2. Analysis

The analysis in this report was prepared using the Energy Information Administration’s (EIA) National Energy Modeling System (NEMS). The reference case for the analysis was based on EIA’s Annual Energy Outlook 2005 (AEO2005), which incorporates only final regulatory action under existing laws.⁶ It should be noted that the projections in the cases in this report are not statements of what will happen but of what might happen, given the assumptions and methodologies used. The reference case projections are business-as-usual trend forecasts, given known technology, technological and demographic trends, and current laws and regulations. EIA does not propose, advocate, or speculate on future legislative and regulatory changes. All laws are assumed to remain as currently enacted; however, the impacts of planned regulatory changes, when defined, are reflected. Consistent with standard EIA practice requiring policy neutrality in baseline projections, the reference case in the AEO2005 did not include pending or proposed actions, such as the proposed Clean Air Interstate and Clean Air Mercury Rules. Neither of these regulations had been finalized prior to the preparation of the AEO2005. However, as requested by the Senators Voinovich and Inhofe, for this report, the reference case has been modified to incorporate the power plant NOx and SO2 emission caps in EPA’s proposed CAIR regulations.

The EIA analysis of mercury control strategies contained in this report, like other EIA analyses, focuses on the impact of the provisions in the bill on energy choices made by consumers in all sectors and the implications of those decisions for the economy. This focus is consistent with EIA’s statutory mission and expertise. The study does not quantify, or place any value on, possible health and environmental benefits of curtailing mercury emissions.

Analysis Cases

At this time, there is significant uncertainty about the degree to which mercury can be removed from some coals. Currently, there are two main approaches being considered for controlling power plant mercury emissions; 1) reducing mercury emissions using technologies primarily designed to remove SO2, NOx, and particulate emissions (often called co-benefit reductions), and 2) reducing mercury emissions using technologies specifically designed to reduce mercury.

Table 3 below provides the mercury removal factors used in recent EIA and EPA modeling work for different power plant configurations and coals. As shown for EIA, the assumed percentage of mercury removed varies from as low as 0 percent for many plant configurations using lignite coal to as high as 95 percent for several plant configurations using bituminous coals. Both sets of factors in Table 3 show that no coal plants using subbituminous or lignite coals are assumed to be able to comply with a 90-percent removal requirement using SO2, NOx, or particulate control technologies (i.e., co-benefit reductions) alone.

In order to continue to meet electric generating requirements and comply with a 90-percent mercury removal requirement at coal plants without using technologies specifically designed to reduce mercury, companies with plants that currently burn subbituminous or lignite coals would have to switch them to bituminous coals and add any needed NOx, or SO2 controls to reduce mercury emissions by 90 percent. This would require major changes in coal supply patterns, because subbituminous and lignite coals together accounted for roughly 50 percent of U.S. coal production in 2003. Alternatively, the companies could reduce their use of coal and increase their use of natural gas and renewable fuels or turn to mercury-specific control technologies.

While many approaches are being considered, the most common technology discussed to remove mercury from coal plants is activated carbon injection (ACI). ACI systems have been widely deployed in other industries, mainly in waste-to-energy plants (municipal solid waste (MSW) plants). In those applications, they have achieved mercury removal rates in excess of 90 percent. However, ACI systems are only now being widely tested on U.S. coal plants and these plants have several characteristics that will tend to make mercury removal more difficult. For example, coal plants are typically much bigger with more flue gas to treat. They also have much lower concentrations of mercury and chlorine in the untreated gas, and it is questionable whether similar removal levels will be achievable for all coals. Sulfur and trace elements in U.S. coals may also pose problems that will have to be resolved. For example, efforts to remove mercury could create corrosive conditions that would damage other parts of the plants. Programs in the Department of Energy’s Office of Fossil Energy are actively exploring these issues.

It should be pointed out that the understanding of this technology is changing rapidly, and EIA normally assumes that this technology will be available in the mid-term as might be required to comply with the 2010 or 2018 mercury emission caps called for in the Clear Skies Act of 2003. Whether current ACI systems for coal plants would meet the analysis request requirement for a “commercially demonstrated technology” for deployment in the 2008 timeframe, particularly if 90-percent removal is required, is unclear.

