Supply Network Design by Using Clustering and Mixed Integer Programming
Nicolás Clavijo Buriticá, John Wilmer Escobar, Rafael Gutiérrez
The Supply Network Design represents one of the high impact strategic decisions in competitiveness for companies. An optimal location of facilities in relation to the capacity of supply and demand, allows a high level service to attend the markets. In this paper, a methodological framework for designing supply networks by joining application of clustering techniques and mathematical programming are presented. The proposed methodology has been tested with real data obtained from a company of non-alcoholic beverages in Colombia. The approach considers three main stages. First, the costumers clustering process is performed by K-means in order to obtain the location for potential Distribution Centers (DC’s). In the second stage, the model for supply network design is performed using a Mixed Integer Programming (MIP) by considering different options to assign DC’s, and finally the valuation of the proposed methodology on a real case. A distribution scheme, which allows enter to new market areas with an efficient strategy to penetrate products to big cities such as Bogotá in Colombia, was found.