Advanced methods for the design of foundations for offshore wind energy structures

Advanced methods for the design of foundations for offshore wind energy structures

With a large proportion of global greenhouse gas emissions coming from energy, the continued uptake of renewable energy sources is vital. Offshore wind forms an important component of the UK's energy supply. This motivates the exploration of novel engineering techniques to continue to increase the commercial viability of offshore wind. This project falls within the EPSRC wind power theme. The research will focus on geotechnical elements of offshore wind turbines, focussing on the behaviour of foundations.
There is scope for the application of advanced computational procedures and data-driven approaches in all stages of an offshore wind project's lifecycle, from site selection to asset protection. The proposed research will focus on the development of these methods for the prediction of behaviour and design of foundations. The research will likely focus on monopile foundations, which form the most common foundation type for offshore wind turbines. The target audience for this research would be practitioners, with a research aim of developing novel, robust analysis techniques that can improve the design of foundations of offshore wind energy structures. The objective of these techniques will be to be rapid, whilst approaching the accuracy of computationally expensive high-fidelity models. This will enable more accurate parametric study and optioneering from designers, creating more optimal, cost-effective designs.
Three-dimensional finite element analysis is typically employed for high fidelity modelling of soil-structure interaction of monopiles. This is computationally expensive, limiting its use for parametric study and optioneering. The research will explore methods to create models which are rapid, approaching the accuracy of three-dimensional finite element analysis. This can be achieved by creating surrogate machine learning models for the three-dimensional finite element analysis. Training data can be generated from the three-dimensional finite element analysis, capturing relevant information about the foundation's behaviour. Machine learning surrogate models can be trained on this data, with the aim of making rapid, accurate predictions. Previous research has applied this approach to make predictions of the foundations macro-element stiffness, which would be applied as a support condition at the base of models of the turbine super-structure.
The student is currently sitting taught modules at the start of the course as part of the EPSRC funded Wind and Marine Energy Systems and Structures centre for doctoral training, so the research carried out may differ from the preliminary aims outlined in this description.