Large-Scale Project Portfolio Selection and Scheduling Problem: A Comparison of Exact Solvers and Metaheuristics

Integrated decision-making regarding the selection and scheduling of a project portfolio, referred to as the project portfolio selection and scheduling problem (PPSSP), can contribute to the more favorable performance of many organizations. However, the optimization of large-scale PPSSP involving a large number of projects and complex constraints remains challenging. While exact solvers offer guaranteed optimal solutions, they are computationally intractable in large-scale PPSSP. On the contrary, metaheuristics search for approximated solutions within a reasonable time. This paper investigates the performance of a commercial exact solver and metaheuristics in addressing large- scale PPSSP within a limited time. In addition, hybrid approaches are proposed to combine the exact solver and metaheuristics in different ways to address PPSSP. Experiments are conducted on PPSSP instances with an increasing number of projects (up to 6000), demonstrating the superiority of metaheuristics in addressing large-scale PPSSP within a short time and the effectiveness of combining metaheuristics in Gurobi in facilitating the optimization process.