Agent-based Simulation Environment for Improving the Planning of Distribution Grids

The recent changes and developments in the electric power system impose new challenges to the distribution system operators not only in the operation, but also in the planning of the grids. The volatile feed-in of distributed generation based on renewable energy sources as well as new and intelligent loads require an appropriate consideration in the distribution grid planning process. With the conventional planning method being dependent on extreme scenarios, the consideration is very limited. Therefore, a new simulation system based on the concept of a multi-agent system is developed and presented in this thesis, permitting not only the consideration of the volatile feed-in characteristics of renewable energy sources but also of the dependencies between the grid users and their environment. Every grid user is modelled as an agent of its own, guaranteeing the preservation of its individual character. The results of the simulation, time series for all relevant system variables, define the new input parameters in the distribution grid planning process. The probabilities of occurrence of loading situations can be derived from the time series. This allows for the first time for a detailed determination of the conditions in the up to now rarely measured medium and low voltage grids. As a consequence, new assumptions for the planning process are derivable, permitting a demand- and futureoriented grid planning and avoiding overdimensioning of the grids.