Data-Driven, Parameterized Reduced-order Models for Predicting Distortion in Metal 3D Printing

TL;DR

This paper presents a data-driven reduced-order modeling framework combining POD and GPR to accurately and efficiently predict distortion in metal 3D printing, significantly speeding up the process compared to deep learning methods.

Contribution

It introduces a novel ROM framework integrating POD with GPR for distortion prediction in LPBF, outperforming deep learning approaches in accuracy and speed.

Findings

POD-GPR predicts distortions within ±0.001mm.

The model achieves approximately 1800x faster computation.

Compared to deep learning, the ROM offers high accuracy and efficiency.

Abstract

In Laser Powder Bed Fusion (LPBF), the applied laser energy produces high thermal gradients that lead to unacceptable final part distortion. Accurate distortion prediction is essential for optimizing the 3D printing process and manufacturing a part that meets geometric accuracy requirements. This study introduces data-driven parameterized reduced-order models (ROMs) to predict distortion in LPBF across various machine process settings. We propose a ROM framework that combines Proper Orthogonal Decomposition (POD) with Gaussian Process Regression (GPR) and compare its performance against a deep-learning based parameterized graph convolutional autoencoder (GCA). The POD-GPR model demonstrates high accuracy, predicting distortions within $\pm0.001mm$, and delivers a computational speed-up of approximately 1800x.