Linear Programming Formulation of Idle Times for Single-Server Discrete-Event Simulation Models

摘要

Mathematical programming representations (MPRs) are discrete-event simulation models represented using math programming. In this paper, we first introduce an MPR formulation for single-server queueing systems based on idle times. We use this formulation to conduct a perturbation analysis to study the effect of imposing a constraint on idle times. We apply the formulation to obtain a linear programming-based gradient estimator for idle times. We demonstrate the integration of optimization into MRP and identify properties of the optimal solution to facilitate finding the optimal solution.