An Acceleration Method Based on Customized Preprocessing for Capacity Expansion Planning in Power Systems

High renewable energy penetration contributes to an increase in seasonal electricity imbalance. Capacity expansion planning (CEP) is essential in alleviating such uncertainty and variability in the power system. However, the network structure and the coupling constraints of different time steps increase the computation complexity of CEP, even for its linear programming formulation. Moreover, the problem could be ill-conditioned, preventing it from being solved efficiently. This paper proposes a customized preprocessing method to handle the numerical difficulties that arise in practical computation. We demonstrate through multiple case studies, from the European case to the Asian case and from one-year CEP to multi-year CEP, that the proposed approach outperforms the direct application of a commercial solver on solving the CEP, significantly reducing the computational time up to 83.3%. Moreover, due to the special structure of the coefficient matrix of the linear program, we also develop a decoupling method based on ADMM and provide the preliminary experiment results.