Metaheuristics for the stochastic hoist scheduling problem

Experts look for solutions to increase production delivery, but industrial systems are generally disturbed by random events. In many cases, there is a lack of information about random events consequences on the production time and quality of the manufactured products. Surface treatment lines are robotic cells with time window constraints and so random events can have important consequences on production management. Products could be damaged by constraint violations and rejected by the control quality department. Moreover, owing to the structure of these systems, variations in transportation times can induce unrecoverable situations (deadlock occurrences, for example). Therefore, experts look for schedules to prevent random events consequences. One of the most difficult operational problems in surface treatment lines is the proper coordination of the material-handling resources management and job input scheduling. Previous related research works concern the makespan minimization problem and do not take into account the random events consequences: this problem is denoted the Hoist Scheduling Problem (HSP). This paper focuses on the Stochastic Hoist Scheduling Problem (SHSP), which is the HSP with random events implying variations in transportation times. It appears that no previous work can be reported concerning such stochastic problems but only deterministic ones. A method is proposed here based on stochastic metaheuristics used to determine the production schedules for which the random events consequences are low. An industrial study that highlights the efficiency of the method is proposed. A company that cannot be named for confidential reasons has provided the test data.