Real-time Allocation Decisions in Multi-echelon Inventory Control

Inventory control is a crucial activity for many companies. Given the recent advances in information technology, there have never been greater opportunities for coordinated inventory control across supply chain facilities. But how do we design efficient control methods and policies that take advantage of the detailed information that is now becoming available? This doctoral thesis investigates these issues within the field of inventory control theory. The objective of the research is: To develop mathematical models and policies for efficient control and increased understanding of stochastic multi-echelon inventory systems, with a focus on allocation decisions and the use of real-time information. This thesis is based on five scientific papers which are preceded by a summarizing introduction. The papers address different types of inventory distribution systems, all consisting of a central stocking facility that supplies an arbitrary number of local stocking facilities (referred to as retailers). The retailers face stochastic end customer demand. The systems are characterized by the presence of real-time inventory information, including continuously updated information on the current inventory levels at different facilities and on the locations of outstanding orders. In Paper I and Paper II we derive and evaluate different decision rules for stock allocation (known as allocation policies) for a central warehouse which applies a time based shipment consolidation strategy. The allocation policy determines how the central warehouse should distribute its stock among different retailers in case of shortages. New allocation policies that utilize real-time information are compared to the commonly used First Come First Served policy which requires less information. In Paper III we shift focus to the delivery policy at a central warehouse which supplies multiple retailers that order in batches. When the central warehouse cannot satisfy an entire retailer order immediately, the delivery policy determines if the order should be shipped in several parts or in its entirety when all items are available. We investigate the value of using a new delivery policy that uses real-time information on when replenishments will arrive at the central warehouse. The information is used to determine the best course of action for each order placed by the retailers. We also study how to allocate safety stocks to all facilities in the system given this new policy. In Paper IV and Paper V we consider a system where retailers may receive emergency shipments from a support warehouse in combination with regular replenishments from a central warehouse/outside supplier. We investigate how safety stocks should be allocated between the retailers and the support warehouse. Furthermore, we evaluate the benefits of tracking orders in real time and using this information in the decision whether or not to request an emergency shipment.