Keywords: wind turbines, foundations, layered soil, stochastic analysis, dynamics.

Monopiles have proved to be cost-effective foundations for offshore wind turbines installed at water depths below 20 m. An example is the 2.0-2.3 MW turbines installed at Horns Rev in the North Sea west of Jutland, Denmark. With growing demands for sustainable energy and limited space for new wind farms onshore and near the coast lines, future wind turbines are to be installed at greater water depths. Here, fewer but larger turbines provide a more economically profitable solution. A critical aspect in the design of such turbines is their response to cyclic loading.

To avoid dynamic amplification of the response, the natural frequency must be well separated from the excitation frequencies related to wind and wave loading. Current codes of practice involve a deterministic design of the foundation assuming a quasi-static load on the structure. The design may be carried out in terms of total safeties and, or partial factors of safety; but none of these approaches provide any insight into the reliability of the estimate on the natural frequency of the structure. As an alternative, the paper suggests a reliability-based design of the monopile foundation, leading to a quantification of the uncertainties of the natural frequency related to the random nature of the subsoil.

The pile is modelled as a beam. The response of the soil is introduced by employing the p-y curve method, i.e. modelling the soil-pile interaction as a Winkler foundation with non-linear horizontal springs. Clay with a lognormally distributed undrained shear strength is considered and the variation of the strength over depth is modelled as a stochastic process [1]. The wind turbine is modelled as a simplified structural system with the main purpose of identifying the first natural frequency. It has been found that the mean value of the wind force has a significant impact on the natural frequency, whereas the variations of the wind and wave loads play a minor part. The probability density for the natural frequency of the wind turbine is determined numerically by Monte Carlo simulation, finding that the eigen frequency is not lognormally distributed. Finally it is concluded that the probability of obtaining a natural frequency significantly below the design value increases with an increase of the correlation length of the undrained shear strength.

1

B.K. Low, C.I. Teh, W.H. Tang, "Stochastic nonlinear p-y analysis of laterally loaded piles", in "Proceeding of the Eighth International Conference on Structural Safety and Reliability (ICOSSAR)", Newport Beach, California, 17-22 June 2001.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £130 +P&P)