Simulation-optimization: A simple approach combining metaheuristics and metamodels

The use of simulation-optimization has increased in recent years. This technique is useful when the objective function and/or any constraints could not be assessed analytically. Beginners in simulation analysis frequently use commercial simulation packages, which include optimizers, to make this analysis. However, many of them see those optimizers as a black-box that searches for an optimum. Simulation-optimization techniques may be time-consuming. In this paper, we propose a simple combination of a metaheuristic and metamodel to show how easy is building simple algorithms that are able to perform this analysis in a competitive time. We have proposed a simple methodology where a simulation model, a metamodel and a metaheuristic work together in a simple loop to obtain solutions for an inventory problem. We also compare the solutions against those obtained by a commercial package.