An Economic Basis for Developing Mercury Control Strategies

Introduction

The U.S. Environmental Protection Agency (EPA), through Title III of the Clean Air Act Amendments (CAAAs), is in the process of reviewing and establishing regulatory standards to control air toxic metal emissions. New standards have already been implemented for several source categories, and more are expected shortly. With respect to utilities, two major studies have recently been completed to determine whether regulations are necessary. Assuming particulate regulations remain the same, most trace elements, with the exception of mercury and selenium, are expected to be controlled effectively by conventional particulate control devices. ¹

The EPA Mercury Study Report to Congress (December 1997) lists total point-source mercury emissions at 158 tons/yr, with coal contributing 72 tons/yr, medical waste incinerators 16 tons/yr, and municipal waste combustors 30 tons/yr.² Coal is now the primary source of anthropogenic mercury emissions in the United States, accounting for 46%. However, on a worldwide basis, the projected increase in coal usage over the next two decades in China, India, and Indonesia will dwarf the current U.S. coal consumption of 900 million tons per year. Consequently, from a worldwide perspective, mercury emissions will increase significantly unless an effective control strategy is implemented. Given the above, this paper will focus discussion exclusively on industrial sources of mercury emissions that utilize coal.

EPA, in its Mercury Study Report to Congress, performed cost estimates of four primary candidates for high mercury removal in utility systems—fuel switching, advanced coal cleaning, activated carbon injection, and carbon filter beds. The cost-effectiveness for mercury control in utility systems was estimated at between $67,000 and $70,000 per pound of mercury removed. Other cost analyses for utility systems in the Report to Congress show costs varying from $4940 to $92,000 per pound of mercury removed.³ Before control strategies can be developed/implemented, utility companies need to know where each technology fits along the cost continuum and what variables under their control affect these costs.

According to the EPA mercury report, more detailed economic analyses, with special emphasis on cost-sensitivity analysis, are required. For example, sensitivity results will help provide the basis for determining critical parameters that should be evaluated in a full-scale test burn. For a given technology, knowledge of which parameters most significantly impact cost will help direct research efforts. Cost analysis should be performed on new technologies throughout their development stages. In addition, detailed economic analyses will address the benefits of cocontrol of multiple pollutants, provide data and models for plant-specific evaluations, and allow potential cost reductions resulting from technology improvements to be evaluated. Research efforts through the Center for Air Toxic Metals (CATM) at the Energy & Environmental Research Center (EERC) will continue to support development and demonstration of new mercury reduction technologies as well as to develop models that can be used to perform economic analyses.

Overview of Control Options

A number of options for reducing emissions of mercury are commercially available, and others are in various stages of development. For utilities, it is uncertain whether reductions could be accomplished cost-effectively with either the widespread application of a combination of existing mercury emission reduction options—energy efficiency improvements, coal cleaning, fuel switching, and retrofitting scrubbers and baghouses—or commercialization of new, developing technologies, including the use of sorbents through direct injection, filter beds, and other contacting methods. Each of these options has the potential to reduce mercury from the environment to a different degree; the cost and ultimate removal efficiency vary considerably. While, from a technical standpoint, some of these options are feasible, from a cost-effectiveness standpoint, they are not. To be cost-effective, options must represent significant improvements to existing technologies or entirely new technologies. However, given the low level of maturity and experience with these technologies, commercial deployment is still several years away. Thus the technical and economic viability of obtaining significant reductions of mercury (90% or greater) are still uncertain.

