Determination of stress concentration factors for shafts under tension

ÖZKAN M. T. , Erdemir F.

MATERIALS TESTING, vol.62, no.4, pp.413-421, 2020 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 62 Issue: 4
  • Publication Date: 2020
  • Doi Number: 10.3139/120.111500
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex
  • Page Numbers: pp.413-421
  • Keywords: Stress concentration factor, tension, finite element analysis, artifidal neural network, SHOULDERED SHAFTS, COLLAR
  • Gazi University Affiliated: Yes


Computer-based design and optimization have become increasingly important in recent years. This paper has investigated the stress concentration factors (SCF) K-t for shoulder filleted shafts with a hole and without a hole. This study contains two types of shoulder filleted shafts, i. e., a stepped bar of circular cross section with shoulder filleted and a tube with filleted shafts under tension stresses. Investigations on SCF that have been carried out in experimental and theoretical studies, were updated and validated for 2 types of shafts. The charts have been converted into numerical value using high precision computer techniques. Dimensional ratios and SCF were determined using previous work charts. This study determines maximum stresses for shoulder filleted shafts by three dimensional finite element analysis (FEA) and artificial intelligence techniques. A set of SCF charts was converted into numerical values and this data was organized and stored in an Excel file. ANSYS models were created and applied the boundary conditions on the models. And also mesh optimizations were performed. Artificial neural networks (ANN) models were designed using previously collected and verified data. Previous works, ANSYS and ANN results were compared to each other. As a result, ANN model and chart results show a good agreement. The usage of ANN model does not require any mathematical formulae or converting the numerical data action for determining the K-t result for shafts. ANN model usage was identified as a very useful and practical method.