PSI - Issue 57

Mohammad F. Tamimi et al. / Procedia Structural Integrity 57 (2024) 121–132 Mohammad F. Tamimi & Mohamed Soliman/ Structural Integrity Procedia 00 (2023) 000 – 000 2

122

1. Introduction Marine structures are susceptible to various deterioration mechanisms due to their exposure to the corrosive marine environmental conditions and fluctuating wave loads. The initiation and propagation of fatigue cracks through the hull can eventually lead to catastrophic failure. In structures employing welded stiffened panels, the stress conditions created by the welded stiffeners induce complex stress conditions that may add challenges to the crack growth prediction process. These structures are also subjected to variable amplitude loading leading to sequence and interaction effects which cannot be ignored. Additionally, as the cracks propagate, they pass through the main plate, welds and stiffeners, leading to multiple crack tips, each with its own stress conditions (Murthy et al., 2007). Research focusing on crack propagation in stiffened panels dates back to the 1960s. Poe (1969, 1971) conducted studies that revealed that the fatigue crack growth rates in panels with riveted stiffeners is lower than that of unstiffened panels. Studies by Nussbaumer et al. (1999), Dexter and Pilarski (2002), and Mahmoud and Dexter (2005) also presented a series of experimental investigations to characterize the crack propagation behavior in welded stiffened panels with different stiffener configurations and load patterns. Despite these insightful studies, the high costs and long durations of experimental testing limit the parameters and number of test specimens that can be investigated, making it a challenge to thoroughly understand the effect of various sources of uncertainty on the crack propagation behavior (Feng et al., 2012). As such, probabilistic approaches have become increasingly important in evaluating the fatigue performance of structures with stiffened panels. Probabilistic fatigue assessment approaches have been particularly useful in formulating optimal design configurations and planning for inspection, maintenance, and repair activities. Notably, researchers such as Feng et al. (2012); Huang et al. (2013); Mahmoud and Riveros (2014); Dong et al. (2018); Tamimi et al. (2023) have developed probabilistic approaches to evaluate the fatigue reliability of stiffened panels in the presence of growing cracks. However, these studies have primarily focused on the uncertainties associated with loads, mechanical properties, and initial crack sizes, with less attention being given to uncertainties related to geometric parameters. Variations between designed and as-built dimensions due to manufacturing tolerances and welding processes can lead to uncertainties in geometric properties (Caiazzo et al., 2017; Hess et al., 2002), which may significantly influence the crack propagation process (Sankararaman et al., 2011). To address this need, this paper quantifies the impact of uncertainties in the input geometric parameters on the crack propagation behavior and fatigue service life of welded stiffened panels. Three-dimensional (3D) finite element (FE) analysis, artificial neural networks (ANNs), and an elastic-plastic crack advancement rule are leveraged to predict the crack growth behavior and fatigue service life. Variance-based sensitivity analysis is conducted to quantify the influence of these parameters on the variability in the fatigue service life.