Optimal control and design of complex systems by simulation and genetic algorithms

In operations research numerous approaches and algorithms exist to solve design and control problems for systems of such different areas like inventories, logistics, transportation, manufacturing etc. Nevertheless, the complexity of real-world systems prevents the application of almost all classical approaches. One method to overcome these difficulties is simulation optimisation where a simulator for the considered system is combined with an appropriate optimisation tool. In our presentation we suggest to combine simulation with the genetic optimisation tool LEO. We briefly discuss the application of that software tool to find optimal order policies for multi-location inventory models and to design an optimal Kanban controlled manufacturing system. Finally, we report on some experiences and further developments.