Pro-Reactive Approach for Project Scheduling Under Unpredictable Disruptions

Existing solution approaches for handling disruptions in project scheduling use either proactive or reactive methods. However, both techniques suffer from some drawbacks that affect the performance of the optimization process in obtaining good quality schedules. Therefore, in this article, we develop an auto-configured multioperator evolutionary approach, with a novel pro-reactive scheme for handling disruptions in multimode resource-constrained project scheduling problems (MM-RCPSPs). In this article, our primary objective is to minimize the makespan of a project. However, we also have secondary objectives, such as maximizing the free resources (FRs) and minimizing the deviation of activity finishing time. As the existence of FR may lead to a suboptimal solution, we propose a new operator for the evolutionary approach and two new heuristics to enhance the algorithm’s performance. The proposed methodology is tested and analyzed by solving a set of benchmark problems, with its results showing its superiority with respect to state-of-the-art algorithms in terms of the quality of the solutions obtained.

I. Introduction

The objective of a standard resource-constrained project scheduling problem (RCPSP) is to find the best sequence of activities, by satisfying all the resource limitations and precedence constraints while minimizing project completion time. In such problems, single-mode resources are considered and are recognized as single-mode RCPSPs (SM-RCPSPs). Multimode RCPSPs (MM-RCPSPs) are an extension of SM-RCPSPs, in which each activity has a number of nonpreemptive execution modes, each of which may be different in terms of resource requirements and duration. Thus, in addition to all the specifications of the SM-RCPSPs, an MM-RCPSP aims to find the best schedule of activities and their best execution modes so that completion time is minimized [1]. However, some of these resources may be unavailable at the time of execution due to an unexpected breakdown, which is known as a resource disruption. Resource disruptions are a critical issue in real-world projects [2]. MM-RCPSP is a well-known NP-hard problem [3], [4]. One of the difficulties with the existing approaches for MM-RCPSPs is that they do not perform consistently over a wide range of problems. When considering MM-RCPSPs with unknown disruptions, they are much harder to solve and there is a lack of effective approaches for this type of problem. In reference to the literature, the approaches are either proactive or reactive and both of them have drawbacks [5], [6]. The assumptions made in those approaches are usually resulted in suboptimal solutions. We believe that an appropriate design and linking of these two approaches will allow us to reduce the effect of those assumptions. The main motivation of this research is that MM-RCPSPs with disruptions is a difficult practical problem, and there is a research gap in the development of its solution approach.