Final Report: Analysis of Dynamic, Flexible NOx and SO2 Abatement from Power Plants in the Eastern U.S. and Texas

EPA Grant Number: R835216
Title: Analysis of Dynamic, Flexible NOx and SO2 Abatement from Power Plants in the Eastern U.S. and Texas
Investigators: McDonald-Buller, Elena , Allen, David T. , Craig, Michael T. , Kimura, Yosuke , McGaughey, Gary , Webster, Mort D.
Institution: The University of Texas at Austin , Massachusetts Institute of Technology , Pennsylvania State University
EPA Project Officer: Chung, Serena
Project Period: June 1, 2012 through May 31, 2016
Project Amount: $500,000
RFA: Dynamic Air Quality Management (2011) RFA Text |  Recipients Lists
Research Category: Air Quality and Air Toxics , Air


Emissions cap and trade programs have been the preferred federal policy instruments for achieving reductions in emissions of nitrogen oxides (NOx) and sulfur dioxide (SO2) from electric generating units (EGUs), motivated by concerns associated with acid deposition and regional ozone and fine particulate matter concentrations. Such programs have contributed to significant season wide and annual emissions reductions of these pollutants over the past three decades. However, pollutant emissions can have markedly variable damages on a daily basis depending on prevailing meteorological conditions. Dynamic air quality management approaches leverage capabilities in air quality forecasting and near real-time operational decisions and offer opportunities to incentivize emissions reductions during time periods that are most conducive to ozone and fine particulate matter formation and accumulation. This project developed methods to evaluate the air quality implications and cost-effectiveness of time differentiated pricing of NOx and SO2 emissions from power plants alone and in combination with other technology and season-wide market-based approaches. Two power systems were considered that differed in generation fuel mix, the Electric Reliability Council of Texas (ERCOT) and the Mid-Atlantic or Classic Pennsylvania-New Jersey-Maryland (PJM) grid. The objectives were to (1) compare time differentiated, season-wide, and combined regulatory designs; (2) explore the resulting emissions reductions considering realistic constraints on EGUs and endogenous technology adoption; (3) demonstrate the implications for predicted regional ozone and fine particulate matter concentrations; (4) investigate opportunities for joint abatement of NOx and SO2 emissions through single-pollutant or multi-pollutant time-differentiated price signals; and (5) evaluate the effects of both time and spatial differentiation in NOx emissions pricing with differentials in price signals between ozone attainment and nonattainment areas.

Summary/Accomplishments (Outputs/Outcomes):

This study developed an integrated modeling framework that included a novel two-stage unit commitment power system model (Control Technology Investment at Nash Equilibrium with Unit Commitment or CONTINU) together with the Comprehensive Air quality Model with extensions (CAMx) to capture short- and long-term generator responses (i.e., redispatching and control technology installations), emissions, cost-effectiveness, and air quality implications of time differentiated pricing policies for NOx and SO2 emissions alone and in combination with season-wide policies in the ERCOT and Mid-Atlantic power systems. Time and spatial differentiation in NOx emissions pricing also was considered through price differentials between ozone attainment and nonattainment areas in each system. Dispatching algorithms in CONTINU accounted for real-world operational constraints, such as minimum load and ramping, and included endogenous decision capabilities regarding control technology (Selective Catalytic Reduction or Flue-Gas Desulfurization) installation. Implementation of an open-loop Nash equilibrium algorithm allowed each generator to evaluate control technology adoption decisions given the decisions of all other generators in the system. Its application was critical to capture the market dynamics of control technology adoption, which affect system-wide prices and therefore the profitability of adoption decisions by generators. Averaged across all monitoring site locations, days with predicted maximum daily average 8-hour ozone concentrations greater than 60 ppb were defined as high ozone days.

Key findings from the analysis of season-wide and time-differentiated policies are summarized below:

