Aerodynamic Properties of an S809 Wind Turbine through CFD Analysis

Dylan B. Gareau, Tara Gholami, Kohl Lounsbury,

Department of Mechanical and Manufacturing Engineering, University of Calgary, AB, Canada
[email protected]
[email protected]
[email protected]

PROBLEM DEFINITION
This paper examines a Computational Fluid Dynamic
(CFD) analysis on an S809 wind turbine blade illustrated
in figure 1. The goal of the CFD analysis was to compare
the aerodynamic forces and moments present on the S809
airfoil when placed in an oncoming air flow. In the
analysis the velocity of the oncoming wind was kept
constant while the angle of attack of the airfoil varied. The
changes in the coefficients of lift, drag, and moment were
obtained from the ANSYS FLUENT solver. The analysis
also examined the results in accordance with three
different viscosity models for comparison reasons.

algorithms that were used for the CFD analysis of the

aerodynamic characteristics of a wind turbine blade in
ANSYS, FLUENT solver.
Fundamental Modelling Assumptions
The first and foremost underlying assumption in
analysing the aerodynamic characteristic of a 3
dimensional wind turbine blade based on its 2 dimensional
airfoils. Such analysis would ignore the effect of
downwash. Downwash is described as the effect of
vortices generated at the tip of a wing, or a blade in this
case. At the tip, the lower pressures on the top surface of
the blade draws air from the bottom side, generating a
vortex that pushes down on the blade and decreases the
amount of lift available at the tip. This is the primary
reason why the 3D lift coefficients are lower than the 2D
lift coefficients for a standard wing (NASA, 2014).
However, in case of a wind turbine, the aspect ratios are
usually very large, meaning that the length of the blades
are much larger than the chords. Due to these high aspect
ratios, the downwash effect at the tip of the blades will not
have any significant impact on the lift generated on the
blade. For this reason, it is justified to use a 2 dimensional
model to accurately analyse the aerodynamic properties of
wind turbine blades (Musial et. al., 1990). Furthermore,
due to the small solidity ratios, meaning that the blades are
located very far apart, the effects of the other two blades
can be ignored in the analysis (Ingram, 2011).

Figure 1: S809 Wind Turbine Blade Profile (NREL,

2014)

RESULTS VERIFICATION STRATEGY

The results of the analysis will be compared to the
results of a similar analysis by Wolfe and Ochs (1997).
who have found the lift, drag, and moment coefficients on
an S809 airfoil both experimentally and through CFD
analysis. The results of Wolfe and Ochs study are outlined
in table 1. To conduct a meaningful comparison, the
parameters of the present analysis were chosen to replicate
the study done by Wolfe and Ochs. This meant that the
analysis was done in 2 dimensions, excluding the effect of
downwash on the tip of the blades. The Reynold
number of the flow was set to 2106 and the chord length
was set to 0.60m. The chosen angles of attack for the
analysis were also accordingly set to 0, 1.02, 5.13, and
9.22. The CFD results in the study by Wolfe and Ochs
were calculated using a k- turbulence model. In the
present CFD analysis, the k- model was used for a point
of best possible comparison, as well as other turbulent
models for further verification.

Operating Assumptions
The analysis is done in steady state. The working
fluid is air, chosen at International Standard Metric
Conditions (15 C and 101.325 kPa). Based on the
viscosity and density of air in these conditions, a chord
length of 0.60m and a Reynolds number of 2106, the free
stream velocity was determined to be 49 m/s. The
roughness constant of the airfoil was not given in the
Wolfe and Ochs analysis. Wind turbine blade roughness
depend on manufacturing techniques, and vary through
contaminating agents such as dust, dirt, ice and insects
(Sagol et. al., 2013). Providing a definite roughness factor
is therefore not a meaningful property. In this analysis the
default roughness constant of 0.5 was utilized.
Meshing Technique
To model the flow around the airfoil, an external shell
was made around the airfoil to represent the far field
boundary. To accurately represent this boundary where
the effect of the airfoil was negligible, the outline of the
shell had to be extended to 20 times the chord length
behind the airfoil and 10 times the chord length above,

MODELING METHODOLOGY
The following underlines the modelling assumptions,
meshing techniques, discretization methods and solver

below and to the front of the airfoil (ANSYS, 2014). The

front side of the shell was formed into a half-circle to
allow for a C type mesh appropriate for modelling flow
around an airfoil. The C-mesh was designed manually and
can be seen in figure 2.

