E-Archive
Shot Peening in the Automotive Industry
in Vol. 25 - November Issue - Year 2024
Shot Peening, Material Defects and Machine Learning
Mario Guagliano
Traditional mechanical design approaches are based on simple theoretical formulas, adjusted with empirical coefficients to suit specific cases. For example, shafts are treated as beams and the maximum critical stress is calculated by multiplying the so called "nominal stress" (calculated using theoretical formulas) by the "theoretical stress concentration factor", a function of the severity of the notch and of the stress concentration induced by the geometric detail. This factor can be found in tables and textbooks for common cases, but for new or unique designs, it must be determined either experimentally or through adaptation from similar situations when experiments are too costly. This approach has shifted with the advent of finite element analysis (FEA), which allows for accurate stress calculations without the need for complex experiments. Finite element methods have led to more refined, analysis-based design approaches, moving away from reliance on simplified formulas. However, even with FEA, materials are often treated as homogeneous, continuous, and defect-free. Apart from the notch effect, in the context of fatigue, traditional design methods rely heavily on empirical coefficients - such as surface factors, size factors, and fatigue notch sensitivity - which considers the effects of complex phenomena on material and component fatigue behavior. Despite this, computational methods still assume a perfect, continuous material.
But materials have defects, the impact of which depends on size, shape, service conditions, environment, and material sensitivity. Surface defects, inclusions and voids, for instance, can significantly affect fatigue strength by acting as pre-existing cracks and providing a starting point for crack propagation. The importance of these defects grows with service life as other failure mechanisms only come into play at higher stress levels. This is particularly true for surface-hardened materials, where the surface is no longer the most critical area.
Developing design approaches that account for these defects is challenging and must be grounded on Fracture Mechanics, which studies how defects propagate. A methodology called "Defect Tolerant Design" has been developed to predict whether a defect poses a risk. While applying this method requires thorough statistical analysis of defect distributions (including their size and frequency), it has been successfully implemented in industries like automotive, where components such as springs and bearings are designed with material defects in mind.
So, what role does shot peening play in this context? Even with defects present, shot peening significantly enhances fatigue strength. Since defects act as pre-existing cracks, the residual stresses induced by shot peening help to prevent crack propagation. This is particularly evident when surface defects are considered, as studies have shown that shot peening can increase the fatigue limit by more than 50% on parts with known defect sizes. For instance, rotating fatigue tests have demonstrated that the fatigue limit of gas-nitrided steel with artificially induced surface defects (created by controlled electro-erosion) was doubled after shot peening, and similar conclusions can be reached in other cases, after an appropriate experimental campaign.
But, in the Age 4.0, with the tools offered by the digital transformation of industry, something more could be achieved. Indeed, the choice of the optimal shot peening parameters is not trivial and depends on many factors, such as the stress state, the statistical analysis of the defect rate and its dimension, the materials, and so on. How about developing a Machine Learning tool able to determine the optimal peening conditions for each single case of interest?
A reasonable amount of data used for training should be available, but I think this should not be a problem. Then, the tool could be used for the analysis of many different cases, leading to an optimized choice of the peening parameters and shortening the time required for the design and application of shot peening.
Contributing Editor MFN and
Full Professor of Technical University of Milan
20156 Milan, Italy
E-mail: mario@mfn.li