  • CONTINU captures generator startups and shutdowns and accounts for minimum load and minimum up and down time constraints, which results in more realistic representations of emissions, costs, and control technology adoption decisions than simple economic dispatch models.
  • Sufficient flexibility existed in the ERCOT and Mid-Atlantic PJM systems, which have marked differences in demand and fuel generation mix, to accommodate time-differentiated emissions pricing without incurring non-served energy.
  • Single-pollutant time-differentiated pricing policies achieved substantial reductions in emissions even at relatively modest price signals in both systems. A $5,000/ton NOx price provided system wide average NOx reductions of 38% and 21% on high ozone days in the Mid-Atlantic PJM and ERCOT systems, respectively.
  • Redispatching from coal- (and biomass-fired generation in Mid-Atlantic PJM) to natural gas-fired generation was the primary mechanism for emissions reductions in both systems with the implementation of time-differentiated pricing regardless of control technology installation choices.
  • A subset of eligible coal-fired generators with higher NOx emissions rates within Mid-Atlantic PJM elected SCR installation at time-differentiated NOx prices at or above $100,000/ton. No generators within ERCOT elected SCR installation, and no generators within either system elected FGD installation regardless of the time-differentiated emissions price signal.
  • Single pollutant time-differentiated pricing policies had co-benefits for joint pollutant reductions but of a different magnitude and cost per ton. On high ozone days, SO2 emissions were reduced by 50% in ERCOT and 75% within the Mid-Atlantic PJM grid at a time-differentiated NOx price of $50,000/ton. In the Mid-Atlantic PJM system, a $50,000/ton SO2 price yielded NOx reductions of 63% ($10,350/ton NOx reduced) on high ozone days; in contrast, a $50,000/ton NOx price yielded reductions of 80% ($6,710/ton of NOx reduced). Incremental costs associated with NOx reductions achieved by SO2 policies alone were more pronounced in ERCOT that had an increased reliance on natural gas-fired generation with higher associated energy costs; an SO2 price of $50,000/ton yielded a 31% ($34,910/ton NOx) reduction in NOx emissions relative to a 53% ($11,410/ton NOx) reduction at the comparable NOx price.
  • On high ozone days, emissions reductions and production costs incurred from time-differentiated pricing were competitive with those of a season-wide program at the same price level.
  • Policy goals must be carefully considered. Season-wide market-based programs have the overall goal of reducing continental-scale pollution and transport to downwind areas and achieve greater cumulative emissions reductions over the summer season at a comparable cost per ton than those of time-differentiated NOx pricing. However, if an objective is to specifically achieve reductions in emissions on days conducive to the formation and accumulation of high pollution levels, then the untargeted application of season-wide program is less cost-effective than a time-differentiated program.
  • The utility of considering time-differentiated pricing as a complement to a season-wide program such as CSAPR was evident for both systems and was particularly pronounced for the MidAtlantic PJM system, which relies more heavily on coal- than natural gas-fired generation. NOx emissions were reduced by an additional 10% - 25% in ERCOT and 20% - 30% in the MidAtlantic PJM system with layered single or joint pollutant pricing policies of $5,000/ton relative to CSAPR alone on high ozone days.
  • Layered policies resulted in reductions in regional mean and seasonal fourth highest MDA8 ozone concentrations that exceeded that of CSAPR in both systems. Benefits were particularly widespread in the Mid-Atlantic PJM region. The median difference in the fourth highest MDA8 ozone concentration at the locations of the EPA’s Air Quality System (AQS) sites was -0.6 ppb with a layered joint pollutant policy versus -0.13 ppb with the implementation of CSAPR alone on high ozone days.
  • With time-differentiated NOx prices of $20,000/ton or higher, median reductions in the seasonal fourth highest MDA8 ozone concentration were -1 to -1.3 ppb at the locations of AQS sites within the Mid-Atlantic PJM system. Individual locations had reductions in the fourth highest MDA8 ozone concentration exceeding -4 ppb and in daily MDA8 ozone concentrations of almost -15 ppb. Reductions were most pronounced in southeastern and southwestern Pennsylvania.
  • Time-differentiated SO2 policies were generally as effective as time-differentiated NOx policies for reducing regional mean MDA8 ozone concentrations on high ozone days in both systems.
  • Median reductions in the fourth highest MDA8 ozone concentrations at AQS sites within ERCOT were markedly smaller than in the Mid-Atlantic PJM system (< -0.05 ppb for all policies considered). Individual locations had reductions in the fourth highest MDA8 ozone concentration that exceeded -1 ppb and in daily MDA8 ozone concentrations of nearly -2 ppb. The locations of reductions in MDA8 ozone concentrations exhibited greater spatial heterogeneity within ERCOT than Mid-Atlantic PJM and were not necessarily in close proximity to AQS sites. The most substantial air quality benefits occurred in northeastern Texas coincident with the location of much of the coal-fired generation in the state.
  • Disbenefits, or increases in ozone concentrations, in response to a given time-differentiated policy scenario were evident but were not widespread in the two systems. Disbenefits were more prevalent under time-differentiated SO2 policies than comparable NOx policies and arose because of two different circumstances: (1) a reduction in NOx emissions that resulted in less titration of ozone or (2) a shift or increase in NOx emissions that contributed to an increase in ozone formation.
  • Under time-differentiated SO2 policies within the ERCOT system, the strong incentive to reduce SO2 emissions from coal-fired generation in northeastern Texas drove the substitution of gas-fired generation that shifted NOx emissions and resulted in isolated instances of increased ozone concentrations in other areas of the state. Disbenefits associated with time-differentiated NOx policies were less than comparable SO2 policies and primarily occurred in the Houston-Galveston area due to NOx titration.
  • Instances of disbenefits under a time-differentiated SO2 pricing policy within the Mid-Atlantic PJM system were associated with shifts in generation that increased NOx emissions, similar to ERCOT, but also reductions in NOx emissions and titration of ozone. Disbenefits of relatively larger magnitude under either time-differentiated SO2 or NOx policies occurred primarily in Pennsylvania associated with reductions in NOx titration.
  • The findings for the two systems suggested the potential utility of evaluating the spatial differentiation of time-differentiated pricing policies for ozone nonattainment areas as well as the relative benefits of joint time-differentiated pollutant pricing policies to mitigate unintended consequences of comparable single pollutant pricing policies.
  • Co-benefits of reductions in 24-hour PM2.5 concentrations were evident with the implementation of time-differentiated price signals for NOx and/or SO2 on high ozone days, but full realization of reductions in peak concentrations may require distinct time-differentiation.
  • A time-differentiated scheme could be implemented as a higher redemption ratio of emissions permits on high ozone days or as an entirely separate trading program layered with a season-wide policy.

