Distribution planning using capacitated clustering and vehicle routing problem

Purpose - The purpose of this paper is to reduce the distribution cost of an Indian cooperative dairy. The reduction of cost was achieved with the application of the clustering method (k-means clustering) and capacitated vehicle routing problem (cheapest link algorithm (CLA)). Design/methodology/approach - Capacitated k-means clustering was used to split delivery locations into similar size groups (i.e. clusters) based on proximity without exceeding a specified total cluster capacity. Each cluster would be served by a local stockist. CLA was then used to find delivery routes from dairy (i.e. depot) to stockist in each cluster and from stockist to all other delivery locations within the cluster. Findings - K-means clustering and CLA suggested optimal delivery routes on which vehicles will run. The complete algorithm was able to provide a solution within 30 s. Practical implications - Clustering of delivery locations and use of heterogeneous fleet of delivery vehicles can result in considerable savings in daily operational cost. Originality/value - Most of the research related to the use of demand clustering to improve distribution routes has been theoretical, which do not take into account real-world limitations like vehicle’s specific limitations. The authors tried to address that gap by taking a real-world case of a cooperative dairy and compared the result with existing distribution routes used by dairy. This work can be used by other dairies or distribution companies according to their scenario.