Simulating Disruptions in Resource-Constrained Project Scheduling Problems with time-varying Resources

Abstract

This abstract presents a simulation-based framework to estimate the impact of disruptions, simulate %their effects, and resolve them in a resource-constrained project scheduling problem with time-varying resources (RCPSP/t). To assess the potential impact of disruptions, the framework employs various complexity measures prior to generating an initial schedule. As the RCPSP/t is a complex combinatorial problem, different heuristics are selected based on problem complexity to obtain a feasible solution. The initial schedule is then executed within a simulation environment, where disruptions are explicitly modeled and introduced at runtime. Disruptions may take the form of resource shortages or increases in activity durations. To reactively resolve disruptions, a novel transformer-based reinforcement learning approach is proposed and compared against a right-shift heuristic baseline and an optimal solution computed via a constraint programming–based approach. The results are analyzed with respect to the estimated disruption impact, enabling a comparison between predicted and realized schedule performance.