Optimizing grid defenses against extreme weather
A new advanced modeling approach demonstrates that integrating proactive grid hardening with reactive recovery strategies significantly reduces the costs of extreme weather-driven power outages.
Why it matters: Extreme weather events are putting unprecedented strain on electricity systems. For example, more than one million customers lost power during Winter Storm Fern in January 2026, and half a million lost power during a blizzard in the Northeast U.S. in February 2026. Relying solely on fixing the grid after extreme weather hits is no longer a viable or affordable strategy, highlighting the need for proactive resilience planning.
Catch up quick: Most current approaches to grid resilience handle proactive decisions (like tree trimming or burying power lines) and reactive decisions (like dispatching repair crews) separately. And traditional planning relies on simplified forecasts or historical averages that often fail to predict the devastating, localized impacts of severe and rare storms.
Action taken: CMU researchers Shixiang (Woody) Zhu and Ramteen Sioshansi developed a novel optimization model that links proactive investments, the uncertain impacts of extreme weather disruptions, and reactive response budgets. The goal is to help utilities make more effective long-term investment decisions while still accounting for the immediate actions needed after extreme weather to restore power. The model was then tested using real-world data from the three major Massachusetts snowstorms in March 2018.
Findings: Optimizing proactive and reactive decisions jointly leads to lower worst-case outage costs compared to traditional, siloed planning methods. In other words, utilities can save money in outages by combining this work. This model allows grid operators to build highly reliable, data-driven uncertainty profiles that adapt to specific, localized regional risks, even when historical data is limited or noisy.
Policy takeaway: Utilities should adopt comprehensive resilience plans that co-optimize both proactive infrastructure upgrades and reactive emergency response capabilities. Transitioning to integrated, adaptive robust optimization ensures that every dollar spent on grid resilience works harder to keep the lights on when the worst happens.
This work was supported by the Electric Power Research Institute, the Carnegie Mellon Electricity Industry Center, and the Scott Institute for Energy Innovation.