Because of the uncertainty about the availability and performance of mercury removal technologies, this analysis includes five mercury control cases with two of the most stringent cases incorporating alternative mercury control technology assumptions. Table 4 describes the cases prepared. The first case, labeled pCAIR, incorporates the proposed SO2 and NOx emission caps of EPA’s proposed Clean Air Interstate Rule, and serves as the baseline case for this analysis. The second case, labeled EPA-Cap incorporates EPA’s proposed mercury cap and trade program, while the third case, labeled EPA-MACT, incorporates EPA’s proposed mercury MACT program. The final three cases incorporate a 90-percent mercury MACT with alternative assumptions about the availability and performance of ACI systems. These alternative cases are included in the most stringent mercury control case to illustrate the sensitivity of the results to the availability and performance of mercury control systems. They are not meant to project the expected evolution of the technology and the case without any ACI systems through 2025 is clearly unrealistic.

Mercury Emissions

In the MACT90, MACT90SL80, and MACT90NoACI cases, all coal-fired power plants would have to take action to remove 90 percent of the mercury in the coal they use, but there is no specific national emissions target. Under these assumptions, mercury emissions are projected to range from 8.9 to 9.9 tons in 2025. Mercury emissions in the no ACI case (8.9 tons in 2025) are lower than the other MACT90 cases (9.8 to 9.9 tons in 2025) because 46 gigawatts of coal plants opt to retire rather than comply with the 90-percent MACT without ACI.

Except for the no ACI case, the NOx and SO2 emissions paths are very similar in the mercury control cases. This occurs because these cases all assume the NOx and SO2 caps as imposed in pCAIR. NOx emissions fall from 4.1 million tons in 2003 to 2.2 million tons in 2025, about half the level expected without pCAIR. SO2 emissions fall from 10.6 million tons in 2003 to between

Mercury Control Compliance

In each of the mercury control cases, mercury emissions are projected to be lowered through a combination of fuel switching between coals and from coal to natural gas and renewables, SO2 scrubber retrofits, NOx SCR retrofits, and ACI technology retrofits.

Fuel Switching

Almost no fuel switching is projected in the EPA-Cap and EPA-MACT mercury control cases. In these cases, there is expected to be a very small shift from coal to natural gas and renewables (Figure 2). However, the impact of a 90-percent MACT strategy on coal usage patterns depends heavily on the performance and commercial availability of new mercury removal technologies. If these technologies are available and able to achieve 90-percent mercury removal on all plant and coal types little fuel switching is projected under a 90-percent MACT strategy. However, if these technologies are not commercialized with this level of performance as in the two 90 Percent MACT cases with limited ACI performance or no ACI altogether, there is projected to be significant switching from coal to natural gas, renewables, and oil. There is also projected to be a dramatic switch in the types of coals used. In 2025, coal generation in these two cases is projected to be between 8 percent and 11 percent below the level projected in the pCAIR case. Conversely, 2025 natural gas generation in these two cases is projected to be between 6 percent and 10 percent higher, while 2025 renewable generation is between 3 percent and 9 percent higher. The shift to natural gas is even more pronounced just after 2008 when the MACT takes affect. For example, natural gas generation in 2010, in the MACT90NoACI case is projected to be 28 percent higher than in the pCAIR case. To meet the more rapid growth in demand for natural gas, companies will have to develop new supplies, particularly for liquefied natural gas (LNG), than otherwise expected. In the pCAIR case, LNG imports are projected to grow from 0.4 trillion cubic feet (tcf) per year in 2003 to 2.5 tcf per year in 2010. In the MACT90NoACI case, they are projected to reach 3.7 tcf in 2010 to meet the larger demand for natural gas. The rapid siting, permitting, and constructing of the terminals needed to meet this growth in LNG may be very difficult. Unfortunately it is unlikely that more than a minimal volume of additional LNG imports over baseline case levels would be able to enter the country by 2008. It is more likely that there would be sufficient time to add additional regasification facilities and that there would be more available liquefaction capacity around the world in the 2009 to 2010 time frame.

The shift in coal usage patterns in the two 90-percent MACT cases with limited ACI performance and no ACI is dramatic, with western coal use falling while eastern coal use increases (Figure 3). Western coal is primarily subbituminous coal from which mercury removal is more difficult than from bituminous coal. As Table 2 indicates, no plant configuration using subbituminous coal is assumed able to comply with a 90-percent MACT using SO2, NOx, or particulate control technologies alone. Mercury removal for a 90-percent MACT from subbituminous coals would require ACI technology. Plants currently using subbituminous coals would have to switch to bituminous coals or to natural gas or would have to retire in order to meet this requirement. In these two cases, western coal production decreases to between 366 and 390 million tons by 2025, compared to between 874 and 914 million tons in the other mercury control cases. There is actually a 4 million ton increase in western coal production in the EPAMACT case, because the proposed standards for subbituminous coals can be readily met in most plants. In contrast to the west, Appalachian and Interior coal production in these cases increases significantly (Figure 4). The increase in Appalachian and Interior coal is not as large as the reduction in western coal because increased natural gas and renewable generation displace some of the coal generation and each ton of western subbituminous coal contains approximately 70 percent as much energy as each ton of eastern bituminous coal. There is also a significant reduction in electricity demand in these cases due to higher electricity prices. The increase in eastern coal production called for in these cases may be very difficult to achieve. Coal production in the east has been declining since 1990 and it may be very difficult to reverse this trend.

