A tabu search procedure for the resource-constrained project scheduling problem with alternative subgraphs

Abstract

This paper investigates the resource-constrained project scheduling problem with alternative subgraphs (RCPSP-AS). In this scheduling problem, there exist alternative ways to execute subsets of activities that belong to work packages. One alternative execution mode must be selected for each work package and, subsequently, the selected activities in the project structure should be scheduled. Therefore, the RCPSP-AS consists of two subproblems: a selection and a scheduling subproblem. A key feature of this research is the categorisation of different types of alternative subgraphs in a comprehensive classification matrix based on the dependencies that exist between the alternatives in the project structure. As the existing problem-specific datasets do not support this framework, we propose a new dataset of problem instances using a well-known project network generator. Furthermore, we develop a tabu search that uses information from the proposed classification matrix to guide the search process towards high-quality solutions. We verify the overall performance of the metaheuristic and different improvement strategies using the developed dataset. Moreover, we show the impact of different problem parameters on the solution quality and we analyse the impact of distinct resource characteristics of alternatives on the selection process.