A supply chain analytics approach for optimizing milk collection routing in multi-depot networks

This study presents a supply chain model for optimizing milk collection routing in multi-depot networks. The problem consists of a fleet of vehicles that leaves their depots (i.e., typically the driver's houses), visits an assigned set of farms to collect the raw milk, and delivers it to the processing plant. This problem has not yet been formulated explicitly in the literature, and it can be classified in the middle between the Team Orienteering Problem (TOP) and the Multi-Depot Vehicle Routing Problem (MDVRP) with heterogeneous vehicles. However, it cannot be reduced to any previously mentioned problems before introducing slight modifications and additional constraints to the mathematical formulation. We introduce a new formulation and propose six heuristic algorithms to minimize the distance covered in milk collection in the dairy sector. The proposed solutions are validated by using new benchmarks and tested in a set of real case applications. Computational experiments on real-life data are performed to investigate the performance of the heuristics varying the milk demand. The results demonstrate the applicability of the proposed approach to the real world and identify the best algorithm in terms of solution quality and computational time.