Modeling and Analysis of a Multi-Echelon Inventory System Using Cost and Service Level KPIs : A Case Study of Centralization and Sourcing Policies for MTS and MTO Items

Sammanfattning

In supply chain management, which is a well-researched subject, a company's warehouse networks can be complex. Distribution networks with more than one level, called echelons, can be referred to as multi-echelon systems. This means that products are stored in at least one location before being distributed to the end customer, in contrast to a single-echelon system where products are shipped directly to the end customer. These multi-echelon networks have often historically emerged with the aim of increasing economic growth, which in turn has led to their being optimized with regard to cost, lead time and service level. A classic tradeoff in multi-echelon theory is to balance between the lower cost and fewer warehouses that centralization provides, and a more costly scenario but with shorter lead time and faster response. The complexity lies in the fact that decisions about centralization and supply flow cannot be analyzed in isolation, since decisions in one part of the network affect cost and service level in other parts of the system. This becomes particularly relevant in global networks that have multiple markets with different customer requirements, and where both make-to-stock (MTS) and make-to-order (MTO) products are handled within the same structure. The purpose of this study is therefore to analyze how a global multi-echelon warehouse network should be designed with respect to centralization and sourcing policies for MTS and MTO products, in order to minimize costs while maintaining required service levels. The study focuses on how local warehouses, regional distribution centers, and a global distribution center should be used in relation to each other. The study was conducted as a quantitative case study of a real global distribution network and was based on historical order and transportation data from the company's network. First, descriptive statistics and linear regression were used to investigate how costs, lead times, and service levels are distributed in the current configuration. Then, an optimization model was developed to identify improvements in network structure under given service requirements. The results show that the network is heterogeneous and that a uniform centralization or sourcing policy is not suitable for the entire system. Not even for the respective product class, MTS and MTO. Cost-savings can be achieved through an increased degree of centralization compared with the current level while maintaining service levels, but not completely centralized. The optimized solution suggests that the number of regional sales company warehouses should be reduced from 16 to 12, with the flow primarily being moved up to the regional distribution centers. The results also show that MTS and MTO products can still be handled within the same hybrid structure, but taking into account their different product characteristics. The main conclusion is therefore that the degree of centralization and sourcing policies should be analyzed jointly, because together they determine how cost, lead time and service level are affected in a global multi-echelon warehouse network. The study contributes to the extent that the degree of centralization and sourcing policy can be improved in a real multi-echelon network, by considering service differences, product logic, and regional differences. Future studies may need to take a clearer grasp of the difference between MTS and MTO products, as well as analyze the complexity of reconfiguring an existing distribution network.