About this Abstract |
Meeting |
2025 TMS Annual Meeting & Exhibition
|
Symposium
|
Additive Manufacturing Modeling, Simulation and Machine Learning
|
Presentation Title |
Additive Manufacturing Process Modeling with Multi-Output Gaussian Processes |
Author(s) |
Sudipta Biswas, Som Dhulipala, Peter German, Andrea M. Jokisaari |
On-Site Speaker (Planned) |
Sudipta Biswas |
Abstract Scope |
The microstructures of the additively manufactured materials, including the grain and subgrain characteristics, are significantly different from conventionally manufactured materials. This influences the properties of the AM products. Idaho National Laboratory’s (INL) Multiphysics Object-Oriented Simulation Environment (MOOSE), specifically the MOOSE Application Library for Advanced Manufacturing UTilitiEs (MALAMUTE) software, provides an ideal platform for developing the multiphysics multiscale model to explore the intricacies of the microstructural evolution during the AM processes within a single framework. Considering that full-fidelity multiphysics models are computationally expensive, especially when these models have to be run numerous times under varied or stochastic process conditions, surrogate modeling with the aid of Multi-Output Gaussian Processes (MOGPs) is utilized for computational efficiency. Dimensionality reduction techniques are combined with MOGPs to significantly reduce their training time and improve their predictive performance. Finally, the outlook for using MOGPs in AM process modeling in conjunction with active learning will be discussed. |
Proceedings Inclusion? |
Planned: |