Lawrence Berkeley National Laboratory

The increasing frequency and severity of wildfires pose significant threats to utility infrastructure and community safety. Wildfires can be ignited due to component failures under extreme weather conditions. System hardening measures aimed at preventing these failures can reduce the probability of endogenous wildfire ignition, introducing decision-dependent uncertainty (DDU). However, existing methods that consider such DDU typically assume independent component failures, neglecting the critical interdependencies between components destroyed by wildfire and those affected by their spread. These interdependencies render such methods inadequate. To address this, we propose a novel formulation that incorporates both wildfire-ignition DDU and a second type of DDU — the interdependency between wildfire-ignited and wildfire-affected components. We develop a two-stage wildfire-preventive decision-dependent resilience enhancement (WDDRE) model for distribution systems. The model optimizes line hardening, distributed generation allocation, and network reconfiguration to enhance system resilience by preventing endogenous wildfire ignition caused by component failures. Numerical experiments demonstrate the effectiveness and superiority of the WDDRE model, showcasing its robustness in managing wildfire stochasticity and the two kinds of DDUs related to planning decisions, significantly reducing potential wildfire risks.