Steel jacket support structures for offshore wind turbine towers made of circular hollow sections (CHS) are becoming a competitive solution compared to monopiles in deeper waters. The current limitation to improving the durability and cost-effectiveness of the CHS structures is the low fatigue endurance of their welded joints. Fatigue-driven design of such structures usually leads to thicker profiles of steel members as well as costly welding of joints. Due to excellent corrosion and fatigue endurance, high strength-to-weight ratio, easy and locally made fabrication, fibre-reinforced polymer (FRP) composite materials have been widely used to strengthen welded CHS joints. Although this strengthening technique can enhance the resistance and fatigue performance of the welded CHS joints, welding still serves as the main way for load transferring in the joints. Therefore, welds remain the source of stress concentration and brittle fatigue failure. Recently, an innovative joining technique, wrapped composite joint, has been proposed at TU Delft. In this joint, the CHS members are connected through the composite wrap. Loads are transferred at the bonded composite-to-steel interface, and welding is completely avoided. While monotonic tensile tests have shown the superior stiffness and resistance of the wrapped composite joint over its welded counterparts, the fatigue performance of the joint still needs to be investigated.
As the first investigation of the fatigue behaviour of the wrapped composite joints, the present study focuses on the most unfavourable failure mode, debonding at the composite-to-steel interface. A typical joint geometry in the jacket support structures, the K-K joint, is chosen as the research object, which is simplified to be the uniplanar X-joint. The general objective is to accomplish knowledge sufficient to predict the debonding behaviour of CHS-wrapped composite X-joints under tensile cyclic loads.
To achieve that goal, the present work starts from the interface level, where the fatigue crack growth (FCG) properties at the composite-to-steel interface are characterised through fracture mechanics experiments, i.e. 4-point bending end notched flexure (4ENF) tests. The steel surface of the specimens is prepared with different roughness levels, and its impact on FCG properties is investigated. The obtained FCG properties provide the basis for predicting crack growth at the joint level. In the wrapped composite joints, friction exists at the composite-to-steel interface due to the confinement by the composite wrap, which may retard the crack growth. This phenomenon is quantified in cyclic tests on joints with simple geometry, i.e. the axial splice joint (A-joint), where the debonding crack growth is monitored through the 3D digital image correlation (DIC) system. At the joint level, tensile cyclic tests are conducted on the wrapped composite X-joints with different surface roughness and at different scales. Post-fatigue static tests are conducted to check the influence of cyclic loads on the residual resistance of the joints. Using the finite element model, the methodology to predict the crack growth and stiffness degradation of the wrapped composite joints is established, which can consider the interaction between debonding on the chord and brace members. The prediction methodology is validated against the test results and used in a probabilistic analysis to explain and reproduce the scattering test results. Finally, the failure criterion of the joints under cyclic loads is proposed to establish the design S-N curves.
The present study found that the surface roughness of the steel tube plays an important role in the FCG properties of the composite-to-steel interface. A minor increase of the surface roughness can significantly improve the joint’s fatigue performance, with the parameter C of the Paris curve decreasing over magnitudes. At the joint level, the wrapped composite X-joints exhibited steady stiffness degradation during the tests due to debonding propagation at the composite-to-steel interface. Joints with reduced surface roughness show deteriorated fatigue performance but still have longer fatigue life over the welded ones. By including friction at the interface, the finite element model gives reduced strain energy release rates (SERR) at the crack front as the crack grows. Thus, the main source of the crack growth retardation is explained and can be quantified. The numerical results match well with test results of X-joints considering different surface roughness, different load levels and scales, and the relationship between debonding on the chord and braces is obtained. By studying the variability of surface roughness and FCG properties, the probabilistic analysis can reproduce scattering of the test results. Finally, the design S-N curves are obtained based on the experimental and numerical results, taking 5% resistance reduction as the failure criterion.
The present study provides a methodology for characterising and predicting fatigue debonding behaviour, not only for wrapped composite joints but also for other large-scale bonded joints with complex geometry, enhancing the application of bonded joints in engineering structures.