Design and Development of Reliable Capacitated Facility Location for Distribution Network Design

Abstract

The most of distribution centers allocation is often implicitly considered unaffected from any disruptions. However, the distribution centers may fail due to natural or economic factors. The reliability is an important factor to averse disruptions in distribution centers. Design of distribution centers with consideration of reliability is beneficial for long-term investment. The aim of this dissertation is to propose a model for reliable distribution centers (DCs) in case of unexpected disruption. The site-dependent failure probabilities and random link disruptions have also utilized in the proposed model. The proposed model has considered three costs such as construction cost, transportation cost and improvising cost. The mixed-integer linear programming model (MILP) has been formulated to provide reliable DCs in case of random failures. Three solution approach lagrangian relaxation, cross decomposition and firefly have been utilized to implement the proposed model. These approaches have been tested on randomly generated datasets for 200 customers and 30 DCs. The experimental results reveal that cross decomposition approach is applied to obtain better results in less time. However, cross decomposition approach performance is not always better than firefly algorithm due to the time required to obtain the cross decomposition bound.