About this Abstract |
Meeting |
2025 TMS Annual Meeting & Exhibition
|
Symposium
|
Additive Manufacturing Modeling, Simulation and Machine Learning
|
Presentation Title |
Machine Learning Assisted Predictions of Thermal and Stress Profiles During Laser-based Additive Manufacturing |
Author(s) |
Aishwarya Manjunath, Venkata Mani Krishna Karri, Amrutha Anantatamukala, Shashank Sharma, Narendra Dahotre |
On-Site Speaker (Planned) |
Aishwarya Manjunath |
Abstract Scope |
The present work attempts to implement a physics-informed machine-learning (PIML) framework for predicting thermo-kinetic and thermo-mechanical evolutions during laser-based additive manufacturing. An experimentally validated 3D Multiphysics finite element (FE) process model is utilized to generate surface contour images related to temperature and state of stress as an input database. The PIML framework is developed on the basis of a conditional GANs machine learning model in a pix2pix framework to obtain images of temperature and stress distribution from numerical values of process parameters, which will be in the form of grid layout images. Where each grid corresponds to a process parameter, such as power, velocity, hatch spacing, layer thickness, thermal conductivity, density, reflectivity, and melting point. The current framework enables accurate prediction of thermo-kinetic and thermo-mechanical evolution in laser-based AM for host of materials such as SS316L, Cu, AlSiMg, and W. |
Proceedings Inclusion? |
Planned: |
Keywords |
Additive Manufacturing, Machine Learning, Other |