A Two-Phase Decomposition Approach for Electric Utility Capacity Expansion Planning Including Nondispatchable Technologies

This paper presents a model and a two-phase algorithm for incorporating nondispatchable technologies or energy sources as decision variables in long-range electric utility capacity expansion plans. The first phase uses a deterministic model to quickly obtain a near-optimal expansion plan. This model employs a step approximation to the time-dependent availabilities of the nondispatchable technologies, and a derating technique to represent forced outages of conventional units. Specialized algorithms that exploit the structure of this deterministic problem are developed. The second phase refines the resulting solution toward the probabilistic optimum. It employs accurate probabilistic production costing techniques to account for forced outages of conventional generators, and uses hour-by-hour simulation to represent the contributions of nondispatchable technologies. This second phase requires few additional iterations in our computational experience. The proposed approach suggests a modification for the existing state-of-the-art methodology electric generation expansion analysis system (EGEAS), particularly for situations in which the cumulant approximation used in EGEAS may be inadequate, and it suggests an improved technique for accelerating Bloom's Generalized Benders' Decomposition algorithm for the conventional equipment capacity expansion planning problem.