Efficiency improvements reduce mercury emissions, as well as emissions of all other pollutants, by decreasing the amount of coal, oil, or gas needed to generate a unit of electricity. Coal cleaning removes mercury from the coal prior to conversion, thereby minimizing emissions. Fuel switching can reduce mercury emissions because some fuels contain lower levels of mercury than others. For example, some coals contain less mercury than others, and fuels such as natural gas and renewables contain very little mercury. Nuclear fuels do not contain any mercury. Back-end controls such as wet scrubbers, baghouses, and sorbent injection offer more operational and fuel flexibility while removing mercury from exiting flue gases. Implementation hurdles to each of these options are plant-specific, depending upon the type of coal being burned, whether the coal is already being cleaned, the type of pollution control equipment already employed at the plant, and a host of other factors. It is anticipated that a combination of several existing measures may produce a significant reduction below current levels, albeit not necessarily cost-effectively. Some generalizations can be made about the relative cost- effectiveness of each emission reduction option to evaluate the fiscal practicality of obtaining reduction. It should be noted that while many options are available to reduce mercury released to the environment, very few have the ability to reduce it significantly, i.e., above 90%.

Combined Sulfur and Mercury Control in a Wet Scrubber

Approximately one-third of fossil utility boiler capacity is currently equipped with wet scrubbers to reduce emissions of SO₂. Wet scrubbers were originally theorized to have the capability of removing essentially all of the mercury from the flue gas, based on the low temperature at which these systems operate. However, while tests indicate that the wet scrubber is highly effective at removing oxidized mercury (HgCl₂), it has relatively little effect on the elemental mercury that is present in the flue gas. Between 85% and 95% reduction of oxidized mercury across a wet scrubber has been consistently measured, while essentially zero removal of elemental mercury has been observed. ^3-6 Removal rates for the oxidized mercury have been further increased by increasing the L:G ratio. For instance, 98% to 99% capture of oxidized mercury was noted at L:G ratios of 120 to 130 gal/1000 acf.⁴,⁷

The ratio of elemental to oxidized mercury that is formed during combustion is a strong function of coal type, with the amount of chlorine present in the coal playing a primary role. In the EPA Mercury Report to Congress, it is reported that 94% of the vapor mercury was oxidized mercury for a bituminous coal, while 32% was oxidized for a subbituminous coal. Other researchers agree with this general trend, with the quantities of oxidized mercury varying with specific coals tested, combustion rig used, gas temperature at measuring point, and measurement technique used.

The mercury removal effectiveness of the wet scrubber appears to depend not only upon the mercury species, but also the scrubber design (mass transfer characteristics, or how effectively the scrubber liquid contacts the vapor). These variations were very apparent in the results from full-scale tests performed by the U.S. Department of Energy (DOE) and EPRI.7 Their results showed a variation in mercury reduction from subbituminous coals from about 10% to 68% and from 45% to 84% for bituminous coals. Therefore, it is difficult to predict with any certainty the amount of mercury reduction to expect across a scrubber for a specific case without measurements of the mercury species present in the flue gas.

Mercury Removal Using Sorbent Injection

The process of mercury control via sorbent injection is relatively simple. A powder-type additive is injected pneumatically into the flue gas between the air heater and the particulate control device. The sorbent, in theory, captures the vapor-phase mercury through physical adsorption and/or chemical bonding. If the sorption occurs rapidly enough, the mercury is trapped by the injected sorbent particles before they reach the electrostatic precipitator (ESP) or baghouse, and the mercury is then collected by the control device along with the fly ash. If most of the sorption occurs before the mercury reaches the device, then good collection should be achieved with either the ESP or baghouse. However, if much of the sorption does not occur upstream, then better control would be expected with a baghouse, because the baghouse serves to some extent as a fixed-bed contactor, providing a gas–solid contact interface unlike that provided by an ESP. Since most utilities are fitted with either an ESP or a baghouse, sorbent injection is applicable to most utilities. This method has the potential to control both Hg⁰ and Hg²⁺, appears easy to retrofit, and would be applicable to both industrial and utility boilers.

