Environ. Sci. Technol. 2006, 40, 3848-3854
Modeling Mercury in Power Plant Plumes KRISTEN LOHMAN,† C H R I S T I A N S E I G N E U R , * ,† ERIC EDGERTON,‡ AND JOHN JANSEN§ Atmospheric & Environmental Research, Inc., 2682 Bishop Drive, Suite 120, San Ramon, California 94583, Atmospheric Research & Analysis, Inc., 410 Midenhall Way, Cary, North Carolina 27513, and Southern Company Services, P.O. Box 2641, Birmingham, Alabama 35291
Measurements of speciated mercury (Hg) downwind of coal-fired power plants suggest that the HgII/(Hg0 + HgII) ratio (where HgII is divalent gaseous Hg and Hg0 is elemental Hg) decreases significantly between the point of emission and the downwind ground-level measurement site, but that the SO2/(Hg0 + HgII) ratio is conserved. We simulated nine power plant plume events with the Reactive & Optics Model of Emissions (ROME), a reactive plume model that includes a comprehensive treatment of plume dispersion, transformation, and deposition. The model simulations fail to reproduce such a depletion in HgII. A sensitivity study of the impact of the HgII dry deposition velocity shows that a difference in dry deposition alone cannot explain the disparity. Similarly, a sensitivity study of the impact of cloud chemistry on results shows that the effect of clouds on Hg chemistry has only minimal impact. Possible explanations include HgII reduction to Hg0 in the plume, rapid reduction of HgII to Hg0 on ground surfaces, and/or an overestimation of the HgII fraction in the power plant emissions. We propose that a chemical reaction not included in current models of atmospheric mercury reduces HgII to Hg0 in coal-fired power plant plumes. The incorporation of two possible reduction pathways for HgII (pseudo-first-order decay and reaction with SO2) shows better agreement between the model simulations and the ambient measurements. These potential HgII to Hg0 reactions need to be studied in the laboratory to investigate this hypothesis. Because the speciation of Hg has a significant effect on Hg deposition, models of the fate and transport of atmospheric Hg may need to be modified to account for the reduction of HgII in coal-fired power plant plumes if such a reaction is confirmed in further experimental investigations.
Introduction Several recent experimental studies provide both circumstantial and direct evidence of reduction of divalent gaseous mercury (HgII) to elemental gaseous mercury (Hg0) in power plant plumes. (HgII corresponds to the measured constituent typically referred to as reactive gaseous mercury, RGM.) First, * Corresponding author phone: (925) 244-7121; fax: (925) 2447129; e-mail:
[email protected]. † Atmospheric & Environmental Research, Inc. ‡ Atmospheric Research & Analysis, Inc. § Southern Company Services. 3848
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 12, 2006
the Mercury Deposition Network (MDN) data (1) along a west-to-east transect from Minnesota to Pennsylvania show no significant spatial gradient in annual mercury (Hg) concentrations in atmospheric precipitation although the Ohio Valley includes several large Hg emission sources located, under prevailing wind conditions, upwind of Pennsylvania. Accordingly, the corresponding annual Hg wet deposition fluxes either show no spatial gradient from Minnesota to Pennsylvania (e.g., in 1998 and 2002), or show a spatial gradient that is governed by differences in annual precipitation (e.g., 2003 and 2004). Possible explanations for this lack of spatial gradient include underestimation of Hg emissions in the upper Midwest (e.g., Minnesota and Wisconsin), high dry deposition rates of HgII in the close vicinity of Ohio Valley sources, and HgII reduction to Hg0 in Ohio Valley power plant plumes (2, 3). Regional models of atmospheric Hg show poor performance in that area because they tend to show a strongly increasing spatial gradient in Hg wet deposition from Minnesota to Pennsylvania; model simulations conducted with lower HgII/Hg0 ratios for power plant emissions showed improved model performance when compared to MDN data (2, 4). Second, experiments