Improving Electricity Peak Demand Forecasts with Measured Data: An Application of PG&E's Residential End-Use Metered Data

Forecasting in an era of integrated resource planning requires end-use·detail, not only for annual energy, but also for hourly loads. While peak and hourly end-use demand forecasting models have been available for some time, measured data to support these activities have been scarce. In late 1984, Pacific Gas and Electric Company began metering domestic appliances for over 700 residential customers. In this paper, we analyze central air conditioner data collected in this project between 1985 and 1989 to develop inputs for an electricity peak demand forecasting model currently in use in California. We describe the structure of the forecasting model, and discuss the requirements the model places on data development. We examine the issues associated with aggregation over days and over geographic regions.