Details

Multi-Worker Multi-Location Assignment (MWMLA): MILP Modelling, Solution Validity, and Performance Evaluation

Year: 2025

Term: Fall

Student Name: Jinfeng Li

Supervisor: Michiel Smid

Abstract: This research project focuses on the Multiple Worker Multiple Location Assignment (MWMLA) problem, a practical resource scheduling problem in which multiple workers depart from different points to serve customers with varying needs in different geographical locations. This research develops an extended mixed-integer linear programming (MILP) model suitable for multi-worker open path systems, based on an improved MDOVRP model proposed by Lalla-Ruiz et al. In this study, the MWMLA model is implemented using PuLP and CBC solvers, and a post-processing program is designed to reconstruct coherent worker paths from the solver output. Feasibility tests using coordinate-based small-scale instances verify that all model constraints—path continuity, demand satisfaction, single-departure behaviour, and sub-tour elimination—are satisfied. Performance experiments using stochastic cost matrices and Euclidean coordinates evaluate the model's scalability. Overall, this study establishes a mathematical foundation for the study of relevant practical scheduling problems and reveals the inherent scalability challenges in structured and multi-demand path problems.