ROBUSTNESS MEASURES AND ROBUST SCHEDULING FOR JOB SHOPS

A robust schedule is defined as a schedule that is insensitive to unfo reseen shop floor disturbances given an assumed control policy. In thi s paper, a definition of schedule robustness is developed which compri ses two components: post-disturbance make-span and post-disturbance ma kespan variability. We have developed robustness measures and robust s cheduling methods for the case where a ''right-shift'' control policy is used. On occurrence of a disruption, the right-shift policy maintai ns the scheduling sequence while delaying the unfinished jobs as much as necessary to accommodate the disruption. An exact measure of schedu le robustness is derived for the case in which only a single disruptio n occurs within the planning horizon. A surrogate measure is developed for the more complex case in which multiple disruptions may occur. Th is surrogate measure is then embedded in a genetic algorithm to genera te robust schedules for job-shops. Experimental results show that robu st schedules significantly outperform schedules based on makespan alon e.