Improving service part allocation at Liander’s multi-echelon structure

Berg, Diederick van den (2013) Improving service part allocation at Liander’s multi-echelon structure.

Abstract:Problem Liander initiated a project, Standaard Op Maat (SOM), to reduce the total supply chain cost and increase customer satisfaction. A part of this project is the revision of the current urgent order fulfilment. The general impression is that a sizeable decrease in inventory costs is possible via proper alignment of the various stock locations for service parts, while maintaining a high service level. Therefore, we stated the following objective: Create a service parts allocation model for urgent orders, which determines the best storage network and order parameters in corresponding warehouses. Current practice Before developing a methodology, the current practice of urgent order fulfillment and its performance is analyzed. We identified the following key problem in the current practice: The exclusive availability of local warehouses to Netcare mechanics employed for upkeep and failure work. The exclusive availability causes unnecessary emergency shipments from the central warehouse for all other types of work that are not upkeep or failure related, while the SKU may be available in a nearby local warehouse. We defined the performance of the service part allocation for urgent orders by costs, as a result of Inventory value and emergency shipments, and the number of StoringsVerBruiksMinuten (SVBM) due to waiting time for spare parts. In order to approach the caused SVBM by Logistics, all expensive SKUs (>€ 100.-) are categorized according to their criticality on the gas or electricity network in the event of a failure. We refer to this value as SVBM’. Models In the developed model there are four different networks for storing a SKU (excluding van): 1) supplier only (consignment stock), 2) central warehouse only, 3) combination of central warehouse and MWs (manned warehouses) and 4) combination of central warehouse, MWs and UWs (unmanned warehouses). Networks 3 and 4 are configured in a two-echelon system, which enables us to utilize the risk pooling effect at the central warehouse. Besides choosing a network, the model also determines the optimal order parameters (minimizing costs while restricting the number of SVBM’) at the corresponding warehouses. In order to find the best solution with the associated decision variables we apply column generation. In order to measure the performance of order parameters in a network, the following assumptions are made: 1) There is no lateral resupply between local warehouses.,2) No orders are backordered; rather, an emergency order occurs from the next higher supply chain level, 3) Customer and replenishment order size is one. The column generation model is used for SKUs valued above € 100.-. The SKUs valued less than € 145.- are all locally stored, irrespective of other SKU’s characteristics, corresponding to networks 3 and 4. A second model is developed for the inexpensive SKUs below € 100.-, based on the assumption that these low valued service parts are always locally stored. This local model saves calculation time in the column generation model and enables the optimization of the replenishment size.
Item Type:Essay (Master)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:52 mechanical engineering
Programme:Industrial Engineering and Management MSc (60029)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page