21 Aug 2021
Intermittent demand, low order frequency, high price variations, outdated bill-of-materials, and many stock-keeping units are all reasons why spare parts planning is a very tough exercise. High service levels to avoid machine downtimes come with high inventory levels or scrapping rates. Balancing these supply chain costs against demanding customer requirements, especially in industrial after-market services, is a very challenging task and is hardly solved with standard supply chain software.
numi collaborates with a global OEM of complex specialist foundation engineering machinery, to tackle these challenges. We proofed in a pilot, that a spare parts planning approach, which includes transactional data from different source systems, machine in-use data, and advanced algorithms have some significant benefits:
After this initial proof-of-value, the pilot will be rolled out globally.
Besides our out-of-the-box algorithms for forecasting, inventory optimization and expected supplier lead time predictions, three success factors are identified and implemented for a first pilot region.
Most machines of the installed base send machine usage data like fuel consumption, running hours, or their location. By connecting these machine data with actual spare parts consumption data, demand patterns and failure rates of specific spare parts could be detected. numi's Machine Learning spare parts prediction algorithm is integrated into regular replenishment decisions at the local warehouses, resulting in a 15% reduction in inventory over the simulation period.
Some older spare parts cannot be predicted with machine data, since these machines are not equipped with an IoT device. For these machines, numi fits the actual item demand distribution instead of using a normal distribution approach as many other software vendors would do. The intermittent demand pattern of spare parts usually does not follow a normal distribution, which causes either high inventory levels with excess inventory or poor customer service.
A further impact on the reduction of inventories was including network rebalancing options into replenishment decisions. In a setup where local warehouses can be replenished from a few centralized hubs or from other local warehouses, automated and cost-optimized sourcing recommendations help to avoid long lead times or unnecessary inventory built up at each supply chain node. The Multi-Echelon-Inventory-Optimization helps to reduce overall network stocks by setting the right inventory parameter at each warehouse and identifying non-mover at various locations.
High-quality spare parts bill-of-materials and machine-in-use data is a prerequisite for our approach, but maintaining bill-of-materials for machines with a product lifetime longer than 15 years can be very challenging and error-prone. Sourcing risks are also much higher in the aftermarket business than at the time of machine production. Parts criticality to keeping a machine running is another dimension that needs to be considered.
All these challenges make it close to impossible to purely trust some advanced algorithms when making costly spare parts stocking decisions. numi follows a collaborative approach, where expert knowledge of parts criticality and sourcing risk is enriched by predictive insights like proactive failure rates. The relevant spare parts planning data can be maintained by the user with numi's workflows and engaging UI and will be augmented by predictive spare parts failures.
This proactive spare parts planning approach has many different side effects besides global inventory reductions. The data can be used for new service revenue potentials like warranty contracts or service agreements with proactive service schedules. Spare parts pricing decisions, to leverage the full potential of spare parts margin for critical and high sourcing risk parts. Or high visibility of the installed base and its KPIs.