Response taxonomy in manufacturing systems rescheduling

Many types of systems occasionally fail to operate in the way they are designed. More a system is complex, less its activities can be defined in detail. This is particularly true in manufacturing systems, in the short term planning of activities. The subject is known as scheduling, dealing with the "correct" arrangement of operations over the time axis, for each product and resource. Modeling the scheduling problem, as a domain of Operations Research, shows several drawbacks. Under an optimality point of view, this is reached only in system with a few machines, under a makespan performance criterion. In facilities comprising a higher number of processing units, optimality must be set aside due to the excessive combinatory leading to time-consuming computation requirements and, guided by dispatching rules, accepting satisfactory solutions instead of optimal. For systems not providing flexibility, the only response is maintenance and recovery. One of the most common sources of disturbance is machine unavailability. The situation analyzed in this paper is rescheduling. It deals with the necessity, if a disturbance appears, to correct schedules once the production is running. The formulation of this problem reaches a higher difficulty level. It has theoretical and practical implications. Therefore, this makes it interesting to focus a broad research field, as noted from a literature review. In this article, the main alternatives in rescheduling are compared, in order to establish a taxonomy of responses. These include the methods to repair the damaged schedule, the methods to construct flexible schedules that resist disturbances, and the studies that reconsider the performance measures. Through the analysis of an example, the different approaches are illustrated.