DeRisk delivers accurate and de-risked models for extreme wave loads which are often dimensioning for offshore wind turbines. This is achieved through physical experiments, numerical modelling and mathematical/statistical methods to obtain a rational chain from the met-ocean data base to the detailed structural loads. Project leader Henrik Bredmose here explains the role of advanced hydrodynamics in this context.
What new knowledge and tools can DeRisk provide?
A key contribution of DeRisk is a data base of wave particle velocities (wave kinematics) computed with a fully nonlinear wave model. Right here, there is a gap in now-a-days standard engineering wave models, which are either valid for small waves (linear) or regular waves (non-overturning, mono-chromatic). Extreme waves occur in a stochastic sea state and the ones causing the extreme loads are usually asymmetric and perhaps also breaking.
Are such models not already used in the oil-gas industry?
In the North Sea, offshore wind turbines are placed at an average depth of about 20m. At this depth the nonlinearity of the waves is much stronger than at deep water and the wave kinematics is thus much more advanced.
What is the link between wave kinematics and loads?
For slender structures as a monopile, the loads are often well estimated by strip-theory models that are based on the undisturbed wave kinematics. Detailed wave-structure interaction and load effects can also be described with CFD (Computational Fluid Dynamics) methods. In DeRisk we’ll work with both types of models and develop them further. The most advanced models deliver insight and accurate computational load predictions which can in turn be used to benchmark the simpler models suitable for engineering design.
How do you ensure that the tools are fast enough for practical use?
We work on many levels to achieve this: The fully nonlinear wave model of DTU Compute and DTU Mechanical Engineering runs on GPU architecture (Graphical Processor Unit) and thus with unprecedented speed. Next, we extract results to a data base, so they can be used without pre-computation. We also work with mathematical uncertainty quantification which is designed speed up standard Monte Carlo simulation strategies by orders of magnitude. We use statistical methods to extract ‘load effect factors’ from the experiments and numerical studies. The final catalogue of load prediction tools will thus span from high-fidelity models to table-look ups of load effects. Eventually, this is condensed to a load evaluation procedure that covers the whole way from the analysis of met-ocean data to the detailed structural loads.
 |
 |