simultaneous use of wind tunnel testing and cfd the...
Post on 14-Aug-2020
3 Views
Preview:
TRANSCRIPT
Suzlon Energy Ltd.
1 Suzlon wind farm in Utah, USA
Simultaneous use of wind tunnel testing and CFD – The numerical wind tunnelC. Mau, G. Tescione, O. UzolSuzlon Blade Science Center
Suzlon Energy Ltd.Outline
2
• Motivation for simultaneous use of WT and CFD
• Initial wind tunnel CFD models
• Wall interference
• Wind tunnel CFD framework
Suzlon Energy Ltd.Motivation for simultaneous use of WT and CFD
3
• Wind tunnel test campaign of very thick airfoils:• Validation of numerical tools or at least get an idea how much they are off• Use of wind tunnel polars within design process
• What is a very thick airfoil?
Suzlon Energy Ltd.Motivation for simultaneous use of WT and CFD
3
• Wind tunnel test campaign of very thick airfoils:• Validation of numerical tools or at least get an idea how much they are off• Use of wind tunnel polars within design process
• What is a very thick airfoil?
Risø C2 extrapolated – t/c = 42% Risø C2 extrapolated – t/c = 59%
Suzlon Energy Ltd.Motivation for simultaneous use of WT and CFD
4
1.8 m 2.74 m
1.2
5 m
TU Delft LTT cross section DWG wind tunnel cross section
Wind tunnels and test conditions :
• t/c = 42% @ TU Delft - Low turbulence wind tunnel (LTT)• t/c = 59% @ Deutsche WindGuard - Large wind tunnel in
Bremerhaven (DWG)• Re ≤ 3e6 - Chord = 0.5 m• High Re need -> conflicts with low blockage/high AR
Suzlon Energy Ltd.Motivation for simultaneous use of WT and CFD
5
• Not much information available in open literature• High blockage• Flow starts to separated @ low AoA for t/c = 59%• No experience with very thick airfoil testing• High expenses for wind tunnel tests• High uncertainty in validity of the test results
Risk mitigation
1. Step -> Run wind tunnel + airfoil in CFD to check what the flow will look like
Suzlon Energy Ltd.Initial wind tunnel CFD models - setup
6
CFD model of t/c = 59% @ DWG
CFD model of t/c = 42% @ LTT
Model settings:
• RANS • incompressible• SST turbulence model• γ-Reθ transition model
Boundary conditions:
• Velocity inlet – Re = 3e6• Pressure outlet• Symmetry BC• No slip smooth wall
Suzlon Energy Ltd.Initial wind tunnel CFD models – First results
7
t/c = 59% - AoA = 15 deg @ DWG – Wall shear stress mag. - WT wall highlighted grey
• Large LE radius leads to corner separation due to wall interference
• Corner separation vortex covers large or total area of the airfoil SS
• High uncertainty in polars extracted from WT tests can be expected
t/c = 42% - AoA = 10 deg @ LTT – Wall shear stress mag. - WT wall highlighted grey
Suzlon Energy Ltd.Wall interference: The horseshoe vortex
8
Flow around buidingsMartinuzzi, R., Tropea, C.; JFE 1993
Scour of cylindrical pylons in water Das et al., 2013
• The adverse-pressure gradient due to the stagnation on the model leading-edge induces separation in the wind tunnel wall boundary layer, resulting in a vortex system around the airfoilGas turbine blade cascades
Goldstein and Spores, 1988
Fuselage wing junction Boermans et al. 1998
Suzlon Energy Ltd.Wall interference: Remedies
9
• Increase model aspect ratio
• Splitter plates
• Boundary layer blowing
• LE strakes
• Boundary layer fences
• Corner VGs
t/c = 59% - AoA = 15 deg @ DWG, Lambda2 criterion
Suzlon Energy Ltd.Wall interference: Remedies - Increase aspect ratio
10
1.2
5 m
TU Delft LTT cross section
• Increase airfoil model aspect ratio:
• WT dimensions fixed • reduce chord would give too low Re
Suzlon Energy Ltd.Wall interference: Remedies - Splitter plates
11
airfoil
Splitter plate
Win
d t
un
nel w
all
• Splitter plates : • Less developed boundary layer, reduction vortex system around the LE • Confines horseshoe vortex between wind tunnel wall and splitter plate• Quite extensive design task -> not doable within time frame• Additional wind tunnel correction needed
Suzlon Energy Ltd.Wall interference: Remedies - Boundary layer blowing
12
• Boundary layer blowing: • Energize BL by injecting additional flow parallel to the
wind tunnel wall and suppress or reduce separation due to horseshoe vortex
• Available @ DWG wind tunnel• Max. blowing ratio (unozzle/u0) of 1.2 -> gives no
improvement• Airfoil is too thick and intersects with possible location
of blowing nozzles
t/c = 59% - AoA = 15 deg with tangential blowing @ DWG top: Lambda2 criterion + Nozzle streamlines bottom left: SS wall shear stress mag. without blowingbottom right: SS wall shear stress mag. with blowing
Suzlon Energy Ltd.Wall interference: Remedies - Corner VGs
13
Corner VGs:
• Energize BL close to wall junction and keep the flow attached• Standard method -> also used for semi thick airfoils• Easy to apply and to test during the measurement campaign
t/c = 59% @ DWG with corner VGs
Suzlon Energy Ltd.Wall interference: Remedies - LE strake
14
Devenport et al., 1992
LE strakes:
• relieves the adverse-pressure gradient on the wind tunnel wall boundary layer
• effectively decrease gradually the curvature radius of the model LE• different design approaches• might be optimize for AoA• rapid prototyping (3D printed)
Suzlon Energy Ltd.
