Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
166 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
42 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

GrainGNN: A dynamic graph neural network for predicting 3D grain microstructure (2401.03661v2)

Published 8 Jan 2024 in cs.CE

Abstract: We propose GrainGNN, a surrogate model for the evolution of polycrystalline grain structure under rapid solidification conditions in metal additive manufacturing. High fidelity simulations of solidification microstructures are typically performed using multicomponent partial differential equations (PDEs) with moving interfaces. The inherent randomness of the PDE initial conditions (grain seeds) necessitates ensemble simulations to predict microstructure statistics, e.g., grain size, aspect ratio, and crystallographic orientation. Currently such ensemble simulations are prohibitively expensive and surrogates are necessary. In GrainGNN, we use a dynamic graph to represent interface motion and topological changes due to grain coarsening. We use a reduced representation of the microstructure using hand-crafted features; we combine pattern finding and altering graph algorithms with two neural networks, a classifier (for topological changes) and a regressor (for interface motion). Both networks have an encoder-decoder architecture; the encoder has a multi-layer transformer long-short-term-memory architecture; the decoder is a single layer perceptron. We evaluate GrainGNN by comparing it to high-fidelity phase field simulations for in-distribution and out-of-distribution grain configurations for solidification under laser power bed fusion conditions. GrainGNN results in 80\%--90\% pointwise accuracy; and nearly identical distributions of scalar quantities of interest (QoI) between phase field and GrainGNN simulations compared using Kolmogorov-Smirnov test. GrainGNN's inference speedup (PyTorch on single x86 CPU) over a high-fidelity phase field simulation (CUDA on a single NVIDIA A100 GPU) is 150$\times$--2000$\times$ for 100-initial grain problem. Further, using GrainGNN, we model the formation of 11,600 grains in 220 seconds on a single CPU core.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (64)
  1. Smith, J. et al. Linking process, structure, property, and performance for metal-based additive manufacturing: computational approaches with experimental support. Computational Mechanics 57, 583–610 (2016).
  2. A generalized field method for multiphase transformations using interface fields. Physica D: Nonlinear Phenomena 134, 385–393 (1999).
  3. A quantitative multi-phase field model of polycrystalline alloy solidification. Acta Materialia 58, 2155–2164 (2010).
  4. The significance of spatial length scales and solute segregation in strengthening rapid solidification microstructures of 316L stainless steel. Acta Materialia 184, 1–16 (2020).
  5. Phase-field modeling of grain evolutions in additive manufacturing from nucleation, growth, to coarsening. npj Computational Materials 7, 1–12 (2021).
  6. The development of grain structure during additive manufacturing. Acta Materialia 211, 116862 (2021).
  7. A 3D cellular automaton algorithm for the prediction of dendritic grain growth. Acta Materialia 45, 2187–2195 (1997).
  8. Understanding Uncertainty in Microstructure Evolution and Constitutive Properties in Additive Process Modeling. Metals 12, 324 (2022).
  9. Computational Materials Science 214, 111692 (2022).
  10. Simulation of metal additive manufacturing microstructures using kinetic Monte Carlo. Computational Materials Science 135, 78–89 (2017).
  11. Miyoshi, E. et al. Ultra-large-scale phase-field simulation study of ideal grain growth. NPJ Computational Materials 3, 1–6 (2017).
  12. Miyoshi, E. et al. Large-scale phase-field study of anisotropic grain growth: Effects of misorientation-dependent grain boundary energy and mobility. Computational Materials Science 186, 109992 (2021).
  13. Effect of strong nonuniformity in grain boundary energy on 3-D grain growth behavior: A phase-field simulation study. Computational Materials Science 127, 67–77 (2017).
  14. Recrystallization and grain growth. Progress in metal physics 3, 220–292 (1952).
  15. Hillert, M. On the theory of normal and abnormal grain growth. Acta metallurgica 13, 227–238 (1965).
  16. On misorientation distribution evolution during anisotropic grain growth. Acta Materialia 49, 2981–2991 (2001).
  17. GrainNN: A neighbor-aware long short-term memory network for predicting microstructure evolution during polycrystalline grain formation. Computational Materials Science 218, 111927 (2023).
  18. Topological changes in coarsening networks. Acta Materialia 130, 147–154 (2017).
  19. Numerical analysis of the vertex models for simulating grain boundary networks. SIAM Journal on Applied Mathematics 75, 762–786 (2015).
  20. Advanced computational vibroacoustics: reduced-order models and uncertainty quantification (Cambridge University Press, 2014).
  21. Surrogate and reduced-order modeling: a comparison of approaches for large-scale statistical inverse problems. Large-Scale Inverse Problems and Quantification of Uncertainty 123–149 (2010).
  22. Airfoil design optimization using reduced order models based on proper orthogonal decomposition. In Fluids 2000 conference and exhibit, 2545 (2000).
  23. Reduced-order modelling for flow control, vol. 528 (Springer Science & Business Media, 2011).
  24. Li, X. et al. A transfer learning approach for microstructure reconstruction and structure-property predictions. Scientific reports 8, 1–13 (2018).
  25. Bostanabad, R. Reconstruction of 3D microstructures from 2D images via transfer learning. Computer-Aided Design 128, 102906 (2020).
