Case Study to Investigate Monitored SO_2 Concentrations Near Coal-Fired Power Plants: Stationary Monitor to Model Comparison

摘要

In 2010, the U.S. EPA promulgated a much more stringent ambient standard for SO_2 with a 1-hour averaging time. U.S. EPA's implementation of this new standard has considered both monitoring and modeling approaches, and the agency plans to address large SO_2 sources over the next few years with a hybrid approach that could include a combination of modeling and monitoring. However, some stakeholders favor the modeling approach for the 1-hour SO_2 National Ambient Air Quality Standards (NAAQS) compliance demonstrations. Modeling with peak emission rates at all times can indicate simulated NAAQS violations in areas with no monitored violations. This type of result was noticed by the Maryland Department of the Environment (MDE) for two coal-fired power plants in the Baltimore, MD area operated by Raven Power: the Crane and Wagner Generating Stations. In spite of modeled NAAQS violations that could be reported with the grossly conservative assumption of constant peak emissions, it was found that at a monitor in the area being modeled, the predicted concentrations exceeded SO_2 measurements by over a factor of 5. It was also evident that the peak predicted concentrations were at areas in the Chesapeake Bay where there are no inhabitants. In the summer of 2013, the Maryland Department of the Environment (MDE) asked Raven Power to assure the agency that actual monitoring would indicate compliance with the 1-hour SO_2 NAAQS (in contradiction to conservative modeling results). Raven Power conducted the monitoring study in two phases, and also supplemented the study with modeling using actual emissions. The initial monitoring phase, lasting one week, involved a set of mobile analyzers placed in vans and on boats to measure concentrations downwind of the plants being studied. The measured concentrations during this phase were generally much lower than the modeling results. A second monitoring phase lasted approximately two months and consisted of measurements at a total of four stationary locations (on land in populated areas), selected on the basis of modeling and the first phase of the monitoring study. This paper summarizes the results of the second monitoring phase by examining the model output in comparison to the measured concentrations at the four monitoring sites, conducting an in-depth analysis for selected periods at each monitor, and discussing the model performance.