What does this research mean for the field?

The Laser Processing Fourier Neural Operator (LP-FNO) accurately predicts three-dimensional temperature fields and melt-pool boundaries in laser welding up to 100,000 times faster than traditional multiphysics simulations. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

To develop a fast and accurate Fourier Neural Operator-based surrogate model (LP-FNO) capable of predicting three-dimensional temperature fields and melt-pool boundaries across laser welding regimes.

August 22, 2026Journal of Intelligent ManufacturingOpen Access

A fast and accurate fourier neural operator-based surrogate for melt-pool prediction in laser processing

View Full Paper

Ask AI

Bookmark

Authors

ABAlix Benoit TIToni Ivas MPMateusz Papierz

Discussion

Member takes

Overview

Computational study demonstrates rapid 3D melt-pool and temperature prediction in laser welding using Fourier neural operators, indicating potential for real-time manufacturing optimization.

Key Points

To develop a fast and accurate Fourier Neural Operator-based surrogate model (LP-FNO) capable of predicting three-dimensional temperature fields and melt-pool boundaries across laser welding regimes.
Trained the LP-FNO model on high-fidelity multiphysics simulations generated with FLOW-3D WELD across conduction and keyhole welding regimes.
Reformulated transient equations into a quasi-steady moving laser reference frame using temporal averaging and a non-dimensional normalized enthalpy approach.
Benchmarked performance and resolution-invariant super-resolution capabilities against coordinate networks, U-Net, and DeepONet architectures.
LP-FNO achieved an average temperature relative error of approximately 2.5% and melt-pool segmentation intersection-over-union scores exceeding 0.9.
Full three-dimensional temperature fields and phase interfaces were computed in tens of milliseconds, running up to 100,000 times faster than finite-volume multiphysics simulations.
Models trained on coarse-resolution data successfully produced accurate super-resolved predictions on finer grids in mesh-converged conduction regimes.

Cite This Study

Benoit et al. (2026) studied this question.

synapsesocial.com/papers/6a895eaeca7ade938187cd34 https://doi.org/10.1007/s10845-026-02917-0

View Full Paper

Ask AI

Bookmark

August 22, 2026Journal of Intelligent ManufacturingOpen Access

A fast and accurate fourier neural operator-based surrogate for melt-pool prediction in laser processing

View Full Paper

Ask AI

Bookmark

Authors

ABAlix Benoit TIToni Ivas MPMateusz Papierz

Discussion

Member takes

Overview

Computational study demonstrates rapid 3D melt-pool and temperature prediction in laser welding using Fourier neural operators, indicating potential for real-time manufacturing optimization.

Key Points

To develop a fast and accurate Fourier Neural Operator-based surrogate model (LP-FNO) capable of predicting three-dimensional temperature fields and melt-pool boundaries across laser welding regimes.
Trained the LP-FNO model on high-fidelity multiphysics simulations generated with FLOW-3D WELD across conduction and keyhole welding regimes.
Reformulated transient equations into a quasi-steady moving laser reference frame using temporal averaging and a non-dimensional normalized enthalpy approach.
Benchmarked performance and resolution-invariant super-resolution capabilities against coordinate networks, U-Net, and DeepONet architectures.
LP-FNO achieved an average temperature relative error of approximately 2.5% and melt-pool segmentation intersection-over-union scores exceeding 0.9.
Full three-dimensional temperature fields and phase interfaces were computed in tens of milliseconds, running up to 100,000 times faster than finite-volume multiphysics simulations.
Models trained on coarse-resolution data successfully produced accurate super-resolved predictions on finer grids in mesh-converged conduction regimes.

Cite This Study

Benoit et al. (2026) studied this question.

synapsesocial.com/papers/6a895eaeca7ade938187cd34 https://doi.org/10.1007/s10845-026-02917-0

View Full Paper

Ask AI

Bookmark