Towards objective evaluation of the accuracy of marginal emissions factors

Marginal Emission Factors are a modeled quantity that measures how a change in demand on the electrical grid affects overall system emissions by considering the carbon intensity of generator(s) that would respond to this change. It is reasonably impossible to record ground truth Marginal Emission Factors, preventing any validation of these models using standard techniques. However, many decision makers are using Marginal Emissions Factors to decide how to optimize load, or where to build new renewable generation to reduce CO 2 emissions consistent with the UN Framework Convention on Climate Change Paris Agreement. This paper provides a rubric of empirical tests to evaluate various Marginal Emission Factor models by considering observable behaviors that marginal Emission Factor models would be expected to emulate. This presents a novel step towards the full validation of these models. The rubric of tests is then applied to seven popular Marginal Emission Factor models. The results of this analysis show that models based on simulated results, rather than real-time information, tend to violate the tests more often but still present value when deciding the best location to site new load or generation. Only one model (WattTime), passes all of the proposed tests, however all models demonstrate general alignment with the expectations of the tests. • Marginal Emission Factors (MEFs) measure the emissions impact of an action on the grid. • MEFs should be used for load shifting or siting renewables to cut total CO 2 emissions. • Nuance is required when selecting an appropriate MEF model for a given use case. • Proposed tests validate MEF models for use when ground-truth data is unavailable. • Tests on seven popular MEF models show general alignment with expectations.