PSI - Issue 57

Inge Lotsberg et al. / Procedia Structural Integrity 57 (2024) 569–580 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

579 11

For a nominal stress in the main plate in specimen 1 of 200 MPa the membrane stress in the doubling plate σ d = 107.38 MPa. For t d = 10 mm and a = 7.0 mm this gives σ w = 153.40 MPa. For 150 800 test cycles for test specimen no 1 this gives an S-N curve for the weld root that is between the W1 and the G-curve (or between FAT 45 and FAT 56). The difference between mean and characteristic S-N curves is accounted for in this assessment. A similar exercise using the calculated notch stress also gives an S-N curve for the weld root slightly above the W1 curve (or FAT 45). The calculated notch stress at the weld root is similar to that of the weld toe. This result indicates that the probability of having a fatigue crack at the root is like that at the weld toe. However, any crack initiation was not detected in the root areas of this as-welded specimen with a throat thickness equal to 7 mm after opening of the connections. This may be explained by a difference in residual stresses in the weld root as compared with the weld toe with typical presence of residual tensile stresses. More test cycles to a fatigue failure were achieved for the toe ground doubling plates but still without significant fatigue cracking from the weld root, Lotsberg (2016). Due to a question of fatigue cracking from the weld root an additional ground specimen with a throat thickness equal to 4 mm was tested. Still only minor crack indications were observed in the weld root of some of the hot spot areas. The effect of residual stresses on fatigue life is difficult to assess without fatigue testing. Therefore, there will still be need for laboratory fatigue testing even if the methodologies for analysis have been significantly improved the last decades.

t d

σ

d

σ s1

σ s2

t p

Fig. 10. Model for calculation of stress in the weld throat.

4. Conclusions Stresses at weld toes of circular and squared fillet welded doubling plates have been derived from measurements of strain at hot spots in a laboratory test of full-scale plate geometry as presented in Lotsberg et al. (2014) and Lotsberg (2016). Fatigue testing until a through crack in the main plate has been performed for comparison with S-N curves used in design recommendations. Specimens in as-welded conditions and after post weld improvements by grinding and ultrasonic peening have been fatigue tested. The purpose of this fatigue testing was to assess the possibility for fatigue cracking from the weld root in post weld improved doubling plates as such plates are not recommended for improvement according to IIW (Hobbacher, 2009). The test results showed fatigue crack growth from the weld toes and only small indication of crack initiation in the weld roots could be seen in some of the welds when the welds were opened after fatigue testing. The number of test cycles until failure for the as-welded doubling plates with a width equal to 150 mm agreed with the recommendations from DNV-RP-C203 (2019), (curve F1), which is the same curve as FAT 63 in IIW, (Hobbacher, 2009). The main purpose of the performed analyses was to assess the sensitivity of finite element modelling that are somewhat different from recommendations in analysis guidelines and compare the calculated values with fatigue test data. Different methods have been used to derive hot spot stresses at the weld toes of the fillet welds in the as