Multi-project scheduling: A benchmark analysis of metaheuristic algorithms on various optimisation criteria and due dates

Abstract

This paper reviews a set of ten existing metaheuristic solution procedures for the resource-constrained multi-project scheduling problem. Algorithmic implementations are constructed based on the description of the original procedures in literature. Equivalence is verified on the original test instances for the original objective and parameters through a comparison with the reported results. An extensive benchmark analysis is performed on a novel, publicly available dataset for a variety of optimisation criteria and due date settings for which the original algorithms have not been tested earlier. The impact of the different objectives, due dates and test instance parameters is analysed and an overall ranking of the metaheuristic solution methods for different situations is discussed. Key insights into the structure of competitive solutions for disparate objectives and due date settings are presented and effective algorithmic components are revealed.