Development of a surrogate model for high-fidelity laser powder-bed fusion using tensor train and gaussian process regression

• DOI: 10.1007/s10845-022-02038-4
• Publication Date: 2022-10-27T16:06:12Z
• Authors: Umesh Kizhakkinan; Pham Luu Trung Duong; Robert Laskowski; Guglielmo Vastola; David W. Rosen; Nagarajan Raghavan
• Keywords: Surrogate model; Uncertainty Quantification

Abstract A multi-physics high-fidelity computational model is required to study the melting and grain growth phenomena in a laser powder-bed fusion (LPBF) additive manufacturing process. The major challenge with long time, which makes it unsuited for any feasible process parameter optimization high dimensional design space. To address this challenge, surrogate models are good option replace model, resulting significant shortening of time at expense an acceptable drop accuracy. In study, tensor train (TT) Gaussian regression (GPR) based methodology proposed develop powder-scale model. An in-house developed used generate training data by simulating microscale different values power. trained TT-GPR can predict thermal history melt pool geometry specified value power, while computation prediction set conditions less than one second. Here we achieve approximate speedup 10 $$^4$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msup> <mml:mrow /> <mml:mn>4</mml:mn> </mml:msup> </mml:math> We provide evidence claim that provides efficiency without compromising