Fracture mechanics based fatigue life prediction for a weld toe crack under constant and variable amplitude random block loading—Modeling and uncertainty estimation

• Published in: Engineering fracture mechanics 2021-02, Vol.242, p.107487, Article 107487
• Main Authors: Leonetti, Davide, Maljaars, Johan, Snijder, H.H. (Bert)
• Format: Article
• Language: English
• Subjects: Amplitudes; Block loading; Crack propagation; Cracks; Cyclic loads; Cyclic R-curve; Fatigue crack growth at weld; Fatigue cracks; Fatigue failure; Fatigue life; Fracture mechanics; Growth models; Life prediction; Linear elastic fracture mechanics; Material properties; Mathematical models; Mechanically short and long cracks; Parameter uncertainty; Short cracks; Stress intensity factors
• ISSN: 0013-7944; 1873-7315
• DOI: 10.1016/j.engfracmech.2020.107487

Propagation of weld toe cracks under cyclic loading is often predicted using fracture mechanics. In as welded condition, most of the propagation life is spent as a short crack, which is known to behave differently than a long crack. Several studies have been conducted with the aim of correlating the fatigue crack growth rate and the threshold condition of small cracks to the well known linear elastic crack driving force parameter ΔK, the stress intensity factor range. In many cases, the application of such models requires the quantification of material properties and model parameters that are difficult to obtain from tests, and therefore scarcely available. The present paper bypasses this inconvenience by making use of the square root of area, area, parameter proposed by Murakami. Successively, a linear elastic fracture mechanics based fatigue crack growth model is formulated for physically short and long cracks under constant and variable amplitude random block loading. The uncertainty of the model parameters is quantified in a frequentist statistical framework. •LEFM is used to predict the fatigue life of a welded joint under CA and VA loading.•The cyclic R-curve is estimated based on existing correlations minimizing the amount of data needed.•The model is successfully applied to two datasets related to welded details.•The uncertainty and scatter of the input parameters is quantified.