Applying heuristics and stochastic optimization for load-responsive charging in a smart grid architecture

There are many approaches to integrate electric vehicles into smart grid architectures by optimizing their charging-and feeding periods. Most approaches, however, were never applied in reality since every field-test requires an existing and expensive infrastructure. In this paper we compare the performance of four different scheduling algorithms. As opposed to other works we only present algorithms that were applied in reality. The test platform used for this experiment is the smart grid at the Europäisches Energieforum (EUREF). We show that stochastic optimization can quickly and significantly improve upon heuristic algorithms alone and also that heuristics as a starting point for optimization lead to significantly faster convergence of the stochastic algorithm applied.