Starting with a carefully defined vocabulary of commonly used terms, the authors proceed to outline the right methods, resources, time commitments, and realistic expectations for V&V. It is made clear that making design decisions using simulation results without V&V represents an area of risk. A section on Risks and Trade-Offs describes why validation is needed for simulation-driven design, whether simulation-based or simulation-informed, and provides guidelines on situations where it might be reasonable to neglect the V&V process. Coupled with V&V is the notion of uncertainty quantification (UQ), a routine concept in well run test laboratories, where a process is developed to quantify the band of error associated with each aspect of a test, or simulation in this case. UQ strengthens the simulation by providing realistic bounds around the simulation result. Section 4 asks a series of important questions which allow the manager to lay out the appropriate resources for the needs of the enterprise. The uncertainties of the simulation tool itself are discussed next, along with steps that must be taken by the software developer to verify the simulation. The validation, usually performed by the simulation user, is described in a simple flow-chart. The validation step is differentiated from the Application Domain, in which a validated simulation process is now used to forecast what might occur under conditions for which there may be no direct experimental corroboration data, such as a new product or use case.
By William L. Oberkampf & Martin Pilch. NAFEMS, 2017. 32 pp.