Routing-based reactive scheduling policies for machine failures in job shops

A scheduling and control system can be viewed as a vital component of modern manufacturing systems that determines companies’ overall performance in their respective supply chains. This paper studies reactive scheduling policies developed against unexpected machine failures. These reactive policies are based on rerouting the jobs to their alternative machines when their primary machine fails. Depending on the subset of the jobs considered for rerouting, the long-term performance of four policies are tested under various conditions. Expecting that these rerouting policies would bring an extra load for a material-handling system (MHS), a dynamic job shop environment was studied with and without a MHS. It is shown that the proper selection of a good reactive policy is based not only on the system characteristics such as utilization, machine down times and frequency of machine failures, but also on the MHS capacity (in terms of speed and number of MH devices). The extensive experiments show that when the MHS is not a bottleneck and/or the down times are long enough to compensate the cost of extra rerouting, rerouting all aOEected jobs to their alternative machines proves to be the best policy. However, when the MHS cannot handle the extra load due to rerouting or the down times are relatively short, then rerouting only the jobs that will arrive to the failed machine during repair performs the best.