Reducing computing time of energy system models by a myopic approach

In this paper, the performance of the existing energy system model PERSEUS-NET is improved in terms of computing time. Therefore, the possibility of switching from a perfect foresight to a myopic approach has been implemented. PERSEUS-NET is a linear optimization model generating scenarios of the future German electricity generation system until 2030, whilst considering exogenous regional characteristics such as electricity demand and existing power plants as well as electricity transmission network restrictions. Up to now, the model has been based on a perfect foresight approach, optimizing all variables over the whole time frame in a single run, thus determining the global optimum. However, this approach results in long computing times due to the high complexity of the problem. The new myopic approach splits the optimization into multiple, individually smaller, optimization problems each representing a 5 year period. The change within the generation system in each period is determined by optimizing the subproblem, whilst taking into account only the restrictions of that particular period. It was found that the optimization over the whole time frame with the myopic approach takes less than one tenth of the computing time of the perfect foresight approach. Therefore, we analyse in this paper the advantages and draw-backs of a change in the foresight as a way of reducing the complexity of energy system models. For PERSEUS-NET it is found that the myopic approach with stable input parameters is as suitable as the perfect foresight approach to generate consistent scenarios, with the advantage of significantly less computing time.