Author: Martin Thormann
Graduate School of Logistics, Faculty of Mechanical Engineering, TU Dortmund University, 44227 Dortmund, Germany
Accurate spare parts demand planning and effective distribution planning is essential for providers of after-sales services in the machine and plant engineering industry to ensure high spare parts availability for maintenance and failure orders (callouts) at a reasonable cost. Low spare parts availability is primarily the result of high uncertainty in spare parts demand, leading to misallocation of parts within after-sales service networks. The lack of spare parts availability causes equipment downtime, resulting in customer dissatisfaction and possible penalty costs for after-sales service providers, if response times are contractually fixed. This paper proposes an approach and planning methods for integrating real-time status information about equipment utilization and service conditions to determine optimal spare parts stocking strategies. For this purpose, spare parts stocking strategies and ordering policies for application in after-sales service networks are analyzed. Furthermore, a binary linear optimization model is developed for the assignment of stocking strategies to spare parts based on real-time demand information of the equipment to be serviced. This method uses data provided by an internationally operating elevator company.
Keywords: after-sales service, demand planning, distribution planning, elevator industry, inventory management, spare parts management
High spare parts availability is essential for providers of after-sales services to efficiently carry out equipment maintenance, troubleshooting, or repair. The spare parts supply is fulfilled via an after-sales service network composed of central (CDC) and regional distribution centers (RDC), branches, and service cars or service technicians, upon which distribution chains are built (see Figure 1). This type of after-sales service network structure is very common in the machine and plant engineering industry, particularly the elevator industry. The spare parts demand and distribution planning requirements in after-sales service networks can vary tremendously among different industries. Figure 2 shows the relevant characteristics describing aftersales service networks and network characteristic forms considered in this research work.
In the networks considered in this research work, aftersales service providers offer product services by undertaking structures (cf. Dekker et al., 2013; Muhammad Naiman Jalil, 2011). Hence, a small number of central entities, such as central and regional distribution centers, serve a large number of branches and service technicians.
To maintain high spare parts availability with a reasonable logistics cost in these large scale network structures, providers of after-sales services need to adequately forecast spare parts demand and effectively make appropriate decisions for the various network entities – such as CDCs, RDCs, branches, and service cars – regarding sourcing, allocating, and disposing of spare parts. These planning tasks can be aligned with demand planning and distribution planning in the supply chain management task model, developed by Kuhn and Hellingrath (2002). Since demand planning is the fundamental basis for subsequent stocking decisions, such as where and how much to stock and when to order, after-sales service provider can only make decisions by integrating both planning tasks (cf. Ihde et al., 1999; Schuh et al., 2013; Klug, 2010; Hayashi et al., 2009). The definitions of these two planning tasks originate from conventional logistics networks, such as retail or production networks, but they are appropriate in the context of after-sales service networks. However, in after-sales service networks, demand and distribution planning of spare parts are even more complex planning tasks, particularly in highly service-oriented manufacturing industries such as the elevator industry. In this industry, the equipment to be serviced (equipment portfolio) comprises equipment from several decades and different manufacturers, leading to high equipment heterogeneity (cf. Hertz & Finke, 2011; Dispan, 2007). Moreover, new installations, high cancellation rates of service contracts, and the consolidation of markets due to mergers and acquisition activities lead to a steady fluctuation in the equipment portfolio (cf. Blakeley et al., 2003; Baumbach, 2004). Consequently, after-sales service providers need to plan for numerous parts in severely uncertain and volatile markets (cf. Cohen et al., 2006; Hertz & Finke, 2011). Therefore, after sales service providers cannot make decisions based solely on historical demand data (cf. Niggeschmidt, 2010; Ihde et al., 1999; Thormann, 2014). Additionally, the highly decentralized network structures increase the complexity in determining optimal decisions that minimize the total logistics cost for the entire service network, such as shipping, holding, handling, or ordering costs, since strong reciprocity exists among the planning decision of each network entity.
