| About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
Verification, Calibration, and Validation Approaches in Modeling the Mechanical Performance of Metallic Materials
|
| Presentation Title |
Predicting Mechanical Properties of Ti-6Al-4V Alloy Using a Physics-Informed Neural Network (PINN) for Crystal Plasticity Modeling |
| Author(s) |
Mohamed Elleithy, Zekeriya Ender Eger, Pinar Acar |
| On-Site Speaker (Planned) |
Pinar Acar |
| Abstract Scope |
This study develops a Physics Informed Neural Network (PINN) to predict the constitutive stiffness tensor of Ti-6Al-4V alloy under varying temperatures and crystallographic orientations. The dual-phase microstructure of Ti-6Al-4V influences the alloy's multiscale mechanical characteristics, rendering it ideal for aerospace applications. By integrating Crystal Plasticity Finite Element (CPFE) modeling with Machine Learning (ML) and physics-informed constraints, this research enhances the prediction accuracy and computational efficiency of the constitutive elastic stiffness tensor. The PINN framework uses temperature, phase fractions, and crystallographic texture descriptors as inputs. It employs a data-driven and physics-informed loss function to ensure physically feasible predictions. The methodology involves generating randomly oriented texture samples constricted to experimental orientations through statistical constraints and training the PINN model to maintain the material's solid-state across rising temperatures. The implementation of the PINN approach accelerates the predictions, reduces experimental costs, and improves the interpretability of results with the incorporation of physics-informed constraints. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
ICME, Mechanical Properties, Machine Learning |