Since mercury in the gas stream from coal combustion is present in trace quantities, usually in the range of 5 to 20 µg/m³ (about 1 ppbv), under ideal conditions, only very small amounts of sorbent may be necessary. Assuming a mercury concentration of 10 µg/m³ and a sorbent-to-mercury mass ratio of 1000:1, the required sorbent loading is 10 mg/m³, which is only 0.1% to 0.2% of a typical dust loading of 5–10 g/m³ (2.2–4.4 grains/scf). However, in reality, significantly higher quantities of sorbent are often needed. Sorbent ratios as high as 100,000:1 may be required, with a corresponding sorbent loading of 1000 mg/m³. The real add rates are the subject of many research and demonstration efforts, with results indicating that sorbent capacity is highly dependent on temperature, mercury species, mercury concentration, and other flue gas constituents. Mercury capture by sorbents is reported to be affected by SO₂, NO_x, HCl, and O₂ and possibly CO, Cl₂, and HF.⁸

Sorbent effectiveness greatly diminishes as flue gas temperatures increase. More experimental data and research are needed to better define the amount of sorbent needed as a function of temperature.⁹ Required rates for new/improved sorbents (including carbons) are hoped to be much lower, especially for higher-temperature applications such as in most utility plants that operate with ESPs at 300 –350°F. Until the flue gas interferences/interactions, temperature effects, sorption capture, mechanisms, and kinetics of reactions with other gases are known, predicting sorbent effectiveness will be very difficult. For now, an empirical approach must be taken, resulting in variable (unknown) sorbent rates that must be applied on a case-by-case basis.

Coal Cleaning

Coal is generally cleaned through the physical separation of particles low in mineral content from particles high in mineral content. Physical separation is typically based on the difference in density between organic coal and the associated minerals or differences in their surface properties. Hydrothermal, chemical, and biological methods may also be used, but these have not yet been applied commercially. The driving force behind much of the development work in coal cleaning has been to increase the carbon (energy) recovery during conventional cleaning. Removal of sulfur has also been an important goal. Another more recent objective has been the development of low-ash-content products to promote new uses of coal. The CAAAs of 1990 have added mercury removal as a target of coal-cleaning research as a preventive approach.

The factors affecting the liberation of mineral matter are themselves complex and difficult to understand. In addition, the heterogeneity of coal makes it difficult to generalize about coal-cleaning processes and expected results. Therefore, the effectiveness of coal cleaning, expressed as amount of ash, sulfur, or mercury removed, is variable. In addition, removal of sulfur and mercury is often limited (to some proportion) to that which is associated with the mineral matter. Physical cleaning does not remove organically associated sulfur or mercury. Therefore, to obtain high removals (75% to 95%) of sulfur and mercury, hydrothermal or chemical methods must be employed.

The EERC, with funding from EPA through CATM, has been investigating the environmental aspects of hydrothermal treatment. The primary objective of the work has been to develop a continuous processing method for removing organic sulfur along with chlorine and selected hazardous air pollutants (HAPs) with subcritical water. This work is an extension of the EERC hot-water-drying process that is currently being used in Alaska as a part of the Arthur D. Little–DOE Clean Coal project to demonstrate the use of coal as fuel for a diesel engine and work performed for EnerTech that led to its development of EFuelM, a clean-burning slurry fuel prepared from municipal solid waste.

Work performed under a project jointly sponsored by CATM, DOE, and the Illinois Clean Coal Institute showed that both sulfur and mercury can be reduced from bituminous coals. Compliant coal (<1.2 lb SO₂/MM Btu) was produced from a 3.5% sulfur coal. In addition, the following reductions in HAPs were noted: 99% mercury, 85% arsenic, 29% selenium, and 53% chlorine. The upgraded product had a heating value of 14,450 Btu/lb. Solids recovery for the process was only 68%; however, the energy recovery was significantly higher, at 90%.¹⁰ The economic analysis presented in the next section is based on these results.

Assigning Costs of Control Technologies

To facilitate easy comparison of options for controlling sulfur and mercury from utility boilers, it would be ideal if the control costs for each technology could be presented as one number, a unitized cost for that option, such as $/ton SO₂ or $/lb mercury removed. However, as the analysis begins, one quickly finds that determining a single, generic number representing the cost of control is difficult. The control costs calculated are dependent upon the specific situation of the plant or utility, such as the boiler size, capacity factor, coal being fired, and whether other plants within the utility are over- or undercontrolled for SO₂, and market assumptions, such as the market value of SO₂ allowances and the utility's policy of selling or banking excess allowances. To address this, a variety of sensitivity analyses are presented to provide an idea of the variability of control costs. For comparison, a base case was chosen to allow a more detailed comparison of control technologies. Some of the underlying assumptions and methods used to calculate the control costs are presented in this section, with the comparison of the results provided in the following section.

