Estimation of Stress Concentration Factor Using Artificial Neural Networks in T-Weld Joints Forced by Bending

Abstract

Stress concentration factor (SCF) is a critical parameter in engineering design and structural analysis and plays an important role in determining the durability and safety of structures. In this study, a data set consisting of 8500 unique data points covering a wide range of geometric structures and parameters was created with the Latin Hypercube method in order to calculate SCF values with a parametric equation. The generated input data was analysed in finite element software by writing an original script recommended as a result of this study, and the resulting data was trained with an artificial neural network. The new parametric equation created at the end of the study has an average error of 4.95%. As a result, in this study, the effect of welding geometry parameters in T-welded joint SCF applications was examined and a parametric equation created for ease of calculation and accuracy was proposed. It is also recommended to apply SCF calculations with the FEA script and the original script used in this study.

Keywords

• finite element analysis
• full automation scripts
• stress concentration factor
• welded joints