A two-stage heuristic method for the planning of medium voltage distribution networks with large-scale distributed generation

With the setting up of regulatory authorities the cost pressure on the electricity distribution companies will increase further in Germany. Due to political reasons the distributed generation (DG) has increased. Therefore, nowadays many distribution network companies face two important tasks: reduction of network costs to improve the network efficiency and integration of the large-scale DG into existing systems. A long-term planning of the distribution networks can provide an optimal target network for the efficiency analysis of distribution networks and the evaluation of effects, which are caused by the DG. Planning of distribution networks without the DG is known as a non-linear mixed integer problem. Additional technical constraints (such as voltage rising effect, short-circuit current, etc.) and stochastic characteristics of the DG make the planning problem become more complex. In this paper we propose a new efficient two-stage heuristic method, which is based on Large Neighborhood Search [1] and Guided Local Search [2], for the planning problem of distribution networks with the DG. Technical modeling and stochastic characteristics of the DG are described. Applications of the proposed method for the planning of real German distribution networks with large-scale DG show that distribution network companies can use this method, not only for planning the future optimal network structure but also for analyzing the effects, which are caused by different DG scenarios.