A Genetic Algorithm for the Flexible Resource-Constraine Multi-Project Scheduling Problem

José Manuel Leitão da Silva Santos

Key information

Authors:

José Manuel Leitão da Silva Santos (José Manuel Leitão da Silva Santos)

Supervisors:

Rainer Kolisch; Tânia Rodrigues Pereira Ramos (Tânia Rodrigues Pereira Ramos)

Published in

December 6, 2022

Abstract

The production planning problem for an engineer-to-order company is addressed. This problem was modeled as a flexible resource-constrained multi-project scheduling problem (FRCMPSP). The FRCMPSP is an optimization problem in the field of project management. It consists of scheduling activities of variable duration to minimize the weighted-earliness-tardiness of several projects. Real-world-based problem instances are solved to find feasible schedules and resource profiles that simultaneously fulfill the constraints of all projects in an aggregated environment. In this work, a hybrid metaheuristic approach for the FRCMPSP is developed and evaluated. It contains a problem-adapted schedule generation scheme, which is embedded into a genetic algorithm (GA). Variations within the GA were analyzed, namely through priority rules and adaptation of operators. The performance of the proposed methods is demonstrated in a computational study with benchmark methods from the literature. It achieved feasible solutions for most of the benchmark test instances (62.5%) considering a diverse subset. The impact of flexibility of resource and due date parameters was evaluated, and resource availability had the largest influence on the objective function.