Inventory management of spare parts : centralisation or decentralisation

Abstract

For many years, spare parts management has been widely discussed because of its various features. Challenges present in their inventory management are infrequent difficult to predict demand, the low demand of items and the large costs in case of a stock-out. Therefore, it appears that spare parts involve different management methods compared to more commonly studied methods of inventory control. Moreover, operating in a maintenance environment, these items can be very critical and require a fast response time to avoid long downtime of technical systems. This is why companies usually set up a multi-level distribution network, the aim being to have the defective product available as quickly as possible.

The purpose of this thesis is to explore the spare parts environment and to suggest a storage location based on the characteristics of these materials. Its guiding research question is: How to determine which spare parts should be stored centrally or locally? An in-depth literature review on these products is conducted to become acquainted with the environment. More specifically, given the wide range of spare parts, this theoretical approach is adapted in order to investigate the specificities of these materials and to obtain methods to distinguish them from each other. In addition, a focus on the most appropriate storage location strategy for spare parts is carried out. These theoretical concepts are transposed to a sample of data from Infrabel, the Belgian railway manager, and the storage site allocations according to the different methods are compared.

Putting the theory into practice, the findings support the hypothesis that the materials held by Infrabel have an intermittent demand pattern but that many of these spare parts have a low value. Next, the application of different methods of assigning storage locations for these products shows opposite trends depending on the dimensions studied. The results indicate that a decentralisation strategy is preferred for the majority of items if the assignment is made based on the criteria of price, predictability of demand and criticality of a part. Conversely, a centralisation strategy is favoured if the allocation of a storage location is based on price, demand variability over time and lead-time. This observation is also valid when the benefits of risk pooling are evaluated.