A second consideration is in the aspect ratio of the

cells near the air foil wall. To maximize accuracy the
mesh was designed to have the lowest possible aspect
ratio cells near the airfoil wall, aiming for nearly square
shaped cells. However, there was an upper limit within
ANSYS on how many cells could horizontally fit across
the top and bottom side of the airfoil. This caused a slight
inconsistencies on cell sizes around the airfoil in high
resolution meshes, as well as increasing the aspect ratio of
these cells as seen in figure 3. However, when results
were compared to Wolfe and Ochs, these high resolution
meshes still produced better accuracies.

Various mesh resolutions were tested for mesh

convergence. The mesh resolutions were categorized and
named in accordance to the number of cell divisions along
the radius of the front section of the shell. The mesh in
figure 1 has 50 divisions along the radius and is therefore
referred to as mesh-50. This mesh was not of high enough
resolution to give accurate results, but serves as a good
tool to illustrate the meshing technique.

Discretization methods
The solution was obtained through implicit
formulation, which is more stable than explicit
formulation and allows for shorter convergence times
(ANSYS,2014). The Least Square Cell Based gradient
method was used as the least computationally expensive
option with reasonably accurate results. For pressure,
momentum, energy, and dissipation rate, the second order
upwind discretization was used. This is because the first
order discretization methods are insufficient to accurately
predict drag (ANSYS, 2014).
Solver Algorithms
The CFD analysis was carried using typical 2
equation viscosity turbulent models. The k- turbulent
model provided a point of best reference to the Wolfe and
Ochs results. This was the least computationally
expensive option, known to produce reasonably good
results for unseparated flows with relatively small
pressure gradients (Bardina et. Al., 1997). To account for
the possibility of flow separation at higher angels of
attack, the analysis was also conducted using the standard
k- model. This model has proven to have better
numerical stability than the k- model, primarily in the
viscous boundary layer near the airfoil wall (Bardina et.
Al., 1997). However, the standard k- model is more
computationally expensive than the k- model. This
analysis will determine if the gain in accuracy is
significant enough to justify using the standard k-
especially when flow separation happens.

Figure 2. Mesh-50 outline around S809 airfoil

There were two important considerations in meshing
technique. Firstly there is an inflation of mesh nodes along
the boundaries of the airfoil, especially the trailing edge
where flow is susceptible to separation, and along the
extension of the chord line to the end of the shell where
vortices are shed. To obtain accurate results it is important
to have y+ values over 30 along the edges of the airfoil,
(Wolfe and Ochs , 1997) and this criteria was met for all
meshes used in this analysis.

The analysis was also conducted using the k- Sheer

Stress Transport (SST) model because it combines
desirable elements of the standard k- and k- model.
The SST model uses zonal weighting of model
coefficients to behave as the k- model near the airfoil
walls where viscous effects dominate. However these
coefficients behave as the k- model near the boundary
edges of the free-shear layers (Bardina et. Al., 1997). The
result is a moderately computationally expensive model
with the increased viscous accuracy near the airfoil where
accuracy is necessary. This report explore the outcome
significance of using different models to conduct the CFD
analysis.
In all cases the convergence criteria of the solution
was a residual smaller than 10^-4 in all parameters. The
program was set to terminate after a maximum of 700
iterations if convergence had not been met.