conducted with a Teflon-lined dilution chamber where the exhaust flue gases from a coal-fired power plant stack are sampled, diluted, and analyzed, showed a lower HgII/Hg0 ratio in the chamber than in the stack (5, 6). Third, aircraft measurement campaigns performed near the Bowen plant in Georgia and the Pleasant Prairie plant in Wisconsin indicate that some conversion of HgII to Hg0 takes place in power plant plumes (6). Finally, ground-level ambient sampling of Hg species (HgII, Hg0, and particulate-bound mercury, Hgp), NOy, and SO2 was conducted with continuous monitors downwind of coal-fired power plants in and around Atlanta, GA and near Pensacola, FL (7-10). The SO2/NOy ratio was used as a signature of individual power plants assuming that this ratio is more or less conserved during transport from the stack to the sampling site. Then, the corresponding speciated Hg measurements were compared with the estimated Hg speciated emissions, based on chemical analyses of coal samples and the EPRI-ICR emissions model. The results from that study suggest that the HgII/(Hg0 + HgII) ratio downwind from several power plants is substantially lower than the HgII/(Hg0 + HgII) ratio estimated for the stack emissions. We investigate here whether our current understanding of the atmospheric chemistry of Hg can explain the changes in HgII/(Hg0 + HgII) speciation from the stack to the downwind ground-level observation point that were observed by Edgerton et al. (7-10). To that end, we used a reactive plume model to simulate several power plant plumes. The resulting HgII/(Hg0 + HgII) and SO2/(Hg0 + HgII) ratios at a downwind point were compared with those from the observations. First, we provide a description of the Reactive & Optics Model of Emissions (ROME). Next, we describe the nine plume events simulated. Then, we present the results of the simulations. Simulations conducted with empirical HgII reduction reactions are also presented. Finally, we discuss the potential implications of HgII reduction in power plant plumes for the modeling of atmospheric Hg.
Description of the Model ROME is formulated according to a two-dimensional grid that travels in a Lagrangian framework according to the mean wind speed. Figure 1 depicts this Lagrangian grid. ROME includes state-of-the-science formulations of plume rise and dispersion, atmospheric transformations, and wet and dry deposition processes (11). 10.1021/es051556v CCC: $33.50
2006 American Chemical Society Published on Web 05/03/2006
sensitivity of the plume model simulation results to the HgII dry deposition velocity is investigated later. The Hg0 dry deposition velocity depends on meteorology and land use/ cover; it may also depend, to some extent, on the ambient Hg0 concentration because a compensation point may exist in the case of deposition on vegetation (15). The value used here is consistent with that estimated with an atmospheresurface exchange model for Hg0 dry deposition over plant canopy (16). ROME has undergone a rigorous performance evaluation for plume dispersion, gas-phase concentrations of several species (SO2, NOx, and NO2/NOx), and plume visibility at several wavelengths (17). ROME was shown to perform systematically better than a standard EPA Gaussian plume model. Therefore, ROME is an appropriate model to use for this analysis of local plume transport, dispersion, transformation, and deposition.
Plume Simulations Nine plume events were selected among those sampled at the Yorkville measurement station (7-10). The events were selected to represent a range of power plants, seasons, and transport times and distances. Selection was also based on availability of ancillary data, magnitude of the SO2 plume concentration, quality of Hg data, and absence of confounding factors, such as precipitation and fog. The web site of the SouthEastern Aerosol Research and Characterization Study experiment (SEARCH) provides the continuous SO2 data and will include the speciated Hg data later in 2006 (www.atmospheric-research.com/studies/SEARCH). For these plume events, the Hg plume increments were