A Non-Iterative Reliability-Constrained Generation Expansion Planning: A Multi-Cluster Risk Surrogate Model

Reliability-constrained generation expansion planning (GEP) is crucial to balance economy and reliability for power systems with high renewable penetration. Most of the existing reliability-constrained GEP methods treat reliability indices as exogenous constraints, which yields subjective and empirical results. Endogenous methods are computationally intractable due to the substantial computational burden caused by massive samples and the complexity of transmission networks. This work develops an efficient non-iterative GEP framework with endogenous reliability assessment built upon a novel risk surrogate model. The implicit relationship among high-dimensional uncertainties, installation capacity, and expected energy not served (EENS) is characterized by the proposed risk surrogate model based on polynomial chaos expansion (PCE), which incorporates network constraints. To enhance approximation precision, a multi-cluster PCE strategy is proposed to construct piecewise analytical formulations that accurately represent the underlying non-convex risk function. Relying on this surrogate model, an efficient non-iterative reliability-constrained GEP framework is proposed, which eliminates iterative solving and balances economy and reliability. Embedding the reliability analytical formulations into the GEP model, the two-level optimization model is transformed into a single-level optimization model. The simulation results demonstrate the effectiveness of the proposed method.