From peak power prices to seasonal storage: Long-term operational optimization of energy systems by time-series decomposition

Long-term operation of energy systems is a complex optimization task. Often, such long-term operational optimizations are solved by direct decomposing the problem into smaller subproblems. However, direct decomposition is not possible for problems with time-coupling constraints and variables. Such time-coupling is common in energy systems, e.g., due to peak power prices and (seasonal) energy storage. To efficiently solve coupled long-term operational optimization problems, we propose a time-series decomposition method. The proposed method calculates lower and upper bounds to obtain a feasible solution of the original problem with known quality. We compute lower bounds by the Branch-and-Cut algorithm. For the upper bound, we decompose complicating constraints and variables into smaller subproblems. The solution of these subproblems are recombined to obtain a feasible solution for the long-term operational optimization. To tighten the upper bound, we iteratively decrease the number of subproblems. In a case study for an industrial energy system, we show that the proposed time-series decomposition method converges fast, outperforming a commercial state-of-the-art solver