Simulation Optimisation: Approaches, Examples, and Experiences

Simulation based optimisation or simulation optimisation is an important field in stochastic optimisation. The present report introduces into that problem area. We distinguish between the non-recursive and recursive approaches of simulation optimisation. For the non-recursive approach we consider three methods, the retrospective, SPO-, and the RS-methods. With the help of a simple inventory problem we discuss the advantages and disadvantages of these methods. As a recursive method we consider in the second part of our report the coupling of simulation with Genetic Algorithms. As an application example we take a complex multilocation inventory model with lateral transshipments. From our experiences with such optimisation problems we finally formulate some principles, which may be relevant in simulation optimisation.