Fatigue initiation from short cracks in thick welded steels

Abstract

In the shipbuilding industry, there is a growing trend toward larger ships, which has led to the increased use of thicker, high-strength steels in the construction of ship and marine structures. This shift has raised concerns about potential failures due to material imperfections arising from welding, external loading, fatigue, and corrosion. Such failures could result in catastrophic environmental pollution or loss of life. While the fracture mechanics and fatigue properties of shipbuilding materials have been widely studied through both experimental and numerical methods, research on thick steel plates, particularly with respect to short flaws, remains limited. To address this gap, both experimental and numerical approaches have been developed to investigate the effects of short cracks on the fracture and fatigue crack growth properties of EH47 and X65 steels, which are commonly used in the shipbuilding and oil and gas industries, respectively. A thick, flat welded plate made from EH47 steel and an X65 steel pipe were selected as the materials for the testing program. SENB specimens of the type Bx2B were manufactured and notched in two different ways: half of the specimens had full-thickness notches, while the remaining specimens had semi-elliptical notches. These specimens were used to conduct fatigue crack growth tests using both a constant loading method (K-increasing) and a K-decreasing method (load-shedding). The surface crack length was measured using the direct current potential drop (DCPD) method. Calibration tests were performed to correlate crack length with potential drop for the semi-elliptical cracks. For the full-thickness notches, the DCPD method was previously calibrated, and the threshold stress intensity factors (ΔKth) were experimentally determined for specimens with semi-elliptical cracks of varying surface crack depths (a-value). The experimental results were compared with the analytical solutions documented in BS7910, and the feasibility of the experiments was verified. It was found that the surface crack length had a significant influence on ΔKth: the longer the surface crack, the larger the ΔKth. However, as the crack width (c-value) approached infinity (i.e., when the semi-elliptical crack progresses into a full-thickness crack), ΔKth decreased instead. Based on the experimental results, numerical models were developed to calculate key fracture mechanical parameters, including stress intensity factors (SIFs), J-integral, T-stress, and plastic zone size (PLZ). These models were verified against empirical solutions, and a theoretical procedure to predict the plastic zone size was derived and calibrated with the numerical results. The calculated fracture parameters were then used to derive fatigue curves and provided a theoretical foundation for explaining the observed fatigue phenomena. Furthermore, numerical models were developed to simulate the fatigue crack growth process, and a variable amplitude loading procedure was generated using Python scripts and the UAMP subroutine. The finite element analysis (FEA) of the fatigue crack growth process was validated against experimental results, with the calculated fatigue crack propagation showing a maximum 10% deviation from the corresponding experiments. This provides a reliable basis for the numerical models in predicting experimental principles and analyzing fatigue behavior in the research. Using the calibrated numerical models, parametric studies were conducted to examine the effects of various parameters, such as the R-ratio, loading methods, and constraint effects, on fatigue crack growth rates (FCGRs). Moreover, the numerically obtained effects of surface crack length on the ΔKth of SENB specimens showed good agreement with the experimental findings. Finally, efforts were made to understand the effects of residual stress on fatigue crack growth rates (FCGRs) using numerical methods. The numerical simulations were validated by comparison with experimental data from the literature and theoretical models outlined in BS7910. Simplified residual stress profiles were incorporated, and the effects of welding residual stress (WRS) on FCGRs were preliminarily investigated. The comparison of residual stress distributions, stress intensity factors (SIFs), and crack growth rates demonstrated a high degree of consistency between the numerical predictions and experimental results, further confirming the reliability of the developed model. The results revealed that WRS has a significant impact on crack growth behavior, particularly during the accelerated crack propagation phase.