Methods for minimizing the combined cost of inventory and maintenance in a stochastic one-machine environment.

摘要

Today's manufacturing environment is heavily dependent on machinery. From automotive to chemical to semiconductor manufacturing, if the equipment is not working, production is halted. Nevertheless, maintenance activities to increase machine availability should not be performed at the expense of production cycle times. Instead, research in the area of the combined scheduling of maintenance and production activities should be performed to reduce total system costs.;The objective of this dissertation is to present methods for the minimization of the expected total inventory carrying cost (ICC) and maintenance cost (M) per unit of process time or ICC + M for a one-machine system. In this work, point estimates of job processing times for different product types and the time to perform preventive maintenance (PM) are used in conjunction with a non-linear cost function that characterizes the cost of a stochastic failure process of a machine. The failure rate of the failure process is assumed to increase with the time since the last maintenance; however, the number of failures can be reduced if PMs are performed at regular intervals. The time that jobs arrive at the tool (job ready times) are initially assumed to be known in advance; however, in the later portion of this research, adjustments are made to address the decrease in effectiveness of the initial heuristic scheduling methods as this assumption is relaxed. The problem decision variables are the job schedules and PM schedules.;In the course of this research, three models and solution techniques are provided to schedule jobs and PM in a dynamic job release environment in the presence of stochastic failures and uncertain job ready times to minimize total system costs. The models are referred to as: (1) the Optimal Maintenance and Production Scheduling (OMPS) branch and bound, (2) the Maintenance and Production Scheduling (MPS) heuristic, and (3) the Window Analysis for Maintenance and Production Scheduling (WAMPS) heuristic.;In this research, the MPS and WAMPS heuristics are shown to reduce the average total cost (ICC + M) when compared to the optimal PM cycle policy from the stochastic model of Co and Araar (1991). This comparison is made over a number of combinations of values for the environmental variables to establish the robustness of the heuristics. Reduction is shown to be up to 25% for some environments. OMPS, on the other hand, uses a branch and bound formulation that allows MPS to be compared to the optimal problem solution, assuming that contiguous maintenance cycles are independent.