Figure 3. Close up of mesh-175 around S809 airfoil

MESH CONVERGENCE
To analyse the quality of mesh resolutions, the results
obtained using each mesh resolution was compared
against the experimental results given by Wolfe and Ochs.
This was done for each turbulent model but only at 0
angle of attack. Table 1 illustrates the percent errors
achieved in coefficient of lift (Cl), coefficient of drag (Cd),
and coefficient of moment (Cm), as mesh quality was

increased. The increase in mesh resolution had to be

significant in order to avoid results that falsely appeared to
be converging to a final value. As seen in table 1, the
percent errors in all coefficients decrease as mesh
resolution increases, for all models. To follow this trend,
the analysis proceeded to increase the resolution until a
finer mesh no longer resulted in significantly lower

Table 1. Mesh Convergence Results

Cl
Turb.
Model
k-

k-

SST

Wolfe's
Exp.
Result

Cd
Error
x 104
890

%
Error
60.6%

Result
0.0280

864

58.8%

0.0263

841

57.3%

0.0261

906

61.7%

0.0280

754

51.4%

0.0263

0.2147

678

46.2%

100

0.2114

645

150

0.1824

175

0.1726

Mesh
100

Result
0.2359

150

0.2333

175

0.2310

100

0.2375

150

0.2223

175

0.1469

0.1469

0.1469

Wolfe's
Exp.
Result

Cm
Error x
104
210

%
Error
300%

Result
-0.0702

193

276%

-0.0688

191

273%

-0.0677

210

299%

-0.0705

193

275%

-0.0664

0.0257

187

267%

43.9%

0.0277

207

355

24.2%

0.0259

257

17.5%

0.0254

0.0070

0.0070

0.0070

percent errors. However, the program crashed when using

mesh-200 (meaning 200 divisions along the radius of the
surrounding shell). Therefore, mesh convergence was

Wolfe's
Exp.
Result

Error x
104
-259

%
Error
58.6%

-245

55.3%

-234

52.9%

-262

59.0%

-221

49.9%

-0.0645

-202

45.6%

296%

-0.0644

-201

45.3%

189

270%

-0.0566

-123

27.8%

184

262%

-0.0540

-97

21.9%

-0.0443

-0.0443

-0.0443

terminated at mesh 175, giving the most accurate

achievable results as seen in Table 1.

Table 2. Comparison of CFD Results with Experimental Results by Wolfe (1997).

Cl
Turb.
Model

k-

k-

SST

Cd

Angle

Result

Wolfe's
Exp.
Result

0.2310

0.1469

1.02

0.3301

0.2716

585

21.6%

5.13

0.7119

0.7609

-490

-6.4%

9.22

1.0599

1.0385

214

0.2147

0.1469

678

1.02

0.3131

0.2716

415

5.13

0.6702

0.7609

-907

9.22

1.0244

1.0385

-141

0.1726

0.1469

257

1.02

0.2651

0.2716

-65

5.13

0.6136

0.7609

-1,473

9.22

0.8953

1.0385

-1,432

-13.8%

Cm
Error x
104

%
Error

Result

Wolfe's
Exp.
Result

0.0070

191

273%

-0.0677

-0.0443

-234

52.9%

0.0297

0.0072

225

312%

-0.0698

-0.0491

-207

42.1%

0.0592

0.0070

522

745%

-0.0742

-0.0609

-133

21.9%

2.1%

0.1140

0.0214

926

433%

-0.0748

-0.0495

-253

51.1%

46.2%

0.0257

0.0070

187

267%

-0.0645

-0.0443

-202

45.6%

15.3%

0.0284

0.0072

212

294%

-0.0659

-0.0491

-168

34.2%

-11.9%

0.0562

0.0070

492

703%

-0.0685

-0.0609

-76

12.4%

-1.4%

0.1081

0.0214

867

405%

-0.0697

-0.0495

-202

40.8%

17.5%

0.0254

0.0070

184

262%

-0.0540

-0.0443

-97

21.9%

-2.4%

0.0276

0.0072

204

283%

-0.0548

-0.0491

-57

11.6%

-19.4%

0.0538

0.0070

468

668%

-0.0592

-0.0609

17

-2.9%

0.1046

0.0214

832

389%

-0.0627

-0.0495

-132

26.7%

Error x
104

% Error

Result

841

57.3%

0.0261

Wolfe's
Exp.
Result

Error x
104

% Error

Roughness height of 0
Spatial Discretization:
Gradient of Least Squares Cell Based
Second order in Pressure
Second order upwind in Momentum
Second order upwind in Turbulent kinetic energy
Second order Upwind in Turbulent Dissipation Rate

VALIDATION OF RESULTS
To analyse the quality of mesh resolutions, the results
obtained using each mesh resolution was compared
against the experimental results given by Wolfe and Ochs.
Table 1 illustrates
The default conditions that we used were:
Turbulent Intensity of 5%
Turbulent Viscosity Ratio of 10
Roughness constant of 0.5

The working fluid was air at International Standard Metric

Conditions (15 C and 101.325 kPa).

