Developing robust distribution networks for future urban planning scenarios

Abstract

Supply chain planning of consumer goods distribution is a constant process with expanding cities and increasing consumer demand. This project is based in the municipality of Ekurhuleni, Gaut- eng, South Africa. The aim of the project is to develop a model that will use data of the predicted population distribution of Ekurhuleni in 2030 to develop a robust supply chain for consumer goods. Three possible development and population layout scenarios that the future of the municipal- ity could embody were investigated. A model was developed to locate a distribution centre (DC) and develop a distribution network, that will be compatible with all three of these scenarios. A literature review was conducted to determine the best practices in the elds of facility location modelling, distribution network development, robust networks and optimisation models. An al- gorithm was developed based on the best practices and the data of the UrbanSim model to solve the problem. The municipality was divided into 1058 zones and based on the algorithm the best zone to locate the DC is zone 538. This zone placement passes with a logical test, since the total cost is lower if the DC is located near the center of the municipality rather than on the outskirts of the municipality. Further veri cation and validation was done on the model. Sensitivity analysis on the model was done by changing certain parameters such as the number of trucks, the capacity of the trucks, and the operating cost per truck. The operating cost per truck has the most in uence on the robustness of the distribution network. A maximum total saving of R 63 279 could made by using this approach to place a DC and develop a distribution network for consumer goods. Thus if the impact on a small scale problem is already this signi cant, the impact on a full scale distribution operation could be tremendous. Future work should consider determining the demand per zone more accurately and in categories since it will impact the given solution.