Preliminary Cost and Performance Models for Mercury Control at Coal-Fired Power Plants

The U.S. Environmental Protection Agency (EPA) has announced it will regulate mercury emissions from coal-fired power plants, with proposed regulations to be issued in 2003. The feasibility and cost of achieving mercury emission reductions is thus a subject of considerable current interest. To assess mercury control options, the Integrated Environmental Control Model (IECM) developed for the U.S. Department of Energy’s National Energy Technology Laboratory (DOE/NETL) has been expanded to include performance and cost models for a variety of mercury control options. These preliminary models are based on a review of recent mercury information collection request (ICR) data, and on the results of pilot plant studies and other data sources employing carbon injection with and without flue gas humidification. Illustrative results using the IECM show that the feasibility and cost of achieving different levels of mercury reduction depend strongly on the fuel type and power plant configuration. In most cases, the presence of a flue gas desulfurization (FGD) unit or a selective catalytic reduction (SCR) system can have a significant (beneficial) impact on mercury removal efficiency and cost. However, because of limitations on the scale and coverage of available data, there is considerable uncertainty in current estimates of mercury control costs and capabilities. Current models and estimates will be refined as new data become available from ongoing programs. BACKGROUND In December 2000, the EPA announced its intention to regulate mercury (Hg) emissions from coal-fired power plants under the air toxics provisions of the Clean Air Act Amendments (CAAA) of 1990. Coal-fired power plants are the largest industrial source of airborne mercury accounting for about one-third of U.S. antropogenic emissions. Mercury controls would have to be installed by 2007 according to the current timetable. Recent legislative proposals for multi-pollutant controls including mercury also have been put forth in the U.S. Congress. Most policy proposals would establish national The EPA/DOE/EPRI Mega Symposium August 20-23, 2001 2 emission caps and allow emissions trading to minimize overall cost. Recent Congressional bills would require mercury reductions to a level of 90% below the 1997 level. While none of these multi-pollutant proposals have yet gained widespread Congressional support, they are nonetheless suggestive of the kinds of requirements that could be imposed on coal-fired power plants in the future. EMISSION CONTROL OPTIONS In general, the methods available to reduce or eliminate power plant emissions of mercury include: (1) switching to a cleaner fuel containing less of the undesirable constituent; (2) installing control technology to reduce or eliminate emissions; (3) improving power generation efficiency to reduce emissions per kilowatt-hour generated; (4) switching to a power generation technology with lower or no emissions; and (5) generating less electricity by reducing demand or by reducing the load factor of “dirtier” plants. The effect of these methods is summarized in Table 1. Table 1. Power plant operations affecting mercury emissions Effect on Mercury Emissions Power Plant Config and Operations Strategy Oxidized Mercury Elemental Mercury Conv. Coal Cleaning Decrease (highly coal specific) Electrostatic Precipitator Some decrease Some decrease Fabric Filter Some decrease Greater decrease Wet SO2 Scrubber Decrease No Effect Spray Dryer/Fabric Filter Some decrease Limited decrease Carbon Adsorption System Decrease (based on pilot-scale studies) The choice of a control strategy is typically dominated by the cost of alternative options. For existing coal-fired plants, coal switching and/or the installation of control technology historically have been the preferred approaches to environmental compliance for criteria air pollutants (SO2, NOx and particulates). For most existing plants, the most promising emission control option for reducing mercury emissions is activated carbon injection (ACI). This involves the injection of powdered activated carbon into the flue gas upstream of the particulate control device. Activated carbon is a specialized form of carbon produced by pyrolyzing coal or various hard, vegetative materials (e.g., wood) to remove volatile material. The technology has been proven in municipal waste combustors, but has not been directly scaled to utility flue gas applications. Mercury is adsorbed onto the sorbent and is then removed along with the fly ash in the fabric filter or electrostatic precipitator (ESP). Moderate removal of mercury is accomplished by ash alone. Circulating the ash in a fluidized bed has been shown to enhance the removal of mercury, particularly with the addition of activated carbon to the circulating fluidized bed . An additional benefit of circulating fluidized beds is the ability to add other sorbents to reduce acid gases. The EPA/DOE/EPRI Mega Symposium August 20-23, 2001 3 Once fuel is combusted, mercury can be identified in primarily two chemical states: elemental (Hg) and oxidized (Hg). The ability of environmental control systems to capture the mercury is dependent on the forms of mercury in the flue gas. The data collected by EPA’s ICR shows that wet scrubbers effectively remove oxidized mercury from the flue gas but are ineffective at removing elemental mercury. Efforts to convert the elemental mercury to an oxidized state using catalysts, such as those used in an SCR to reduce NOx emissions, have resulted in higher capture rates of mercury in scrubbers . MERCURY EMISSION CONTROLS Mercury control technologies, such as sorbent injection, have not yet been installed commercially at coal-fired power plants. However, data from smaller-scale tests indicate that mercury capture efficiency may be strongly affected by temperature, coal properties, and interactions with other environmental control systems. In the sorbent injection process, fine solids such as activated carbon are injected into the flue gas to adsorb or interact with gaseous mercury species. The adsorbed solids are then collected in a downstream particulate collector such as an electrostatic precipitator or fabric filter. For existing coal-fired plants with only a particulate collector such as an ESP (the predominant plant configuration), mercury control is nominally achieved by injecting activated carbon upstream of the ESP. To achieve high