Research On Inventory Routing Problem With Stochastic Demand

In current market environment, companies often get involved in competition through participating in supply chain. In the whole process of supply chain management, logistics management has received extensive attention from both enterprise and academia. In traditional logistics management, Retailer Managed Inventory (RMI) is the main inventory management strategy. However, under RMI, the demand information is invariably enlarged with supply level increasing, i.e., Bull Whip Effect, which causes inventory of each level in supply chain and thus cost increase. To overcome this problem, scholars come up with another inventory management strategy, Vendor Managed Inventory (VMI), which manages inventory systematically and integratedly, and thus overcome Bull Whip Effect effectively. For suppliers, how to make efficient decisions in the process of distribution and inventory management simultaneously is crucial to implementing VMI. Inventory Routing Problem (IRP) becomes the hot topic in research under the pressure of optimizing distribution and inventory coordinately.IRP is a kind of typical NP-hard problem and it becomes more challenging for decision-makers when demand is uncertain. The current research often assumes that demand is deterministic. However, the real demands faced by companies are stochastic, and thus research on stochastic demand inventory problem (SDIRP) is pressing. According to different distribution method, SDIRP falls into four categories, i.e., SDIRP with direct deliveries (SDIRPDD), SDIRP without vehicle constraint (i.e., the distribution center serve all the customers with one vehicle), SDIRP based on fixed partition policy (FPP), and SDIRP with general structure. The main contents are as follows:Chapter1first analyzes the research background and meaning of IRP, and gives the general mathematical expression of IRP. The difference between vehicle routing problem (VRP) and IRP is also analyzed. Further, IRP is categorized according to its inherent characteristics and the related literature is reviewed according to demand feature, distribution method, and solving algorithm. The drawbacks of current research on IRP are concluded, and the technique course and the main research contents are also given in this chapter.Chapter2studies SDIRPPDD, proved that the optimal stable policy of SDIRPDD with and without vehicle constraints is (s,S) form, and gives the corresponding algorithm. Then, compared with a kind of often-used strategy through a specific numerical example, the algorithm is confirmed efficient. Finally, the number of vehicles in logistics system with direct delivery is analyzed using the above algorithm.Chapter3studies SDIRP when customers’demands are relatively small compared with the capacity of vehicle. In some specific condition, the structure of optimal inventory policy is proven that it is (s’,S’) form similar to SDIRPDD’s and the lower and upper bound for the parameters in the optimal strategy are given. Further, two heuristics are proposed to solve this problem. Finally, the systematic average operating costs under the above two algorithms are compared with those under fixed delivery route policy. And the applicability of fixed delivery route policy is also discussed.In Chapter4, based on fixed partition policy, SDIRPs with periodical and continuous replenishment policy are considered respectively. The corresponding customers partitioning algorithms are designed and the corresponding optimal inventory policy form is proven that is (T,S) and modified (s,S) respectively. Further, the corresponding heuristics are designed and the efficiency of different partitioning algorithms is discussed through a numerical example. Based on above, a kind of SDIRP under partition policy using dynamic routing policy is researched according to relevant conclusions in chapter2.In Chapter5, SDIRP with general structure is studied. A decomposing algorithm based on (s,S) policy and modified C-W is designed in this chapter, which decomposes SDIRP into SDIRPDD and VRP. For SDIRPDD, with applying the conclusions in Chapter2, the stable inventory policy is derived. For VRP, the modified C-W saving algorithm is proposed. Finally, based on this algorithm, this chapter proposes an improved algorithm to solve SDIRP with time windows and the efficiency of the above two algorithm are confirmed through numerical example.The concluding part summarizes the whole thesis and gives the future research direction.