A Two-Stage Heuristic Procedure for Solving the Long-Term Unit Commitment Problem with Pumped Storages and Its Application to the German Electricity Market

Abstract

In electricity systems unit commitment problems (UCP) target at a proper scheduling and coordinating of thermal plants, renewable energies, and storages. The need for fast solution methods has been growing in line with recent changes in the electricity system’s environment and complexity, in particular with the increasing share of volatile renewable feed-ins. In order to meet this need even for large-scale systems a decomposition methodology for the UCP is suggested within this paper. Our two-stage decomposition first performs an isolated dispatching of thermal plants using a greedy algorithm, rule-based algorithms and local search based steps, followed by a re-optimization stage in order to incorporate energy storages into the final solution. The comparison of the iterative two-stage heuristic with commonly used approaches based on mixed integer linear programming shows outstanding results in terms of solution time and solution quality. Besides typically used test instances, the heuristic is applied to comprehensive case studies of the German electricity market, where (near-) optimal solutions can be derived for a yearly planning horizon with hourly time steps with computational effort of a few minutes using a standard PC.