Vous êtes sur la page 1sur 2

Dave Corson - Altair Engineering, Inc.

Paul Lees - PAX Streamline

Validation of High Fidelity CFD Modeling Approach for


Utility Scale Wind Turbines
Abstract
A methodology is presented to use a commercially available finite element based flow solver, AcuSolve™, to study utility scale
wind turbine aerodynamics employing advanced, high fidelity modeling techniques. The wind turbine studied here is based
upon the NREL 5 MW offshore design. This machine uses a horizontal axis, 3-bladed rotor with a diameter of 126 meters. Due
to the large size of this rotor, the simulations present challenges to CFD practitioners to develop accurate, efficient, and robust
modeling approaches. Using the techniques described in this document, steady state Reynolds Averaged Navier Stokes (RANS)
and fully transient Detached Eddy Simulation (DES) results were computed for a range of wind speeds and rotor RPMs. Rotor
thrust, torque, and power were resolved and compare favorably to accepted results obtained from researchers at laboratories
and academic institutions. Flow structures were also identified and compared for different wind speeds using the commercial
post processing package, FieldView.

Solution Process Modeling Methodology


A key aspect in this work is not only validation of the flow solver for Simulations were constructed using the process described in Figure 1.
utility scale turbines, but the development of a robust and easy to A thorough mesh sensitivity study was performed to ensure grid
use modeling approach. To facilitate this requirement, we focus on independent results for the simulations. The final mesh for the full
a process that exploits 3-d CAD modeling, automated unstructured rotor contained approximately 13 millions nodes and 55 million
meshing, automated solution set-up, and automated post processing. elements. The following images illustrate the shape of the blade and
The work process is illustrated below: a representative unstructured mesh.

3-d Solid Model of Turbine Geometry

CAD based meshing ensures accurate representation


of complex airfoils

Customizable Graphical User Interface Figure 2: Blade geometry for the 5 MW rotor models.
Python scripting used to automate the set-up of the
meshing controls, boundary conditions, and solver
parameters

Unstructured Mesh Generator

Unstructured meshing technology operates directly


on the underlying CAD model

AcuSolve Flow Solution

FVX scripting enables automated batch processing of


CFD results

Post Processing Using FieldView Figure 3: Unstructured mesh used for wind turbine
simulations. Note the structured edge meshing on the
Figure 1: Flow chart illustrating the simulation process leading and trailing edges as well as the anisotropic
surface triangles. These techniques provide an
efficient method of resolving the pressure field on the
surface without the need to use structured meshing
Copyright © 2011 Altair Engineering, Inc. All trademarks are property of their respective owners.
Two operating points were selected for simulation. The lower power The power and thrust for each of the simulations was computed and
case corresponds to a 9 m/s wind speed and 10.3 RPM rotational compared to the results published by Riso1.
speed. The higher power case entails a wind speed of 11 m/s and
a rotational speed of 11.9 RPM. Steady RANS simulations using
the Spalart-Allmaras turbulence model were performed for both
conditions, while a full sliding mesh DES simulation was performed
only for the lower power case. All simulations were performed
using a 64 core AMD Opteron cluster with an Infiniband message
passing network. Steady state simulations of the full rotor model
required approximately 10 hours of compute time on the cluster to
reach a steady state solution. Figure 6: Power and thrust comparisons between AcuSolve
and Riso simulations.

Results The AcuSolve results compare well to the Riso simulations, indicating
The steady RANS solution provides detailed information about the that the unstructured meshing/finite element solution methodology
performance of the rotor. The local pressure field on the high and low provides accurate results for this application. Additionally, the DES
pressure side of the blade for the lower power case is shown approach is found to provide similar results as the steady RANS
in Figure 4. simulations. For this application, the additional compute cost of
the DES approach is not warranted if integrated quantities such as
power and thrust are the only items of interest. However, this also
implies that the DES approach produces accurate results and can be
used for inherently transient applications such as acoustic and fluid-
structure interaction simulations.

Conclusions
An unstructured grid based CFD modeling methodology has been
developed and successfully used to simulate the flow around a utility
Figure 4: Surface pressure distribution on the wind
turbine blade. scale wind turbine rotor. The total power and thrust predicted by
the simulations compare favorably with results obtained by other
The CFD solution is also successful at capturing the detailed research groups. To facilitate the ease of performing the full rotor
flow structures in the wake of the turbine. Adequately capturing simulations, all gridding was performed using fully automated
these features requires highly accurate numerical methods to unstructured mesh generation techniques, and post processing was
propagate the wake downstream without the need to use excessive performed using automated batch processing.
compute resources.
The wealth of insight provided by CFD simulations gives designers
and engineers the opportunity to rapidly investigate advanced design
concepts and establish the improvements in efficiency, reliability,
and cost effectiveness that are required to propel wind power
technology into the future. This validation effort represents an
important step in achieving these improvements and can easily be
extended to encompass more complex physics such as sheared wind
conditions, gust events, and fluid structure interaction.

Figure 5: Flow structures in the wake of the 5 MW rotor models. References


The image on the left depicts iso-surfaces of the Q-criterion 1. UPWIND, Aerodynamics and aero-elasticity. Rotor aerodynamics
colored by velocity magnitude. This clearly illustrates the root
and tip vortices as well as the trailing edge vortex sheet.
in atmospheric shear flows. Niels N. Sorensen. Wind Energy
The image on the right is a cut plane showing contours of Department, Risoe National Laboratory,
vorticity in the wake of the rotor. www.risoe.dtu.dk/rispubl/art/2007_140_paper.pdf.

Vous aimerez peut-être aussi