REACTIVE SCHEDULING OF PROJECTS WITH UNRELIABLE RESOURCES

摘要

The Resource Constrained Project Scheduling Problem (RCPSP) is a well-known complex scheduling problem. In real world situations, the parameters of projects are vulnerable to uncertainty, change or disruption, which necessitates that the initial baseline schedule must be revised. In this work, we have proposed two different mathematical models for both the Discrete Time-Based Reactive Approach (DTRA) and the Event-Based Reactive Approach (EBRA), in an attempt to reduce the number of variables required in representing RCPSPs. The EBRA and DTRA are, therefore, to simultaneously determine the recovery start time, the resource profile, and the duration of each activity in the recovery schedule, in order to minimize the makespan for a single disruption, as well as a series of disruptions, without having any advance activity interruption information. To test the proposed approach, a set of thirty-activity test instances from the Project Scheduling Library (PSPLIB) and one real-life scheduling problem are solved with randomly generated disruption events. Computational experiments were also conducted to analyse the effects of different factors that relate to the disruption recovery process. The experimental study reveals that the re-optimization process can reduce the revised makespan, as compared to simple right shifting of affected activities, and the level of improvement depends on the duration of the affected activities, their resource requirements, their relationships with forwarding activities, the disruption duration and the amount of disrupted resources.