PSI - Issue 22

Yang Ai et al. / Procedia Structural Integrity 22 (2019) 70–77

71

2

Y. Ai et al. / Structural Integrity Procedia 00 (2019) 000 – 000

1. Introduction In order to reduce cost and time-consuming of full-scale testing of engineering components, normally smooth and smaller specimens are designed for reality test. In general, the characterization of both notch and size effects is essential for a valid transfer from tested specimens with small and smooth geometry to full-scale components [1]. Furthermore, as a fairly empirical way, high safety coefficients has been commonly utilized for this transfer combined the experimental data and structural design [2, 3]. However, deterministic analysis is insufficient for fatigue analysis [4], because it is not sufficient to quantify the statistical uncertainties and material variability [5 – 8] and cannot explain well the origin of dispersions [9, 10]. Therefore, based on the stress gradient effect, size effect and uncertainty, a valid method which ensure the transfer from a limited size to a component level is still lacking and highly expected in reality. As one of the most commonly used approaches for fatigue strength evaluation taking into account the geometrical and statistical size effects [11 – 13], highly stressed volume (HSV) approach was originally developed by Kuguel [14]. The probability for crack initiation to start increases with the critical volume, which leads more easily to fatigue failure. In addition, Weibull distribution has been normally applied for prediction of fatigue strength and fatigue life distribution [15, 16]. Note from Castillo et al. [17, 18] that the Weibull model supported by experimental data were utilized to investigate the statistical size effect. With this logical reasoning and its general simplicity, the combination of HSV approach with Weibull model have become an adequate tool for reliability evaluation of engineering components. By combining the HSV approach with Weibull model, this paper develops a probabilistic procedure for modeling the notch fatigue and size effect of engineering components. Firstly, based on the HSV concept, a relation of local region with the same stress level is applied for transferring fatigue life distribution among specimens with different geometries. Then, a developed model based on Weibull model is performed in taking the influence of both statistical and geometrical size effects into account. Experimental data of two alloys are used for model validation. Finally, conclusions are summarized. Nomenclature , Maximum local stress %

Empirical parameter of scale σ , % Scale parameters of stress Stress concentration factor Maximum normal stress Stress ratio ( ) Cumulative distribution function of fatigue failure Shape parameter Scale parameter Predicted volume 0 Reference volume Model coefficient , , Parameter of model coefficient Empirical constant of model Number of loading cycles 0 Reference number of loading cycles