Structural Engineering and Mechanics

Techno-Press (테크노프레스)
권호 표지
DOI QR코드

DOI QR Code

Structural analysis and design using generative AI

  • Moonsu Park (Korea Atomic Energy Research Institute) ;
  • Gyeongeun Bong (Department of Mechanical Engineering, Korean Advanced Institute for Science and Technology) ;
  • Jungro Kim (Department of Mechanical Engineering, Korean Advanced Institute for Science and Technology) ;
  • Gihwan Kim (Korea Atomic Energy Research Institute)
  • Received : 2024.06.21
  • Accepted : 2024.08.05
  • Published : 2024.08.25
⟨ Previous Next ⟩

Abstract

This study explores the integration of the generative AI, specifically ChatGPT (GPT-4o), into the field of structural analysis and design using the finite element method (FEM). The research is conducted in two main parts: structural analysis and structural design. For structural analysis, two scenarios are examined: one where the FEM source code is provided to ChatGPT and one where it is not. The AI's ability to understand, process, and accurately perform finite element analysis in both scenarios is evaluated. Additionally, the application of ChatGPT in structural design is investigated, including design modifications and parameter sensitivity analysis. The results demonstrate the potential of the generative AI to assist in complex engineering tasks, suggesting a future where AI significantly enhances efficiency and innovation in structural engineering. However, the study also highlights the importance of ensuring the accuracy and reliability of AI-generated results, particularly in safety-critical applications.

Keywords

Acknowledgement

We extend our heartfelt gratitude to Prof. Phill-Seung Lee of the Korea Advanced Institute of Science and Technology (KAIST) for providing valuable insights that greatly contributed to this research.

References

  1. Bathe, K.J. (2006), Finite Element Procedures, Prentice Hall, New York.
  2. Beck, A., Flad, D. and Munz, C.D. (2019), "Deep neural networks for data-driven LES closure models", J. Comput. Phys., 398, 108910. https://doi.org/10.1016/j.jcp.2019.108910.
  3. Choi, H.G. and Lee, P.S. (2023), "Towards improving the 2DMITC4 element for analysis of plane stress and strain problems", Comput. Struct., 275, 106933. https://doi.org/10.1016/j.compstruc.2022.106933.
  4. Guo, X., Li, W. and Iorio, F. (2016), "Convolutional neural networks for steady flow approximation", Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, New York, NY, August.
  5. Hughes, T.J.R. (2012), The Finite Element Method: Linear Static and Dynamic Finite Element Analysis, Dover Publications, Mineola, New York.
  6. Jung, H., Lee, C. and Lee, P.S. (2023), "Strain-smoothed polygonal finite elements", Struct. Eng. Mech., 86(3), 311-324. https://doi.org/10.12989/sem.2023.86.3.311.
  7. Jung, J., Jun, H. and Lee, P.S. (2022), "Self-updated four-node finite element using deep learning", Comput. Mech., 69(1), 23-44. https://doi.org/10.1007/s00466-021-02081-7.
  8. Jung, J., Yoon, K. and Lee, P.S. (2020), "Deep learned finite elements", Comput. Meth. Appl. Mech. Eng., 372, 113401. https://doi.org/10.1016/j.cma.2020.113401.
  9. Kim, S. and Lee, P.S. (2018), "A new enriched 4-node 2D solid finite element free from the linear dependence problem", Comput. Struct., 202, 25-43. https://doi.org/10.1016/j.compstruc.2018.03.001.
  10. Ko, Y., Lee, P.S. and Bathe, K.J. (2016), "The MITC4+ shell element and its performance", Comput. Struct., 169, 57-68. https://doi.org/10.1016/j.compstruc.2016.03.002.
  11. Kochkov, D., Smith, J.A., Alieva, A., Wang, Q., Brenner, M.P. and Hoyer, S. (2021), "Machine learning methods for data-driven turbulence modeling", Proceedings of the National Academy of Sciences, USA.
  12. Lee, C. and Lee, P.S. (2018), "A new strain smoothing method for triangular and tetrahedral finite elements", Comput. Meth. Appl. Mech. Eng., 341, 939-955. https://doi.org/10.1016/j.cma.2018.07.022.
  13. Lee, Y., Lee, P.S. and Bathe, K.J. (2014), "The MITC3+ shell element and its performance", Comput. Struct., 138, 12-23. https://doi.org/10.1016/j.compstruc.2014.02.005.
  14. Ling, J., Kurzawski, A. and Templeton, J. (2016), "Reynolds averaged turbulence modelling using deep neural networks with embedded invariance", J. Fluid Mech., 807, 155-166. https://doi.org/10.1017/jfm.2016.615.
  15. Raissi, M. (2018), "Deep hidden physics models: Deep learning of nonlinear partial differential equations", J. Mach. Learn Res., 19(25), 1-24.
  16. Raissi, M., Perdikaris, P. and Karniadakis, G.E. (2019), "Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations", J. Comput. Phys., 378, 686-707. https://doi.org/10.1016/j.jcp.2018.10.045.
  17. Sirignano, J. and Spiliopoulos, K. (2018), "DGM: A deep learning algorithm for solving partial differential equations", J. Comput. Phys., 375, 1339-1364. https://doi.org/10.1016/j.jcp.2018.08.029.
  18. Tompson, J., Schlachter, K., Sprechmann, P. and Perlin, K. (2017), "Accelerating eulerian fluid simulation with convolutional networks", Proceedings of the 34th 741 International Conference on Machine Learning, Sydney, Australia, July.
  19. Zhang, Z.J. and Duraisamy, K. (2015), "Machine learning methods for data-driven turbulence modeling", In Proceedings of the 22nd AIAA Computational Fluid Dynamics Conference, Dallas, TX, USA, June.
  20. Zienkiewicz, O.C. and Taylor, R.L. (2000), The Finite Element Method: Solid Mechanics, Vol. 2, Butterworth-Heinemann.