Supply Chain Issues
Current USAF Mission Capable and Aircraft Availability rates highlight the need for continued improvement and innovation. Large portions of our assets remain grounded because of supply chain and maintenance issues. Lead times, even for parts which have a standard procurement process, are often counted in years rather than days or weeks. Advanced Manufacturing (AM) seeks to solve this problem by producing or repairing parts on a more consistent, predictable, and expedited timeline.
The complexity of the Air Force supply chain poses a significant challenge in identifying parts that can be effectively produced using AM. A thorough analysis of the supply chain and the feasibility of AM for each part is necessary, but given the vast number of parts, a manual analysis is not feasible. This makes the development of Artificial Intelligence (AI) and Machine Learning (ML) tools essential to automate this process.
Preliminary results showing success in predicting the probability that a part will be in backorder in the near future. Note that the future backorder status of most parts is correctly predicted with high confidence, and better results are expected as we continue to develop these ML models.
The Artificial Intelligence Solution
SEEKR, a custom software application that EXPANSIA developed to integrate disparate data sources and enable quick, comprehensive, and predictive supply chain visibility, uses AI and ML to study each part’s viability for Advanced Manufacturing in two ways:
- Identify, “Can AM be used for each part?” by assigning probabilities based on similarity to parts which engineers have determined can be produced using AM.
- Identify, “Should AM be used for each part?” by making stock on hand forecasts and computing probabilities that each part will go on backorder before an order placed today would be delivered.
These algorithms are being trained on the inventory history of over a million parts as well as approximately three thousand internal Part Reports completed by mechanical engineers detailing a parts viability assessment for AM.
Word clouds in the shape of a C-5 Galaxy showing words which frequently occur in part names which can or cannot be produced using Advanced Manufacturing. This leverages Natural Language Processing (NLP) and is a portion of the AI and ML at work in SEEKR, helping to identify patterns and trends in part naming conventions and informing the development of more effective AI and ML models.
The IMPACT of AM
By identifying which parts can and should be produced with Advanced Manufacturing, SEEKR allows engineers to focus their efforts on parts which will be produced using AM rather than wading through a sea of parts that will not. This targeted approach will significantly enhance supply chain resilience by reducing procurement lead times, increasing part availability, and ultimately improving aircraft availability and mission capability rates.
Furthermore, SEEKR’s capability to identify potential alternative sources of supply empowers the USAF with critical leverage to renegotiate and optimize traditional supply chain contracts, potentially leading to shorter lead times, better pricing, and more reliable deliveries. Collectively, this enhances the USAF’s operational readiness and force effectiveness.
Artificial Intelligence and Machine Learning are the ultimate force multipliers. Their impact goes well beyond keeping aircraft in the air; they are also able to directly increase the efficiency and productivity of airmen and civilians on the ground.