Stabilized Benders decomposition for energy planning under climate uncertainty

DOI10.1016/J.EJOR.2024.01.016arXiv2208.07078MaRDI QIDQ6507958FDOQ6507958

Authors: Leonard Göke, F. Schmidt, Mario Kendziorski 

Abstract: This paper applies Benders decomposition to two-stage stochastic problems for energy planning under climate uncertainty, a key problem for the design of renewable energy systems. To improve performance, we adapt various refinements for Benders decomposition to the problem's characteristics, a simple continuous master-problem, and few but large sub-problems. The primary focus is on stabilization methods, specifically comparing established bundle methods to a quadratic trust-region approach for continuous problems. In a quantitative case-study stabilization accelerates Benders decomposition by a factor of four to ten. Among the investigated stabilization methods, the trust-region approach demonstrates the best performance, followed by the level bundle method, which is on average 33% slower. The use of inexact cuts achieved an average reduction of computation time by 28%. Thanks to parallel computing our refined algorithm outperforms the deterministic equivalent if planning covers more than four weather years. Finally, we propose additional enhancements to the algorithm. First, heuristic methods to narrow down the solution space of the master-problem. Second, approximations of the sub-problems to determine inexact yet valid cuts faster.

This page was built for publication: Stabilized Benders decomposition for energy planning under climate uncertainty

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6507958)