Improving Solar and Load Forecasts by Reducing Operational Uncertainty

Homeowners and businesses in California have installed numerous solar photovoltaic (PV) systems because of California’s Renewable Portfolio Standard requirements as well as the decreasing costs of PV. The California Independent System Operator (California ISO), who operates California’s electric grid, does not measure behind-the-meter PV generation. The California ISO and the electric utilities are facing the uncertainty associated with PV generation profiles. The California ISO is conservatively forecasting and scheduling excess regulation and spinning reserves because of this PV uncertainty. They must extend their load forecast models to better predict when customers rely on the grid to meet their electricity requirements versus relying on their behind-the-meter PV generation. This project addresses this issue by advancing the state of the art in solar energy forecasting as it relates to the operation of the California electric grid. It undertook four specific technical tasks:

1. Investigate supplementing the California ISO’s current real-time solar data feeds.
2. Improve the California ISO’s solar production forecasts.
3. Investigate alternative net load forecasting methods to improve integrating PV generation forecasts with an operational net load forecast, and
4. Estimate the monetary value of the alternative net load forecasts and develop a framework for optimizing their use.

This research identified improvements necessary for real-time solar data and forecasts, and alternative methods of net load forecasting that provide value to the grid and its stakeholders. The research team also identified additional areas of future research. The California ISO has adopted the findings of this research and implemented these methods, providing savings for ratepayers and other stakeholders. In addition, the Australia Energy Market Operator, the New York Independent System Operator, and the Independent Electricity System Operator in Ontario Canada have implemented variations of these findings.