Fatigue life assessment of steel specimens with weld ends and complex cracking scenarios by crack propagation numerical analyses

• Longitudinal and transverse joints with weld ends were experimentally tested in a previous study. • Paris laws were experimentally derived from CT specimens extracted from the considered joints. • Crack propagation analyses were performed for both joints according to LEFM and the EIFS concept. • The long crack propagation phase of transverse joints was explained on the basis of K I versus a trend. • Fatigue life estimated by crack propagation analyses were in good agreement with the experimental results. In a previous work, steel welded specimens consisting of a tube inserted into a plate joined by four intermittent fillet welds were fatigue tested under axial loading. Two geometries were analysed, namely longitudinal and transverse , having intermittent weldments parallel and perpendicular to the applied load, respectively. Fatigue strength estimations based on the Peak Stress Method (PSM) identified crack initiation locations, but underestimated the fatigue life of transverse joints due to their prolonged crack propagation phase. In the present investigation, linear elastic fracture mechanics was adopted to simulate crack propagation by means of finite element analyses using Abaqus® and FRANC3D® software packages. Material-specific Paris’ law parameters were experimentally calibrated both for 25MnCr6 Q&T (tube) and S355JR (plate) steels using Compact Tension (CT) specimens. Initial cracks with 0.1 mm radius were located at the experimentally observed initiation sites, and numerical simulations were performed according to the Equivalent Initial Flaw Size (EIFS) concept suggested by the IIW guidelines. When using the most accurate simulations in terms of weld bead geometry and initial cracking scenario, fatigue life-to-failure estimations were in good agreement with the average experimental lifetime with deviations between −34 % and + 45 % for transverse joints and between −17 % and −12 % for longitudinal joints.