Trace Elements in Two Pulverized Coal-Fired Power Stations

Jan 24, 2001 - Jacob D. McDonald , Richard K. White , Tom Holmes , Joe Mauderly .... Rainer Backman , Ingwald Obernberger , Thomas Brunner , Markus ...
0 downloads 0 Views 188KB Size
Environ. Sci. Technol. 2001, 35, 826-834

Trace Elements in Two Pulverized Coal-Fired Power Stations KRISTOFFER SANDELIN* AND RAINER BACKMAN Åbo Akademi University, Process Chemistry Group, Lemminka¨isenkatu 14-18 B, FIN-20520 Turku, Finland

Beside major pollutants (particulates, carbon, sulfur, and nitrogen oxides), coal combustion generates emissions of potentially toxic trace elements. The current work focuses on predicting the fate of eight trace elements (As, Cd, Hg, Ni, Pb, Se, V, and Zn) in power stations that fire pulverized coal and are equipped with flue gas scrubbers. The core of the study is global equilibrium analysis carried out with the aid of three extensive databases. The first set of equilibrium constants describes conditions prevailing in the furnace and the flue gas duct, while the second set describes reactions in the flue gas scrubber. Melting behavior of ash and solubility of trace elements within the slag are described as a third set of data. To test the modeling approach taken in this paper, the predicted overall partitioning of trace elements is compared with measured data from two full-scale facilities. The results of the study indicate that As, Cd, Ni, Pb, V, and Zn are captured in the fly ash, and that the fate of these element correlates with the overall particle capture of the power plants. Calculations for the flue gas scrubber facilities show that nonvolatile trace elements are likely to dissolve in the scrubber solution, and that capture of these elements likewise is correlated with the overall particulate behavior. Theoretical predictions of the melting behavior indicate that As, Ni, Zn, and to some extent Pb are likely to dissolve in the molten ash.

Introduction The fate of potentially toxic trace elements in combustion systems has received an increasing interest from both regulatory authorities and scientists. In response to the 1990 Clean Air Act Amendments, the U.S. Environmental Protection Agency has initiated extensive research to be able to establish information upon which future regulation of emissions from coal-fired power plants can be based (1-3). Regulations concerning emissions for cadmium and lead have also been agreed upon within the United Nations Economic Commission for Europe 1998 (4). The Toxic Release Inventory (5, 6) of the U.S. Environmental Protection Agency requires power plants burning coal and oil to estimate and report their emissions and has during recent years put more wind in the sails for modeling emissions of trace elements within the streams of power stations. Great efforts have been made in developing computer models for estimating distributions and emissions of trace elements within coal-fired power plants. Models of this kind are represented by the Emissions Factors Handbook, the EFH model (7), Power Plant Integrated * Corresponding author phone: +358 2 215 4036; fax: +358 2 215 4780, e-mail: [email protected]. 826

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 5, 2001

FIGURE 1. Configuration of the Dutch power station.

FIGURE 2. Schematic picture illustrating the theoretical reactor model. Systems: Chemical Emission Studies, the PISCES model (8), and the recent model by Helble (9). In contrast to describing toxic metal emissions, there are models/research aiming at a theoretical description of factors influencing the behavior of metals within streams of combustion systems. In this work, equilibrium analysis (10-15) and simulation of aerosol dynamics (16-18) have proven to be an important tool. To model trace element behavior, it would however be important to understand the role of surface reactions occurring in the flue gas duct. Recent research (19-23) has shed some light on these processes, but much fundamental work is still to be done for a qualitative description of the mechanisms. An excellent model for predicting trace element distributions in power plants has been presented by Rizeq et al. (24). In their work, they combined several submodels describing the vaporization, particle entrainment, aerosol dynamics, and air pollution control. An essentially more simple model for predicting distributions in power units has been presented by Sandelin and Backman (25). In this paper, their “reactor method” is tested on two coal-fired power stations, and their approach is extended to include the scrubber unit. In addition, the melting behavior of the ash and the solubility of trace elements within the liquid slag is briefly investigated. The goal of this study is to combine global equilibrium analysis with the reactor method to get an initial understanding of the complicated chemical processes occurring in combustion systems.

