Evaluating a queueing approximation for the machine interference problem with two types of stoppages via simulation optimization

An optimization framework for the evaluation of queueing approximations is proposed and illustrated on a queueing approximation for the two-type stoppage machine interference model. The optimization framework combines infinitesimal perturbation analysis and stochastic approximation in order to optimize the approximation’s error function using simulation. The optimization procedure attempts to find the worse case error of the approximation and the associated parameter values. The optimization framework is compared against the standard factorial design approach for queueing approximation evaluation. The results indicate the potential of evaluating queueing approximations using an optimization framework. In addition, the performance of the machine interference approximation is shown to be worse than previous results indicated using factorial design approaches. This research illustrates to practitioners how queueing approximation evaluation is performed and should indicate to researchers new directions to take in queueing approximation evaluation methodology.