PSI - Issue 57

Gustav Hultgren et al. / Procedia Structural Integrity 57 (2024) 428–436

429

2

Hultgren & Barsoum/ Structural Integrity Procedia 00 (2023) 000 – 000

Nomenclature

t

Plate thickness

, , A B L

W

Weld area, width and length

Combined plate and weld height

Reference area, width and profile length Euler-Mascheroni constant, 0.5772 

0 0 0 A ,B ,L

Greek symbols

EM c

h



Weld height

Distribution shape parameter



1 I L m

First stress invariant Attachment length

Flank angle

0 

Characteristic fatigue strength Principal stresses, 1 2 3      Signed von Mises stress criterion

Basquin exponent and slope

1 2 3 , ,   

n

The number of cycles to fatigue failure

SvM 

Model failure probability

Equivalent stress Reference stress Variation in logC

f p f p

equ  ref  logC 

Failure probability based on nominal stress

nom

r

Weld toe radius

Digital systems for quality assurance of welded joints have made it possible to assess local variations in complete 31 welds fast and reliably according to the geometrical definitions stipulated in weld quality standards, Stenberg et al. 32 (2017); Renken et al. (2021); Ottersböck et al. (2021). Digitalisation is a significant step forward for process 33 monitoring of the weld quality, as traditional audits have proven insufficient to capture the quality variations, 34 Hammersberg et al. (2010). 35 Finite element simulations of the idealised weld geometries are a common technique for studying the influence 36 that the local weld geometry has on the fatigue strength. However, the setback with this technique is that simple 37 mathematical functions cannot define the true weld toe geometry since perfect transitions with constant radii are not 38 seen in real welds, Hou (2007). Welded joints can include variations such as ripple lines, asymmetry, convexity, and 39 competing radii in the weld toe that change the local stress state compared to an idealised weld geometry. Finding 40 representative idealisations for such variations does then become a non-trivial task. 41 The digitalisation of quality inspection makes it possible to further study and correlate the actual weld geometry 42 and all of its local imperfections with the real, local stress distributions. Hou (2007) did an early study on the weld toe 43 plasticity for non-load-carrying joints using simulations of scanned welds and showed a correlation between the weld 44 toe plasticity and the major crack locations. 45 Probabilistic methods are great tools for evaluating variability in stochastic processes, such as the influence of the 46 actual weld geometry. Lang and Lener (2016), Niederwanger et al. (2020), and Hultgren et al. (2023) have successfully 47 deployed such probabilistic methods for the fatigue evaluation of welded joints. The effect of measured local weld 48 geometry and its variability is here further studied based on the method and specimens presented in Hultgren et al. 49 (2023), together with analytical models for the determination of stress concentration factors (SCFs). Detailed 50 information on the weld geometry is available in the publication, gathered from the commercial quality assurance 51 system Winteria® together with the equivalent probabilistic stress based on the weakest-link theory. 52 2. Experimental work 53 Eighteen single-pass, non-load-carrying, tee joints from the second series in Hultgren et al. (2023) were used in the 54 present work to conduct a more in-depth study of the impact of the measured local weld geometry. The plates from 55 which these specimens were cut were welded with different welding settings to promote a greater geometry variation, 56 making them suitable for further studies. All specimens, with the nominal dimensions highlighted in Fig. 1, were 57 produced out of 6 mm S960 strip steel, where one of the two welds was post-treated using TIG-dressing to promote 58

Figure 1. Dimensions of the non-load-carrying, tee joint, fatigue test specimen.