SYNOPSIS
The purpose of this document is to propose a topic of
study for the ENME 547 final project presentation. The
topic of study that will be pursued is as follows: the
computational fluid dynamics (CFD) analysis of moving
air acting upon a stationary wind turbine blade. The
purpose of this study is to analyse the resultant forces

acting upon the stationary wind turbine blade as a function

of the airs velocity and the blades angle of attack. This
document also contains the following additional
information: background knowledge and brief overview of
wind turbine operating principle, objectives for pursuing

Table 3. Comparison of CFD Results with CFD Analysis by Wolfe and Ochs (1997)
Cl
Turb.
Model
k-

k-

SST

Cd

Cm

angle

Result

Wolfe's
k-
Result

0.2310

0.1324

986

74.5%

0.0261

0.0108

153

142%

-0.0677

-0.04

-277

69.3%

1.02

0.3301

0.2494

807

32.4%

0.0297

0.0110

187

170%

-0.0698

-0.0426

-272

63.8%

5.13

0.7119

0.7123

-4

-0.1%

0.0592

0.0124

468

377%

-0.0742

-0.0513

-229

44.7%

0.2147

0.1324

823

62.2%

0.0257

0.0108

149

138%

-0.0645

-0.04

-245

61.3%

1.02

0.3131

0.2494

637

25.5%

0.0284

0.0110

174

158%

-0.0659

-0.0426

-233

54.7%

5.13

0.6702

0.7123

-421

-5.9%

0.0562

0.0124

438

353%

-0.0685

-0.0513

-172

33.5%

0.1726

0.1324

402

30.4%

0.0254

0.0108

146

135%

-0.0540

-0.04

-140

35.0%

1.02

0.2651

0.2494

157

6.3%

0.0276

0.0110

166

151%

-0.0548

-0.0426

-122

28.7%

5.13

0.6136

0.7123

-987

-13.9%

0.0538

0.0124

414

334%

-0.0592

-0.0513

-79

15.3%

Error x
104

% Error

Result

Wolfe's
k-
Result

Error x
104

% Error

Result

Wolfe's
k-
Result

Error
x 104

% Error

TOPIC OF STUDY
The topic of study will be restated in this section so
that it is explicitly clear to the reader. At this point is it
unclear how the velocity of the air or the angle of attack
affects the resultant forces that act on the specific turbine
blade geometry shown in figure 1. CFD analysis will be
used to model the flow of moving air passing a stationary
turbine blade with geometry shown in figure 1, to
determine the resultant forces that act upon the turbine
blade. This topic of study is relevant to the course material
because it involves using numerical methods to solve fluid
dynamics.

this topic of study, methodology o accomplish

analysis, projected timeline, and additional references
related to the topic of study.

APPLICATION OF RESULTS
Using the results from a 2 Dimensional airfoil analysis
to design a 3 Dimensional wind turbine blade has been
commonly practiced traditionally. To this day, this method
is the chosen industry approach in designing wind turbine
blades in accordance with blade element theory (Liu &
Janajreh, 2012). Blade element theory divides a 3
dimensional blade in ten to twenty sections along the
length of the blade. Each section is then averaged to its 2
dimensional airfoil. The Cl, Cd, and Cm data at various
Reynold number and various angles of attack for a chosen
airfoil is then used to decide the desired twist angle for
each of the blade sections (Emrah, 2011). A 3 dimensional
blade can therefore be designed from the results of the
current analysis, if it is extended to include more
Reynolds numbers.

INTRODUCTION AND BACKGROUND

Wind turbines are becoming more common
throughout the world today. Society is shifting towards
green forms of energy, placing increased emphasis on
renewable power sources. As a result, wind power has
been increasingly more important as part of a countrys
energy profile. In the past, the cost to power ratio was a
large deterrent to developing wind power, but significant
research has and continues to be done to further the
effectiveness of these turbines.
CFD analysis has contributed significantly to this
field, furthering the understanding of wind turbines by
allowing complex analysis of the airflow.

