Options for EIA to Publish CO2 Emissions Rates for Electricity

This report reviews the use cases for emissions rate data and the options available to the Energy Information Agency in how to publish data on both average and marginal emissions rates.

Download

Date

Aug. 11, 2022

Authors

Karen Palmer, Brian Prest, Seth Villanueva, and Stuart Iler

Publication

Report

Reading time

7 minutes

Abstract

Demand for data on the CO2 intensity of US electricity consumption is growing as governments and private companies seek to understand the emissions effects of both their electricity consumption and their clean energy investment choices. The desire for transparent and consistent data on electricity emissions rates led the US Congress, in the Infrastructure Investment and Jobs Act, to call for the US Energy Information Administration (EIA) to publish spatially and temporally granular electricity emissions rate data, beginning in late 2022. This report reviews the use cases for emissions rate data and options available to EIA to publish data on both average and marginal emissions rates, and it presents relevant findings and recommendations for EIA to consider.

Executive Summary

Decarbonization efforts in the US consist of a mix of government policies to promote clean sources of energy and improve energy efficiency and voluntary actions by private actors to reduce their CO2 emissions. Understanding both the likely and actual benefits of both policies and private actions requires information about the carbon intensity of different sources of energy including electricity. Demand for data on the CO2 intensity of US electricity production and consumption is growing as governments and private companies seek to understand the emissions effects of both their electricity consumption and their clean energy investment choices. Emissions rate information is also important for Scope 2 emissions accounting under the GHG Protocol Corporate Standard1 that provides a way of tracking progress toward clean energy targets that many companies have declared.

The desire for transparent and consistent data on electricity emissions rates led the US Congress, in the Infrastructure Investment and Jobs Act (IIJA), to call for the US Energy Information Administration (EIA) to publish spatially and temporally granular electricity emissions rate data, beginning in late 2022. Specifically, the IIJA calls on EIA to report on hourly operating data, including “where available, the estimated marginal greenhouse gas emissions per megawatt hour of electricity generated” within each balancing authority and by pricing node. The law also calls on EIA to harmonize its electric system operating data with GHG and other relevant data collected by EPA or other federal agencies, as well as data collected by state or renewable energy credit registries. The resulting integrated data set should include net generation data and “where available, the average and marginal greenhouse gas emissions by megawatt hour of electricity generated within the boundaries of each balancing authority,” to be offered on a real-time basis through a publicly accessible application programming interface (API).

This report reviews the use cases for emissions rate data and options available to EIA to publish data on both average and marginal emissions rates. Our findings and recommendations for EIA’s consideration are listed below.

Summary of Findings

  • The emissions metrics requested of EIA in the IIJA are relevant to both retrospective emissions accounting (typically hourly average emissions rates) and near-term operating decisions (short-run, hourly marginal operating emissions, at the balancing authority or nodal level, and in real time). These are the most common applications, although in certain cases, average emissions rates can be used to inform operational decisions and marginal rates can be used for retrospective purposes. • The primary two types of approaches used by providers of real-time operating marginal emissions rate data are (1) economic dispatch-based approaches that reveal the characteristics of the marginal generator(s) at specific locations to which estimated emissions rates can be applied; and (2) statistical and machine learning models that predict marginal generator type(s) at a point in time based on how generation (and hence emissions) will change in response to changes in load under different grid conditions, including load levels, prices, and weather.
  • The two approaches to calculating marginal emissions rates—economic dispatch-based and statistical—have trade-offs. Economic dispatch-based approaches reflect the algorithms used by grid operators and can yield granular nodal estimates, but they require access to detailed data often known only to grid operators. Statistical approaches are more scalable to new regions because they rely primarily on publicly available data, but they can struggle to produce reliable estimates at the very disaggregated nodal level.
  • All marginal emissions rate estimation methodologies are imperfect for one reason or another. For example, marginal emissions estimates produced by economic dispatch-based approaches are strictly valid only for small changes in load (say, 1 MW or less). A change in load large enough to change which generator is marginal will lead to a different marginal emissions rate, and the magnitude and direction of this difference are not necessarily known without more detailed information on the dispatch stack and transmission constraints (or reduced-form representations of them by fuel type). Estimates produced by statistical models inevitably involve assumptions and approximations that also lead to imperfect estimates.
  • Approaches to calculating marginal emissions rates that yield geographically and temporally detailed data require aggregation to create hourly estimates and estimates at broader spatial resolutions, such as at the balancing authority or sub-balancing-authority level. When transmission is constrained, multiple generators will be on the margin; in those cases, one approach to spatial aggregation (used by ISO New England (ISO-NE)) would be taking weighted averages, weighing the emissions rate estimates for those generators by the amount of load served by each. Translating sub-hourly values into hourly could be done using different approaches. ISO-NE uses both a time-based approach and a marginal load-based approach to weighting five-minute estimates to provide two different estimates of hourly marginal emissions rates.
  • Providing information on the emissions rates for the residual mix is challenging for EIA in the near term because information on privately procured clean energy (outside the transactions directly involving the utilities from which it typically collects information) is limited.

