Traditional planning approaches often rely on deterministic rules such as reserve margin or N–k contingency criteria, which provide simplicity but may underestimate the risks associated with equipment failures and large-scale disruptions. To address these limitations, this work first develops a suite of probabilistic and stochastic optimization frameworks that explicitly account for uncertainties in generation and transmission availability. A probabilistic reliability-constrained expansion planning model is proposed to optimize both long-term investment decisions—such as the timing, location, and capacity of generation additions or retirements—and short-term operational decisions including unit commitment and economic dispatch. Formulated using Generalized Disjunctive Programming, the model captures the dual role of backup generators, which can act as reserves or contribute directly to electricity production. To enable large-scale applications, a bilevel decomposition algorithm is designed, equipped with logic-based tailored cuts that significantly improve convergence.
The framework is further extended to integrate transmission planning and environmental policies, including renewable energy targets and carbon reduction goals. A simplification strategy is developed to identify critical nodes and generators whose failures most significantly impact system performance, thereby reducing computational burden without sacrificing accuracy. Application to San Diego County illustrates the interplay between reliability and policy-driven planning: larger dispatchable capacity requires higher reserves due to greater failure risks, whereas higher renewable penetration reduces additional reserve needs because of the inherently low failure rates of renewable technologies.
Building on these foundations, a two-stage stochastic programming model is developed for resilience-oriented planning under disruption scenarios. The model determines both proactive strategies, such as line hardening and distributed resource deployment, and reactive operational responses, including re-dispatch and network reconfiguration. A scenario reduction–based decomposition algorithm ensures computational feasibility while capturing the probabilistic nature of extreme events. Results highlight the value of explicitly incorporating disruption uncertainty into long-term planning to minimize losses and enhance system resilience.
These contributions demonstrate how probabilistic reliability modeling and stochastic resilience planning can strengthen the robustness, sustainability, and adaptability of future power systems in the face of growing uncertainty and extreme events.