15Pressure probes location Transition line Separation line
Wall interference: Remedies – LE strake @ t/c = 59%
t/c = 59% - AoA =15 deg @ DWG without LE strake SS wall shear stress mag.
t/c = 59% - AoA =15 deg @ DWG with LE strake SS wall shear stress mag.
Suzlon Energy Ltd.
15
t/c = 59% - AoA =15 deg @ DWG Raw IR-camera image - SS from wind tunnel test with strake
Airfoil
Pressure probes location Transition line Separation line
• Strake straightens out transition line
• Strake reduces wash down on the SS and gives a more straight seperation line around the midspan
• Good agreement of the transition and separation line between experiment and CFD
t/c = 59% - AoA =15 deg @ DWG without LE strake SS wall shear stress mag.
t/c = 59% - AoA =15 deg @ DWG with LE strake SS wall shear stress mag.
Wall interference: Remedies – LE strake @ t/c = 59%
Suzlon Energy Ltd.
16
w/o strakes and vorner VGs (left):• Corner vortices and separation on a ”W” line• Curved transition line• On the mid-span (PP location) separation is
delayed
With strakes and corner VGs (right):• Corner vortices suppressed• Straight transition line• Separation is more uniform (but it starts on the
mid-span)
Wall interference: Remedies – LE strake & corner VGs @ t/c = 59%
t/c = 59% - AoA =15 deg @ DWG Raw IR-camera image - SS from wind tunnel test without strake and corner VGs
t/c = 59% - AoA =15 deg @ DWG Raw IR-camera image - SS from wind tunnel test with strake and corner VGs
Suzlon Energy Ltd.
17
t/c = 42% - AoA =10 deg @ LTT Raw IR-camera image - SS from wind tunnel test with strake
Pressure probes location Transition line Separation line
• Different design approaches for the LE strake where also tested in CFD for the t/c = 42 % airfoil
• LE strake gave no improvement in CFD and it was also confirmed during the WT test
• LE strake was no option for t/c = 42 % airfoil
• Another solution had to be found
t/c = 42% - AoA =10 deg @ LTT with LE strake SS wall shear stress mag.
Wall interference: Remedies – LE strake @ t/c = 42%
t/c = 42% - AoA =14 deg @ LTT without LE strake SS wall shear stress mag.
Suzlon Energy Ltd.
18
Wall interference: Remedies - Boundary layer fences
Boundary layer fences:
• Does not prevent HS vortex but it confines it• Prevent the HS vortex to travel spanwise causing
the corner vortices• different design approaches• works for several AoA until limit• rapid prototyping (laser cut)
Suzlon Energy Ltd.
19
Wall interference: Remedies - Boundary layer fences
• Corner vortex is caught between wind tunnel wall and fence
• transition line is straightened out
• Suppresses wash down on the SS
• Separation is more uniform (but it starts on the mid-span)
Pressure probes location Transition line Separation line
t/c = 42% - AoA =14 deg @ LTT without LE strake SS wall shear stress mag.
t/c = 42% - AoA =14 deg @ LTT with LE strake SS wall shear stress mag.
Suzlon Energy Ltd.
