Rescheduling affected operations - a purely predictive approach

Abstract

Rescheduling is used to minimize the impact on the system performance when disruptions are present. Affected operations rescheduling (AOR), is often mentioned in the literature on rescheduling. This method generates an initial deterministic job shop schedule offline, which is updated online in response to machine breakdowns. Only operations directly or indirectly affected by a disrupt is rescheduled. In this paper, we formulate a purely offline AOR approach for job shops. Based on a time-optimal schedule, the proposed method generates sequence-based constraints resulting in the same system performance. The realized schedule is identical for both approaches. Also, the proposed approach applies to any disrupt that might cause delays in the system. In right-shift rescheduling (RSR), all remaining operations are postponed if disrupts occur. A formal proof is presented to show that AOR will always perform better than or equal to RSR in the face of disruptions. For analyzing the effect of disrupts, an analytical measure of the makespan is introduced, which depends on possible delays in the system.