Content area
Public Safety Power Shutoffs (PSPS) mitigate wildfire risks by deactivating grid components in high-risk areas and redispatching generators to minimize load shedding. Effective PSPS planning requires coordination across time scales and managing uncertainties in grid demand and wildfire forecasts. This work presents four key contributions to enhance PSPS planning.
First, it introduces deterministic and two-stage mean-CVaR stochastic frameworks to optimize generator commitment and transmission line de-energization strategies. Comparing these frameworks to the deterministic Optimal Power Shut-off (OPS) model highlights that two-stage stochastic planning yields lower expected costs. Risk aversion to total economic costs and wildfire risk mitigation create a Pareto optimal front, resulting in different first-stage decisions for high- and low-demand scenarios.
Second, I extend our model to incorporate the Large Fire Probability (WLFP), which correlates more strongly with observed wildfire ignition probabilities (WIP) than the commonly used Wildland Fire Potential Index (WFPI). By mapping WLFP to wildfire ignition log probabilities (log(WIP)), a constraint with Bernoulli random wildfire ignitions becomes a tractable linear constraint. Simulations on the IEEE RTS 24-bus test system reveal that WLFP-based risk assessment reduces expected real-time costs by an average of 14.8% compared to WFPI-based approaches.
Third, a mixed-integer linear program (MILP) efficiently approximates a distributionally robust PSPS to model decision-dependent wildfire-driven failure probabilities. This framework balances costs and wildfire risk under varying levels of distributional robustness, defined by a parameter, $\kappa$. The socioeconomic impacts of PSPS are analyzed using the IEEE RTS 24-bus test and a section of the reduced 240-bus WECC system, focusing on the high-risk areas of Los Angeles County.
Finally, this work establishes a foundation for managing distribution-level grid resources during wildfire events using a decentralized deterministic economic dispatch. The proximal message passing (PMP) algorithm solves a relaxed Second Order Cone (SOCP) AC-OPF problem for distribution networks with distributed energy resources (DERs). PMP minimizes device operating costs, penalizes power flow constraint violations, and reduces supply-demand mismatches on a modified IEEE 13-bus system. Extending this work to incorporate wildfire risk is left for future research. By integrating stochastic frameworks, advanced wildfire risk metrics, and decentralized optimization, this research enhances power system resilience and wildfire mitigation, laying the groundwork for future innovations in PSPS planning.