Abstract:
A supply chain is a set of facilities connected together in order to provide products to customers. The supply chain is subject to random failures caused by different factors which cause the unavailability of some sites. Given the current economic context, the management of these unavailabilities is becoming a strategic choice to ensure the desired reliability and availability levels of the different supply chain facilities. In this work, we treat two problems related to the field of supply chain, namely the design and unavailabilities management of logistics facilities. Specifically, we consider a stochastic distribution network with consideration of suppliers' selection, distribution centres location (DCs) decisions and DCs’ unavailabilities management. Two resolution approaches are proposed. The first approach called non-integrated consists on define the optimal supply chain structure using an optimization approach based on genetic algorithms (GA), then to simulate the supply chain performance with the presence of DCs failures. The second approach called integrated approach is to consider the design of the supply chain problem and unavailabilities management of DCs in the same model. Note that, we replace each unavailable DC by performing a reallocation using GA in the two approaches. The obtained results of the two approaches are detailed and compared showing their effectiveness.
Machine summary:
"The second approach (integrated approach) consists of defining the optimal supply chain structure and simulate its performance in the presence of DCs failures at the same time in order to minimize the total generated cost.
Used variables and notations We use the following variables and notations for the mathematical formulation of the considered problem: The used notations are: /: Set of retailers indexed by i; /: Set of suppliers indexed by k; /: DC located at retailer j; /: Global demand of retailer i; /: Daily demand of/; /: Global demand variance of retailer i; /: Fixed annual cost of locating/; /: Per-unit shipment cost from / to retailer i; /: Inventory holding cost per unit per year in/; /: Fixed order cost (including transport fixed cost) placed by / at the supplier k; /: Per-unit shipment cost (purchase and transport costs) from supplier k to/; /: Mean lead-time in days from supplier k to/; /: Variance of lead-time from supplier k to/; /: Desired service level at DCs; /: Standard normal variate such that/; / : Total cost of / unavailability; / : Number of / unavailability during step/; / : Mean cost of /unavailability; / : Global generated cost in the step/; / : Unavailability management cost of the integrated approach; : Unavailability management cost of the non-integrated approach.
These results represent the supply chain structure (located DCs and selected suppliers) and the unavailability management cost for the two proposed approaches ( / and /) obtained for five different scenarios simulation.
Firstly, we used a non-integrated approach which consists of using a genetic algorithms based optimization to solve a stochastic distribution network problem design where the strategic decisions of suppliers selection, DCs location and retailers allocation are integrated in the same model."