ABSTRACT
For geological sequestration of carbon dioxide (CO2) separated from pulverized coal combustion flue gas, it is necessary to adequately evaluate the potential impacts of flue gas impurities on groundwater aquifers in the case of the CO2 leakage from its storage sites. This study estimated the flue gas impurities to be included in the CO2 stream separated from a CO2 control unit for a different combination of air pollution control devices and different flue gas compositions. Specifically, the levels of acid gases and mercury vapor were estimated for the monoethanolamine (MEA)-based absorption process on the basis of published performance parameters of existing systems. Among the flue gas constituents considered, sulfur dioxide (SO2) is known to have the most adverse impact on MEA absorption. When a flue gas contains 3000 parts per million by volume (ppmv) SO2 and a wet flue gas desulfurization system achieves its 95% removal, approximately 2400 parts per million by weight (ppmw) SO2 could be included in the separated CO2 stream. In addition, the estimated concentration level was reduced to as low as 135 ppmw for the SO2 of less than 10 ppmv in the flue gas entering the MEA unit. Furthermore, heat-stable salt formation could further reduce the SO2 concentration below 40 ppmw in the separated CO2 stream. In this study, it is realized that the formation rates of heat-stable salts in MEA solution are not readily available in the literature and are critical to estimating the levels and compositions of flue gas impurities in sequestered CO2 streams. In addition to SO2, mercury, and other impurities in separated CO2 streams could vary depending on pollutant removal at the power plants and impose potential impacts on groundwater. Such a variation and related process control in the upstream management of carbon separation have implications for groundwater protection at carbon sequestration sites and warrant necessary considerations in overall sequestration planning, engineering, and management.
INTRODUCTION
Carbon dioxide (CO2) from fossil fuel combustion is the largest source of U.S. greenhouse gas emission and accounts for nearly 80% of global warming potential weighted emissions in 2004.1 Emissions from this source category increased at an average annual rate of 1.3% and grew by 20% in the period from 1990 to 2004, during which CO2 emissions from fossil fuel combustion were responsible for most of the increase in emissions in the United States. Among the CO2 emissions from fossil fuel combustion in 2004, coal combustion for electricity generation accounts for approximately 34% of total U.S. emissions. Therefore, CO2 emissions reduction from coal- fired power plants is critical to controlling the atmospheric CO2 level in global warming mitigations.
Because carbon accounts for approximately 70 – 80% of fossil fuels, the quantity of CO2 produced in coal combustion for electricity production amounts to 10 –12% (v/v) of the flue gases generated. On average, the combustion of coal results in the release of 90 kgCO2/MJ (210 lbCO2/MBtu), as compared with 70 kgCO2/MJ (162 lbCO2/ MBtu) for distillate fuel oil and 50 kgCO2/MJ (116 lbCO2/ MBtu) for natural gas.2 For the future need in management and control of CO2 emissions, flue gas of coal-fired power plants is scrubbed to separate CO2 for subsequent long-term storage in underground reservoirs, such as depleted oil and gas fields, deep brackish aquifers, coal seams, and other geological formations.1 Previous studies have focused on CO2 plume migration in subsurfaces and interactions with groundwater and geological materials. Little definitive information is available on the effect of sulfur oxides (SOx), nitrogen oxides (NOx), mercury, and other impurities on their interactions with groundwater in case of leakage from their storage site. However, the potential impacts and related upstream contaminant removal from coal-fired power plants are important factors in the chemistry of separated CO2 streams and the identification of the best combination of separation technologies for use with respect to flue gas constituents.
The objective of this study is to estimate flue gas impurities, including SOx and in particular sulfur dioxide, (SO2), nitrogen dioxide (NO2), hydrogen chloride (HCl), and mercury included in the CO2 stream separated from a CO2 control unit. Flue gas impurities in the separated CO2 stream from coal-fired power plants are estimated for five cases with a different combination of air pollution control devices (APCDs) and a potential CO2 control unit. Among the five cases, estimation was made for monoethanolamine (MEA)-based CO2 absorption, which requires strict acid gas control. The current status of commercial MEA absorption technologies and the potential environmental impacts of CO2 leakage on groundwater aquifers are also identified in this paper.