SO2 Scrubber Retrofits

A significant amount of SO2 scrubbers are projected to be added in all cases in order to comply with the SO2 emission caps established in pCAIR (Figure 5). For example, by 2025, 128 GW of coal-fired power plants are expected to be retrofitted with flue gas desulfurization scrubbers in pCAIR case. Even though SO2 scrubbers also contribute to mercury removal, similar levels of SO2 scrubber retrofits are expected in most of the mercury control cases. In the EPA-Cap, EPAMACT, and MACT90 cases, between 1 and 5 additional gigawatts of capacity are projected to be retrofitted with SO2 scrubbers. However, without commercialized mercury removal technologies capable of 90-percent removal, SO2 scrubbers are projected to play a much bigger role in reducing mercury emissions under a 90-percent MACT. In the MACT90SL80 and MACT90NoACI cases, coal plant operators would have to limit their coal use to bituminous coal and add SO2 scrubbers and NOx SCRs in order to comply with the 90-percent mercury MACT. As a result, in the MACT90SL80 case, SO2 scrubbers are expected to be retrofitted to an additional 29 gigawatts of capacity, while in the MACT90NoACI case, an additional 75 gigawatts of capacity are expected to add them.

NOx SCR Retrofits

The story for NOx SCR retrofits is similar to that of SO2 scrubber retrofits. All cases are projected to add a significant number of retrofits to meet the NOx emissions targets specified in pCAIR (Figure 6). Approximately 133 gigawatts of capacity are expected to add SCRs in the pCAIR case. In most of the mercury control cases, only a small amount of additional capacity is projected to add them. For example, in the EPA-Cap, EPA-MACT, MACT90, and MACT90SL80 cases, SCR retrofits grow to between 135 and 138 gigawatts of capacity. Only in the case with a 90-percent mercury MACT and no ACI technology available are significantly more SCRs expected. In the MACT90NoACI case, SCR retrofits are projected to be added to 184 gigawatts of capacity.

Activated Carbon Injection Retrofits

Fuel Prices

The minemouth price per ton of bituminous coals is generally higher because they have more energy per ton than subbituminous coals. Delivered coal prices per Btu to the power sector in 2025 are only projected to be between 30 and 36 percent higher than in the pCAIR case.

A similar pattern for natural gas wellhead prices is projected in the in the two cases with limited ACI performance or no ACI altogether - the price increase will be relatively large when the program first begins, but they will moderate over time as new resources are developed and brought to market. In 2010, natural gas wellhead prices are projected to be between 14 percent and 26 percent higher than in the pCAIR case. By 2025 the increases ranges from 2 percent to 5 percent.

Electricity Prices

The electricity price impacts of controlling mercury emissions generally increase with the level of mercury removal required (Figure 10). In 2010 and 2025, national average electricity prices in the EPA-Cap and EPA-MACT cases are projected to be less than 0.5 percent higher than prices in the pCAIR case. The size of these changes reflects the relatively modest mercury emissions reductions called for in the EPA-MACT case and the impact of the mercury safety valve which limits the mercury emissions reductions in the in the EPA-Cap case.

The electricity price changes in the 90-percent MACT cases are larger and very sensitive to the assumptions about the performance and availability of ACI mercury removal technologies. When ACI technologies are assumed to be available and able to achieve 90-percent mercury removal for all plant and coal types, a 90-percent MACT is projected to lead to electricity price impacts similar to those in the EPA-Cap and EPA-MACT cases. In the 90-percent MACT cases that assume limited ACI performance or no ACI, the increase in electricity prices in 2010 are projected to range between 18 percent and 22 percent. By 2025, the electricity price increases in these cases are projected to be smaller, ranging between 4 percent and 7 percent. The stronger impacts in 2010 result from the sharp impact on coal and natural gas markets when the MACT requirement takes effect in 2008. The shift from subbituminous and lignite coals to bituminous
coals, natural gas, and renewables is so rapid that prices increase sharply in the near term. Over time, new supplies of the various fuels can be developed and prices would be expected to moderate. However, caution should be used when viewing the price results in these cases, because predicting market responses to the fuel consumption shifts expected in these two cases has considerable uncertainty.

