Assessment of Discrete Event Simulation Software for Enterprise-wide Stochastic Decision Problems

The increasing complexity of the enterprise has motivated the utilization of mathematical programming methodologies to deal with many of the decisions faced by companies. However, in many cases the combinatorial complexity of the decision problems is exacerbated by uncertainty in many of the model parameters, requiring the use of scenarios or stochastic programming strategies. Unfortunately, the size of practical problems often makes the use of stochastic programming and stochastic dynamic programming infeasible from a computational perspective. This situation motivated the development of a set of computational strategies known as simulation-optimization as an alternative way to help decision makers in obtaining robust and near-optimal solutions. However, the selection of a discrete event simulation (DES) software to implement such approaches involves a balancing act. The ideal product flexible enough to represent the multiple levels of decision making while allowing the use of any optimization methodology, is not yet available. In this work, we report on the implementation of a multi-stage screening process to evaluate and select the most suitable software for simulation-optimization. A set of criteria was developed and a list of 52 commercial packages was compiled. Nine firms were invited to participate in the final stage of the study, which consisted in analyzing the ability of their software to address a generic problem that included all the criteria. An individual analysis of the seven packages that accepted the invitation was conducted and a quantitative comparison is presented.