CARBON CAPTURE TECHNOLOGIES FROM COAL-FIRED POWER PLANTS
CO2 Absorption Processes
A wide range of technologies is currently under consideration for separation and capture of CO2 from coal combustion gases, including absorption, adsorption, membrane, and cryogenic processes.3–5 Among these technologies, chemical absorption uses chemical solvents reacting to form weakly bonded intermediate compounds that are then recovered by applying heat. Alkanolamines are the most widely used solvents, and new solvents are under study and development to selectively capture and separate CO2 from coal-fired flue gas streams more energy efficiently and effectively. These solvent candidates include aqueous solutions of promoted potassium carbonate using piperazine,6 reclaimed magnesium hydroxide,7 ammonia,8 and ionic liquids.9
The amine-based absorption processes were developed approximately 60 yr ago as a nonselective solvent to remove acid gas components such as hydrogen sulfide and CO2 from natural gas streams. Typically, 75–90% of CO2 is captured using this technology, producing an almost pure (>99%) CO2 product stream. The exact recovery choice is an economic tradeoff, and a higher recovery will lead to greater capital and operating costs. This process is effective for dilute CO2 streams such as coal combustion flue gases, which typically contain 10 –12% (v/v) of CO2. The most widely used alkanolamines include MEA, diethanolamine (DEA), and methyldiethanolamine (MDEA). Another amine, 2-(aminoethyl)ethanolamine (AEEA), has recently received attention under investigations.10,11 Among these amines, MEA is the most widely studied for CO2 capture from all coal-fired power plants in the United States12–14 because of its higher reactivity. Disadvantages of MEA include its high energy consumption for desorption relative to DEA and MDEA, and a relatively low (~50%) limitation on its maximum CO2 loading capacity compared with MDEA, which has a 100% equilibrium loading capacity. Other drawbacks include evaporation losses due to its high vapor pressure and more severe corrosion problems than other alkanolamines, for which corrosion inhibitors are commonly applied when used in higher concentrations.
Impacts of Flue Gas Impurities on MEA-Based CO2 Absorption
There is little information available on flue gas impurities included in the CO2 stream separated by different CO2 solvents in the research and development stage. However, the interactions of other flue gas components with MEA are relatively well documented. All of the MEA entering the stripper (MEA regeneration column) is not fully regenerated because of irreversible degradation of MEA in reaction with oxygen (O2), NO2, and SOx, which leads to numerous operational problems such as foaming, fouling, increased viscosity, and formation of unregeneratable heat-stable salts in the amine. O2 concentrations in the range of 3–12 vol % in typical flue gas streams are known to induce oxidative degradation of alkanolamines, resulting in severe corrosion.15,16 However, proprietary O2 inhibitors are available that can make alkanolamines tolerant to O2.16
SO2 control is particularly important for the MEA- based process because its concentration is higher than any other acid gas components in the flue gas; the typical SO2 range is 500-3000 parts per million by volume (ppmv).16,17 In general, wet SO2 scrubbers are capable of removing 80 –95% of the SO2 and are typically operated at the upper end of the removal efficiency range.18 Although deep desulfurization with greater than 99% SO2 removal is achievable, such an operation is not economically favorable for the sole purpose of flue gas desulfurization (FGD). However, SO2 levels higher than approximately 10 ppmv at the inlet of the MEA-based CO2 absorber can create process problems, including foaming, corrosion, fouling, plugging, and solvent loss. Therefore, commercial MEA processes set a maximum of 10 ppmv of SO2 as a feed specification to keep solvent consumption (~1.6 kg MEA/t of CO2 separated) and make-up costs at reasonable values.19,20 MEA makeup is reported to fall in the range of 0.5–3.1 kg MEA/t of CO2 separated, which accounts for approximately 10% of the CO2 capture cost.16 MEA costs approximately $1250/t MEA in the current market place.16 Dry FGD technologies using alkaline sorbents are incapable of meeting the deep desulfurization requirement for subsequent MEA-based CO2 separation because of their less than 90% SO2 removal capability under normal operation conditions.18
