Two-Echelon Vehicle Routing Problem With Mobile Satellites For Perishable Goods: A Casestudy Of Aba Cooltrans

Abstract

It is widely accepted that the Two-echelon vehicle routing problem (2E-VRP) addresses the intricacies of distribution networks by incorporating larger vehicles for the first echelon and smaller vehicles for the second echelon. Goods supplied from the distribution centers to serve customers are transferred at intermediate points between two echelons called “satellites”. Recently there has been a spate of interest since it helps reduce overall operating costs, improve last-mile delivery efficiency, and maximize resource usage. This study dedicates the concept of synchronizing mobile satellites that consider the first echelon vehicles at the customer location to be satellites and allow the second echelon to not return to the departure satellites. The problem reflects real-world logistics challenges for the perishable goods of a company in Vietnam. A mixed-integer linear programming model was used to find exact solutions with the objective of minimizing transportation costs and the variants of capacity, time windows, and synchronizing mobile satellites. A Hybrid Genetic Algorithm and Tabu Search Algorithm were proposed to approach a larger-scale problem. This study investigated the use of benchmark instances from the University of Vienna and real data from the company in Vietnam using the combination of trucks and motorbikes. The computational experiment was presented to validate the model and figure out that Hybrid Genetic Algorithm have better performance in terms of time and the optimal solution. From that, the new approachs is implemented to generate the optimal routes that assit the studied company in costs- and time-saving. Taking the concept of synchronizing mobile satellites in 2E-VRP into account for the first time is the solution for many complex logistics distribution networks, especially for perishable goods. Although this study has generated a big contribution to the literature review, there remains a need for an efficient approach that can solve more constraints that reflect real-world challenges.