significant (10). They ranged from 110 to 660 pg/m3 for Hg0 and corresponded to 6 to 45% of the Hg0 background value which varied among the plume events from 1.47 to 2.26 ng/m3; they ranged from 5 to 147 pg/m3 for HgII and were at least a factor of 5 above the HgII background value which varied among the plume events from 0 to 3.6 pg/m3 (the background values were defined from the values measured before and after each plume event). These nine plume events and their meteorology are listed in Table 1. Power plant emissions (continuous emission monitor data) and stack information were provided by Southern Company. Table 2 summarizes the stack characteristics of the four power plants studied here. Because the plume events analyzed here could not be forecast, Hg emissions could not be sampled at the times of those plume events. Instead, Hg emissions were based on analysis of coal samples representing coal burned on the day of the plume event and the amount of coal consumed. The Hg speciation was estimated based on an EPRI analysis of the EPA Information Collection Request (ICR) data (18). (The EPA ICR database consists of Hg speciation sampling results at over 80 U.S. coal-fired power plants; EPRI developed a regression analysis of those ICR data that takes into account coal type, boiler configuration, and emission control equip-
FIGURE 1. Schematic representation of ROME. Plume rise is calculated according to the initial momentum and buoyancy of the stack gases. Plume transport is simulated in a Lagrangian framework according to the mean wind speed and direction. Plume dispersion in the cross-wind directions can be simulated using several options, including a secondorder closure algorithm, a first-order closure algorithm, empirical dispersion coefficients, or user-specified dispersion coefficients; here, we used the option of selecting user-input values of the horizontal and vertical dispersion coefficients because the measured concentrations of SO2 could be used to characterize plume dilution. Chemical transformations include gas-phase reactions of VOC, NOx, and SO2 simulated with the Carbon-Bond Mechanism (CBM-IV), aqueous-phase oxidation of SO2 and NOx simulated with 30 reactions and 31 equilibria, and gas-phase and aqueous-phase reactions of Hg species. Hg transformations are simulated with the AER mechanism (12). Dry deposition was simulated using velocities of 0.5 and 0.005 cm/s for HgII and Hg0, respectively. The HgII dry deposition velocity is consistent with that used in our other models of the atmospheric fate and transport of Hg at regional, continental, and global scales. For example, the regional/continental scale model TEAM uses a dry deposition velocity that is a function of meteorology and land use/cover; its mean value over land is 0.5 cm/s (13). We use a dry deposition velocity of 0.5 cm/s for HgII dry deposition over land in our global model (12). These dry deposition velocity values were selected for consistency with nitric acid (HNO3). Data on HNO3 dry deposition suggest a range of 0.06 to 5 cm/s with a geometric mean value of 0.5 cm/s (14). The
TABLE 1. Meteorology and Descriptions of the Nine Simulated Events
event name
trajectory length (hours)
wind speed (m/s)
temperature (°C)
relative humidity (%)
clouds
Bowen 6/27/01 Bowen 7/20/01 Gaston 12/7/01 Wansley 12/13/01 Hammond 12/29/01 Bowen 2/9/02 Bowen 6/17/02 Bowen 7/5/02 Bowen 7/21/02
6 3 11 2.5 2.5 13 3 4 6
2.3-3.1 2.0-3.0 1.3-3.6 3.1-3.9 4.8-5.2 0.2-4.7 0.9-2.4 0.0-3.6 1.3-2.8
26-31 31-34 16-21 15-16 8-11 4-13 28-29 31-33 30-37
53-72 89-99 74-93 100 45-48 37-80 36-43 40-48 90
no yes no yes no no no no no
VOL. 40, NO. 12, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3849
TABLE 2. Stack Characteristics of the Four Power Plants
power plant
latitude
longitude
stack height (m)
Bowen Stack 1 Bowen Stack 2 Gaston Stack 1 Gaston Stack 2 Hammond Wansley
34.1256 34.1256 33.2442 33.2428 34.2653 33.4167
84.9192 84.9192 86.4567 86.4589 84.6269 85.0333
305 305 229 229 229 305
stack diameter (m)
exit temperature (°C)
exit velocity (m/s)
7.6 7.6 7.2 9.1 6.7 7.6
