Perturbation analysis and optimization of a flow controlled manufacturing system

Abstract

We use Stochastic Fluid Models (SFM) to capture the operation of threshold-based flow control policies in manufacturing systems without resorting to detailed discrete event models. By applying Infinitesimal Perturbation Analysis (IPA) to a SFM of a workcenter we derive gradient estimators of throughput and buffer overflow metrics with respect to flow control parameters. It is shown that these gradient estimators are unbiased and independent of distributional information of supply and service processes involved. Moreover, they can be implemented using actual system data, which enables us to develop simple iterative schemes for adjusting thresholds (hedging points) on line so as to optimize an objective function that trades off throughput and buffer overflow costs.