levels of mercury control, substantial amounts of carbon injection are required, increasing the load on the particulate collector. Thus, a larger ESP, or a second collector (e.g., a baghouse filter), is needed to achieve allowable particulate emission levels if carbon injection is used for mercury control. The use of water injection to humidify the flue gas can reduce the activated carbon requirement and the associated load on the particulate collection device. As indicated earlier, the presence of a wet lime or limestone FGD system also reduces emissions of air toxics including mercury. Thus, power plants already equipped with a wet FGD system can achieve mercury emission reductions at substantially lower costs. For plants burning eastern bituminous coals, limited data from the ICR suggests that the presence of an SCR system together with a wet FGD system can eliminate altogether the need to inject activated carbon while achieving high levels of mercury control. On the other hand, for plants without a wet FGD system, the addition of SCR appears to have little or no effect on mercury capture efficiency. Additional research is clearly needed to better understand these observations. The particle size of the activated carbon also has been shown to influence the effectiveness of the sorbent injected . Reducing the particle size will potentially increase the mass transfer between the gas and the solids by increasing the interfacial area of the sorbent. However, due to limited available data on particle size effects, the IECM does not currently consider particle size. EVALUATING FEASIBILITY AND COST Analyses of competing options for environmental control are typically carried out using computer models to simulate or optimize the electric utility response to new requirements The EPA/DOE/EPRI Mega Symposium August 20-23, 2001 4 or policy proposals. Figure 1 depicts several types of modeling tools that are currently used for analysis. One type of model (to be elaborated in this paper) evaluates emission control options and costs at the level of a single plant or facility. This type of model is able to incorporate a fairly high level of technological detail and site-specific factors, while offering fast turnaround time and minimum data requirements. The plant-level model typically draws upon results of more detailed process-level models and data for individual plant components. Figure 1. A hierarchy of models for policy analysis Options for a single facility (feasibility, efficiency, emissions, cost) Multi-facility (or multi-sector) optimization or simulation Integrated assessment models (incl. impacts) Other models are designed to analyze multiple facilities and multiple time periods. These models are more complex and data intensive. Typically they treat power plants as aggregates of representative facilities of a given type or class. While these models have less technological detail, they incorporate a wider variety of interactions such as interfuel substitutions, energy demand forecasts, electric power dispatching, and macroeconomic impacts. The National Energy Modeling System (NEMS) used by the U.S. Department of Energy (DOE) for its Annual Energy Outlook is an example of this class of model. Another class of models depicted in Figure 1 is integrated assessment (IA) models. These large-scale models link anthropogenic emissions to the environmental consequences and impacts of proposed policy measures. Typical applications of IA models include assessments of acid deposition, ambient ozone concentrations, and atmospheric CO2 levels. IA models attempt to represent the complex couplings between emissions, atmospheric processes, and resulting impacts at the regional, national or global scale, for time periods ranging from decades to a century or more. In principle, the different types of modeling and assessment tools shown in Figure 1 can draw upon one another to form an overall hierarchy of analytical capabilities able to address a broad spectrum of questions. In the present paper, the emphasis is on the “bottom-up” plant-level model. This perspective is needed to develop a careful understanding of plant-level factors that influence the feasibility and cost of multipollutant emission control strategies. It is important that large-scale “top-down” models in turn adequately represent such factors and interactions in their more aggregated representations of power plant technologies. The EPA/DOE/EPRI Mega Symposium August 20-23, 2001 5 THE IECM MODELING FRAMEWORK The Integrated Environmental Control Model (IECM) developed for the U.S. Department of Energy by Carnegie Mellon University provides plant-level performance, emissions and cost estimates for a variety of environmental control options for coal-fired power plants. The model is built in a modular fashion that allows new technologies to be easily incorporated into the overall framework. A user can then configure and evaluate a particular environmental control system design. Current environmental control options include a variety of conventional and advanced systems for controlling SO2, NOx, particulates and mercury emissions for both new and retrofit applications. The IECM framework now is being expanded to incorporate a broader array of power generating systems and carbon management options . Technology Performance Models The building blocks of the IECM are a set of performance and cost models for individual technologies that can be linked together to configure a user-specified power generating system. Figure 2 shows a schematic of the integrated approach, which links the various technology types together. Each technology area represents one or more individual technologies. The process performance models employ mass and energy balances to quantify all system mass flows including environmental emissions. The energy requirements of each technology also are modeled and used to calculate the net efficiency of the overall plant. Details of current models can be found in published papers and reports and the software is publicly available. Typically, each process performance model has approximately 10 to 20 key input parameters, depending upon the complexity and maturity of the technology. Figure 2. Schematic diagram of the IECM technology types Combined SOx/NOx Removal Advanced Particulate Removal Coal Cleaning Combustion Controls Flue Gas Cleanup & Waste Management SO2 Removal Particulate Removal Mercury Removal NOx Removal NOx Rem. Technology Cost Models For each technology