PSI - Issue 75

Philippe Thibaux et al. / Procedia Structural Integrity 75 (2025) 546–554 P. Thibaux et al. / Structural Integrity Procedia (2025)

553

8

detection of a crack and its breakthrough is smaller when the loading is in-plane. In the previous sections, it has also been highlighted that a crack appearing at the saddle points on the chord leads to the formation of a very large crack and a decrease of the stiffness of the joint. There is likely a significant load redistribution during crack growth in out-of-plane bending. In the in-plane bending test, the crack remains in comparison smaller, reaching a total length of 450mm and only marginally affecting the stiffness of the sample. Load redistribution is therefore limited. The difference in load redistribution is likely causing the main difference between both crack growth behaviour. For the safety of the total structure, the difference of load redistribution will have significant consequences. A jacket structure is more displacement controlled, meaning that the loads acting on a crack appearing at the saddle will decrease, while they will remain relatively constant for a crack appearing at the heel because there is no stiffness reduction. A crack appearing at the saddle points will therefore be more resilient, because there is more time between appearance of a crack and failure, giving more time for detection by inspection, and also because the load redistribution will reduce the crack growth velocity by decreasing the applied forces. The fatigue strength measured at the heel / brace and at the saddle points on the chord are rather similar, although the thickness of the brace is 25mm and the thickness of the chord is double. For comparison, the results were extrapolated to 2 million cycles using a Basquin law with a slope of 3. The slope of 3 was found suitable to describe fatigue tests on tubular joints (Thibaux 2025). A value of 127MPa is found for the in-plane test, while 122MPa is found for the out-of-plane test. The thickness exponent correction of 0.25 would predict a difference of 19%. Consequently, either the thickness correction is not suitable to predict the fatigue strength at the heel and the chord, or the fatigue strength itself is different between the heel and the saddle. The difference in behaviour can be caused either by a difference in the local weld geometry or by the loading mode. The heel of a jacket node is a complex geometry, with a small opening angle and limited access; the welder has therefore less opportunity to make a smooth joint, leading to smaller weld toe angles, and larger local stresses. The loading is also quite different between the chord and the brace, with the loading at the chord being almost entirely in bending, while it is mixed bending and membrane loading on the brace. One can therefore expect a faster crack growth through the thickness on the brace. This result should be interpreted cautiously as the number of failure points is limited (only one on each component). However, if this result is significant, it means that either the thickness correction is not directly applicable to tubular joints, or that the hot spot stress method should be corrected for a failure through the brace.

500

OPB Init. IPB Init.

OPB fail. IPB fail. T-design

OPB RO IPB RO

Hot spot Stress Range (MPa) T-mean

50

50000

500000

5000000

Cycles (-)

Fig. 7. Comparison of the experimental results of the in-plane bending (IPB) and out-of-plane bending (OPB) with the design and mean curve of the T-category of DNV-RP-C203. Run-outs are indicated with open symbols, breakthrough cracks with filled symbols and moment of crack detection by strain gauges with grey symbols.