Policy‐comparing Simulation Experiments: Design and Analysis

One of the principal uses of simulation in business and industry is to compare alternative policies or decision rules for the operation and control of complex systems. The immediate objective of such simulation studies is usually quite simple: to discover the best of the several policies under consideration. The measure of effectiveness used is also usually straightforward, often dollars of cost or profit. The basic experimental design that will produce the best comparison of alternatives is generally agreed upon: all policies should be compared under, to the extent possible, identical experimental conditions, i.e., when randomness is involved, all alternatives in a simulation experiment should be simulated using a common sequence of randomly‐generated events.