Reactions between SO2 entering the CO2 absorber and MEA are known to lead to formation of heat-stable salts (e.g., isothiocynatoethane and tetrahydrothiophene) and to generate MEA solution waste in the process.16 –31 These heat-stable salts or degradation products are removed at the bottom of the heat exchanger known as the reclaimer and disposed of as hazardous chemical waste, leading to increased disposal costs (~$175/t waste).16,27 The purpose of the reclaimer located in the CO2 stripper section is to remove contaminants such as heat-stable salts, suspended solids, acids, and iron products from the circulating MEA solution, thus reducing corrosion and fouling in the MEA system. In most commercial MEA processes, some of the MEA lost by heat-stable salt formation is recovered by adding sodium carbonate to the reclaimer.27 Electrodialysis was also proposed by Union Carbide to remove the heat-stable salts and considered as effective in reducing solvent losses.28
Typically sulfur trioxide (SO3) is present at a level of no more than 20 –30 ppmv in coal combustion flue gases.17 It is also well known that SO3 concentration increases over vanadium oxide catalysts used for selective catalytic reduction (SCR).21,22 Approximately 50% of SO3 is known to be removed by wet FGD scrubbers. Fine submicron sulfuric acid mist droplets less than approximately 0.05 µm are not easily captured by wet FGD scrubbers.21 In addition, HCl is present up to 100 ppmv in flue gases,23 and 90 –95% of HCl is removed by wet FGD scrubbers.16 SO3 and HCl are also likely to form heat-stable salts similar to SO2 in MEA absorption16; however, little information is available on the effects of SO3 and HCl on the MEA solution.
Total NOx in the coal combustion flue gas typically consists of approximately 95% nitric oxide (NO) and 5% of NO2.17,24 NO does not react with amines under normal operating conditions, but NO2 in the range of 10 – 40 ppmv does react to form heat-stable salts.16,25,28 Thus, NO2 concentration at the inlet of the MEA process is reported to be limited at or below approximately 10 ppmv.30 This specification is usually met by the use of low NOx burners (LNBs) with a SCR unit.16 The application of LNBs and SCR units can typically achieve 40 – 60% and 75– 85% NOx removal, respectively.24
Removal of mercury is not well characterized for CO2 absorption systems but is anticipated to be similar to wet scrubber systems. Elemental mercury is not captured by wet scrubbers because of its very low solubility in the aqueous solutions used for commercial wet FGD systems. On the other hand, oxidized mercury is readily captured because of its high solubility although mercury re- emission problems have been reported.26 In a recent metal analysis of MEA reclaimer wastes, mercury was also found to be present in the reclaimer bottoms at the level of approximately 1 part per billion by weight (ppbw) but was not detectable in the lean MEA (<0.02 ppbw), indicating removal of mercury from the separated CO2 stream.27
Estimation of Flue Gas Impurities in Separated CO2 Streams
As described in the previous section, flue gas components including O2, SOx, NO2, and HCl have adverse impacts on MEA absorption processes. As air pollution generated by coal-fired power plants becomes more stringently regulated, SOx, NOx, and mercury emissions will be more strictly controlled regardless of their impacts on CO2 capture technology. The U.S. Environmental Protection Agency (EPA) issued the Clean Air Interstate Rule (CAIR) on March 10, 2005, which will permanently cap SO2 and NOx emissions from 28 eastern states and the District of Columbia, where most coal-fired power plants are located across the United States.32 The rule will reduce SO2 and NOx emissions by over 70 and 60%, respectively, from 2003 levels in 2015. In addition, the Clean Air Mercury Rule (CAMR) was also issued on March 15, 2005, to reduce mercury emissions from 50 to 38 t/yr by 2010 during Phase I and to 15 t/yr by 2018 during Phase II.33 However, the United States Court of Appeals for the District of Columbia Circuit in the state of New Jersey vacated the CAMR and CAIR in February and July, 2008, respectively. These court decisions will eventually lead to stricter revised rules.
