Success Story

Forecasting Method Makes Electric Grid Operations More Stable and Efficient

Neighboring power organizations exchange electric power to keep the electricity grid operating as efficiently and reliably as possible. Utilities rely on past trends to predict how much power they’ll need to generate in the next few minutes to hours. But accurate forecasting is challenging because it can be affected by weather, renewable energy fluctuations, and the instability of neighboring power organizations. With little time for corrective action, incorrect forecasts can raise operational costs and affect stability of the regional power market. Put simply, the ability to forecast accurately is crucial for the efficiency and safety of the power grid.

Pacific Northwest National Laboratory (PNNL) developed a software tool that makes much more accurate predictions than what most utilities are using now. To address the uncertainty associated with forecasting tasks, the software combines forecasts from multiple scenarios into one that is more reliable and accurate. This way, utilities can operate closer to the mark, improving grid reliability and efficiency, while minimizing penalties.

It was not long before this innovative forecasting method caught the attention of a heavy hitter in the industry. PJM Interconnection, of Valley Forge, Pennsylvania, is the largest grid operator in North America, serving over 61 million people. PJM licensed the software in 2013. When PNNL evaluated the software on PJM’s data, the software reduced forecast errors by up to 48 percent, which PNNL estimated could save PJM’s customers about $75M annually. These impressive results demonstrate the tremendous potential of this forecasting method to reduce the price volatility in the nation’s power market.

This work was supported by PNNL’s Lab-Directed Research and Development internal funds. Under an Agreement for Commercializing Technology, PJM funded PNNL to customize the software for PJM’s operating environment.