مهسا پشنگ

Objective : This research aims to propose a large-scale vehicle routing model for the distribution network of a food industry product and apply the model in a real-world case study. Methods : A mathematical model is formulated to minimize the total variable transportation costs.  Considering the complexity of the model, a constrained clustering algorithm is used to decompose the problem. Then, vehicles are assigned to demand clusters according to their capacity. Finally, each cluster's symmetric traveling salesman problem (TSP) is solved using a genetic algorithm. The parameters of the proposed genetic algorithm were calibrated based on its widespread application in solving symmetric TSPs. A conservative approach was adopted to ensure the solution's validity by evaluating a worst-case scenario considering the highest node demands.  Results : By applying the proposed algorithm to the case study, over 2,000 demand nodes across Tehran were grouped into 91 clusters. Then, based on the demand level of each cluster, the vehicles are assigned, consisting of 26 small and 65 large cars. Within each cluster, the assigned vehicle followed an optimized route among the nodes, designed based on the optimal tour generated by solving the cluster-specific TSP using the genetic algorithm, and then returned to the central warehouse. Conclusion : Comparing the results with the current situation, the size of the proposed transportation fleet showed a 40% reduction. Additionally, reducing fleet size and optimizing the routes improved the total distribution network costs by 25%. Given the model's computational efficiency, this improvement is considered satisfactory.