Optimized operation of long-term storages considering a scalable modelling depth

The control and structural expansion of decentralized energy systems are very challenging due to the volatility of renewable energies and progressive structural changes. For balancing out seasonal fluctuations, conversions into heat or gas in combination with long-term storages are frequently discussed approaches. In context of an optimal conceptual synthesis of such systems, investigations regarding the operation and design require a large time period of at least one year. In order to solve such optimal control problems, an immense calculation time is required. This contribution presents a multistep approach which determines the optimal operation strategy in an iterative way and is capable of reducing the calculation effort. In the first step, a rough optimization incorporating a low modelling depth is performed. Especially in combination with a rough time discretization, dynamic short-term storages (e.g. electrical batteries) can become irrelevant from an optimization point of view. Therefore, the considered system can be virtually reduced by several state and control variables resulting in a significantly reduced computation time. In a second optimization, the optimal control problem is constrained using the results of the previous step. Especially the obtained values for the state of charge of the long-term storage improve significantly the quality of the second optimization. While in the first step, the dynamic programming is utilized to solve the optimal control problem in one instance, the second step uses the mixed integer linear programming to solve multiple short time periods of the optimal control problem in a sequential way. Results are presented on the basis of a simple test scenario where the electrical energy supply of a residential quarter is investigated using real photovoltaic data of one year, a modelled fuel cell system as long-term storage and an electrical battery storage as short-term storage