Aside from these future regulations of SO2, NOx, and mercury, a previous case study on MEA-based CO2 capture suggests that more stringent control of these impurities (especially SO2) in coal combustion flue gas makes subsequent CO2 capture more cost-effective because SO2 removal can significantly reduce the expensive reagent (MEA) costs (~$1250/t MEA) and also benefit the SO2 credit (~$550 – 600/t SO2 as of November 2007) traded under the National Acid Rain Control Program.16 Previous trade-off studies showed that the total carbon separation cost could be minimized by reducing the SO2 concentration to below 10 ppmv at the CO2 absorber inlet, which would require removing more than 99% SO2 and most of the acid gases in wet FGD systems.16,34 Recent case studies with advanced MEA systems also showed that flue gas bypass would be the most cost-effective CO2 control method for variable CO2 capture levels.34,35 In both studies, approximately 90% CO2 capture efficiency with the MEA system was estimated to be the most cost- effective control level from existing typical 500-MW pulverized-coal power plants, with a fraction of flue gas bypassing the MEA system. A comprehensive economic study including a sensitivity analysis was also conducted for an existing 500-MW pulverized-coal power plant in comparison with the no-CO2 capture case.34 Its result showed an almost linear decrease in the incremental cost of electricity from 3.92 to 1.35 cents/kWh when the flue gas capture proportion decreases from 90 to 30% (i.e., with an increase in flue gas bypass from 10 to 70%) while achieving 90% CO2 removal in the MEA system.
Figure 1 shows a schematic of a typical coal-fired power plant, including a CO2 capture and separation unit. In this study, the compositions of the flue gas impurities potentially included in the CO2 stream recovered from a CO2 stripper were estimated by taking into consideration the type of coals, the presence of individual APCDs, and their typical performances. Cases 1– 4 in Table 1 considered different combinations of SO2 and NOx control devices in conjunction with an absorption-based CO2 control unit in Figure 1. Case 5 considered a commercial MEA-based CO2 control unit, which requires low SO2 levels (<10 ppmv) before CO2 separation. Either a LNB or a SCR unit was considered for NOx control because the NO2 level requirement for CO2 control can be easily met by either one of these two NOx control technologies. Typical APCD removal efficiencies of acid gases and mercury vapor in the flue gas and the likelihood of heat- stable salt formation by MEA are summarized in Table 2. The performance values of the individual APCDs for the estimated removal of individual components are also selected from the literature and listed with their references in Table 2. Little information is available on the forma- tion (i.e., identification of individual salts and their formation rates) of heat-stable salts generated from the CO2 stripper using a MEA solution in Case 5. A recent study conducted in Europe using a pilot plant29 indicates that approximately 75% of SO2 can be removed in the form of heat-stable salts when approximately 10-ppmv SO2 enters the CO2 absorber. No further information is available on salt formation rates for other acid gases although those acid gases have been reported to form such heat-stable salts.36
Figure 1. Schematic of a typical coal-fired power plant with an absorption-based CO2 control system.