module in the IECM, associated cost models are developed for total capital cost, variable operating costs, and fixed operating costs. These elements are combined to calculate a total annualized cost based on a consistent set of user-specified financial and lifetime assumptions. Normalized cost results, such as costs per kilowatt (or kilowatt-hour) of net capacity, and the cost per ton of pollutant removed or avoided, also The EPA/DOE/EPRI Mega Symposium August 20-23, 2001 6 are calculated. Cost models typically have about 20 to 30 parameters per technology, including all indirect cost factors and unit costs. An important feature of the cost models is that they are explicitly coupled to the process performance models. Thus, capital costs depend on key flowsheet variables such as mass or volumetric flow rates, and important thermodynamic variables such as temperature or pressure. Annual operating and maintenance (O&M) costs also are linked to mass and energy flows derived from the process performance model. Characterization of Uncertainties An important feature of the IECM is the capability to rigorously characterize and analyze uncertainties. In addition to conventional deterministic (single-valued) calculations, the IECM allows any or all model input parameters and output results to be quantified probabilistically. This allows the interactive effects of uncertainties in many different parameters to be considered simultaneously. Stochastic analysis thus provides quantitative insights about the likelihood of various outcomes, allowing users to more rigorously address questions such as: • What is the likely cost (or cost savings) of a particular emission control strategy relative to other options? What are the potential risks such as shortfalls in performance or overruns in cost? • Which control methods and technologies are most suitable for a given plant? Are there particular markets or applications that are likely to be most attractive for a given approach? • Which parameters contribute most to overall uncertainty in performance and cost? What are the potential payoffs from targeted research and development to reduce key uncertainties? User-Friendly Operation The IECM was designed to provide sophisticated modeling capabilities with quick turnaround time (seconds per run), transparency, and ease of use. A user-friendly graphical interface provides the capability to configure an analysis, set key parameter values (and their uncertainties), and get results in either probabilistic or deterministic form. A variety of graphical, pictorial, and tabular reports are available via the interface. Figure 3 shows several screen shots from the IECM’s current graphical user interface. MERCURY CONTROL MODULE Activated carbon injection with the option of water spray injection is the first mercury capture module in the IECM. Version 3.4 of the IECM software was released in May 2001 with the addition of this new module. This technology enhances baseline mercury removal with additional removal to allow flue gas treatment to meet a user-specified emission constraint or percent reduction requirement. The EPA/DOE/EPRI Mega Symposium August 20-23, 2001 7 Figure 3. Sample screens from the current IECM graphical user interface The adsorption of mercury onto the activated carbon sorbent is a physical rather than a chemical process, whose effect increases as the temperature decreases. Spray cooling of the flue gas is an effective method for reducing the temperature of the flue gas stream. In most cases this reduces the amount of sorbent required for mercury capture and therefore the cost of mercury control. Understanding the potential effects of retrofitted controls that affect the baseline removal may further reduce the required sorbent and the total cost of pollution control. The combination of base level controls and enhanced controls provide a useful approach for evaluating the relative merits of retrofitting a power plant to meet specific emission constraints, while minimizing the overall cost of abatement. As will be shown in the illustrative examples in the next several sections, the uncertainty features of the IECM add a unique dimension to the analysis of mercury control not provided by other models. Performance Model The IECM incorporates a dual approach to the removal of mercury in coal-fired power plants. The amount of mercury required for capture is determined by an emission constraint specified by the user. Because mercury is adsorbed in small amounts by The EPA/DOE/EPRI Mega Symposium August 20-23, 2001 8 bottom ash and fly ash, each abatement technology incorporates a level of mercury capture in the absence of special treatment. To reach higher levels of mercury capture, an activated carbon injection and optional water spray injection model has been added to the IECM. This additional injection system is referred to from this point as the mercury control module. The mercury control module is assumed to build on (add to) the baseline removal rate. Mercury Emission Constraint The level of removal of any flue gas constituent is determined by an emission constraint in the IECM. Nominally, the mercury capture is specified as an overall percentage reduction or removal efficiency. The level of removal refers to the total fraction of mercury that must be removed after the economizer and prior to the stack. The mass flow rate of mercury emitted into the flue gas is directly proportional to the concentration of mercury in the coal. However, each power plant component between the economizer and the stack contribute to the removal of mercury even without the addition of a mercury module. Each of these baseline removals must first be determined and factored together before consideration of the mercury module. The necessary amounts of activated carbon and water are then calculated so as to remove any additional mercury necessary to meet the emission constraint. Baseline Mercury Removal Baseline mercury removal is defined as total removal of mercury without the addition of a carbon injection mercury module. These baseline removals are measured across the inlet and outlet of flue gas abatement technologies designed to remove other constituents in the flue gas, as shown in Equation (1). Examples of these baseline technologies include cold-side ESP, fabric filter, wet lime/limestone FGD, spray dryer, and SCR. Note that mercury removal efficiency (%) is based on total (oxidized plus elemental) mercury removed from the flue gas and is defined as