PSI - Issue 79

Abubakr E.S. Musa et al. / Procedia Structural Integrity 79 (2026) 206–216

210

2.4. Experimental results Table 2 summarizes the results of the experimental tests on the steel tubes with varying levels of localized corrosion, quantified by the parameter ∆ t/t, which is the reduction in thickness divided by the thickness. Two specimens were tested for each level, and their maximum loads (P max ) were recorded. The table also presents the average P max for each imperfection level, revealing a decreasing trend in the load capacity with increasing thickness reduction; for instance, a 2.0 mm reduction in thickness (T4 specimen) resulted in a 22% decrease in the maximum load compared to the reference specimen (T1). Table 2 Specimens details No. Specimen Designation ∆ t/t P max (kN) Average P max (kN) 1 T1-1 0.0 327.0 322.5 2 T1-2 318.0 3 T2-1 0.33 259.0 271.0 4 T2-2 283.0 5 T3-1 0.50 252.0 253.5 6 T3-2 255.0 7 T4-1 0.67 229.0 250.0 8 T4-2 271.0 3. Finite element simulation To ensure accurate FE analysis, it is crucial to account for two distinct types of imperfections relevant to the current problem. These are initial imperfections and intended imperfections introduced to simulate thickness reduction caused by corrosion. The developed FE models should explicitly incorporate both types to capture the structural behavior realistically and provide reliable results. These imperfections are discussed in detail in the following sections. In the present FE analysis, a mesh with 100 divisions along the tube's circumference and 100 along its longitudinal axis was employed, resulting in 10,100 nodes and 10,000 elements. A mesh sensitivity study was conducted, confirming that this discretization is sufficient to achieve a converged solution. 3.1. Initial Imperfections The initial imperfections refer to the inherent irregularities present in the cylindrical tube before introducing thickness reduction imperfections. These include out-of-roundness, thickness variations, uneven cylinder cuts, variations in weld depressions, and residual stresses, among other factors. Additionally, eccentric loading is another critical imperfection, though it is not geometric in nature. These imperfections can significantly diminish the tube's load-carrying capacity, making it essential to incorporate them into the FE model of tested tubes to avoid overestimating their predicted performance using FE analysis. Various researchers have investigated the effect of initial imperfections by introducing equivalent geometric imperfections into FE models (Ashraf et al. (2006), Gardner and Nethercot (2004), Zhao et al. (2015a, 2015b)). Typically, these imperfections are modeled using eigenmodes derived from linear buckling analysis, scaled by a factor ranging between 0.1t and 0.5t (where t is the tube thickness) (Gardner and Nethercot (2004)). In some cases, varying thicknesses are employed for calibration purposes to improve accuracy. However, in most of these studies, no intended imperfections are incorporated, such as the thickness reduction introduced to simulate corrosion damage in the present study. This unique aspect of the present study motivates the authors to explore alternative approaches to address both initial and intended imperfections effectively.