Using state-of-the-art physics-based models, Veracity’s scientists in Perry, UT, are developing innovative approaches for reducing uncertainty and error in modeling service-life estimates of solid rocket motor (SRM) propellant. Veracity is developing the Missile Inventory Management System (MIMS) to use real-time sensor data in physics-based models to calculate service life for individual missiles. The goal is to reduce uncertainty by a minimum of 10 percent, resulting in greater confidence in the reliability of SRM weapon systems. MIMS is built to accommodate internet-of-things sensors feeding data into a standards-based Application Programming Interface (API). It then uses cloud-based infrastructure to perform complex calculations to automatically update the service-life assessments.
This project supports the Air Force Research Lab (AFRL), Rocket Propulsion Division, Edwards AFB Motor Aging and Surveillance Technology (MAST) Broad Agency Announcement.
Aging and Surveillance
At Veracity, we employ state-of-the-art analysis tools for aging and surveillance (A&S) of energetic materials. We support our customers throughout the A&S life cycle:
- Failure mode and effects analysis (FMEA)
- Test plan generation
- Data processing
- Chemical evolution modeling
- Safe-life and service-life calculations
- Reliability modeling
At each phase we work closely with material experts and engineers to ensure current scientific understanding and product familiarity.
We create advanced material models for energetic materials. Our experience includes creating models for polymeric binders, particulate reinforcement, and particulate agglomeration. We provide greatest value by quickly prototyping and verifying material models and then implementing them in production-ready code. At Veracity, we build models using Python, C++, and various other programming languages. We also have built models for integration into third-party analysis tools, such as user-defined mechanical behavior subroutines (UMAT) for Abaqus (finite element analysis tool).
Machine Learning
Veracity knows that understanding reliability and maintainability issues is a critical concern for the U.S. military commands that manage weapons systems and platforms. Many weapons systems are a complex collection of subsystems and components, with tracking occurring through many different data streams. However, a trove of unused data exists that is difficult to process to predict reliability and maintainability issues. Machine learning, often referred to as artificial intelligence (AI), offers an efficient means of processing this unused data.
Prior to becoming part of Systems Planning and Analysis, Veracity completed a phase I Small Business Innovative Research effort that used augmented machine learning to model rocket system reliability. Using notional data, we demonstrated that the methodology we developed reduced the average error in predicting overall weapon system reliability for a single missile by 65 percent. It also reduced the error in predicting reliability among all missiles by 35 percent. For individual missile components, our models can reduce error occurrence by 36 percent to 90 percent.
The Veracity approach is generic enough to cover arbitrary contribution levels from both traditional subject matter expert (SME)-informed reliability analysis and machine learning. As a result, Augmented Machine Learning can be used for all reliability analyses. In fact, given its generic nature, our approach for combining machine learning with traditional SME-informed models may be applied outside of reliability modeling in areas such as inventory and parts forecasting.