Decision making in advanced manufacturing

Abstract

In this dissertation, we demonstrate how operations research (OR) methods can be effectively implemented in advanced manufacturing environments in order to im- prove decision-making processes. Most manufacturing processes inevitably involve difficult decisions which can be formulated as optimization problems. Applying an- alytical methods to these problems and selecting an efficient solution approach can improve significantly manufacturing performance. In this dissertation, we illustrate how OR methods can contribute to advanced manufacturing through two real-world applications.

Next, we consider the application of OR methods to improve spare parts in- ventory management for complex equipments, when real-time sensor data provides partial information about the status of each machine. While maintenance and spare parts inventory decisions have generally been considered separately, our inventory model incorporates the condition of the machines into the inventory decision. For many types of machinery, it is common to have a well-defined maintenance sched- ule. However, some machines may degrade at an unexpectedly rapid rate and thus require maintenance prior to the planned maintenance event. This degradation may be captured by real time sensor data. We develop a periodic review inventory model which incorporates demands that may arise from both scheduled maintenance and unexpected degradation.

We begin by considering a wafer fab scheduling problem. In a wafer manufac- turing, queue-time constraints occur when a set of consecutive process steps must be completed within a fixed time window. Violations of queue-time constraints can lead to rework, scrap and longer cycle times. Given a job arriving at a queue-time zone, a key decision to be made is whether and when the job should be allowed to enter the zone. To assist in making this decision, we propose two alternative mixed integer linear programming models of the local scheduling problem within the queue-time zone.