Reconstruction of Molten Pool Flow Fields from High-Speed Video Using Physics-Informed Neural Networks

Abstract

Molten pool flow significantly influences bead geometry, defect formation, and solidification microstructure in wire-arc directed energy deposition (DED). However, direct flow sensing remains an open challenge due to extreme temperatures, intense arc illumination, and complex flow patterns. Traditional research has relied either on computationally intensive fluid dynamics simulations, which require assumed boundary conditions and lack empirical anchoring, or on vision-based tracer particle tracking, which yields sparse, noisy, and physically unconstrained velocity estimates. This work bridges these two paradigms by reconstructing physics-constrained 2D surface flow fields from high-speed imaging of a GTAW-based wire-arc DED process under distinct flow regimes: quasi-steady convection and periodic droplet impact. The pipeline begins with homography correction to calibrate the oblique camera view, followed by dense optical flow extraction to provide raw velocity observations. For quasi-steady pools, a physics-informed neural network (PINN) reconstructs the surface velocity field by jointly enforcing data fidelity, incompressible Navier-Stokes momentum equations, no-penetration boundary conditions, and an arc-forcing prior through composite loss minimization. For pools subject to periodic droplet impact, a phase-conditioned time-dependent PINN is further developed to resolve the cyclic flow evolution throughout the impact period. This work provides a unified framework that bridges high-speed visual sensing and fluid mechanics, enabling physics-grounded flow diagnostics for wire-arc DED processes.

Physics-informed, boundary-constrained Gaussian process regression for the reconstruction of fluid flow fields
Adrian Padilla-Segarra, Pascal Noble, Olivier Roustant +1
International Journal for Uncertainty Quantification1 Jun 2026Peer-reviewedLicence not statedFull text2 citedStatistical & nonlinear physicsHAL
Reconstruction of Magnetic Fields with a Physics-Informed Neural Network
P. Adhikari, R. Bhattarai, J. Brewington +9
arXiv16 Oct 2026PreprintCC BY 4.0Full textApplied physics & instrumentationarXiv
Rapidity-Coupled Spin Dynamics in Pulsed Laser Fields from Physics-Informed Neural Networks
N. S. Akintsov, A. P. Nevecheria, S. N. Andreev +1
arXiv17 Sep 2026PreprintarXiv licenceFull textOptics & photonics Plasma physicsarXiv
Non-linear mechanical field reconstruction coupling recurrent neural networks with physics-informed graph neural networks
Manuel Ricardo Guevara Garban, Yves Chemisky, Étienne Prulière +3
arXiv13 Oct 2026PreprintCC BY 4.0Full textComputational physicsarXiv
Physics-informed Kolmogorov-Arnold networks (PIKANs) for noise-robust reconstruction of boiling heat flux from infrared measurements
Siavash Khodakarami, Parimala Vardhan Vydyula, Aniruddha Bora +2
arXiv1 Oct 2026PreprintCC BY 4.0Full textComputational physicsarXiv

Cite

BibTeX RIS

BibTeX

@misc{cao2026reconstruction,
  title = {{Reconstruction of Molten Pool Flow Fields from High-Speed Video Using Physics-Informed Neural Networks}},
  author = {Yue Cao and Qin Su and Dung Hoang Tien and Van Anh Nguyen},
  howpublished = {arXiv},
  year = {2026},
  month = sep,
  doi = {10.48550/arxiv.2609.32126},
  eprint = {2609.32126},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.32126},
}

RIS

TY  - UNPB
TI  - Reconstruction of Molten Pool Flow Fields from High-Speed Video Using Physics-Informed Neural Networks
AU  - Yue Cao
AU  - Qin Su
AU  - Dung Hoang Tien
AU  - Van Anh Nguyen
PY  - 2026
DA  - 2026/09/26
PB  - arXiv
DO  - 10.48550/arxiv.2609.32126
UR  - https://arxiv.org/abs/2609.32126
AB  - Molten pool flow significantly influences bead geometry, defect formation, and solidification microstructure in wire-arc directed energy deposition (DED). However, direct flow sensing remains an open challenge due to extreme temperatures, intense arc illumination, and complex flow patterns. Traditional research has relied either on computationally intensive fluid dynamics simulations, which require assumed boundary conditions and lack empirical anchoring, or on vision-based tracer particle tracking, which yields sparse, noisy, and physically unconstrained velocity estimates. This work bridges these two paradigms by reconstructing physics-constrained 2D surface flow fields from high-speed imaging of a GTAW-based wire-arc DED process under distinct flow regimes: quasi-steady convection and periodic droplet impact. The pipeline begins with homography correction to calibrate the oblique camera view, followed by dense optical flow extraction to provide raw velocity observations. For quasi-steady pools, a physics-informed neural network (PINN) reconstructs the surface velocity field by jointly enforcing data fidelity, incompressible Navier-Stokes momentum equations, no-penetration boundary conditions, and an arc-forcing prior through composite loss minimization. For pools subject to periodic droplet impact, a phase-conditioned time-dependent PINN is further developed to resolve the cyclic flow evolution throughout the impact period. This work provides a unified framework that bridges high-speed visual sensing and fluid mechanics, enabling physics-grounded flow diagnostics for wire-arc DED processes.
ER  -

Abstract

Related

Cite