Title: The capacitated vehicle routing problem revisited: using fuzzy c-means clustering

Authors: Henrique Ewbank; Peter Wanke; Henrique L. Correa; Otávio Figueiredo

Addresses: COPPEAD Graduate School of Business, Federal University of Rio de Janeiro, Rua Pascoal Lemme, 355 – Cidade Universitária, Rio de Janeiro, RJ, Brazil ' COPPEAD Graduate School of Business, Federal University of Rio de Janeiro, Rua Pascoal Lemme, 355 – Cidade Universitária, Rio de Janeiro, RJ, Brazil ' Crummer Graduate School of Business, Rollins College, 1000 Holt Ave., Winter Park, FL 32789, USA ' COPPEAD Graduate School of Business, Federal University of Rio de Janeiro, Rua Pascoal Lemme, 355 – Cidade Universitária, Rio de Janeiro, RJ, Brazil

Abstract: This paper proposes to simplify complex distribution scenarios and find near-optimal solutions by applying a heuristic approach for solving the capacitated vehicle routing problem with a homogeneous fleet using fuzzy c-means as the clustering technique. A memetic algorithm determines the number of clusters and an improved fuzzy c-means algorithm allocates customers to routes. When benchmarked with other methods and compared with 50 known instances from the literature, it indicated an error average of less than 3%. Due to the nature of the errors studied, a tobit regression has been applied to predict the average percent error in terms of the characteristics of the demand and the distance of each customer. Results also suggest that kurtosis and skewness of the distances among all customers, capacity of the vehicles and standard deviation of the demand could be used to predict the average percent error.

Keywords: capacitated vehicle routing problem; CVRP; cluster-first route-second heuristic; fuzzy logic; homogeneous fleet; tobit regression.

DOI: 10.1504/IJLSM.2019.103513

International Journal of Logistics Systems and Management, 2019 Vol.34 No.4, pp.411 - 430

Received: 30 Sep 2017
Accepted: 23 Apr 2018
Published online: 08 Nov 2019 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article