001). 3. U.S. Environmental Protection Agency, Hazardous Waste Combustion Frequently Asked Questions, (March 26, 2002). 12

13 4. Johansen, V.C, Hawkins, G. J., Mercury Speciation in Cement Kilns: A Literature Review, PCA R&D Serial 2567 (Portland Cement Association, 2003). 5. Senior, C.L., Sarofim, A.F., Zeng, T., Helble, J.J., Mamani-Paco, R., Gas-phase transformations of mercury in coal-fired power plants, Fuel Process. Technol., 2000, 63, Gaspar, J.A., Widmer, N.C., Cole, J.A., Seeker, W.R., Study of Mercury Speciation in a Simulated Municipal Waste Incinerator Flue Gas, paper presented at the International Conference on Icineration and Thermal Treatment Technologies, Oakland, California, May 12-16, Sprung, S., Technological Problems in Pyroprocessing Cement Clinker: Cause and Solution. (Beton-Verlag, 1985). 8. EPA Air Toxics Website - Utility Toxics HAP Study, 9. U.S. Environmental Protection Agency, Hazardous Waste Combustion: NODA Documents, December 19, Richards, J. Compilation of Cement Industry Air Emissions Data for 1989 to 1996, SP125, Portland Cement Association, Skokie, Illinois Portland Cement Association, An Analysis of Selected Trace Metals in Cement Kiln Dust, SP109T, Skokie, Illinois, Owens, W.D., Sarofim, A.F., Pershing, D.W. The use of recycle for enhanced volatile metal capture, Fuel Process. Technol., 1994, 39,