Predictive Control for Dynamic Scheduling and Maintenance in Open-Shop Manufacturing Systems

This paper addresses an open-shop scheduling problem within a manufacturing system modeled using Mixed Integer Linear Programming, which accounts for the progressive wear of machinery after job processing. The open-shop problem is transformed into a flow-shop problem, involving the enumeration of all possible job paths and the selection of paths that minimize the makespan. The research provides a comparison among an offline algorithm, unable to perform machinery maintenance and disregarding increases in processing times; an offline algorithm subject to machinery maintenance constraints after processing times exceed a certain threshold; and a dynamic algorithm utilizing Model Predictive Control concepts to reroute jobs at each relevant event, including unexpected job arrivals, as well as the initiation and completion of machine maintenance. The dynamic algorithm also considers the potential unavailability of machinery due to maintenance and deterioration. Its effectiveness is assessed through a simulated case study of a flexible manufacturing system, revealing a notable improvement in makespan – approximately a 30% reduction compared to the offline scenario without maintenance. This showcases its potential for real-world applications in the manufacturing industry.