| About this Abstract |
| Meeting |
2024 TMS Annual Meeting & Exhibition
|
| Symposium
|
Additive Manufacturing Fatigue and Fracture: Towards Rapid Qualification
|
| Presentation Title |
Rapid Prediction of Fatigue-performance Heat Maps in Additively Manufactured Metals by Integrating Physics-based and Data-Driven Modeling |
| Author(s) |
Krishna Prasath Logakannan, Ashley Spear |
| On-Site Speaker (Planned) |
Krishna Prasath Logakannan |
| Abstract Scope |
Predicting (micro)structure-property relationships using data-driven approaches is limited by the amount of high-fidelity training data available, especially for microstructure-sensitive fatigue behavior. A fast Fourier transform (FFT)-based crystal plasticity method provides an opportunity to generate high-throughput data. The objective of this work is to enable rapid prediction of spatially dependent fatigue properties in additively manufactured build volumes by leveraging an FFT framework in conjunction with deep learning models. Realistic additively manufactured microstructures of stainless steel 316L generated previously using a multi-physics framework are modeled here using an FFT framework to predict fatigue indicator parameters (FIPS) throughout microstructural subvolumes, which can be visualized using heat maps. Generated high-throughput data are leveraged to train a 3D convolutional neural network. The overall framework presented here can be used to assist with design and qualification for metal additive manufacturing by enabling rapid predictions of hotspots in fatigue-critical parts. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Additive Manufacturing, Mechanical Properties, Machine Learning |