Three control technologies with the ability to remove significant quantities of mercury were compared: sorbent injection, wet scrubbing, and coal cleaning. Sorbent injection is used primarily to control mercury and has little or no effect on other priority pollutants. Therefore, all costs for sorbent injection were attributed to the mercury removal, and the cost of mercury control was simply the total cost associated with injection divided by the amount of mercury removed. The control cost is a function of the flue gas temperature, sorbent add rate, sorbent cost, required mercury removal, and the species of mercury present. For this analysis, the total control costs included both a fixed cost of installing the injection system and a variable cost consisting primarily of sorbent cost, but also including operating and maintenance (O&M) costs. Sorbent add rates of 3.8, 7.0, and 18.7 lb/mmdscf flue gas at a cost of $0.55/lb sorbent were used for the 25%, 50%, and 90% mercury reduction cases, respectively. These add rates were based on a flue gas temperature of 315°F. Annual costs were calculated as a function of boiler size and converted into unitized costs of cents/kWh and $/lb Hg removed. Sensitivity analysis was performed using the cost of the sorbent and the amount of sorbent added as variables.

Comparing the mercury costs for wet scrubbing and coal cleaning becomes more challenging. First, both of these technologies control multiple pollutants, with sulfur and mercury being the primary ones. Second, on average, each technology controls to a different level, i.e., although conventional coal cleaning is relatively inexpensive, it may remove only 20% of the sulfur and mercury, whereas the more expensive scrubbing may remove 95% of the sulfur and 75% of the mercury. The question of how to treat a technology that does not result in compliance also needs to be considered. Third, both of these technologies respond differently to different coals. For example, coal cleaning can remove a much higher percentage of sulfur from a coal with mostly pyritic and inorganic sulfur, but has difficulty cleaning coals with significant quantities of organically associated sulfur. Fourth, for the scrubber the unitized costs will vary significantly as a function of plant size, capacity factor, and the inlet sulfur content of the coal being fired. A scrubber designed for a low- to medium-sulfur coal and a small plant can cost upwards of five times that for a large plant firing a high-sulfur coal. Fifth, other benefits will accompany the control technology, such as lower ash disposal costs and lower operating and maintenance costs due to a lower ash content in the cleaned coals, and should be accounted for.

The methodology used in calculating the cost of control is summarized in Table 1. First, the annual cost for implementing the control technology was calculated. This cost includes both the levelized capital cost and the annual O&M cost. The annual cost was calculated over a range of plant sizes from 100 to 1000 MW and for capacity factors of 0.65 and 0.80. Next, the total amount of sulfur removed for each technology was calculated. For the case of the scrubber, it was assumed that 95% sulfur removal was obtained in the treated stream. The cost of lower SO₂ removal was calculated by assuming a partial bypass of the flue gas, maintaining the 95% removal in the treated flue gas, and sizing a smaller scrubber. For the coal cleaning, the removal efficiencies and average treatment cost shown in Table 2 were used. A fuel premium cost was calculated based on the higher coal price, accounting for the lower coal usage due to the Btu enhancement.

The market price for SO₂ was used to determine the value of the SO₂ removed. Over the past year, this price has risen from about $100 to $200/ton SO₂, with $200 being the current price. Therefore, $200/ton was assigned as the value for SO₂. Sensitivity analysis was performed over a range of $100 to $250/ton. The value of the SO₂ removed was then assigned as a credit and subtracted from the annual cost of control in a attempt to put technologies that undercontrol on the same basis as those that overcontrol for SO₂.

