Loading...

Effect of Initial Surface Roughness on the Residual Stress Profile Induced by Shot Peening Process using Finite Element Method and Machine Learning Algorithms

Nazemi Azad, Farbod | 2023

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57293 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Farrahi, Gholamhossein; Chamani, Mohammad
  7. Abstract:
  8. Shot peening process, which induces compressive residual stress near the surface, is one of the most effective methods for enhancing surface properties. Shot peening on smooth surfaces increases surface roughness, which is a factor that can reduce fatigue life. Conversely, the distribution of compressive residual stress contributes to improved fatigue life. Therefore, examining the initial surface conditions and the effects of average surface roughness on residual stress distribution is essential.In this study, due to the limited attention given to initial surface roughness in previous research, a finite element model of the shot peening process was developed considering the actual topography of the rough surface. This numerical model includes the selection of an appropriate material model due to the presence of cyclic plasticity and strain rate effects, the rough surface profile of the target component, and the random impact of the shot particles. To extract the actual topography of the rough surface, a laser profilometer was used for machined parts. The validation of the numerical results was carried out using previous studies in the area of residual stress.Subsequently, the effects of impact velocity and shot diameter, surface coverage percentage, contact parameter (friction coefficient), and initial surface roughness were investigated. The results of the finite element model indicated that with an increase in the velocity and diameter of the shots, as well as within a specific range of surface coverage, both the depth of the compressive layer and the maximum compressive residual stress are increasing. The effect of the friction coefficient became negligible beyond a certain range due to insignificant changes in the distribution of residual stress.Regarding the effect of initial surface roughness, which was analyzed for smooth surfaces and average roughness values of 9.7, 16, 22.1, and 35.4 microns, the results showed that with an increase in surface roughness, the maximum compressive residual stress decreased by approximately 18.7%, the depth of maximum compressive residual stress reduced by 40%, and the depth of the compressive layer decreased by 10%. Finally, using the results from the finite element model for different surface conditions, four different machine learning models—Artificial Neural Network, Support Vector Regression, K-Nearest Neighbors, and Random Forest Algorithm—were trained to predict the distribution of residual stress. The performance of each model was compared based on various error metrics such as root mean square error, mean absolute error, and coefficient of determination. Among these methods, the Random Forest model exhibited the highest coefficient of determination (96.91%) on the test data, while the Artificial Neural Network model had the lowest (92.74%)
  9. Keywords:
  10. Shot Peening ; Finite Element Method ; Residual Stress ; Surface Treatment ; Machine Learning ; Initial Surface Roughness

 Digital Object List

 Bookmark