The challenge
A major West Coast utility needed a better way to predict event-level reductions across its DR portfolio. It needed to balance intermittent production from solar resources, meet regulatory obligations, and fulfill its goal of using DR as a virtual power plant.
The utility had developed an in-house forecasting tool that based predictions of load-shedding performance on historical averages. The tool was useful but didn’t provide the necessary accuracy. The utility needed a more granular, dynamic approach.
The solution
The key to more accurate forecasts? Modeling individual consumer behavior rather than depending on annual program averages. The utility turned to E Source and its AI-powered suite of solutions to combine historical customer data, smart meter data, and the utility’s proprietary cache of third-party consumer data and develop a granular, individual forecast for every customer enrolled in DR. E Source then aggregated those microforecasts to provide program-level predictions.
The results
With E Source’s help, the utility boosted the performance of its DR programs by 30% and was provided with the precise data needed to apply DR more locally on the grid. The program runs daily during DR season to optimize the utility’s portfolio of DR resources, comprising 350,000 customers.