Coal cleaning also offers a number of benefits to the utility in addition to SO₂ reduction in the form of lower operating maintenance costs resulting from reduced ash content and lower waste disposal costs. In this analysis, these benefits were recognized and their value was attributed to determine the overall costs of control using the various coal-cleaning technologies. The credit allowed for improvement in O&M was $7.21/ton of ash removed (based on Coal Quality Impact Model results).¹¹ A credit of $20/ton of ash removed was assumed for reduction in waste disposal costs. Further, since it was assumed that the coal-cleaning plant is independently owned, the coal premium and cost savings, for practical purposes, were not a function of boiler size. The credit for ash reduction was subtracted from the annual cost of control to give an annual cost of control that was attributed to mercury control.

In addition to performing the analysis over a range of boiler sizes and capacity factors, three different coals were used to evaluate the impact of coal sulfur level on control costs. Table 3 summarizes the properties of the three test coals. For those cases where it was necessary to select a single condition to allow a comparison of multiple technologies, the base case of a 500-MW plant operating at a capacity factor of 0.65 firing Ohio coal was used. For these base cases, the SO₂ allowance was assigned a value of $200/ton and the scrubber was designed to treat the entire flue gas stream. Mercury removal in the scrubber was assumed to be 75% for the base cases.

Comparison of Control Technology Costs

The assignment of the costs associated with installing and operating a scrubber for mercury control is the most complex of the three technologies compared. The cost of scrubbing varies considerably with both boiler size and the level of sulfur in the coal. The installed cost of a scrubber to remove 90% of the SO₂ from the three comparative coals varied from a high a $450/kW for a 50-MW plant to $120/kW for a 500-MW plant. This relates to a very broad range of control costs as shown in Figure 1 ($1560 to $231/ton SO₂). The cost of scrubbing the 500-MW plant firing the Illinois No. 6 coal is further reduced to $198/ton when a capacity factor of 0.80 is used rather than 0.65.

Figure 1

The variations in both annual cost and unitized cost for scrubbing for mercury control are also significant when the control costs are normalized by adding the appropriate credit for SO₂ reductions in an effort to allocate costs for mercury control. One way of analyzing the costs associated with scrubbing for mercury control is to design a scrubber to treat only a portion of the flue gas stream so that a target mercury reduction can be achieved. This may or may not result in meeting SO₂ compliance levels. Figure 2 presents a case where it is assumed that 80% of the mercury in the flue gas is in the oxidized form and that 95% of that is captured in the scrubber. Target removal levels of 25%, 50%, and 75% of the inlet mercury were met by bypassing a portion of the flue gas stream. The target level of 75% represents the case with no bypass. To achieve 25% and 50% removals, significantly smaller scrubbers would be required, and control costs would likewise be smaller. As seen in Figure 2, the levelized cost of Hg control would be 0.81 cents/kWh for 75% removal in a 100-MW plant and 0.06 cents/kWh for a 1000-MW plant. However, if less stringent mercury control is required, the costs are reduced considerably, to as low as 0.02 cents/kWh for 25% Hg reduction in a 1000-MW plant.

Figure 2

An important factor with scrubbing for mercury control is the amount of oxidized mercury in the gas stream. In Figure 3, the cost of achieving 50% mercury reduction in the gas stream was calculated as a function of the percent of oxidized mercury. Less flue gas bypass and bigger scrubbers were required as the percent of oxidized mercury decreases. This results in levelized costs increasing from 0.54 to 0.72 cents/kWh for the 100-MW case as the percent of oxidized mercury decreased from 80% to 60% and an increase from 0.10 to 0.13 cents/kWh for the 1000-MW case. Figure 4 presents these same costs calculated on a $/lb-Hg-removed basis. For the 100-MW plant, costs increased from $111,289 to $148,386/lb Hg as the oxidized mercury was reduced from 80% to 60% and from $19,592 to $26,123/lb Hg for the 1000-MW case.

