MULTI-CRITERIA DECISION-MAKING MODEL ON CLOSE-OPEN MIXED VEHICLE ROUTING PROBLEM FOR MIDDLE-MILE DELIVERY OPTIMISATION

Abstract

Middle-Mile Delivery (MMD) is perceived to carry less importance in supply chain management and has the least potential for improvement, prompting many companies to skip this step in logistics. Although MMD is studied as part of the Vehicle Routing Problem (VRP), it receives less attention than last-mile delivery. Nevertheless, MMD is typically more predictable, presenting greater opportunities for enhancements. An optimised middle-mile distribution network can reduce transportation costs and delivery times. The main challenges of MMD include route distance, the locations of distribution centres, and delivery duration. Routing is an essential element of logistics, significantly contributing to economic growth. Inefficient routing may result in elevated expenses, especially for courier and logistics companies. Nodes, also known as distribution centres, are critical components in the distribution system. The initiation or termination of these nodes is frequently restricted by corporate constraints, rendering such modifications challenging. Consequently, optimisation initiatives must prioritise the selection of nodes according to their relevance to the company's comprehensive delivery process. This study presents a hybrid approach for the Close-Open Mixed Vehicle Routing Problem (COMVRP), which addresses both open and closed routes while integrating Multi-Criteria Decision Making (MCDM). The objective is to reduce the overall delivery distance. We propose a refined Genetic Algorithm (GA) that integrates with the Analytical Hierarchical Process (AHP) and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). In managerial decision-making, the AHP-TOPSIS method is used to improve an initial COMVRP generated by the nearest neighbour algorithm. The AHP technique focuses on criteria weights, whereas TOPSIS emphasises delivery centres’ performance as the priority node. This solution set achieves ideal GA performance, displaying minimal route distances with external vehicle deployment. The calculation results also showed that the proposed model reduced the total route distance by 4.86%, which exceeded the standard COMVRP model with 28.03% less than the current postal delivery system.