Verification and validation in industrial simulation

Simulation, as a method of analyzing and predicting the performance of complex systems, is commendably Protean, able to increase understanding of and contribute to the improvement of systems in a broad spectrum of contexts. These contexts include, but are certainly not limited to, assessing the structural adequacy of mechanical systems (mechanical or civil engineering), predicting the behavior of circulatory systems (meteorology), evaluating military strategies (defense), increasing the efficiency of business processes (business process reengineering), configuring computer and communications networks (computer systems design), and enhancing the reliability, productivity, and economy of operation of manufacturing systems (industrial engineering). Users of simulation in any of these contexts share the challenges of verification and validation. Any simulation model must be verifiable as operating in the way intended by the modelers, valid as a representation of the actual physical system under design or study, and credible as a recipient of trust and confidence from the managers who look to the simulation for guidance in making decisions involving uncertainty and economic risk. In this paper, we describe the verification and validation techniques typically used within simulation studies of industrial processes. Such studies may, for example, improve the allocation of buffer storage within an assembly line, determine an appropriate number of pallets to be used within a recirculating pallet loop, or assist in determining routing policy for automatic guided vehicles. Via case studies, we illustrate the proper role of verification and validation, not as incidental increments to a simulation project, but as basic requisites thoroughly integrated into such a project. In conclusion, we then outline promising areas for future advances in development and usage of verification and validation techniques.