Integration Of Artificial Neural Network And Finite Element Method For Prediction Of Elastic-Plastic Deformation Behaviour Near Crack Tips

• Author(s) / Creator(s)

  • Mortazavi, S.N.S.
  • Ince, A.

• It has been widely accepted that stress and strain fields near the crack tip govern crack propagation behavior. In most cases, an elasto-plastic analysis is required to determine local stress and strain fields around the crack tip due to the high stress concentration. However, complexities of such analysis lead researchers to either employ modified elastic analyses to approximately address stress/strain fields near the crack tip, or consider cracks as micro notches and use the Neuber rule as an approximation method to estimate elasto- plastic stress/strain fields from elastic stress/strain fields in the vicinity of micro notch tips. Unfortunately, both approaches have limitations to provide generalized solutions. The present work aims to develop robust artificial neural network (ANN) models to obtain elasto-plastic stress, strain, and displacement fields near the crack tips by means of a numerical elastic solution rather than a complex elasto-plastic solution. In order to do so, two separate finite element models (FEMs) are implemented to analyze a cracked specimen, made of stainless steel (SS304), under mode (I) of loading in both elastic and elasto-plastic states. ANN models are developed to learn the relationship between elastic and elasto-plastic behavior of the material in the presence of cracks. The elastic and elasto- plastic FEMs are employed to generate the input and output numerical data, respectively, to train and validate the constructed ANN models. The results show that well-trained ANN models can efficiently and accurately predict the elasto-plastic stress, strain, and displacement fields around the crack tips on the basis of numerical elastic finite element solution under monotonic loading conditions.

  Part of Proceedings of the Canadian Society for Mechanical Engineering International Congress 2022.

• Date created

  2022-06-01

• Subjects / Keywords

  • finite element analysis
  • artificial neural network
  • stress
  • strain
  • displacement prediction

• Type of Item

  Article (Published)

• DOI

  https://doi.org/10.7939/r3-2q0x-yv88

• License

  Attribution-NonCommercial 4.0 International