AERMOD performance evaluation for three coal-fired electrical generating units in Southwest Indiana

摘要

An evaluation of the steady-state dispersion model AERMOD was conducted to determine its accuracy at predicting hourly ground-level concentrations of sulfur dioxide (SO_2) by comparing model-predicted concentrations to a full year of monitored SO_2 data. The two study sites are comprised of three coal-fired electrical generating units (EGUs) located in southwest Indiana. Thesites are characterized by tall, buoyant stacks, flat terrain, multiple SO_2 monitors, and relatively isolated locations. AERMOD v12060 and AERMOD v12345 with BETA options were evaluated at each study site. For the six monitor-receptor pairs evaluated, AERMOD showed generally good agreement with monitor values for the hourly 99th percentile SO_2 design value, with design value ratios that ranged from 0.92 to 1.99. AERMOD was within acceptable performance limits for the Robust Highest Concentration (RHC) statistic (RHC ratios ranged from 0.54 to 1.71) at all six monitors. Analysis of the top 5% of hourly concentrations at the six monitor-receptor sites, paired in time and space, indicated poor model performance in the upper concentration range. The amount of hourly model predicted data that was within a factor of 2 of observations at these higher concentrations ranged from 14 to 43% over the six sites. Analysis of subsets of data showed consistent overprediction during low wind speed and unstable meteorological conditions, and underprediction during stable, low wind conditions. Hourly paired comparisons represent a stringent measure of model performance; however, given the potential for application of hourly model predictions to the SO_2 NAAQS design value, this may be appropriate. At these two sites, AERMOD v12345 BETA options do not improve model performance.