STEAM: A scalable data-driven surrogate modeling framework for part-scale scanwise thermal process simulation of laser powder bed fusion
Source abstract
Abstract
Simulating scanwise thermal behavior at part scale in laser powder bed fusion (LPBF) is essential for understanding defects and optimizing process parameters, but it remains prohibitively expensive for real engineering parts. While finite element method (FEM) based process simulations provide valuable insights, the steep computational cost limits their routine application. This work introduces a scalable data-driven surrogate modeling framework for scanwise LPBF thermal simulation through deep learning called STEAM (Scanwise Thermal Emulator for Additive Manufacturing) to predict the transient thermal history of complex parts. The key features behind the proposed framework include 1) a compact dataset constructed from geometrically distinct modular blocks capturing representative thermal patterns, and 2) a combination of temporal and spatial features employed as model inputs to capture key physical phenomena influencing thermal behavior. Temporal features describe how each node interacts with the moving heat source over time, while spatial features capture the effect of the surrounding geometry on local heat conduction and boundary conditions. The model is validated on 2.5D test geometries as much as ~10 times larger than the training cases with different geometric complexity. On large-scale test blocks, STEAM achieves an R² of 0.97, and 29.61 °C mean absolute error, with a 1237–3648× speedup compared with FEM using four GPUs (i.e., 309–912× per GPU). Importantly, performance improves with increasing model complexity, which is promising for further scaling STEAM to larger and more complex LPBF domains. These results demonstrate that the STEAM framework provides a promising path toward fast and accurate thermal prediction in real-world LPBF parts. • A scalable data-driven surrogate model for part-scale scanwise thermal process simulation of LPBF is presented. • The training set consists of compact yet geometrically diverse blocks that capture representative heat transfer patterns. • Temporal features describe node–heat source interactions, while spatial features reflect the influence of surrounding geometry on local heat conduction and boundary conditions. • The model is evaluated on unseen test blocks that are as much as ~10 times larger than the training blocks. • The model predicts thermal histories with an R² of 0.97 across millions of nodes and thousands of time steps.
Source: openalex_inverted_index · Quality: baseline
Research context
OpenAlex topics: Additive Manufacturing Materials and Processes · Additive Manufacturing and 3D Printing Technologies · Radiative Heat Transfer Studies
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
References (60)
Show source-provided references
1. A review of Laser Powder Bed Fusion Additive Manufacturing of aluminium alloys: microstructure and properties
2. Melt pool temperature and cooling rates in laser powder bed fusion
3. A review on in-situ process sensing and monitoring systems for fusion-based additive manufacturing
4. In situ synchrotron X-ray imaging of 4140 steel laser powder bed fusion
5. Data-fusion for in-situ monitoring and molten state identification during LPBF of NiCoCr medium-entropy alloy
6. Continuous comprehensive monitoring of melt pool morphology under realistic printing scenarios with laser powder bed fusion, 3
7. Local porosity prediction in metal powder bed fusion using in-situ thermography: a comparative study of machine learning techniques
8. Accurate detection of local porosity in laser powder bed fusion through deep learning of physics-based in-situ infrared camera signatures
9. S. Hinnebusch, D. Anderson, B. Bostan, A.C. To, In-Situ Infrared Camera Monitoring for Defect and Anomaly Detection in Laser Powder Bed Fusion: Calibration, Data Mapping, and Feature Extraction, (2024).
