PSI - Issue 57

Mohammad F. Tamimi et al. / Procedia Structural Integrity 57 (2024) 121–132 129 Mohammad F. Tamimi & Mohamed Soliman/ Structural Integrity Procedia 00 (2023) 000 – 000 9

5.2. Sensitivity Analysis of the Fatigue Service Life

A sensitivity analysis, using the ANN-based simulation model, is conducted to identify key parameters impacting the fatigue service life of stiffened panels. The analysis considers several geometric parameters including the main panel thickness, stiffener characteristics, and spacing between stiffeners. The parameters and their descriptors are summarized in Table 2. The number of cycles required for the crack to propagate across the panel is used as the response parameter (fatigue service life). A variance-based sensitivity is conducted to assess the effects of the variability in these parameters on the fatigue service life are evaluated. The UQ-Lab MATLAB toolbox (Marelli & Sudret, 2014) computes Sobol’s sensitivity indices for these parameters. The effect of variability in each input parameter on the variability of the fatigue service life is analyzed, with results presented in Figure 8. It's shown that the main panel thickness is a key influence on fatigue service life variability for both stiffener types, contributing approximately 49% to total variance. Following this, stiffener web properties (height and thickness) have a significant effect, then the spacing between stiffeners. Flange properties (width and thickness) contribute minimally, with indices less than 3%. Table 2. The statistical descriptors of the considered random variables.

Variable

Unit mm mm mm mm mm mm

Mean

COV 0.019 0.019 0.05 0.05 0.03 0.05

Distribution

Reference

Stiffeners Height Flange Width

101

Normal Normal Normal Normal Normal Normal

(Hess et al., 2002) (Hess et al., 2002) (Hess et al., 2002) (Hess et al., 2002) (Hess et al., 2002) (Hess et al., 2002)

76

Stiffener Thickness (Web) Stiffener Thickness (Flange)

8 8

Stiffener Spacing Main Panel Thickness

381

13

Fig. 8. The total sensitivity indices of the individual input parameters associated with the geometrical parameters for the investigated stiffened panels. 6. Conclusions This paper quantifies the effect of different sources of uncertainty on the crack propagation behaviour in welded stiffened panels. It employed an integrated approach that combines 3D FE analysis, ANN, and an analytical crack growth model to quantify the fatigue service life. The developed ANN-assisted simulation approach was capable of accurately predicting the fatigue crack growth behaviour in welded stiffened panels tested experimentally in literature. The effect of the variability in the geometric parameters of the stiffened panel on the crack propagation behaviour is