| About this Abstract |
| Meeting |
2022 TMS Annual Meeting & Exhibition
|
| Symposium
|
Additive Manufacturing of Large-scale Metallic Components
|
| Presentation Title |
Prediction of Large Domain Thermal History and Molten Pool Shape Using the Surrogate Modeling and Machine Learning |
| Author(s) |
Corbin Grohol, Yung C. Shin |
| On-Site Speaker (Planned) |
Yung C. Shin |
| Abstract Scope |
This study is concerned with predicting accurate temperature fields in a large domain using a surrogate modeling technique and machine learning. Though high-fidelity modeling has been demonstrated to provide accurate representations of the resulting geometry and temperature field, simulation of large scale as-built components is not feasible, even with massively parallelized computing. Instead, a surrogate modeling approach is demonstrated using a lower-fidelity model to extract features for use in a Gaussian process regression and implementing an active learning algorithm to determine when the high-fidelity model needs to be simulated to improve modeling results. Using such an approach, the high-fidelity model computational load can be decreased significantly, increasing calculation throughput. With this approach, accurate molten pool shapes of a large domain are predicted with affordable computational time. The validation results provide evidence that this method is effective and provides reasonable accuracy. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Additive Manufacturing, Machine Learning, Modeling and Simulation |