Having said that, currently the preferred method of

conducting airfoil analysis is using experimental results.
This is due to the relatively large error present in average
CFD analysis and the extremely high computational cost
of high accuracy CFD analysis (Liu & Janajreh, 2012).
However, CFD analysis with accuracies such as the one
presented in this study is often used as a first point of
reference for testing a new airfoil design because it is
relatively quick and cheap.

Wind Turbine Operation Principle

Wind turbines function by converting the kinetic
energy of moving air into electric energy. This is
accomplished by exploiting the conservation of energy
principle. A key component to wind turbine design is the
aerodynamic profile of the turbine blades. The profile of
each blade acts as an airfoil, generating different air
pressures along the sides of the blade. One side of the

METHODOLOGY
Due the complexity of modelling all blades of a wind
turbine, this analysis will focus on a single stationary
blade. The blade will be stationary in order to analyse the
moving air in the blades relative frame of reference as
opposed to the absolute frame of reference. The program
ANASYS will be used to model the geometry of the
turbine blade. In particular, the blade profile to be
analysed is S809, which is a standard profile used in
existing wind turbines (Wolfe & Ochs, 1997). Figure 1
shows the cross sectional geometry of a S809 turbine
blade profile.

turbine blade has a longer length than the other side. This
causes air to flow faster over the longer side, due to the
continuity of the airstream. According to the conservation
of energy principle, an increase in the fluids velocity will
result in lowering the fluids pressure. This subsequently
causes the fluid on the longer side of the turbine blade to
have a lower pressure. The pressure difference acting
along the sides of the blade generates a net force. It is this
force that results in the rotational motion of the rotor. The
rotation of the rotor is then used to power an electric
generator that produces electrical energy.
Factors Affecting the Blade Profile Effectiveness.
Wind turbines are capable of rotating their blades to
optimize the angle of attack. This is the angle between the
oncoming flow and the blade, specifically the chord line
of the blade. This has a significant effect on how the air
flows past the blades, subsequently affecting the forces
exerted on the profile.
The magnitude of the winds velocity will also have
an effect on these forces. Higher speeds will often
generate increased forces on the turbine blades.

Once the geometry has been modelled, the CFD

functionality of ANSYS will be used to generate a mesh
that encompasses the fluid around the model. FVM
numerical methods will then be used to analyse and
document the resultant forces that act on the turbine blade.
Assumptions
In order to complete this analysis, a number of
assumptions must be made: steady state flow,
incompressible flow, fluid acts as a Newtonian fluid,
neglected thermal effects, no slip along boundary walls,
turbine blade geometry does not change along the length
of the blade, two dimensional flow, irrotational flow,
neglected gravitational effects, and the turbine blade
develops a turbulent boundary layer.

OBJECTIVES
The objectives for this project are as follows: to
develop an in-depth understanding of the CFD
functionality of the program ANSYS, to display an
understanding of FVM numerical methods to solve
complex problems related to fluid mechanics, to display
an understanding of boundary conditions and how to
implement them in CFD analysis, and to display an
understanding of the material learned in ENME 547. One
objective for this analysis is to find a specific blade angle
of attack that will result in a maximum resultant force for
a blade with geometry shown in figure 1. Another
objective for this analysis is to examine the relation
between the velocity of the moving air and the resultant
force acting on the blade.

Comparison with Literature

To compare the validity of the results, this analysis
will be compared with the papers listed in the reference
section. In particular, Wolfe and Ochs (1997) have
conducted a similar analysis of S809 aerodynamic
characteristics, publishing how lift and drag coefficients
vary with angle of attack.

Table 1. Projected timeline

Task
Finalizing topic
Finish Abstract
Make the physical model and mesh on ANSYS,
Finalize assumptions/ boundary conditions/ operating conditions/
discretization method and apply to model on ANSYS
Conduct the CFD analysis with ANSYS
Start report
Start PPT presentation
Finalize PPT presentation
Practice Presentation
Present
Finish final report
ESTIMATED TIMELINE
The project will follow a set of internal deadlines for
key milestones, as summarized in table 1.