Simple Reactor Method The fate of trace elements in two power plants burning bituminous coal is examined in this paper. The first unit is located in The Netherlands, and the second one is located in Finland. A diagram of the configuration of the Dutch unit is shown in Figure 1. This diagram should be compared to Figure 2, which schematically illustrates the theoretical reactor method of this study. The exercise in predicting the equilibrium chemistry of trace elements was conducted by dividing the power plants into four hypothetical reactors or “boxes”. The first reactor simulates conditions occurring in the furnace. The composition of a hypothetical fuel is 10.1021/es000035z CCC: $20.00

 2001 American Chemical Society Published on Web 01/24/2001

introduced together with an appropriate amount of air and is equilibrated at a “furnace mean temperature”. This reactor generates an equilibrium ash and a flue gas. On the basis of the bottom ash to fly ash ratio of the specific power plant, a fraction of the condensed phases is withdrawn from the first reactor as an “equilibrium bottom ash”. The rest of the condensed phases are together with the flue gas sent to the second reactor, where the chemistry in the flue gas duct is predicted. This reactor generates an “equilibrium fly ash”. The main part of the equilibrium fly ash is withdrawn from the process, but a small fraction is assumed to pass through the electrostatic precipitator as particulate slip. The amount and composition of the dust that passes through the particulate control device is estimated from the particulate removal efficiency of the specific power plant and the composition of the equilibrium fly ash from the second reactor. In this paper, condensed phases will be referred to as particulate matter, particulates, or dust. The term aerosol is applied when used in original literature references. The third reactor predicts conditions in the flue gas treatment facilities. Flue gases from the second reactor are then introduced into the third reactor together with the dust. Water, limestone, and other desulfurization reagents are added and equilibrated at the adiabatic temperature. This reactor generates equilibrium information about the solid byproduct, the wastewater effluent, and the flue gases. An additional simulation is done to study the chemistry of a prescrubber unit. Flue gases from the second reactor are on that occasion introduced into the third reactor together with the dust and river water, and they are equilibrated at the adiabatic temperature of the system. Condensed phases and water are withdrawn as prescrubber reject, and the water vapor-saturated flue gases continue for removal of sulfur in the fourth reactor.

Equilibrium Analysis Chemical composition in each reactor is determined by global equilibrium analysis, minimizing Gibbs free energy. A computer program “ChemSage” (26) is used for the purpose. Three sets of equilibrium constants are used within this study. The first set of equilibrium constants is applied in the first two reactors simulating the conditions in the furnace and the flue gas duct. This set of data has previously been described in the literature (25, 27). The second set of equilibrium constants is used for simulating conditions prevailing in the wet scrubber unit. A third set of data is applied to investigating the melting behavior of the ash and the solubility of trace elements in the liquid slag. The first set of data utilizes thermodynamic constants from the Scientific Group Thermodata Europe (SGTE) database for pure substances (28). It contains equilibrium constants for reactions among the main components (C-O-H-S-N-Cl), the ash forming elements (Al-Ca-Fe-K-Mg-Na-Si), and the trace elements (As-Cd-Hg-Ni-Pb-Se-V-Zn). In all, the description includes 21 elements, one single mixture phase including 304 gaseous species, and a total of 393 condensed phases. All species are taken into account simultaneously, except for gaseous cadmium hydroxide, Cd(OH)2(g), which is believed to have an inconsistent numerical value. In a previous work (27), the equilibrium distribution of trace elements in the C-O-H-S-Cl-As-Cd-Hg-Ni-Pb-Se-V-Zn system has been studied. Within the work, the gaseous species Cd(OH)2(g) was found to be stable at significant concentrations (a few percent of the total amount of cadmium) already at 500 °C, and this species was dominant at 600-1100 °C. The species was however not found in data bases other than the SGTE database for pure substances (28). Since the Cd(OH)2(g) species was classified with the lowest priority in the SGTE database and since it was found unexpectedly stable, it was decided that this species should not be included in the

