Grant PUD is a customer-owned utility with a mission to deliver safe, efficient, and reliable electricity and fiber services across central Washington. With a total of 55,000 meters and a new peak demand record of over 1,000MWs, Grant PUD makes managing energy demand efficiently a key priority. In 2024, as part of a multi-year initiative to enhance commercial value and deliver greater benefits to the utility, county, and ratepayers, Grant PUD prioritized improving its load forecasting capabilities to optimize procurement strategies and wholesale operations.
Before partnering with Amperon, Grant PUD relied on a forecast provider that did not appear to be investing in model improvements. "I know behind the scenes, [they] primarily relied on linear regression approaches that risked overfitting, especially as our territory experiences load growth and extreme weather," said Paul Dietz, Senior Manager of Forecasting and Market Analytics. Sure enough, his model performance concerns were realized during the MLK Weekend cold snap in 2024.
In addition to this acute weather event, Grant PUD recognized upcoming challenges given new mandates of SPP’s Western States’ Resource Adequacy Program (WRAP). Their previous vendor provided only a 6-day forecast, which fails to meet upcoming 7-day forecast mandates. These concerns could not be left unaddressed, and it became clear that Grant PUD needed a new forecasting vendor. This led Grant PUD to Amperon, which now delivers accurate forecasts while serving as a responsive, dependable partner.
Grant PUD’s Business Intelligence and Market Analytic steam seamlessly integrated Amperon’s forecasts into daily workflows through the web-enabled platform and modern API to support load management, real-time control room activities, and load balancing efforts. “Partnering with Amperon has been a critical step forward for our division. We made a commitment to achieve timely and accurate load forecasts, and by integrating their forecasts into our workflows, our teams—from control rooms to market balancing—can confidently inform realtime operational and financial decisions,” says Dietz.
Amperon’s hybrid regression/ML-driven model provides hourly load predictions extending up to 15 days out. Unlike traditional regression models, Amperon’s methodology combines proven regression approaches and novel techniques that machine learning has recently enabled to continuously calibrate and learn from update sin weather and load data, ensuring forecasts remain accurate even in rapidly changing conditions. In addition, Amperon’s tailored model for Grant PUD incorporates 13 weather points, capturing microclimates and seasonal variations that influence demand. This level of granularity allows the utility to manage load fluctuations and improve its response to extreme weather.
Since implementing Amperon’s solution, Grant PUD has seen improvements in forecast accuracy. Even during extreme weather months that previously disrupted operations in 2024, Amperon’s average monthly day-ahead MAPE and daily peak demand day-ahead APE stayed below 2% while previous vendors delivered day-ahead MAPEs exceeding 10% during MLK Weekend 2024. This improvement allows Grant PUD to better manage wholesale positions and mitigate financial risks.
“Enhancing our load forecasting is a vital step in modernizing our wholesale market strategy,” says John Mertlich, Chief Commercial Officer. “By leveraging advanced analytics, we optimize resource planning, mitigate financial risk, and ensure reliability for our customers as the energy landscape evolves." With Amperon’s forecasting solution and market experts, Grant PUD is well-positioned to navigate continued load growth and extreme weather.