In the first step of the development of the distribution planning approach, spare parts stocking concepts have been analyzed to determine appropriate stocking strategies for the considered multi echelon after-sales service networks. In this context, spare parts stocking strategies are defined as a readily designed combination of stocking, sourcing, and disposal decisions of all autonomously acting network entities as well as the corresponding supply relationships of a distribution chain. These stocking strategies have to be applied for each spare part within the after-sales service network. By matching appropriate stocking strategies with the network-specific requirements, a strategy tool kit with applicable stocking strategies alternatives has been developed, which can be applied by each distribution chain. However, after sales service providers need to consider all decisions of each part and each distribution chain simultaneously to find stocking strategies that minimize the total logistics cost for the considered network. This is because strong reciprocity exists between the decisions for each part and the distribution chain. For instance, following the notation of Figure 1, the cost-optimal stocking strategy for CDC RDC1 B1 might be to stock a particular part in the branch storage, whereas the network wide cost-optimal stocking strategy considering all distribution chains would be to stock this part in the central distribution center and to supply the branches on demand. Moreover, capacity restrictions exist at the branch and mobile storage locations. Therefore, only a set of spare parts has to be stored at the branch storage stage, which minimizes the network-wide logistics costs. For this reason, making these decisions successively or in an isolated manner might result in a loss of pooling potential and higher logistics costs. Thus, aftersales service providers need to simultaneously consider all possible decision alternatives within the network to satisfy each network entity in terms of the required service level and in minimizing the total logistics costs. Due to the variety of spare parts, network entities, and available decisions, an optimization model is used to find stocking strategies for each spare part and distribution chain. For the integration of the developed stocking strategies into an optimization model, the stocking strategies have been logically broken down to echelon-dependent inventory management decision alternatives (EDIMDA). EDIMDA will be defined as applicable combinations of sourcing, stocking, and disposal decision alternatives that can be taken on each distribution stage (echelon) for each part. They determine whether to stock a spare part or not, what inventory policy to choose, and from where to source a spare part for each network entity. Since EDIMDA represent decisions for only one part and echelon, it is possible to quantitatively evaluate these decisions and model them in the optimization approach.
Based on the integration of real-time demand information and the EDIMDA, the integrated spare parts distribution planning approach follows three main steps. In the first step, all considered EDIMDA are parameterized for each stocking location and spare part with a developed heuristic applying the METRIC approximation (cf. Muckstadt & Sapra, 2010 or Axsäter, 2006). The parameterization facilitates the quantitative evaluation of each EDIMDA with developed cost models in the second step. The final step involves a binary linear optimization model that simultaneously considers all parameterized and evaluated EDIMDA of the parts for each stocking location in the network by minimizing total cost and considering capacity and time constraints.
Since this approach has been developed for tactical distribution planning, the method should be applied after an interval of several months, after significant change in demand has occurred. For this, service providers need to retrieve the real time equipment status information and forecast the demand as presented by Thormann (2014). With the updated real time demand information, the three steps of the distribution planning method should be initiated. After that, service providers need to check the distribution planning results with the current distribution plan and implement appropriate adaptation measures, such as the adaptation of stocking policies or disposal parameters, or the reallocation of spare parts within the after-sales service network.
This paper presents an approach for the integration of real-time demand information and planning of spare parts distribution, and appropriate planning methods that aftersales service network providers can apply to find costoptimal spare parts stocking strategies. The paper illustrates the relevant approaches in the literature, and introduces spare parts stocking concepts and ordering policies. Based on this analysis, a spare parts stocking strategy tool kit and the derived EDIMDA are proposed. Finally, the paper presents the three steps of the method of the distribution planning approach, including a heuristic for approximating, parameterizing, and evaluating EDIMDA, and a binary linear optimization model for choosing EDIMDA, which can also be solved for large problem instances. A regional distribution stage can be easily integrated by adding the required constraints, since the EDIMDA are already defined. Moreover, due to the modular structure of this approach, it can be easily extended to more echelons in a similar way, when developing further EDIMDA and adjusting the optimization model. This could be considered for further research, although more than three distribution stages are not very common in after-sales service networks. However, two main questions remain for further research. The first issue is either the determination of events, such as a significant change in spare parts demand, that will prompt the necessity to optimize distribution planning or the determination of appropriate time intervals for applying the distribution planning method. The second issue is that this approach assumes centralized planning: all network entities support the objective of minimizing the total cost of the after-sales service network. However, due to common existing profit center structures in large scale companies, this assumption does not always hold true. Therefore, further research should also address how the respective network entities can be incentivized such that centralized decisions are supported and satisfied. Answering this question could yield a very promising integrated spare parts distribution planning approach for providers of after sales service networks.
Martin Thormann received a bachelor’s degree in logistics from the faculty of mechanical engineering at the TU Dortmund University, Dortmund, Germany in 2011 and a master’s degree in supply chain engineering from the H. Milton Stewart School of Industrial and Systems Engineering at the Georgia Institute of Technology, Atlanta, USA in 2012. He gained practical experience in logistics and supply chain management while working on consulting projects for several internationally operating companies. He was a student research assistant at the Chair of Factory Organization at the TU Dortmund University and at the Fraunhofer Institute for Material Flow and Logistics. Since November 2012, he has been working on his dissertation project at the TU Dortmund University, Dortmund, Germany under a doctoral scholarship. His research focuses on spare parts demand planning and distribution planning in after-sales service networks. Mr. Thormann is member of the Council of Supply Chain Management Roundtable Germany e. V. He received a scholarship from the CSCMP Global Education e.V. for his study abroad.