Internal Deadline
Sept .23
Oct .09
Oct. 19
Oct. 19
Nov.9
Nov. 12
Nov. 12
Nov.27
Nov.27 Nov.30
Dec.1 Dec.5
Nov.27

External Deadline
Oct.10

Dec. 1

REFERENCES
Gangele, A., and Ahmed, S., 2013, "Modal Analysis
of S809 Wind Turbine Blade Considering Different
Geometrical and Material Parameters", Journal of The
Institution of Engineers (India): Series C, ISSN 22500545, 07/2013, Volume 94, Issue 3, pp. 225 228.

http://www.sciencedirect.com/science/article/pii/S136403
2113001366
ANSYS (2014) Workshop 04: Fluid Flow Around the
NACA0012 Airfoil. ANSYS, Inc.
Bardina, J. E., Huang, P.G., Coakley, T.J., (1997).
Turbulence Modelling Validation, Testing, and
Development. NASA Technical Memorandum 110446.

Kaminsky, C., Filush, A., Kasprzak, P., and Mokhtar,

W., 2012, "A CFD Study of Wind Turbine
Aerodynamics", Technical Report HTL-26, CFD-4, Grand
Valley State University, Grand Rapids, MI. Retrieved
from
http://www.cdadapco.com/sites/default/files/conference_proceeding/pdf/
Kaminsky.pdf

Liu and Janajreh, (2012). Development and

application of an improved blade element momentum
method model on horizontal axis wind turbines.
International Journal of Energy and Environment
Engineering 3:30. doi: 10.1186/2251-6823-3-30

National Renewable Energy Laboratory (NREL),

2014, "NREL's S809 Airfoil Graphic and Coordinates",
S809 Airfoil Shape, University of Illinois at UrbanaChampaign.,
Champaign,
IL.
Retrieved
from
http://wind.nrel.gov/airfoils/shapes/S809_Shape.html

Emrah Kulunk (2011). Aerodynamics of Wind

Turbines, Fundamental and Advanced Topics in Wind
Power., Dr. Rupp Carriveau (Ed.), ISBN: 978-953-307508-2,
InTech,
Retrieved
from:
http://cdn.intechopen.com/pdfs-wm/16241.pdf

Poudel, R. C., and Kato, Y., n.d., "A Steady-state CFD

Analysis of S809 Airfoil with 3D Grid", American
Institute of Aeronautics and Astronautics, Northern
Arizona University, Flagstaff, AZ. Retrieved on October
8,
2014
from
http://www.academia.edu/7670589/A_Steadystate_CFD_Analysis_of_S809_Airfoil_with_3D_Grid
Wolfe, P., Ochs, S., 1997, CFD Calculations of S809
Aerodynamic Characteristics, AIAA-97-0973. Retrived
from
http://energy.sandia.gov/wp/wpcontent/gallery/uploads/AIAA-97-0973-SAND20118298C.pdf
Xie, Y., Chen, J., Qu, H., Xie, G., Zhang, D., and
Moshfeghi, M., 2013, "Numerical and experimental
investigation on the flow separation control of S809 airfoil
with
slot",
MATHEMATICAL
PROBLEMS
IN
ENGINEERING, ISSN 1024-123X, 2013, Volume 2013,
pp. 1 14
NASA., (n.d.). Downwash Effects on Lift. Retrieved
from
http://www.grc.nasa.gov/WWW/k12/airplane/downwash.html (November, 2014)
Ingram, G., (2011). Wind Turbine Blade Analysis
using Blade Element Momentum Method. Version 1.1.
Retreived
from
https://community.dur.ac.uk/g.l.ingram/download/wind_tu
rbine_design.pdf
Musial, W.D., Butterfield, C. P., Jenks, M. D., (1990).
A Comparison of Two- and Three- Dimensional S809
Airfoil Properties for Rough and Smooth HAWT Rotor
Operations.
Retrieved
from
http://www.nrel.gov/docs/legosti/old/3603.pdf
Sagol. E., (2013) Issues Concerning Roughness on
Wind
Turbine
Blades.
Retrieved
from