Customized Benders Decomposition for Unit Commitment Integrated Generation Expansion Planning

Generation expansion planning (GEP) is a crucial tool in guiding the growth of electric power systems towards achieving goals of carbon peaking and neutrality. Accurately capturing the high-density variations of renewable energy sources (RES) necessitates integrating the unit commitment (UC) decision process into the planning investment stage. However, the UC-integrated GEP model typically comprises millions of variables and constraints, rendering it unsolvable by commercial solvers directly. To tackle this large-scale optimization problem, this paper presents a customized Benders Decomposition approach, partitioning the UC-integrated GEP model into a master problem optimizing investment variables and several sub-problems representing annual operation. The optimality cuts and feasibility cuts, fundamental to the decomposition algorithm, are explicitly formulated, and the solving procedure is provided. Case studies on a simplified dataset representing the Electric Reliability Council of Texas demonstrate efficiency of customized Benders decomposition approach. This approach efficiently addresses the large-scale GEP problem over a 20-year planning horizon, entailing 820,200 general integer variables, 1,189,141 continuous variables, and 2,595,441 constraints, with a solution time of under 30 minutes.