In today's global market, managing the entire supply chain efficiently becomes a crucial factor for a successful business. Performance efficiency of a supply chain network depends on how the inventories are managed across the entire network. Inventory management in a supply chain network is a complex problem due to the nature of interdependencies among different nodes of the network, and can rarely be solved using closed-form mathematical solutions. These problems can be broadly classified in to two categories: single-echelon and multi-echelon. In single-echelon inventory control problems, the focus is on determining the appropriate level of inventory for an individual unit within the supply chain network. On the contrary, multi-echelon inventory optimization takes a holistic approach by focusing on the correct levels of inventory across the entire supply chain network. The goal of this research is to use stochastic modeling approach to develop a scalable multi-tier supply chain model that can accommodate multiple inventory items, and to experiment with the model to study and compare its behavior under single-echelon vs. multi-echelon inventory systems. A genetic algorithm based multi-objective optimization method is used to optimize model's behavior with two conflicting objectives: minimizing average inventory across the end to end supply chain and maximizing overall fill rate or service level. The results show that the solutions generated using multi-echelon optimization can be quite different than the solutions generated using single-echelon optimization. Under single echelon settings, network behaves as a decentralized system and as a result, entire supply chain network suffer with higher inventory levels and lower fills rates. In contrast, multi echelon network behaves as a centralized system and provides lower inventory levels while maintaining higher fill rates for the entire supply chain network. This makes sense since the former takes a far-sighted systems level view of the problem as against the short-sighted individual unit level approach taken by the latter. However, distribution centers failed to provide optimal values when performing under multi echelon configuration. In the best interest of the system as a whole, distribution centers have to compromise on their individual performance.
展开▼