Multi-objective algorithms to multi-mode resource-constrained projects under mode change disruption

Abstract

The present paper studies the Multi-mode Resource-Constrained Project Scheduling Problem (MRCPSP) wherein tasks have different execution modes, each implying a duration and different resource requirements. During execution, the project may encounter some problems that may cause schedule deterioration or even unfeasibility. Its utmost goals are firstly the proposition of a set of new disruption measures that can be used in new objective functions, and then the resolution of a bi-objective problem minimizing the project makespan and a disruption measure. The latter is solved by an Evolutionary algorithm and a new reactive multi-objective heuristic. Considering that the project is in progress, the paper focuses mainly on problem feasibility, Pareto Front performance, and algorithm CPU time. Given that the feasibility problem of static MRCPSP is NP-Complete (Kolisch & Drexl, 1997), we transform the bi-objective MRCPSP problem to a multi-objective problem by including a penalty to cope with the potential violation of non-renewable resource constraints, and enlarge the search space. We show that schedule adjustment is necessary for the problem solving process. A diverse set of non-time-consuming operators is also included in the algorithm.