Back to the programme printer.gif Print



Wednesday, 12 March 2014
16:30 - 18:00 Wakes: Do we need different models for onshore and offshore wind farms?
Resource Assessment  


Room: Ponent
Session description

The accurate prediction of the wake effects within and between wind farms is vital to wind farm design and to provide a prediction of the energy output. To date, the vast majority of wind farms have been designed using engineering models which have been tuned and validated using experimental data. As wind farms become larger, empirical correction upon empirical correction are being developed upon the basis of scarce and perhaps erroneous experimental data. Perhaps this is the appropriate time to question if this is the right and only approach.

In the session advanced models and observations will be described and discussed.

Learning objectives:

  • Describe what wakes are, how they can be seen in observations and how they are modelled
  • Identify different models, state-of-the-art and more classical ones
  • See wakes in observations and understand the related issues
Lead Session Chair:
Lars Landberg, DNV GL – Energy (Garrad Hassan), Denmark
Reza S. Abhari ETH Zurich, Switzerland
Co-authors:
Samira Jafari (1) F P
(1) ETH Zurich, Sonneggstrasse 3 CH-8092 Zürich, Switzerland

Printer friendly version: printer.gif Print

Abstract

Estimation of array loss in onshore and offshore wind farms using RANS

Introduction

The operation of turbines in wakes decreases the wind farms’ output up to 20% and increases fatigue loads on downstream turbines up to 80%. The consequence is reduced lifetime of turbines and an increased cost of energy [1]. An accurate prediction of wind speed and turbulence in wakes is necessary for successful micrositing in wind farms. Engineering models often fail when applied over a wide range of atmospheric and topographic conditions [2-9]. The present work demonstrates that the routine use of Computational Fluid Dynamics code with three-dimensional field models is possible for predictions in onshore and offshore wind farms.

Approach

Novel immersed turbine models, which model simulate the turbines’ three-dimensional flowfield, are integrated in our in-house preconditioned multistage Reynolds-Averaged Navier Stokes solver in connection with a k-ω turbulence model [10-12]. This approach enables the wind flow and turbine wake to be simulated simultaneously, and allows accounting for atmospheric conditions and/or topography (onshore or offshore). In comparison to prior works that have simulated offshore wind farms and farms located in complex terrain [13-16] with three-dimensional field models, the present work is computationally more efficient as the complete wind rose, with one Cartesian grid, can be simulated in less than one week.

Main body of abstract

In this paper, the model is used to simulate the wind flow and turbine wakes in the onshore wind farm, Mont Crosin, which is located in complex terrain [10], and in two offshore wind farms, Lillgrund [17] and Horns Rev [18]. These wind farms are comprised of 16, 48 and 80 wind turbines, respectively. For each test case, the simulations are performed over the entire wind rose with 10-degree intervals. The simulation of the entire wind rose uses a single Cartesian grid and the simulation takes less than one week. The predicted output powers of the turbines show very good agreement with the measured SCADA data. For each simulation, the evolution of the wakes, and their associated wind speeds, directions, and turbulence intensities are also detailed through the whole wind farm. These simulations highlight the impact of the wakes of the different turbines and the site-specific characteristics of the wakes; the latter is the primary reason why engineering wake models are of limited utility for the development of future wind farms. The sensitivity of the power loss to small changes in the direction and turbulence intensity of the incoming wind is also investigated. Lastly, the directional array loss is predicted, and is shown to agree very favourably with available measurements. Therefore, this work demonstrates the feasibility of routinely using Computational Fluid Dynamics for the accurate assessment of wind farm projects. This approach is preferred as the need for lengthy and possibly inaccurate tunings in engineering models is eliminated.



Conclusion

Predictions of the array loss of onshore and offshore wind farms, with different turbines and topography, are shown to compare favourably to measurements for different wind directions. The simulations of the entire wind rose use a single Cartesian grid and take less than one week. Therefore, this work demonstrates the feasibility of routinely using Computational Fluid Dynamics, which accounts for atmospheric and topographic conditions, for the accurate assessment of offshore and onshore wind farms. This approach is preferred as it includes the different atmospheric and topographic conditions that may be encountered in the broad range of wind farms.


Learning objectives
• Understand approaches to wake modelling
• Grasp ideas of Computational Fluid Dynamics methodology
• Gain knowledge of microscale Computational Fluid Dynamics simulations
• Be familiar with characteristics of wind turbine wakes and differences between atmospheric flows in onshore and offshore
• Understand how to compare simulations and measurements