The concentrations of the impurities in separated CO2 were estimated for the five control cases listed in Table 1 using the typical performance values listed in Table 2, and the results are summarized in Table 3. Please note that maximum flue gas impurities except for CO2 are included in a separated CO2 stream when a low CO2 concentration is present. For example, when 4.4 wt % of a maximum SO2 concentration is included in the separated CO2 stream under Case 1, 95.5% of CO2 is present in the stream. For mercury estimates, only oxidized mercury was considered for the estimation because elemental mercury is not readily soluble in the aqueous phase. Flow rates of individual flue gas components were added to the impurity calculation for a typical 500-MW pulverized- coal power plant in Table 3. Among the five cases shown in Table 3, Case 1 has the highest concentrations of acid gases and mercury in the separated CO2 stream because no SO2 and NOx control options are deployed. Among Cases 1– 4, Case 4 has the smallest impacts because the flue gas is controlled with both FGD and LNB/SCR processes before CO2 separation. The results for Cases 1 and 3 without a wet FGD system indicate that a maximum 4 wt % of SO2 could be included in the separated CO2 stream when no salt formation is assumed. Alternatively, the SO2 level could go further down to 1 wt % when 75%
Table 1. CO2 control cases with respect to SO2 and NOx control.
salt formation is assumed. Meanwhile, Cases 2 and 4 with a wet FGD system show that hundreds or thousands of parts per million by weight (ppmw) SO2 could be included in the separated CO2 stream. Other impurity acid gases include SO3, NO2, and HCl at a level less than 0.1%. Mercury concentrations are greatly affected by the presence of a wet FGD system and can be less than 30 ppbw when it is available. Case 5 assumed no (0%) and 75% heat-stable salt formation from SO2 for a MEA-based CO2 control system. For the 75% salt formation case, 75% of SO2 was assumed to form salts and be recovered from the reclaimer bottoms. Thus, the remaining 25% was assumed to be included in the separated CO2 stream. When SO2 in flue gases is controlled below 10 ppmv before CO2 separation, and 75% of the SO2 is recovered as heat-stable salts from the reclaimer bottoms, the least amount (~34 ppmw) of SO2 was estimated to be included in the separated CO2 stream. When no salt formation was assumed under the same 10-ppmv inlet condition, 135 ppmw of SO2 was estimated to be included. Other impurity gases— including SO3, NO2, HCl, and mercury of Case 5—are expected to be less than those of Case 4 because such deep desulfurization usually requires a secondary SO2 scrubber and can also further remove such soluble pollutants.
POTENTIAL IMPACTS OF FLUE GAS IMPURITIES ON GROUNDWATER
The gas impurities in a liquefied CO2 stream (Table 3) can interact with groundwater when in contact at a geological sequestration site. There are numerous investigations and publications on the site selection, carbon injection, and storage,3,37– 42 yet few have examined the potential groundwater interactions with a CO2 stream of different compositions, especially those from coal-fired power plant flue gas. Large-scale CO2 leakage from the sequestration sites and subsequent migration to the surface and near-surface environment has not been reported.43 However, an inherent probability exists for such a leakage to occur because of the imperfect geologic seal of a storage site and the pressure gradient and CO2 buoyancy differences.44 For example, at the West Pearl Queen pilot site near Hobbs, NM,45 an average CO2 leakage rate determined from tracer testing was approximately 0.0085% of the total CO2 sequestered per annum, which was below the U.S. Department of Energy Carbon Sequestration Program leakage limit of 0.01%/yr. The principal mechanisms for vertical migration of the gas are diffusion in microseepage and convection for high leakage flow rates through reactivated fractures, including manmade conduits such as abandoned or seal-failing boreholes.44 –55
Table 2. Typical performance values for removal of flue gas components by SO2, NOx, and CO2 control systems.