Regional electricity prices are generally expected to follow the national pattern, with small changes expected in the EPA-Cap and EPA-MACT cases, and larger changes in the cases with a 90-percent MACT, particularly those where commercialized technologies capable of removing 90-percent of the mercury from all plant and coal types are not assumed to be available (Figures 11, 12 and 13). In the two 90-percent MACT cases without commercialized technologies capable of 90-percent mercury removal on all plant and coal types, electricity prices are projected to be significantly higher in all regions, particularly in the near term. The largest regional price increases, in absolute terms, are expected in the MAPP, SPP, and RA regions, which all rely heavily on subbituminous coal. In these three regions, the 2010 electricity price increases are projected to approach 2.5 cents per kilowatthour. In percentage terms, their electricity prices in 2010 are projected to be as much as 45 percent, 36 percent, and 33 percent higher, respectively, than in the pCAIR case projections. By 2025, the price changes in these cases are projected to moderate as new fuel supplies are developed, but prices in the MAPP and SPP are still projected to be more than 0.8 cents per kilowatthour than those in the pCAIR case.

Resource Costs

The relative impact on resource costs in each of the cases generally follows the electricity price impacts discussed previously. In general, the resource cost impacts of controlling mercury increase as the mercury removal required grows. The one exception to this rule occurs in the EPA-MACT case, where the increase in resource costs is larger than in the EPA-Cap case, even though mercury emissions are lower in the EPA-Cap case. The discounted resource costs and safety valve payments are projected to be $2 billion in the EPA-Cap case and $8 billion in the EPA-MACT (Figure 11). This counter intuitive result occurs because the cap and trade strategy in the EPA-Cap case allows power plant operators to reduce mercury emissions at the plants where it is most economical. Conversely, in the EPA-MACT case, some plants where it is relatively expensive to reduce mercury emissions are required to install equipment to make emissions reductions.

In the 90-percent MACT cases, the assumptions about the availability and performance of ACI technologies lead to a wide range in projected resource cost impacts. In the 90-percent MACT case where ACI technologies are assumed to be available and able to achieve 90 percent mercury removal for all plant and coal types, discounted resource costs are projected to increase by $22 billion, roughly 10 times the cost increase projected in the EPA-Cap case and 3 times the cost increase expected in the EPA-MACT case. As noted above, even with these higher resource costs, the electricity price impacts of the MACT90 case are similar to those in the EPA-Cap and EPA-MACT cases. This occurs because the coal plants that incur these costs are not the plants that will set electricity prices during most hours. The resource cost impacts of a 90-percent MACT with limited ACI performance or without ACI are much higher still, ranging between $261 billion and $358 billion. The higher costs in these two cases reflect the costs of shifting away from relatively inexpensive subbituminous and lignite coals to other more expensive fuels. As mentioned previously, caution should be used when interpreting the near-term market price impacts from these scenarios.

Uncertainties

As with any projection, especially those that look out beyond a few years, there are considerable uncertainties. It is impossible to predict how existing generation or emissions control technologies might evolve in cost and performance or what currently unknown technologies might emerge to play unexpectedly important roles in the market. Of particular concern in this analysis are the cost and performance of technologies to remove mercury.

In recent years, substantial information has been gathered on the factors influencing mercury emissions at existing plants - i.e., mercury content of coal, coal rank, coal chlorine content, power plant particulate, SO2, and NOx control systems, etc. - but significant uncertainty remains. Experts at the EPA and the U.S. Department of Energy have different views on the mercury removal rates that should be assigned to particular plant configurations using various coals. Often their analyses use the same data sources, but because of variability in the data and their interpretation, they reach different conclusions. The understanding of what contributes to mercury emissions will likely improve in coming years as research efforts continue, but the outcome of these efforts is unknown.

One particular area of uncertainty concerns the role that NOx control devices, SCRs, play in removing mercury from lower rank coals (subbituminous and lignite). Evidence suggests that when combined with a wet scrubber for SO2 removal, they do enhance mercury removal in plants using bituminous coals. The same has not been found to be true for the lower rank coals but research is ongoing. Another area of uncertainty is the cost and performance of mercury removal systems. Supplemental fabric filter systems using ACI are expected to be a key technology in removing mercury. Tests of such systems have demonstrated their ability to remove mercury from bituminous coals, but full-scale tests on subbituminous and lignite coals are only now being performed.

Tables 3 and 4

Notes