Modelling of skirted foundations for offshore wind turbines
MetadataVis full innførsel
The thesis focuses on the modelling of the response of shallow skirted foundations supporting offshore wind turbines (OWT). OWT-structures need to be optimized in design to reduce the cost of energy from offshore wind production. The optimization is demanding since OWTs are dynamically sensitive structures exposed to loads with a complex load frequency content. To capture the coupling between non-linear actions and reactions from wind, wave, turbine controller, structural dynamics and foundation response, it has become industry standard to carry out so so-called integrated dynamic analyses in the time domain. However, the foundation response of a skirted foundation is typically modelled in a simplified manner in design. This is unfortunate as research, including studies performed in this PhD, show that foundation behaviour significantly influences the dynamics of OWTs. Accurate geotechnical analyses, which typically model the soil by continuum elements and complex non-linear constitutive models, are unsuitable for integrated time-domain analyses, as they significantly increases the computation time. Foundation modelling by macro-elements is a promising method that balances the requirements of accuracy and computational efficiency. A macro-element represents the whole foundation and the soil by a non-linear formulation at one point. This point can for example be connected to the structure at seabed. Existing research and development of macro-elements has been largely based on results from small scale 1g-tests, centrifuge tests and a few field tests. This gives the models a firm scientific basis. However, the research has been less focused on the challenges of practical usage of the foundation models, and nor the adaption to other conditions than those considered in the model tests. This prevents usage of macro-elements in design, leaving practitioners with only the simplified analysis options. A macro-element is proposed in this PhD-thesis aiming to address some of these challenges. The macro-element formulation is based on results from finite element analyses of skirted foundations. These results are presented as a separate study of foundation behaviour subjected to general loading. The NGI-procedure accounting for cyclic loads is used to define the soil behaviour, and the study investigates the capacity and stiffness response of the foundation. The study considered combined vertical, horizontal and moment loads and combined static and cyclic loading. The results are used to define a procedure for estimating the response of a skirted foundation subjected to general loading, accounting for cyclic soil degradation at foundation level. The study also reveals how the load- displacement response for pure vertical displacement, horizontal displacement and rotation, denoted uniaxial response, can be used to estimate general load paths. Thus, the uniaxial response is considered to define the fundamental behavioural characteristics of the behaviour for a specific foundation. This idea is brought on into the development of a macro-element, by requiring the uniaxial response curves as model input. The macro-element is formulated within the multi-surface plasticity framework. The framework is well suited for modelling of cyclic response, as it has a memory of the recent history through the back stress vector, which contain the coordinates of the origin of the yield surfaces. The Piecewise linear hardening makes the macro-element flexible in the sense that different load-displacement curves can be reproduced. The hardening is anisotropic based on interpolation between the uniaxial responses using the flow direction vector. Requiring the uniaxial responses as input makes the macro-element behaviour site-specific and able to capture the variations in response due to non-homogeneous profiles. This makes the macro-element more suitable for practical use than other macroelements presently available in the public domain. The macro-element response is compared with finite element analyses and a field test, and the agreement is found to be very good. Assessment of accumulated displacement is addressed outside the integrated analyses. For this reason, the macro-element was formulated with a pure kinematic hardening rule such that no accumulation of displacements occur during a cyclic history. Special attention is given to the problem of numerical ratchetting, which is reported to be a problem in several cyclic models. Testing showed that the macro-element is rigorously formulated with respect to ratchetting. The macro-element is also formulated to include internal flexibility of the foundation. This addresses recent full scale measurements of a jacket founded on caisson foundations (Suction Bucket Jacket) where the caisson flexibility contributed significantly to the total stiffness. The flexibility is included in the formulation as an elastic correction, implemented as a series coupled threedimensional spring. The macro-element is found to include the effect of caisson flexibility with sufficient accuracy when compared to FEA results. The PhD-thesis address the challenges of foundation modelling from the perspective of practitioners. The proposed procedures and the macro-element can be used in practical design, and should contribute to improved accuracy in the prediction offoundation behaviour in integrated analyses.