Operational Outage, Ignition and Wildfire Risk Modeling at Pacific Gas and Electric Co. (Invited Presentation)

PG&E operates near 100,000 miles of overhead electric lines in Northern and Central California that can fail during weather events and present risk of wildfire ignition when fuels are receptive. Operational grid mitigations such as proactive deenergizations and sensitive relay settings effectively reduce ignition probability but lead to increased customer outages. Thus, it is imperative to deploy accurate operational risk models to drive daily decisions to mitigate risk. This talk focuses on our fifth-generation machine learning outage, ignition and wildfire probability models trained by coupling a WRF-based 35-year climatology of weather, dead and live fuel moisture data with sub-daily fire occurrence and PG&E outage and ignition data.