Cyclic axial behaviour of piles in sand: 3D FE modelling using SANISAND-MS

Abstract

Wind energy is now a popular competitor among other energy sources all around the world. The offshore wind industry has progressed in recent years, with larger wind turbines being installed in deeper oceans. The construction of such large-scale wind farms necessitates more modern foundation design technologies to increase operational safety while also lowering total structure set-up costs. The environmental load applied to offshore piles are of great complexity. Currents, wind, waves, and even earthquakes are very common dynamic loads in an offshore loading environment. Of course, when a wind turbine is working normally, it also has significant operation loads. The design of offshore wind turbine support structures often involves some universal criteria, e.g., the pull-out capacity of jacket structures on piles. Wind turbine foundation capacity is determined by the qualities of the offshore soil as well as the properties of support structure configurations. Therefore, it is necessary to take account of the potential cyclic impacts of soil-structural interaction to guarantee dependable responses of the wind turbine structure. This thesis aims at evaluating soil−structure interaction of offshore wind turbine foundations under cyclic loading, with emphasis on the tension capacity of axially loaded displacement piles, under different load conditions (cyclic-to-average ratios) on Fontainebleau NE34 sand in France. A newly developed constitutive soil model SANISAND-MS (2018) is applied to model sand stress-strain evolution. In this thesis, the soil is assumed a homogeneous linear elastoplastic material for the sake of simplicity. The SANISAND-MS constitutive model used in this thesis can capture sand ratcheting after considerable cyclic loading cycles. Furthermore, drained and undrained compression triaxial tests performed at DTU GEO−Lab were used to calibrate the model parameters of the constitutive model for Fontainebleau NE34 sand. The finite element model adopted here is built in an open-sourced platform, the OpenSees. The Small-strain approach is adopted in the finite element modelling part. The pile is simplified as a wished-in place which does not include the installation effect and the time effect after the installation and before cyclic tests. Finally, the modelling results are compared to the experiment results recorded by Tsuha et al. (2012). Clear stable, metastable, and unstable response types are recognized in the model results. However, the initial stress state of the sand at the soil-pile interface differs a lot compared to the experiment results. This is the consequence of not including the pile installation effects in the finite element modelling. Recommendations are given to use large-strain soil modelling techniques to include the pile installation process.