equilibrium prediction. In the current database, all solid phases are used within given temperature limits of the data except for 2CaO‚V2O5 alone, which is extrapolated above 1015 °C (1288 K), i.e., constants in the Cp function are used above their temperature limit. Data for gaseous species are however extrapolated. No solid or liquid mixture phases are included in the first set of equilibrium constants. The second reactor utilized the same database with restrictions for sulfurcontaining species that are described with “apparent equilibrium”, i.e., an assumption of a metastable state. A prediction including all sulfur species forecasts the element to be almost entirely captured within the ash (29). Apparent equilibrium approach is used since predictions including the whole database give results that are inconsistent to observations in typical coal-fired units (8, 30, 31). Ash analyses from boilers firing pulverized coal show that sulfur is captured to a lesser extent, typically only to a few percent. Condensed sulfur species are therefore deleted from the mass balance, except for sulfates of cadmium, mercury, and lead, which are included in order to describe capture of cadmium and lead in the fly ash. Compositions in the flue gas treatment reactors are predicted with the second set of equilibrium constants, which include data for gaseous, aqueous, and solid phases. Selected constants are taken mainly from the HSC database (32). In all, the description includes 21 components and two mixture phases including 19 gaseous and 59 aqueous species. Only pure elements, (hydr)oxides, sulfates, and carbonates are considered in addition to common precipitates (Ca3(AsO4)2, PbO‚PbCO3, and PbO‚PbSO4) as condensed material. The total number of condensed phases is 54. Table S1 (see Supporting Information) shows a complete list of the species included in the second set of data. The chemical equilibrium composition of the flue gas treatment facility is determined at the adiabatic temperature. The temperature is calculated by giving the exact composition, temperatures, and pressures of the incoming streams and by assuming no exchange of heat (∆H ) 0). Main gaseous components are entered as N2, CO2, H2O, O2, SO2, and HCl. Ash forming elements are entered as metal oxides, and the trace elements are entered in their elemental form. Carbon, chlorine, and sulfur of the condensed streams are given as aqueous CO2, HCl, and gaseous SO3. Water is entered as liquid water. The temperatures of the streams are set to typical values of the specific plant, and the pressure is set to 1 bar. An apparent equilibrium approach is applied in the predictions for the prescrubber unit, i.e., sulfur with oxidation state +VI is prohibited. Calculations including the whole database give results that are inconsistent to observations in the absorption vessel of the Dutch unit (30). Analysis shows that very little sulfur is absorbed in the prescrubber unit. A prediction including all sulfur species predicts however that the element to be captured within the prescrubber, and we assume that sulfur is oxidized to a very small extent under the prevailing conditions. To investigate the melting behavior of the ash, a third set of equilibrium constants was extracted from the FACT database (33) and combined with data from SGTE (28). The description of the melting behavior is simplified by assuming iron oxides as inert material and by presuming magnesium to react as calcium. All gaseous species and condensed pure species except PbCl4(g) are included for the C-O-H-S-ClAl-Ca-K-Na-Si-As-Cd-Ni-Pb-Zn system from the FACT database. Gaseous lead tetrachloride was excluded since calculations of the coal-moisture-air system showed that this species would be unexpectedly stable, i.e., dominant at room temperature. The calculation included reactions among C, O, H, S, Cl, and Pb. The third database includes N2 as an inert gas. The equilibrium description of the evaluated “FACT, slag-a database” is included with some exceptions: NiS and NiCl2 are excluded and As2O3, NiO, PbO, and ZnO are included VOL. 35, NO. 5, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

827

TABLE 1. Characteristic Parameters of the Dutch and Finnish Power Stations parameter

Dutch unit

Finnish unit

electrical output (MWe) boiler type burners combustion temp (°C) heating value (MJ/kg) fuel ash content (wt %) S content of fuel (wt %) N content of fuel (wt %) chlorine content (ppmw) particulate control particulate control efficiency (wt %) SOx control main scrubber reagent NOx control ash partitioning (%) bottom ash economizer ash

600 wall fired 36 1400 29 ∼12 ∼1 1.53 1300 cool side ESPa