10. Predicting defects in laser powder bed fusion using in-situ thermal imaging data and machine learning
11. Defect prediction of localized regions in laser powder bed fusion based on multi-layer photodiode image fusion10.1016/j.measurement.2025.117413 ↗
12. Unveiling the layer-wise dynamics of defect evolution in laser powder bed fusion: Insights for in-situ monitoring and control
13. A study of modeling assumptions and adaptive remeshing for thermomechanical finite element modeling of the LPBF process10.1007/s00170-021-07362-2 ↗
14. Exploring the Flash Heating method in additive manufacturing for residual stress prediction: a comparative study with diffraction results from X-ray and neutron techniques
15. Experimental validation of thermo-mechanical part-scale modeling for laser powder bed fusion processes
16. A study of the influence of novel scan strategies on residual stress and microstructure of L-shaped LPBF IN718 samples10.1016/j.matdes.2022.110386 ↗
17. Accelerating high-fidelity thermal process simulation of laser powder bed fusion via the computational fluid dynamics imposed finite element method (CIFEM)
18. A new finite element model for welding heat sources10.1007/BF02667333 ↗
19. A review of multi-scale and multi-physics simulations of metal additive manufacturing processes with focus on modeling strategies
20. Enabling Part-Scale Scanwise process simulation for predicting melt pool variation in LPBF by combining GPU-based Matrix-free FEM and adaptive Remeshing
21. Digital twins for rapid in-situ qualification of part quality in laser powder bed fusion additive manufacturing
22. Toward the digital twin of additive manufacturing: Integrating thermal simulations, sensing, and analytics to detect process faults10.1080/24725854.2019.1701753 ↗
23. A spatiotemporal two-level method for high-fidelity thermal analysis of laser powder bed fusion10.1016/j.finel.2022.103815 ↗
24. A highly efficient computational approach for fast scan-resolved simulations of metal additive manufacturing processes on the scale of real parts
25. B. Williams, S. Cremaschi, Novel Tool for Selecting Surrogate Modeling Techniques for Surface Approximation10.1016/B978-0-323-88506-5.50071-1 ↗
26. Gaussian process-based surrogate modeling framework for process planning in laser powder-bed fusion additive manufacturing of 316L stainless steel10.1007/s00170-017-1045-z ↗
27. Multi-fidelity surrogate with heterogeneous input spaces for modeling melt pools in laser-directed energy deposition
28. Harnessing on-machine metrology data for prints with a surrogate model for laser powder directed energy deposition
29. A mean–variance estimation bidirectional convolutional long short-term memory surrogate model predicting residual stress and model error for laser powder bed fusion
30. Data-driven prediction of the high-dimensional thermal history in directed energy deposition processes via recurrent neural networks
31. A Real-Time Iterative Machine Learning Approach for Temperature Profile Prediction in Additive Manufacturing Processes
32. Data-driven modeling of thermal history in additive manufacturing
33. Thermal field prediction for welding paths in multi-layer gas metal arc welding-based additive manufacturing: a machine learning approach10.1016/j.jmapro.2021.02.033 ↗
34. Towards a generic physics-based machine learning model for geometry invariant thermal history prediction in additive manufacturing10.1016/j.jmatprotec.2021.117472 ↗
35. Multi-layer thermal simulation using physics-informed neural network
36. Machine learning for metal additive manufacturing: predicting temperature and melt pool fluid dynamics using physics-informed neural networks10.1007/s00466-020-01952-9 ↗
37. Single-track thermal analysis of laser powder bed fusion process: parametric solution through physics-informed neural networks10.1016/j.cma.2023.116019 ↗
38. Surrogate modeling of melt pool temperature field using deep learning
39. Time-efficient surrogate models of thermal modeling in laser powder bed fusion
40. Part-scale thermal process modeling for laser powder bed fusion with matrix-free method and GPU computing
41. Finding the limits of single-track deposition experiments:an experimental study of melt pool characterization in laser powder bed fusion10.1016/j.matdes.2023.112069 ↗
42. On the Evaluation of Sequential Machine Learning for Network Intrusion Detection
43. Effects of vector length on the melt pool morphology and grain structure characteristics of Alloy 718 in laser powder bed fusion10.1016/j.jmatprotec.2024.118317 ↗
44. Analysis of gradient vanishing of RNNs and performance comparison10.3390/info12110442 ↗
45. A review of recurrent neural networks: LSTM cells and network architectures10.1162/neco_a_01199 ↗
46. Comparing LSTM and GRU Models to Predict the Condition of a Pulp Paper Press
47. Vaswani · Atten. Is. All You Need · 2023
48. On the computational complexity of self-attention
49. A review on weight initialization strategies for neural networks10.1007/s10462-021-10033-z ↗
50. High-performance deep learning library
51. Finite element modellingand simulations of piezoelectric actuators responses with uncertainty quantification10.3390/computation6040060 ↗
52. Matplotlib: A 2D graphics environment10.1109/MCSE.2007.55 ↗
53. LSTM and GRU Neural Networks as Models of Dynamical Processes Used in Predictive Control: A Comparison of Models Developed for Two Chemical Reactors10.3390/s21165625 ↗
54. GRU and LSTM Comparison for Black-Box Modeling of Power Electronic Converters
55. Using LSTM and GRU neural network methods for traffic flow prediction
56. Energy Load Forecasting using Deep Learning Approach-LSTM and GRU in Spark Cluster
57. Multi-GPU acceleration of large-scale density-based topology optimization10.1016/j.advengsoft.2021.103006 ↗
58. Strategies to parallelize a finite element mesh truncation technique on multi-core and many-core architectures10.1007/s11227-022-04975-6 ↗
59. Multi-GPU Acceleration for Finite Element Analysis in Structural Mechanics10.3390/app15031095 ↗
60. Mills · 2002