Carbon migration mechanisms in geological formations can be found in several comprehensive review publications.44–49
Low-rate leakage and diffusive carbon gas migration as observed at the West Pearl Queen pilot site in New Mexico could lead to geochemical changes in overlying groundwater formations. As the escaped CO2 contacts groundwater en route to the surface, a complex hydrological interaction and dynamic equilibrium of chemical constituents occurs, leading to groundwater quality changes.56–62 In addition, dissolved CO2 forms carbonic acid, causing changes in the pH of the groundwater aquifer and mobilization of metals, sulfate, and chloride. For example, on the basis of CO2 transport and geochemical modeling, Wang and Jaffe61 investigated the effect of releasing pure CO2 from a point source at the 100-m depth into a shallow groundwater aquifer containing the lead-bearing mineral galena (PbS). They concluded that, in weakly buffered formations, the escaping CO2 could mobilize sufficient lead from the mineral into groundwater to pose a health hazard over a radius of a few hundred meters from the single-point CO2 source. The negative impact was sustained over a wide range of CO2 concentrations at the depth. The modeling further indicates that the lead dissolution is highly dependent on groundwater alkalinity and pH buffering capacity, and a calcite-bearing alkaline aquifer could significantly decrease the secondary effect of CO2 dissolution.
The interactions between CO2 stream and groundwater at a leaking carbon storage site are potentially complex depending on several factors in aquifer properties and the water-gas two-phase transport and equilibrium. Because the aquifer properties are highly variable in space and often location-specific, accurate characterization of the potential interactions requires hydrological and geochemical characterization of the entire geological sequence above the carbon storage reservoirs, geophysical mapping of site geological structures, and two- and three- dimensional computer modeling of carbon movement.37,38,55,63. The complexity increases when other flue gas impurities in the carbon stream are considered. For example, as shown in Case 4 in Table 3, even if there is a CO2 control unit that does not require an SO2 concentration below 10 ppmv at the inlet of the CO2 control system, the SO2 and HCl levels in the CO2 stream could be as high as 2400 and 44 ppmw, respectively. At these levels, these two impurity gases that are converted into sulfuric acid and hydrochloric acid in contact with groundwater could result in significant changes in groundwater pH and, hence, undesired geochemical interactions.61 Furthermore, the maximum mercury concentration for Case 4 in Table 3 is estimated to be 27 ppbw in the CO2 stream. The subsequent mercury partitioning between groundwater, the migrating CO2 stream, and aquifer minerals will determine mercury concentrations in the impacted groundwater. Evaluation of such partitioning for mercury is highly location-specific, but nevertheless necessary when the impacted groundwater could have a concentration increase above the EPA drinking water standard at 2µg/L.
The potential risk from CO2 leakage could be reduced by controlling the impurity levels in the upstream carbon separation at the coal-fired power plants. As shown in Table 3, the compositional properties of the CO2 stream separated from a CO2 control unit depend on combinations of APCDs and their performances. Each is associated with a set of engineering variables determining the system functionality, removal performance, energy consumption, capital and operational demands. As for hazardous waste deep well injection and storage,56 –59 there may exist optimum points at which a carbon stream with optimum impurity compositions can be achieved for minimal impacts on groundwater, reduced energy payback in CO2 separation and contaminant removal, and favorable capital investment and operational requirements. This process optimization requires further engineering studies.
CONCLUSIONS
The primary objective of this study was to quantify flue gas components other than CO2 included in separated CO2 stream, which might adversely impact groundwater quality when the sequestered CO2 stream migrates to the groundwater aquifer. A case study was conducted for the estimation of flue gas components in terms of different combinations of APCDs and their typical performances reported in the literature. The MEA-based chemical absorption was also used as a model process that separates CO2 from coal combustion flue gas. Although most of the typical system performance parameters are available, the information on the reaction rates of heat-stable salt formation between MEA and acid gases (or compositional analysis of the flue gas impurities included in the separated CO2) can rarely be found from the literature. For the MEA-based CO2 absorption process, most of the process licensors limit the SO2 level below 10 ppmv before CO2
Table 3. Estimated concentrations and flow rates of flue gas components in separated CO2 stream on the basis of a typical 500-MW pulverized coal-fired power plant.
Advertisement
The 10 largest coal producers and exporters in Indonesia:
