Wildfire Risk-Informed Preventive-Corrective Decision Making under Renewable Uncertainty

Researchers Satyaprajna Sahoo and Anamitra Pal have published a novel paper titled 'Wildfire Risk-Informed Preventive-Corrective Decision Making under Renewable Uncertainty' on arXiv. The work addresses a critical vulnerability in modern power grids: the intersection of increasing wildfire frequency and the growing penetration of variable renewable energy sources like wind and solar. Unlike traditional single-asset failures, wildfires can simultaneously threaten multiple transmission lines and substations, creating cascading outages that rapidly degrade system stability. The researchers' key insight is that wildfire precursors—dry and windy conditions—are predictable with high confidence at least 24 hours in advance, creating a window for proactive intervention.

Their solution is a sophisticated, two-stage AI optimization model. The first stage is a day-ahead 'preventive' unit commitment and power flow plan, made using wildfire risk forecasts. The second is a real-time 'corrective' control system that adjusts operations as conditions evolve. This stochastic formulation explicitly accounts for the uncertainty of renewable generation output, which can fluctuate wildly during the same weather events that fuel fires. The model was validated on a reduced 240-bus model of the complex US Western Interconnection grid. Results demonstrate that this coordinated approach can significantly boost grid resilience across varying levels of wildfire risk without sacrificing economic efficiency, offering a blueprint for utilities operating in fire-prone regions.