Figure 3

Figure 4

All of the previous analyses were performed on the assumption that SO₂ allowances were worth $200/ton. They also assumed a medium-sulfur (3.43%) Ohio coal. Mercury control costs are very dependent upon the assumed value of the SO₂ allowances and somewhat dependent upon the level of sulfur in the coal. Figure 5 compares the cost of mercury control for the three comparative coals as a function of the cost of SO₂ allowances. For the purposes of these calculations, it was assumed that all of the flue gas was treated and that the mercury capture was 75% in the scrubber. As the cost of the SO₂ allowances increases, more credit is given for sulfur reduction, with the result being a lower calculated cost of mercury control. The data in Figure 5 show a dramatic difference in the costs associated with mercury control for the three different coals. A scrubber for a high-sulfur coal has the ability to remove significantly more sulfur than a low-sulfur coal, and, therefore, a much larger credit can be given to a high-sulfur coal for SO₂ allowances. The result is that a lower percentage of the costs need to be attributed to mercury control, and the calculated costs for mercury control will be significantly lower for the higher- sulfur coal. Another important point becomes evident from Figure 5. When the cost of SO₂ allowances becomes equal to the cost of controlling SO₂, $231/ton for the Illinois coal, the mercury is removed at no cost. Calculations actually show a negative value for those cases. More cases where a negative mercury control cost is calculated will be shown for the coal-cleaning options.

Figure 5

The levelized cost of coal cleaning, especially when done off-site, independent of the utility, does not vary with the size of the boiler or the capacity factor, simplifying the cost analysis. Like the case of the scrubber, however, costs will vary with the level of sulfur in the coal based on the larger amounts of SO₂ allowances being generated for a high-sulfur coal as compared to a low-sulfur coal. A comparison of the calculated costs for mercury control for the three test coals is given in Figure 6. The negative values for conventional cleaning and hydrothermal treatment imply that the cost of removing sulfur was calculated to be less than the $200/ton SO₂ allowance cost.

Figure 6

Figure 7 presents the costs associated with sorbent injection for mercury control. The control cost is somewhat sensitive to the unit size, but not to the same degree as scrubbing because of the smaller contribution of fixed (installation) costs for the sorbent injection system. This analysis shows a cost ranging from 0.205 cents/kWh to 0.145 cents/kWh to obtain 90% mercury removal as the size of the unit increased from 100 to 1000 MW. The control costs for lower mercury reductions are significantly lower: 0.152 cents/kWh for 90% reduction in a 500-MW plant compared to 0.068 and 0.041 cents/kWh for 50% and 25% reduction, respectively. As mentioned earlier, sorbent effectiveness is sensitive to add rates and temperature. For example, the cost difference between operating at 300 and 325°F can result in an increase in cost of control by 50%.

Figure 7

Figure 8 normalizes these same costs on a $/lb-Hg-removed basis, yielding some interesting results. On this basis, the cost per pound mercury is less for 90% removal in a 100-MW plant than for 50% removal, which is less than for 25% removal. This may lead one to believe that it is cheaper to remove 90% mercury for this case; however, the data in Figure 7 reflect the true costs, showing that the 90% removal would cost two to three times that compared to the lower removal efficiencies. Even at 1000 MW, normalizing the data on a $/lb-Hg-removed basis shows the 50% removal case to be cheaper than the 25% removal. The primary reason for this phenomena is that the fixed costs do not vary considerably as a function of the desired removal efficiency, while the amount of mercury removed does. Comparing the 25% and 50% removal cases for a 100- and 1000-MW installation, the proportion assigned to fixed costs is $17,425/lb Hg and $3550/lb Hg for the 25% case compared to $8715/lb Hg and $2980/lb Hg for the 50% case, respectively. Consequently, care should be taken when comparing control methods on a $/lb-Hg-removed basis. Costs presented on a cents/kWh basis more truly represent the cost the utility will incur when implementing the control strategy.