Key findings from the analysis of time and spatially differentiated policies are summarized below:

  • Implementing time and spatially differentiated NOx pricing via differentials between ozone attainment and nonattainment areas on high ozone days resulted in changes in generation and NOx emission patterns relative to layered time-differentiated or season-wide policies. The responses in both ERCOT and Mid-Atlantic PJM were complex and highly sensitive to the magnitude of the NOx price differential between the areas. Three factors contributed to the effects: (1) the interregional transfer of generation from nonattainment to attainment areas; (2) the intraregional shift from coal- to natural gas-fired generation within each region, and (3) the shifting to lower emitting units within a single fuel type.
  • Differentials in NOx emissions pricing below $5,000:$25,000 (i.e., $X,000:$Y,000 to represent that a NOx price of $X,000 per ton is imposed in attainment regions on high-ozone days, and a price of $Y,000 is imposed in nonattainment regions on high ozone days) resulted in reductions in MDA8 ozone concentrations in attainment and nonattainment areas that were generally comparable to or exceeded the benefits of a time-differentiated only policy, but at greater costs in the ERCOT and Mid-Atlantic PJM systems.
  • Increasing the price differential above $5,000:$25,000 had complex effects in both systems. Reductions in ozone benefits in attainment areas were accompanied by considerably higher system-wide costs than lower price differentials. The effects in nonattainment areas were not uniform across grid cells with AQS monitors, with some locations experiencing additional benefits while others had diminishing air quality benefits as NOx emissions due to generation in neighboring attainment counties increased.
  • These results suggested that time and spatially differentiated policies must be carefully designed and evaluated to determine if the relative benefits exceed that of time-differentiated only strategies and if implemented whether their effectiveness would be improved by extending price differentials to include nearby attainment counties with large upwind generators that routinely affect nonattainment area air quality.

Journal Articles on this Report : 2 Displayed | Download in RIS Format

Other project views: All 11 publications 2 publications in selected types All 2 journal articles
Type Citation Project Document Sources
Journal Article Craig M, McDonald-Buller E, Webster M. Technology adoption under time-differentiated market-based instruments for pollution control. Energy Economics 2016;60:23-34. R835216 (Final)
  • Full-text: ScienceDirect-Full Text HTML
  • Abstract: ScienceDirect-Abstract
  • Other: ScienceDirect-Full Text PDF
  • Journal Article McDonald-Buller E, Kimura Y, Craig M, McGaughey G, Allen D, Webster M. Dynamic management of NOx and SO2 emissions in the Texas and Mid-Atlantic electric power systems and implications for air quality. Environmental Science & Technology 2016;50(3):1611-1619. R835216 (2014)
    R835216 (Final)
  • Abstract from PubMed
  • Full-text: ES&T-Full Text HTML
  • Abstract: ES&T-Abstract
  • Other: ES&T-Full Text PDF
  • Supplemental Keywords:

    electricity generation, emissions trading, ozone, particulate matter, Texas, CAMx;

    Progress and Final Reports:

    Original Abstract
  • 2012 Progress Report
  • 2013 Progress Report
  • 2014 Progress Report