Evaluation of uncertainty on scheduling algorithms in printed wiring board manufacturing

Unrelated parallel machine scheduling problems have been well studied for several single-objective cases under deterministic processing conditions. While these problems are appropriate for some manufacturing environments, there are situations where multiple objectives need to be considered. This dissertation analyzes problems based on a Japanese printed wiring board manufacturer's drilling operation which is subject to processing time variability and random equipment breakdowns (usually in the form of jams). The manufacturer is interested in five key performance objectives: makespan, number late, total overtime, average finishing time of machines, and utilization. This research presents and evaluates optimization, hybrid optimization/dispatching, and dispatching algorithms designed to address this scheduling problem. In addition, these algorithms are modified to include hedges to buffer against the uncertainty present in the manufacturing process. Periodic and event-driven predictive-reactive rescheduling techniques are also considered and their performance is compared to hedged and non-hedged schedules. The algorithms are evaluated using discrete-event simulation with symmetric and skewed distributions to determine how each is impacted by processing time variability, and to determine which algorithms perform best under various levels of uncertainty.