Evaluation of an Inventory Policy in a Divergent Multi-Echelon System with Upstream Demand

Abstract: The purpose of this master thesis is to evaluate the performance of an inventory policy in a one warehouse, multiple retailer inventory system with end customer demand at all stock locations. The objective is to compare the performance of a multi-echelon model with stock rationing, compared to real inventory data which is based on uncoordinated single-echelon optimization. The comparison is carried out by simulations, where the focus is put on expected service levels and expected inventory levels. The multi-echelon model with service constraints in Berling and Marklund (2013) is used with the inclusion of a virtual retailer which only serves the end customer demand at the central warehouse (upstream demand). The virtual retailer approach is used to approximate a critical level policy at the central warehouse. This means that when the stock on hand at the warehouse falls to or below the critical level, only customer orders are satisfied while retailer orders are backordered. The results show that the multi-echelon model greatly outperform the uncoordinated solution in terms of the ability to reach target service levels. This is particularly evident when customer order sizes are large. Furthermore, the virtual retailer approach is shown to overestimate the critical level which leads to excess stock. However, the multi-echelon model still holds on average 10% less inventory at the central warehouse when both models achieve the target service level. Finally, the sensistivity analysis illustrates that a critical level policy has the potential to reduce the total inventory with up to 25% but the potential reductions diminish as the fraction of upstream demand increases.