---

Modeling and Learning Locational Emission Rates for Low-Carbon Power System Planning and Operation

PI and co-PIs: Yize Chen (University of Alberta; Canada); Yuanyuan Shi (University of California San Diego; United States); Feng Zhao (ISO New England; United States)

Funding amount: $150,000

Project overview: As the biggest segment of greenhouse gas contributions, power sector emissions are an essential target in achieving net-zero goals. However, the carbon footprint of using electricity is not the same everywhere or every hour: it depends on which power plants are running, and how power flows across the grid. Most grid operators do not publish real-time, locational carbon intensity, largely because there isn’t a computationally efficient, market-compatible way to calculate it in fine geographic detail. This project aims at creating a machine learning-based tool to rapidly predict system-wide emissions and extract location-specific marginal carbon signals, enabling fast forecasting and decision support in partnership with electric power system operators. The project stands to help utilities, policymakers, and large flexible electricity users in shifting demand and planning investments toward cleaner electricity, and will also create a new benchmark comprising both grid data and emission models. By making “where and when electricity is cleaner” measurable at operational timescales, this work will turn carbon-aware planning and operations into a practical capability, ultimately helping reduce power-system emissions while maintaining reliability and affordability.

Power & Energy Impact Assessment