A Multi - product capacitated vehicle routing problem model with seasonality for the stationery industry

Abstract

This dissertation presents an optimisation approach to deliver stationery commodities with highly seasonal demand. The problem is structured as a capacitated vehicle routing problem (CVRP) consisting of one distribution centre to numerous customer locations (one-to-many) with multi-products, where each customer requires various product mixes. A mixed-integer linear programming model was used to formulate the capacitated vehicle routing problem, and the Gurobi solver with customised heuristics algorithms was used to arrive at the solution. Intending to gain the advantage of separating the distributing points according to geographical areas, K means clustering was engaged. The solution generated through this model determined that annual distance has diversified by 44% between the peak and off-peak periods. The main findings show that the annual distance savings is around 28%, while the annual capacity saving is 22% when using the CVRP model compared to current practices. Further, it is determined that it is adequate to have 15 and 53 trips per week starting and ending at the distribution centre for off-season and season, respectively. Moreover, the route sequence of every vehicle was illustrated cluster-wise and season-wise separately. Two experiments were done, changing vehicle capacity and time horizon to have better outcomes, which will be additional guidance for the organisation. The dissertation offers a guide to improving the use of optimising techniques in the distribution network aiming at seasonal demand variations while providing a sound basis for future research directions.