Stress concentration factors for bending with symmetric opposite notches in thin beam evaluated by FEM and ANN

Abstract In machine design, numerous tables are utilized, but only a few are used intensively. The development of scientific techniques and software has allowed frequently used tables to become more user-friendly. This study focuses on converting previously nondigitized experimental data into a more practical format using advanced computer technology and software. Fatigue is the primary cause of failure in machine parts. During the design phase, features like holes, channels, grooves, and protrusions disrupt the continuity of the part, and dimensionless part size parameters play a crucial role in the design process. This research investigates bending with symmetric opposite notches in a thin beam, based on data compiled by Peterson that had not been digitized before. Initially, the curves from the tables were digitized and organized while adhering to the original parameters. A parametric Finite Element solid model was created, and analyses were conducted using the obtained parametric values. Each curve from the original table was validated through regression methods using Finite Element results. Furthermore, an Artificial Neural Network (ANN) model was developed based on the original curve data. The original curves, finite element results, and ANN model were compared, and a new model proposal was made.