Predicting Distributed Solar Power Production for Utilities

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 6 January 2015: 8:45 AM
224B (Phoenix Convention Center - West and North Buildings)
Julia M. Pearson, NCAR, Boulder, CO; and S. E. Haupt, T. L. Jensen, C. Burghardt, T. C. McCandless, T. Brummet, and S. Dettling

Distributed solar installations, such as rooftop solar panels, impact a utility's electrical load by decreasing the apparent load in the case of net metering. This impact, known also as a load cutout, will continue to grow as more distributed solar power is installed. In order to help a utility efficiently plan energy production to meet load and to balance resources across the grid, the National Center for Atmospheric Research has developed a forecast system targeted at predicting the distributed solar power production using Dynamic Integrated foreCast (DICast®) predictions of meteorological variables as well as aggregated photovoltaic (PV) production observations. This algorithm first uses a statistical learning technique to predict the percentage of solar power generated at specific sites across Colorado and then upscales the forecast to nearby installations. In this paper we will present an overview of the algorithm as well as present a statistical evaluation of the forecasts' performance.