Preemptive resource leveling in projects

Abstract

As a well-known NP-hard problem in project scheduling, the resource leveling problem (RLP) has attracted many researchers’ attentions. In the RLP, a typical assumption is that activities are non-preemptive during project execution, which means that activities cannot be interrupted once they have been started. However, preemption is not uncommon in project management practice and existing studies already show that it is beneficial to consider preemption when leveling resource usage. Therefore, we investigate the preemptive resource leveling problem and design a genetic estimation of distribution algorithm (GEDA). To analyze the performance of the GEDA, we conduct extensively computational experiments on 2160 randomly generated instances. We also examine the impacts of various factors on the GEDA. Comparative experimental results show that the GEDA outperforms the existing meta-heuristic algorithm.