  26. Goodfellow, I. et al. Generative adversarial nets. Advances in neural information processing systems 27 (2014).
  27. Yang, Z. et al. Microstructural materials design via deep adversarial learning methodology. Journal of Mechanical Design 140 (2018).
  28. Lee, X. Y. et al. Fast inverse design of microstructures via generative invariance networks. Nature Computational Science 1, 229–238 (2021).
  29. Long short-term memory. Neural computation 9, 1735–1780 (1997).
  30. Accelerating phase-field-based microstructure evolution predictions via surrogate models trained by machine learning methods. npj Computational Materials 7, 3 (2021).
  31. Accelerating phase-field predictions via recurrent neural networks learning the microstructure evolution in latent space. Computer Methods in Applied Mechanics and Engineering 397, 115128 (2022).
  32. Yang, K. et al. Self-supervised learning and prediction of microstructure evolution with convolutional recurrent neural networks. Patterns 2, 100243 (2021).
  33. Learning two-phase microstructure evolution using neural operators and autoencoder architectures. npj Computational Materials 8, 190 (2022).
  34. Learning nonlinear operators via deeponet based on the universal approximation theorem of operators. Nature machine intelligence 3, 218–229 (2021).
  35. A generalized vertex dynamics model for grain growth in three dimensions. Modelling and Simulation in Materials Science and Engineering 18, 015010 (2009).
  36. Vertex models for two-dimensional grain growth. Philosophical Magazine B 60, 399–421 (1989).
  37. Three-dimensional microstructural evolution in ideal grain growth—general statistics. Acta Materialia 48, 1297–1311 (2000).
  38. Physics-embedded graph network for accelerating phase-field simulation of microstructure evolution in additive manufacturing. npj Computational Materials 8, 201 (2022).
  39. Graph neural networks for an accurate and interpretable prediction of the properties of polycrystalline materials. npj Computational Materials 7, 103 (2021).
  40. Graph neural networks for efficient learning of mechanical properties of polycrystals. Computational Materials Science 217, 111894 (2023).
  41. Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907 (2016).
  42. The graph neural network model. IEEE transactions on neural networks 20, 61–80 (2008).
  43. Morris, C. et al. Weisfeiler and leman go neural: Higher-order graph neural networks. In Proceedings of the AAAI conference on artificial intelligence, vol. 33, 4602–4609 (2019).
  44. Wieder, O. et al. A compact review of molecular property prediction with graph neural networks. Drug Discovery Today: Technologies 37, 1–12 (2020).
  45. Duvenaud, D. K. et al. Convolutional networks on graphs for learning molecular fingerprints. Advances in neural information processing systems 28 (2015).
  46. Neural message passing for quantum chemistry. In International conference on machine learning, 1263–1272 (PMLR, 2017).
  47. Crystal graph convolutional neural networks for an accurate and interpretable prediction of material properties. Physical review letters 120, 145301 (2018).
  48. GC-LSTM: Graph convolution embedded LSTM for dynamic network link prediction. Applied Intelligence 1–16 (2022).
  49. Pareja, A. et al. Evolvegcn: Evolving graph convolutional networks for dynamic graphs. In Proceedings of the AAAI conference on artificial intelligence, vol. 34, 5363–5370 (2020).
  50. dyngraph2vec: Capturing network dynamics using dynamic graph representation learning. Knowledge-Based Systems 187, 104816 (2020).
  51. Vaswani, A. et al. Attention is all you need. In Advances in neural information processing systems, 5998–6008 (2017).
  52. Pinomaa, T. et al. Process-Structure-Properties-Performance modeling for selective laser melting. Metals 9, 1138 (2019).
  53. Dendrite-resolved, full-melt-pool phase-field simulations to reveal non-steady-state effects and to test an approximate model. Computational Materials Science 207, 111262 (2022).
  54. Growth competition of columnar dendritic grains: A phase-field study. Acta Materialia 82, 64–83 (2015).
  55. Takaki, T. et al. Competitive grain growth during directional solidification of a polycrystalline binary alloy: Three-dimensional large-scale phase-field study. Materialia 1, 104–113 (2018).
  56. Linking phase-field and atomistic simulations to model dendritic solidification in highly undercooled melts. Interface Science 10, 121–136 (2002).
  57. Shi, Y. et al. Masked label prediction: Unified message passing model for semi-supervised classification. arXiv preprint arXiv:2009.03509 (2020).
  58. Efficient numerical algorithm for multiphase field simulations. Physical Review E 73, 016703 (2006).
  59. Phase-field simulation of the columnar-to-equiaxed transition in alloy solidification. Acta Materialia 54, 2015–2026 (2006).
  60. Pinomaa, T. et al. Phase field modeling of rapid resolidification of Al-Cu thin films. Journal of Crystal Growth 532 (2020).
  61. Three-Dimensional Numerical Simulation of Grain Growth during Selective Laser Melting of 316L Stainless Steel. Materials 15, 6800 (2022).
  62. Zhang, Z. Improved adam optimizer for deep neural networks. In 2018 IEEE/ACM 26th International Symposium on Quality of Service (IWQoS), 1–2 (Ieee, 2018).
  63. Active learning in materials science with emphasis on adaptive sampling using uncertainties for targeted design. npj Computational Materials 5, 1–17 (2019).
  64. Neural network ensembles, cross validation, and active learning. Advances in neural information processing systems 7 (1994).

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now