Optimal design of experiments with simulation models of nearly saturated queues

The paper develops an approach to the optimal design of experiments with simulation models of nearly saturated queues, suggested in Cheng and Kleijnen (Oper. Res. v.47, 1999). This approach is based on the introduction of a regression model of the input-output behaviour of the underlying simulation model. In this way the problem is reduced to a special regression experimental design problem. An analytical solution is found under assumptions for the regression function and the behaviour of the output variance. In paricular D-optimal designs are considered and the solution given for this case