Carbon-Aware Scheduling of Flexible Electricity Loads via Fuel-Mix-Derived Grid Carbon Intensity: An ISO New England Case Study

Grid carbon intensity varies hour to hour, so a load that must finish by a deadline but not at a specific hour, such as electric-vehicle charging, heat-pump operation, or batch computation, can be shifted to a cleaner time at no cost to delivered energy. We present a transparent, fully reproducible pipeline that derives hourly, lifecycle-based grid carbon intensity for ISO New England from public fuel-mix data and documented emission factors, and that computes the exact minimum-carbon schedule for real deferrable loads: mixed-integer optimization is used where a shared per-hour allocation budget binds, and a provably optimal greedy placement where it does not. Over the full 2019 ISO New England year, hourly carbon intensity ranges from 116.7 to 389.4 gCO2/kWh; in an idealized re-timing benchmark that holds delivered electrical energy fixed and does not model building thermal dynamics or comfort bounds, carbon-aware shifting of a residential heat-pump load avoids 8.29% of its average-basis emissions under a shifted-load allocation budget of three times its median active load (12.70% without one; 12.19% when, instead, total hourly draw is held within the building’s observed peak). The attainable saving is set primarily by two operational quantities we characterize, deadline slack and the shared allocation budget, and it is strongly seasonal, ranging from roughly 10% in the milder shoulder seasons to about 3.1% in January, where the limit is the allocation budget rather than a shortage of clean hours: January’s ceiling without an allocation budget is the highest of any month. It is robust across buildings and to the least certain emission factor (between 5.57% and 7.70%). In our experiments, a weather-blind climatology forecast retains about 88% of the savings without an allocation budget and 91.5% of the savings under the allocation budget.