Load forecasting is as complex as ever, combining econometrics, emerging but uncertain large loads, customer-sited DERs, and policy-driven electrification. E3 forecasts electricity demand for utilities, regulators, and system planners, from the aggregate system peak impacts to adoption of individual technologies through to hourly load on a specific feeder. We project how quickly customers take up electric vehicles, heat pumps, rooftop solar, storage, and efficiency measures, develop the hourly load shapes that go with each, and combine the two into the energy and peak forecasts that feed a resource plan, a distribution plan, or a rate case. We also forecast new large loads, including data centers and industrial customers, where the drivers are different and the growth arrives in much larger increments.
As the grid becomes more constrained with new load growth we can provide an actionable range of peak load scenarios including key drivers and trackable markers to know how to plan under uncertainty in real time.
Technology adoption forecasts start from customer economics. We use Bass diffusion, S-curve, and logit models calibrated using historical. Then we bring in technology costs, incentives, tax credit changes, tariffs, and local market conditions to capture the impact on adoption of economics across several potentially competing technologies. We deliver a range of scenarios rather than a single number to account for policy and market uncertainty. Each technology then gets its own hourly shape based on factors including heating service demand, weather, and charging constraints derived from real driving data. Adoption components may be regarded individually or combined and diversified to build a complete load forecast that accounts for real-world dispersion.
Once we know how much load there is, we can work on the question of where that load occurs. Forecasting Anywhere, built with Integral Analytics, distributes adoption across a service territory using machine learning and parcel-level data on building type, square footage, zoning, and land use. The results aggregate to census tract, feeder, substation, or disadvantaged community boundaries, and load directly into the tools planners already use, including LoadSEER, CYME, and Synergi. We also build climate-informed forecasts from downscaled climate models. Many engagements end with the forecasting tool in the client’s hands so staff can update assumptions and run sensitivities between planning cycles.
Tory Clark
Eric Cutter
