About this Abstract |
Meeting |
MS&T23: Materials Science & Technology
|
Symposium
|
Materials Processing and Fundamental Understanding Based on Machine Learning and Data Informatics
|
Presentation Title |
Physics-informed Machine Learning for Crystal Plasticity Model Calibration of Ti-7Al Alloy |
Author(s) |
Mohamed Elleithy, Ender Eger, Arulmurugan Senthilnathan, Mahmudul Hasan, Pinar Acar |
On-Site Speaker (Planned) |
Mohamed Elleithy |
Abstract Scope |
We develop a physics-informed machine learning (ML) model to identify the crystal plasticity (CP) parameters of Ti-7Al alloy, a candidate aerospace alloy for jet engine components. We address this challenge by solving an inverse problem that aims to obtain the optimum slip and twin parameters by minimizing the differences between the experimental data and CP model predictions for the deformed texture. For such a problem requiring excessive computing times, physics-informed ML models perform more efficiently than conventional ML models by improving accuracy, computational efficiency, and explainability. We apply a new method, the Physics-Informed Neural Network (PINN), which incorporates the underlying problem physics through the loss function definition. Here, we demonstrate the application of PINN to a small-data problem driven by a CP model that needs to satisfy the physics-based constraints of the microstructural orientation space while obtaining the slip and twin parameters of Ti-7Al alloy. |