A hybrid crow search-based method for solving the closed loop supply chain network design

This thesis develops a hybrid optimization framework for large-scale facility location and Closed-Loop Supply Chain Network Design (CLSCND). The main contribution is a decomposition-based method that combines a Binary Crow Search Algorithm (BinCSA) with Mixed-Integer Linear Programming (MILP) exact refinement: BinCSA explores facility-opening decisions, a fast allocation procedure generates feasible assignments, and MILP is invoked only when needed to refine promising solutions. This design preserves scalability while retaining the reliability of exact optimization. The framework is validated on the Single-Source Capacitated Facility Location Problem (SSCFLP), a single-objective CLSCND model, and a bi-objective CLSCND model minimizing total cost and net CO2 emissions. For SSCFLP, the method reproduces proven optima on small, medium, and large OR-Library instances and maintains small optimality gaps on extreme-large instances with practical runtimes. For single-objective CLSCND, it achieves solution quality comparable to that of CPLEX while substantially reducing computation time, with average runtime dropping from 2,416 s to 158 s on two-period instances and from 3,118 s to 647 s on five-period instances. A further contribution is the extension of the framework to bi-objective optimization by integrating ε-constraint search with BinCSA-based seeding and exact MILP refinement. The proposed method achieves Pareto-set quality similar to that of a pure ε-constraint baseline, reduces runtime by about six times, and produces nearly twice as many refined non-dominated solutions, all MILP-certified. Overall, the study shows that selective exact refinement embedded in metaheuristic search is an effective and scalable strategy for solving large CLSCND problems.<p></p>

Author dc:creator

• Wangyue Xu (21039587)

• Facility Location Problem
• Closed-Loop Supply Chain Network Design
• Binary Crow Search Algorithm
• Multi-Objective Optimization

Statement dc:rights

• All rights reserved
• Open Access after 2027-11-05

Identifier

10779/exe.32125312.v1

OAI identifier oai:identifier

oai:figshare.com:article/32125312