|M.Sc Student||Shiry Varem|
|Subject||A Robust Approach to the Integrated Inventory Replenishment,|
Lateral Transshipments, and Routing in a
Single-Commodity Supply Chain
|Department||Department of Industrial Engineering and Management||Supervisors||Dr. Naseraldin Hussein|
|Professor Emeritus Ben-Tal Aharon|
|Full Thesis text|
Inventory control is an essential part in operations management. We consider a system comprised of retailers facing uncertain demand. In each period, replenishments are made, endogenously, and order-up-to quantities are set. Lateral transshipments are performed after demand is realized but not materialized, in order to minimize predicted excess and shortage of inventory costs at the retailers. However, since transshipments have a non-negligible cost, an important decision on routing costs arises. Lateral transshipments literature provides recommendations on what quantities to deliver directly between retailers. Those recommendations usually cannot be performed “as is” and managers must plan the route by themselves, which can be a complex task. Finding the optimal route to perform the transshipments is different from finding an optimal route for supplying goods (Traveling Salesman Problem) or from pick-up and delivery problems.
Another important aspect at hand is the need to deal with uncertainty in the demand. While typically stochastic optimization is the approach, here we address the case where no distribution is given or can be deduced, which is the case when there aren’t any or enough or reliable or on-time past demand data or it cannot be attributed to a known distribution.
In our research, we studied the integrated problem of inventory replenishment, lateral transshipments, and routing in a setting in which the demand is uncertain. We propose a solution that simultaneously minimizes the system’s total predicted cost including inventory (holding and shortage) and transportation. To cope with uncertainty in demand, in such a rather complex setting, we employ the Robust Optimization methodology which guarantees feasibility for every possible occurrence of the demands, with a user defined uncertainty set.
To validate our approach advocated in this study, we performed a numerical study in which we compared our results to benchmarked models. Our model performed better in all the instances examined, with average reduction in cost of 16%.
We also performed sensitivity analysis and explored how sensitive the objective function is to change in several parameters. It was found that the system’s cost is weakly sensitive to most of the cost parameters, yet, is highly sensitive to changes in order-up-to levels.
This research thesis presents a tool and a methodology to help lower inventory levels, reduce transportation costs, and enables managers to implement the transshipment plan using optimal route.
Practically speaking, given the right advanced information systems, the proposed method can be implemented in retailing systems, such as chains of electrical devices, pharmacies, mobile phones, etc. It can also be implemented for balancing levels of raw material, tools, and spare parts and in a multi-project setting.