Summary of Recommendations

Average Emissions Rates

  • EIA has a clear path forward for producing average emissions rates (both production- and consumption-based) through an extension of the methodology underlying its recently released hourly national average emissions rate estimates.2 EIA is already producing production-based average emissions estimates at the national level, and the same methodology could be applied at the balancing authority level, with appropriate adjustments to assumed emissions factors. In extending those methods to produce consumption-based average emissions rate estimates at the balancing authority (rather than national) level, EIA should account for interchange between regions—for example, by using flow-tracing algorithms, as mentioned in this report. Other improvements can be made on an ongoing basis—for example, by using more regionally (e.g., balancing authority level instead of national) and temporally (e.g., seasonal or monthly instead of annual) granular and representative estimated emissions factors.

Marginal Emissions Rates

  • For marginal emissions rates, EIA has several reasonable options that we would endorse. Our preferred approach would be to request that the data be reported by the balancing authorities, from which EIA routinely collects grid operating data in real time. PJM and ISO-NE are balancing authorities that are already producing marginal emissions rate estimates, and some others could produce analogous estimates. Below are three potential approaches that a balancing authority could take to produce such estimates, listed in order of preference, contingent on implementability, which may vary by balancing authority.
    1. Adopt a methodology similar to that of PJM and ISO-NE, which use their economic dispatch algorithms to identify marginal generators.
    2. Extend the balancing authority’s existing methodologies that identify marginal sources of generation (e.g., MISO reports data on marginal fuel types) and then apply emissions factors. In some instances, EIA could perform such calculations in house—for example, by taking MISO’s reported fuel on the margin dataset and applying emissions factors, just as EIA does in its calculation of average emissions rates.
    3. Partner with existing data providers to develop such estimates (e.g., the model used by the California Self-Generation Incentive Program).
  • It is unlikely that all balancing authorities will be able to produce marginal emissions rate estimates right away; therefore, EIA would need additional sources to publish nationally comprehensive data on marginal emissions Although some existing methods (such as the statistical methods discussed in this report) can produce geographically comprehensive estimates of marginal emissions rates for all balancing authorities in the contiguous 48 states, quantitative comparisons of their estimates are insufficient. Thus, we recommend that an independent organization work with third-party providers of emissions rate data and compile a comprehensive set of marginal emissions rate estimates, then use those data to quantify their similarities and differences and assess their relative reliability. This compilation and quantitative comparison of data would allow data users to better understand the differences among alternative estimates. It would also enable data providers and other analysts to improve, compare, validate, stress-test, and harmonize their methods and data products.
  • Such a public quantitative comparison would inform EIA’s task of providing marginal emissions rate information in regions where balancing authorities may not yet be able to produce the necessary data. With a public data comparison in place, EIA could then consider publishing data provided by third-party organizations, such as those discussed in this report, with clear and appropriate data definitions, descriptions, and caveats.
  • If multiple approaches to estimating a particular emissions rate metric are available and deemed reliable, EIA should consider presenting more than one set of estimates so that data users and researchers, at EIA or elsewhere, can compare the various methods to understand their differences. When presenting multiple estimates for the same metric, EIA should include clear definitions of the data, include caveats, and note the underlying methodology (e.g., economic dispatch, statistical). These recommendations would apply regardless of the entity—EIA or an independent third party— presenting the standardized estimates.

All Emissions Rates

  • Although there are not yet standards for emissions-related data, balancing authorities, utilities, private companies, and government agencies can work now to build flexibility into their data publishing systems (e.g., the static files they publish and the APIs they maintain) so that they can more easily adapt those systems to any future data standards. This is relevant both for data that are used as inputs to the estimation of emissions rates (e.g., information on dispatch) and for the publication of emissions rate estimates themselves (e.g., average and marginal emissions rates). EIA should consider making its emissions rates and other data conform, as applicable, to any future standards. That would facilitate the transferability and scalability of data and applications among regions and promote data transparency and reliability.
  • For emissions rate data specifically, data publishers should be transparent and provide documentation for their metrics, including the conceptual basis for each metric and the underlying methodologies and data sources. Such documentation ensures that users understand what the data mean and how they should be used, while also setting the stage for future standardized data definitions that might incorporate (and differentiate) the metrics provided by different entities.
  • As part of the IIJA’s required interagency data harmonization process, EIA should coordinate with EPA, including the Green Power Partnership, which has expertise in voluntary renewable procurement, to ensure common understanding of the scope of data collection necessary to develop estimates of residual mix emissions rates.

Authors

Related Content