Study On Inventory-Location For Distribution Network Design

Distribution refers to the approaches which take the products from the supplier stage to the customer stage in the supply chain. In Supply Chain Management (SCM), Distribution Network Design (DND) has always been a key strategic issue of an enterprise with extensive research. The distribution business of the products is operated and achieved by the distribution network, therefore the performance of the enterprise DND has a direct impact on the total logistic cost and service level of the enterprise.In generally, DND includes three different decision levels: strategic decisions (facilities location), tactical decisions (inventory policy) and operational decisions (transportation). The DND problem is typically solved following a sequential approach, in which tactical/operational decisions are simplified or omitted, and once the DND is solved, then these omitted decisions may be incorporated. A wide family of Facility Location Problems (FLP) solutions is used to solve the DND, where parameters, demands, constraints and the objective function are considered to be deterministic. However, the FLP as strategic entities are not capable of dealing with the interactions between facility location and other levels' decisions, such as inventory control decisions. Thus, in order to meet customer demands with lowest total cost, DND should be from the perspective of overall situation of the enterprises, to integrate one or more of inventory control, facilities location and other decisions into a system to optimize, basing on the idea of systematization and integration. And for designing optimal distribution network, DND should also weight synthetically transportation, production, inventory, facilities location and other relating costs to minimize the total system cost. Therefore, my research and application of facility location problem integrated inventory management and/or other decisions are presented and discussed in this thesis based on the theory of this field.Firstly, the thesis analyses background motivation, introduces the reasons of choosing this topic, and proposes the research objection and research route. Then the thesis introduces relevant concepts and character of the distribution network, and discusses the factors which would impact DND and the components of the cost of distribution network. Moreover, the thesis reviews the origination, evolution and theories of traditional FLP. Based on these, the thesis summarizes further the research results of FLP applied in DND in the last ten year, and the literatures involving Inventory-Location Model (ILM), which is the core issue in the thesis.Secondly, the thesis introduces the basic Single-stage ILM, and analyses the properties of this model which are different from the traditional FLP. Then, the thesis develops a joint inventory-location model, which considers the market selection decisions. This model is formulated as a nonlinear integer programming model that is NP-hard. Therefore, a heuristic solution approach is presented based on Lagrangian Relaxation and Greed Algorithm. Moreover, numerical examples are given to verify the efficiency of this algorithm.Thirdly, in order to design the distribution network to cope with continually changing market environment in a long strategy horizon, the thesis addresses two multistage inventory-location models, one of which considers the capital budget, and the other is not. In these models, explicitly handles parameter uncertainty by allowing parameters to be described by discrete scenarios, each with a specified probability of occurrence. For solving these models, a heuristic solution approach is also presented based on Lagrangian Relaxation and Greed Algorithm. Moreover, numerical examples are given to verify the efficiency of this algorithm.Fourthly, the thesis studies an inventory-location model under inflation. The growing inflation rates at home and abroad have the tremendous impact on the world economy and the human's daily lives, thus for subsistence and development of the enterprises, it is critical to design optimal distribution network which should tackle high inflation. Three genetic algorithms are presented for solving our model, based on different coding forms, and the effectiveness of these algorithms is analyzed through different sizes of instances. The results of the analysis show that, for minimizing the present value of the total cost under the growing inflation rate, the optimal policy would need to increase the number of the built distribution centers.Fifthly, the thesis studies the contents and characters of the distribution network for deteriorating items. Then from the perspective of inventory models, the thesis develops an inventory-location model for deteriorating items which assumes the deteriorating rate is a known constant. Two algorithms are presented: Lagrangian Relaxation Algorithm and Genetic Algorithm. Moreover, the thesis develops a more realistic but more complex product-inventory-location model, and designs an adaption of Genetic Algorithm to solve it. Furthermore, numerical examples are given to illustrate the application of the algorithm, and show that GA is an efficient solution approach for this problem.Finally, the thesis gives conclusion to the contents and the innovative achievements of the research, and presents the future scope, the purpose and the prospect of this topic in further studies.