Figure 8

One way to compare the different technologies is to look at the maximum reduction that each option can provide and the corresponding cost. Figure 9 shows such a comparison between the cost of wet scrubbing, conventional coal cleaning, froth flotation, hydrothermal treatment, and sorbent injection for mercury control. The reader should note that this may not provide a good comparison since each technology is not able to meet the same level of mercury reduction and, as such, is restricted to applications where less than maximum is required. For a 500-MW plant operating at a capacity factor of 0.65 and assuming an SO₂ allowance value of $200/ton, Figure 9 shows that sorbent injection is the most cost-effective when no credit is taken for SO₂ allowances and becomes the most expensive when credits of $250/ton SO₂ are given to the other control technologies. Of the other technologies, conventional coal cleaning is the most cost- effective method for controlling mercury, followed by hydrothermal treatment, froth flotation, and wet scrubbing. It should be noted that the more costly options generally provide higher levels of mercury reduction. Also, of these technologies, only wet scrubbing and hydrothermal treatment have the potential of controlling SO₂ emissions to the required year 2000 level of 1.2 lb/MMBtu set in the 1990 CAAAs. The other technologies would require additional treatment or the purchase of SO₂ allowances.

Figure 9

Recognizing that each technology does not have the same level of mercury or SO₂ control causes some difficulty in making a fair comparison among the technologies. In an attempt to put all technologies on the same basis, the total annual cost of achieving a 1.2-lb/MMBtu sulfur limit and 90% mercury reduction was calculated for each control strategy. It was assumed that SO₂ allowances would be purchased where necessary and that similar mercury allowances would be available and also be purchased as necessary to meet 90% Hg reduction. In Figure 10, the annual cost for the various control technologies was calculated as a function of the price of SO₂ allowances, assuming the cost of a mercury allowance was $20,000/lb Hg removed. For this case, sorbent injection, conventional cleaning combined with buying credits, and buying credits are the low-cost options when the cost of SO₂ credits is low (less than $100/ton). As the cost of allowances increased, hydrothermal treatment became the most economically attractive option, with conventional cleaning and sorbent injection costs steadily increasing but remaining as second-choice alternatives. At around $200/ton SO₂, the cost of scrubbing approximates that of sorbent injection, buying credits, and froth flotation. At and above $250/ton SO₂, the scrubbing option becomes an economically attractive option.

Figure 10

In Figure 11, the cost of achieving 1.2 lb/MMBtu SO2 and 90% mercury control is plotted as a function of the cost of fictitious mercury allowances. The "allowances" can represent an add-on control technology. For example, for the conventional coal-cleaning case, sorbent injection or some other technology could be used to increase the mercury removal efficiency from the base of 21% to the target of 90%. The impact of the coal cleaning combined with sorbent injection on the total cost of control could then be determined. Also, although EPA has not proposed a plan for mercury control, if it uses allowances in a similar manner as it has for SO₂ control, Figure 11 can help determine the impact of the price of these allowances on the total control cost. Using this analysis, hydrothermal treatment is the most cost-effective treatment method for all but very low mercury allowance costs. The high sulfur and mercury removal efficiencies of hydrothermal treatment offset its high cost ($20/ton fuel premium). Of the other technologies, the low cost of conventional cleaning makes this option attractive even though considerable mercury and sulfur allowances would need to be purchased. Sorbent injection to 90% Hg reduction combined with buying SO₂ credits becomes more attractive than conventional cleaning as mercury allowance costs increase above $30,000/lb Hg and more attractive than the other technologies compared in this analysis for allowance costs above $15,000/lb Hg. Finally, based on the assumptions of this analysis, with the exception of buying credits, scrubbing appears to be the most expensive option, mainly due to high capital and operating costs. However, it does offer some flexibility with regard to fuel choice.

Figure 11

Future Direction

This project is ongoing, and, therefore, the results presented in this paper represent a progress report rather than a final conclusive comparison of air toxic metal control technologies. Other mercury reduction options and technologies will also be evaluated, including other sorbents and sorbent-contacting methods, scrubbers, and fuel switching. The cost and the effectiveness of the various control technologies will be compared and different mercury reduction strategies presented as work progresses.

