Renewable energy and modernization of power operation demand Independent System Operators (ISOs) to solve ever more complex and larger programming problems to securely and economically schedule power resources. A key step in the scheduling process is the unit commitment (UC). In a hydro-dominated system, this process also involves managing reservoirs and is called hydrothermal UC (HTUC). Due to its size, non-convex nature and strict solution-time requirements, HTUC remains a challenging problem to ISOs even in its deterministic form. Relying on a commercial solver to directly tackle this problem might require a careful configuration of the solver's parameters, and, given the generality of commercial solvers, it might not fully exploit all characteristics of the HTUC. As an alternative, researchers have recently proposed dual-decomposition strategies for addressing the UC, which enables solving subproblems in parallel. However, dualizing a non-convex problem generally results in a duality gap and requires adhoc techniques for recovering a primal solution. In this paper, we propose a novel cooperative multi-search Benders decomposition approach to solve the deterministic HTUC. Our approach benefits from being a primal method and, at the same time, it exploits the structure of the HTUC to solve subproblems in parallel. To assess the effectiveness of our proposal, we use the state-of-the-art solver Gurobi as a benchmark, and we perform our experiments on 25 cases of a large-scale system with over 1,000 generating units and 7,000 buses. On average, our approach is more than 15 times faster than Gurobi and it is able to solve the HTUC problems in under 20 min to a 0.1 gap.
Report 1. 02/2021.