ITC Infotech undertook work to help optimize inventory keeping in complex asset-intensive industries. By combining simulation, machine learning, and optimization, they demonstrated effective asset management and inventory optimization for rotable/repairable spares that balances service levels and inventory costs.
The usefulness of expendables, which are disposed of or recycled at the end of their operational life, is related to how long they operate before failing. This time is often quantified generally as the mean time before failure—MTBF. For repairable parts, it is also necessary to consider repair time, which depends on defect type, work center availability, and repair staff availability.
All things considered, the lifecycle of a repairable can be broken into five stages: working, breakdown, replacement, repair, and in-store.
To understand how to optimize inventory management for repairable spare parts, it is necessary to model the life cycle of the parts and the system they are used in.
With an overall system model, it is possible to measure end-to-end performance. It can show the effects of part failure rates, as well as the time needed to replace a part, repair it, and return it to inventory.
Related to the system model is a model that helps accurately determine part failure rate. ITC Infotech created a failure model that considered several factors: running life, machine used, failure history, maintenance history, and operating conditions.
The failure model produces its predictions for part-failure-rate using a decision-tree based machine learning library in Python (learn more about connecting to Python in this Pypeline webinar video). The predictions pass back into the simulation where the OptQuest engine determines the optimal spare part count.
The system architecture consists of five main components: the current state of the system, the baseline input state, part failure rate prediction using Python, a simulation model, and the spare part count optimization.
By modeling a system in this way, ITC Infotech can optimize the number of spare parts to hold in inventory. As a result, they can ensure machine availability and establish policies that balance inventory costs against service levels.
The system was described at the AnyLogic User Conference 2019 in India by Sumit Kumar. The presentation recording is available to watch in our video library.
You can stay up to date with AnyLogic developments by subscribing to our newsletter. And also, don’t miss the chance to meet with others in the AnyLogic community – check out our upcoming events.
