Energy system models are limited in their scope and level of disaggregation by the availability offast computing hardware. While improvements in hardware and solver developments have led toan increasing size of solvable models, problems with high temporal and geographical resolutionremain difficult to solve in one loop. In this paper, we evaluate the use of rolling planning as aspeed-up method for energy system models. In a stylized model, we highlight potential issuesthat occur at the boundary of optimization horizons, especially regarding time-linking constraintssuch as energy storage balances. In multiple configurations of the energy system model WILMAR-JMM, we investigate the tradeoff between solution quality and problem size / solution time thatcharacterize the use of rolling planning.
Energy system model rolling-horizon optimization linear programming