Final Report: Dynamic Electricity Generation for Addressing Daily Air Quality Exceedances in the US

EPA Grant Number: R835219
Title: Dynamic Electricity Generation for Addressing Daily Air Quality Exceedances in the US
Investigators: West, J. Jason , Blumsack, Seth , Vizuete, William
Institution: University of North Carolina at Chapel Hill , Pennsylvania State University
EPA Project Officer: Chung, Serena
Project Period: June 1, 2012 through May 31, 2014 (Extended to May 31, 2015)
Project Amount: $250,000
RFA: Dynamic Air Quality Management (2011) RFA Text |  Recipients Lists
Research Category: Air Quality and Air Toxics , Air


Air quality is forecast daily in the United States. This information can inform daily management decisions in the electric power sector, attempting to avoid daily air quality standard exceedances by redispatching generation from regions strongly influencing the exceedance to elsewhere. Here, we propose to design and evaluate a dynamic management system for the electric power sector with the goal of avoiding daily exceedances of air quality standards for ozone in the eastern United States. We will first demonstrate this dynamic system for a selected episode, evaluating choices in how this system might operate, and incorporating model uncertainty into the design. We will then demonstrate its application over a full summer season, and evaluate this system based on costs, fuel consumption, greenhouse gas emissions, electrical system reliability, success in avoiding local ozone exceedances, and changes in ozone and PM2.5 over the eastern United States, considering environmental justice.

Summary/Accomplishments (Outputs/Outcomes):

Research at the University of North Carolina (UNC) has focused on modeling ozone using CAMx, and in estimating ozone sensitivities using DDM. In doing so, we have completed a series of scoping simulations in which we address several questions that would be particularly important for designing a dynamic air quality management system. James McCann had set up the CAMx and DDM simulations and explored these questions for a single episode, August 4, 2005, focusing on Pittsburgh, PA, and nearby areas. Evan Couzo then reanalyzed the earlier results for Aug. 4, and conducted a similar analysis for an additional episode, August 13, 2005, focusing again on Pittsburgh, PA, and nearby areas. By analyzing two episodes we now have greater confidence in the applicability of these results more generally. Our main findings from this research include: 

  • Peak ozone concentrations respond to power plant emissions by a few ppb, which may be enough to prevent exceedance of the ozone standard in at least some cases.
  • When removing power plant emissions from multiple facilities, ozone at a receptor location generally responds most strongly to emissions from a few large facilities (<6) close by. In several cases, a single power plant is responsible for over half of the sensitivity. This suggests that the number of power plants that would need to be affected to reduce ozone at a single location is small.
  • Turning off emissions for the entire day before a high ozone episode provides significant benefits compared to turning off emissions just the night before. Turning off emissions more than one full day in advance does not provide significant additional benefits.
  • DDM estimates of ozone sensitivities for power plant emissions provide reasonable estimates of sensitivities that agree well with estimates derived from brute force simulations. Using a higher-order sensitivity technique (HDDM) provides estimates of sensitivity that are more accurate when compared to brute force results, but at significantly greater computational costs. 
A complete journal article on this work has been submitted to the Journal of the Air & Waste Management Association
The Penn St. team has led the analysis of power grid operations. This has involved using a large-scale model of the power grid on two spatial scales covering the Mid-Atlantic region and the Eastern half of the United States. The plant-level ozone sensitivities developed by the UNC team were utilized by the Penn St. team to evaluate decision rules for adjusting power plant output from coal-fired power plants following the prediction of an ozone exceedance event. The decision rules considered essentially rank-ordered coal-fired power plants in two ways. First, we rank-ordered power plants based on the absolute magnitude of the ozone sensitivity, so that plants with the highest sensitivities would be shut down first. Second, we rank-ordered power plants based on a measure of cost per unit of concentration abatement. This essentially meant that the most expensive coal-fired power plants (per unit of reduction in ozone concentration) would be shut off first. 


The findings of the power grid analysis, which are currently in preparation for submission, can be summarized as follows: 
  • For the particular episode in the Pittsburgh area that was analyzed, turning off sufficient coal-fired power generation in the Mid-Atlantic region to ameliorate ozone exceedances in Pittsburgh yielded infeasible power flow results. In other words, our simulations suggest that shutting down enough coal-fired power plants in the Mid-Atlantic region to address a summertime ozone exceedance in Pittsburgh could threaten the reliability of the regional power grid. This conclusion is naturally sensitive to the magnitude of the ozone exceedance, but we did observe infeasibility in the power grid model when it was asked to reduce concentrations in Pittsburgh by 2.5 ppb or more. 
  • If the dynamic air quality management system was expanded to include the entire eastern United States, then ozone exceedances in Pittsburgh could feasibly be addressed through controlling coal-fired power plant output. The most critical units to control, our simulations found, were actually outside of the Mid-Atlantic region. Our findings suggest that successful implementation of the proposed air quality management scheme would require coordination between regional power grid operators. 
  • Because the ozone sensitivities are dominated by a few large power plants, and because these plants tended to be relatively inefficient (i.e., have higher operating costs) we did not always observe large differences in outcomes between the cost-based and sensitivity-based decision rules. 

In completing this work, a significant task was to harmonize the database of generating facilities in the dispatch model with the air quality model, which required us to match facilities by locations, compare estimates of emissions in the two databases, and adjust for sources that do not match. This task has been completed, making it possible to do further analyses more efficiently in the future. 

Beyond these two studies, we will consider continuing this work by further developing the decision rules, and by applying a dynamic system to a full summer in the eastern United States. A third paper may be written that will address the legal and management opportunities and barriers to the application of a dynamic management system, considering the complexity of electricity generation and management in the United States. We are actively pursuing applying for continued funding through other government sources (especially NSF) and through the possibilities of direct sponsorship by industry. 

Journal Articles on this Report : 1 Displayed | Download in RIS Format

Other project views: All 2 publications 1 publications in selected types All 1 journal articles
Type Citation Project Document Sources
Journal Article Couzo E, McCann J, Vizuete W, Blumsack S, West JJ. Modeled response of ozone to electricity generation emissions in the northeastern United States using three sensitivity techniques. Journal of the Air & Waste Management Association 2016;66(5):456-469. R835219 (Final)
  • Abstract from PubMed
  • Full-text: JAWMA-Full Text PDF
  • Abstract: JAWMA-Abstract & Full Text HTML
  • Supplemental Keywords:

    Ozone, electrical generation, dispatch model, PM2.5, particulate matter, air quality

    Relevant Websites:

    Dr. J. Jason West
    William Vizuete
    Seth Blumsack

    Progress and Final Reports:

    Original Abstract
  • 2012 Progress Report
  • 2013