Conclusion

Many options and strategies are available for reducing the quantities of mercury released to the environment from utility plants. However, many of these options either provide only a small to moderate reduction in mercury and/or have a very high associated cost. In comparison to other industries such as municipal solid waste, even the lowest-cost estimates are still a factor of 5 to 10 higher for utilities. Assuming significant reduction of mercury is needed, such as 90%, simple or stand-alone (nonintegrated, nonsynergistic) control options are not viable or cost-effective. Additional research, developmental work, and pilot-/full-scale testing are needed before most options can be considered commercially viable. Preliminary conclusions from this project are highlighted below:

1. Based on today's knowledge, technologies, and implied assumptions, cost-effective options or solutions for mercury reduction for utility plants do not exist in comparison to options and costs available to other industrial sources. Many options provide only low to moderate reductions at a premium cost.
2. Given available technologies, the cost of mercury control for utilities is a factor of 5 to 10 times higher compared to other source categories.
3. Options for significant mercury reduction (i.e., 90%) are costly.

The cost of mercury control is extremely sensitive to unit-specific parameters such as fuel type, unit size, unit configuration, capacity factor, emission control equipment, equipment layout and space availability, value of SO2 allowances, etc.

1. The cost of mercury control for smaller units (<100 MW) as compared to large units (>1000 MW) is projected to be 1 to 3 times higher on a cents/kWh or $/lb-Hg-removed basis.
2. Options involving control of multipollutants, i.e., mercury and SO₂, may provide cobenefits but are extremely sensitive to the value of SO₂ allowances, the cost of alternative mercury control options, and the proportion of oxidized mercury.

Wet scrubbing is a relatively expensive method of controlling mercury, except in the case of large plants operating at high-capacity factors using high-sulfur and chlorine coal with a high proportion of oxidized mercury.

1. Installing a scrubber for mercury control alone is not cost-effective, unless the market for SO₂ allowances is favorable and excess allowances can be generated.
2. Sorbent injection and hydrothermal treatment are two control options that have the potential to achieve high removal efficiencies.

Based on preliminary results, coal upgrading using hydrothermal treatment offers significant reductions in mercury (>90%), sulfur, and ash. The benefits obtained from the reduced sulfur and ash partially offset the cost of the hydrothermal treatment.

Conventional coal cleaning is a very cost-effective method for mercury removal; however, it can only achieve low-to-moderate levels of mercury reduction. In addition, over 70% of eastern coal is currently cleaned, leaving little margin for additional mercury reduction.

Advanced coal-cleaning methods, including froth flotation and selective agglomeration, can achieve moderate reductions in mercury; however, they are not cost-effective relative to other options.

New technologies and/or sorbents in the developmental stage show promise for lower-cost mercury reduction options.

For those plants with existing scrubbers, methods to preoxidize the mercury to enhance the mercury removal in the scrubber should be examined.

Mercury reduction costs for utilities need to be verified using plant-specific data, since small deviations in fuel characteristics and operating conditions can significantly impact the cost of control.

Costs for mercury reduction are likely to decrease as more knowledge and experience are gained.

The intended final product of this work is a computer model that can be used for strategic planning of combined air toxic and SO₂ control. This work should provide guidelines regarding the type of information/measurements that are required to evaluate control options specific to a particular plant. The results should also be useful to technology developers and researchers by showing the sensitivity of cost-effectiveness to various parameters and indicating areas where further research can most dramatically impact bottom-line cost. Follow-up papers/reports on this topic will be published by the EERC.

About the Author: John H. Pavlish is a senior research manager and associate director of a Multiyear, multimillion dollar Center for Air Toxic Metals Program at the Energy & Environmental Research Center of the University of North Dakota.

1. Pavlish, J.H.; Benson, S.A. "A Summary of Key Air Toxic Results from the Center for Air Toxic Metals (CATM)," Presented at the EPRI–DOE–EPA Combined Utility Air Pollutant Control Symposium, Washington, DC, Aug. 25–29, 1997.