Integrated vehicle scheduling and routing policies for cross-dock systems

• Main Author: Agustina, Dwi
• Other Authors: Lee Ka Man
• Format: Theses and Dissertations
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering::Industrial engineering::Operations research; DRNTU::Engineering::Industrial engineering::Supply chain; DRNTU::Engineering::Systems engineering
• Online Access: https://hdl.handle.net/10356/68868
• Institution: Nanyang Technological University

Cross-docking has become an increasingly popular logistics strategy in which products from several suppliers are consolidated into a single shipment to enable FTL, rather than LTL, shipments. The cross-dock warehouse ships the received goods directly to the customers, with little or no storage in between. The advantages of implementing cross-docking include shortened delivery time, improved customer satisfaction, and increased cost savings. Cross-docking depends heavily on the use of information technology and sophisticated planning software that coordinate the physical product flows and information flows. However, unavailability of effective IT systems is one of the biggest barriers faced by companies implementing cross-docking. In addition, contemporary research only considers vehicle scheduling, product consolidation and routing issues separately, leading to sub-optimal solutions. This research takes a step towards integrating operations planning issues at a cross-dock, and develops vehicle scheduling and routing policies at a cross dock warehouse using an integrated model. There are four key contributions of this research. (1) First is a comprehensive literature review of cross-dock operations planning issues. Extensive literature review in cross-docking is done and research gap identified. The main research gap identified and addressed in this research is the integration of scheduling and vehicle routing policies at cross docks. This research focuses on delivery vehicle scheduling and routing problem at a cross-dock facility by combining these two problems, and considering the customer delivery time window and the product consolidation decision. (2) Second contribution of this research is a model for vehicle routing and scheduling problems at a cross-dock, namely VRSP, and modelled as a MILP. The model consists of suppliers, inbound trucks, cross-dock facility, outbound trucks, and customers, and tries to determine the vehicle routing and scheduling policies jointly. The objective of VRSP is to determine the schedule of inbound and outbound vehicles at the cross-dock and the routing of outbound vehicles to deliver products within customer specified time windows. VRSP aims to minimize the total cost, which consists of the earliness penalty cost, the tardiness penalty cost, the inventory holding cost, and the transportation cost of deliveries. The performance of VRSP is experimented using CPLEX Solver version 12.3 running on an Intel® Celeron® CPU N2840 2.16 GHz computer. The experiments show that CPLEX Solver can solve the VRSP in a reasonable and acceptable time only for small scale problems. The CPLEX Solver takes a long time to solve medium size problems, and is unable to solve large-scale, real-life size problems. (3) The third contribution of this research is a modified model incorporating customer zones and hard-time windows to eliminate late and early delivery costs and substantially reduce the solution space. Through experiments using the same software and computer platform as VRSP, it is shown that the modified model (VRSP-CZHTW) can be solved for medium to large scale, real-life problems within an acceptable time. Only when the number of customers reaches five hundred or higher does the Solver reach its time limit without a solution. (4) The fourth and final contribution of this research is a meta-heuristics (TS algorithm) based approach to solve the original VRSP. Further experiments are done to validate and test the TSA-VRSP on an Intel® Celeron® CPU N2840 2.16 GHz computer. The algorithm is coded using MATLAB version 7.9.0. The experiments show that TSA-VRSP can be solved even for large scale problems in an acceptable time. The experiments also show that TSA-VRSP has better performance than the two previously developed methods (VRSP and VRSP-CZHTW, using CPLEX Solver). Thus, TSA-VRSP has the potential to be the foundation of a decision support tool for cross-dock managers, and help develop effective and integrated vehicle routing and scheduling policies for day-to-day operations.