130.0 136.7 128.3 143.3 156.7 157.2
20.7 27.1 25.6 19.5 25.3 20.4
TABLE 3. Emissions (lb/h) of Major Chemical Species for the Nine Plume Evets event
Hg0
HgII
Hgp
SO2
NOx
HCl
Bowen 6/27/01 Bowen 7/20/01 Gaston 12/7/01 Wansley 12/13/01 Hammond 12/29/01 Bowen 2/9/02 Bowen 6/17/02 Bowen 7/5/02 Bowen 7/21/02
0.064 0.068 0.073 0.014 0.012 0.028 0.068 0.053 0.046
0.075 0.102 0.045 0.017 0.009 0.056 0.085 0.083 0.068
0.002 0.003 0.001 0.0005 0.0003 0.001 0.002 0.002 0.002
46172 44349 30442 12656 5486 36167 47145 47350 43465
9737 10094 6253 3309 1811 9589 8453 8209 10184
1016 1504 288 277 181 1970 1158 1707 854
ment.) Uncertainties for speciated Hg emissions have not been formally evaluated but are likely to be far greater than SO2 and NOx emission uncertainties. HCl emission estimates were also based on coal analysis. Table 3 presents the emissions of Hg (speciated), SO2, NOx, and HCl for these nine events. As a check on the Hg emission speciation, we compared the Hg speciation estimated with the EPRI-ICR method with the results of a stack sampling program conducted at the Bowen power plant in September and October 2002 (19). The Ontario Hydro method was used for Hg speciation. The stack sampling results showed that HgII represented from 69% of total Hg at the units without selective catalytic reduction (SCR) to 87% at the unit with SCR (two
FIGURE 2. Locations of the monitoring site and power plants. 3850
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 12, 2006
of the four units at the Bowen power plant had SCRs that were operated during the ozone season). The EPRI-ICR Hg speciation for the six plume events of the Bowen power plant show HgII ranging from 53 to 66% of total Hg (Table 3); therefore, the HgII fraction of the Bowen power plant could be underestimated here. Meteorological data collected at the Yorkville site were used for eight of the events. One event did not have any on-site data; so meteorological data from the Atlanta Airport were used instead. None of the events had significant precipitation. Figure 2 is a map of the study area delineating the location of the Yorkville sampling station and the four power plants included in the study: Bowen, Hammond,
TABLE 4. Comparison of HgII/(Hg0 + HgII) Ratios in Emissions, Ambient Measurements and Model Simulations plume simulation HgII/(Hg0 + HgII)
plume event
emissions HgII/(Hg0 + HgII)
ambient measurements HgII/(Hg0 + HgII)
base simulation
sensitivity simulation HgII dry deposition × 10
sensitivty simulationa HgII f Hg0
sensitivity simulationa HgII + SO2 f Hg0
Bowen 6/27/01 Bowen 7/20/01 Gaston 12/7/01 Wansley 12/13/01 Hammond 12/29/01 Bowen 2/9/02 Bowen 6/17/02 Bowen 7/5/02 Bowen 7/21/02
0.54 0.59 0.38 0.55 0.44 0.67 0.55 0.61 0.59
0.12 0.06 0.03 0.07 0.12 0.11 0.21 0.11 0.21
0.52 0.59 0.35 0.53 0.40 0.54 0.53 0.59 0.57
0.35 0.48 0.13 0.37 0.18 0.13 0.33 0.35 0.32
0.18 0.37 0.07 0.32 0.18 0.01 0.24 0.19 0.12
0.30 0.01 0.17 0.38 0.32 0.04 0.19 0.06 0.05
a
The additional reaction to the base chemical kinetic mechanism is listed below.
Wansley, and Gaston. Plume trajectories for each of the events from the power plant to the Yorkville measurement station were calculated using the HYSPLIT model (20, 21). The HYSPLIT trajectories are backward, isobaric trajectories at 250 and 500 m agl calculated for the hour of peak SO2 ( 1 h. Those trajectories are included in the Supporting Information. The trajectories indicate that transport times ranged from 2.5 to 13 h. For modeling purposes, the plumes were divided into 10 columns transverse to the plume centerline and 10 layers. The columns were assumed to be symmetric with respect to the plume centerline. The layer depths were as follows: 50, 50, 100, 100, 150, 150, 200, 200, 500, and 500 m. As mentioned above, the plume widths were input so that the simulated SO2 plume concentration matched the measured maximum SO2 concentration at the Yorkville site. The implicit assumption with this approach is that the model correctly represents oxidation and deposition of SO2 between the stack and the measurement site. Oxidation of SO2 to sulfate is slow (