20
Wall interference: Remedies - Boundary layer fences
• Corner vortex is caught between wind tunnel wall and fence
• transition line is straightened out
• Suppresses wash down on the SS
• Separation is more uniform (but it starts on the mid-span)
Pressure probes location Transition line Separation line
t/c = 42% - AoA =14 deg @ LTT Raw IR-camera image - SS from wind tunnel test without fence
t/c = 42% - AoA =14 deg @ LTT Raw IR-camera image - SS from wind tunnel test with fence
Suzlon Energy Ltd.
21
Upfront CFD simulations of the wind tunnel:
• identified potential risks before the test campaign• were used to find solutions, which extend the validity and quality of
the tests• saved time in the wind tunnel for the actual test campaign
Wall interference remedies:
• LE strakes in combination with corner VGs on a 59% airfoil succeded in reducing the horseshoe vortex and improving spanwise homogeneity of transition and separation lines
• BL fences on a 42% airfoil succeded in bounding the horseshoe vortex preventing corner separations and improving spanwise homogeneity
Wall interference: Remedies - conclusions
Suzlon Energy Ltd.Wind tunnel CFD framework
22
• Development of a ”push button” wind tunnel CFD framework • Used for 1 to 1 comparison to test results after the wind tunnel test campaign is completed• Tune and validate CFD models and setup
Suzlon Energy Ltd.Wind tunnel CFD framework - Input
23
Input:
• Geometry: airfoil shape, wind tunnel geometry (LTT, DWG, LM, Poul la Cour) or 2.5d• Test conditions: Re, T, ρ, Tu, TVR, AoA, clean/dirty, chord• Booleans and list for pressure probes at the airfoil• Booleans and geometry parameters for potential addons, e.g. slat, LE strake, fence, VG’s, etc.• Mesh size settings for global and local scaling, e.g. local refinement around transition location• Solver settings: steady/transient, time step size, inner iterations, number of iterations/time steps to average over
Suzlon Energy Ltd.Wind tunnel CFD framework - Output
24
1D Output:• Cl, Cd based on pressure probes and surface
integral• Transition location• VG drag penalty and forces• …
2D Output:• cp distribution along pressure probe location• cf distribution at different spanwise positions• Velocity distribution downstream of VG array• …
3D Output:• Airfoil surface streamlines• Velocity and pressure field around the airfoil• Flow field downstream of VG array• …
Suzlon Energy Ltd.Wind tunnel CFD framework - Examples
25
• Airfoil tested @ DWG• Running AoA sweep in 3D CFD• Tuning of CFD model (mesh settings, model parameters, etc.)• Good agreement in terms of Lift and transition location
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
0 2 4 6 8 10 12 14 16
AoA
Cl
raw Exp. PP
3D CFD PP
corr. Exp.
2D CFD
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0 2 4 6 8 10 12 14 16
AoA
Cd
raw Exp. PP
3D CFD PP
corr. Exp
2D CFD
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 2 4 6 8 10 12 14 16
x/c
AoA
Transition SS/PS
Exp. SSExp. PS3D CFD SS3D CFD PS2D CFD SS
Suzlon Energy Ltd.Conclusions
26
Before wind tunnel test – CFD First:• Identify risks for testing of unconventional geometries• Find solutions, which extend the validity and quality of the tests• Helps to find preferable position of pressure probes• Saves time in the wind tunnel for the actual test campaign
After wind tunnel test:• 1 to 1 comparison to wind tunnel test data• Helps to find explanations, why there are deviations between tests and simulations • Very useful to tune the settings and models of 3D CFD setup
Suzlon Energy Ltd.
Suzlon wind farm in Utah, USA
Thank you!
Simultaneous use of wind tunnel testing and CFD – The numerical wind tunnelC. Mau, G. Tescione, O. UzolSuzlon Blade Science Center
Suzlon Energy Ltd.Using IRT for transition location detection
25
• Very accurate detection of transition location from WT test
• Live monitoring allows to detect disturbances and help to increase the validity of the test results
Raw images processing:
• find max. grey scale gradient along each horizontal pixel line
• Transition location is dertermined with help of calibration images correlating pixel and chordwise position
Suzlon Energy Ltd.Using IRT for flow visualization
26
AoA = 7.5 deg
AoA = 12.5 deg
AoA = 15 deg0.0
0.5
1.0
1.5
2.0
2.5
0 5 10 15 20 25 30
Cl
AoA
Exp. PP
CFD PP Fine
Risø C2 extrapolated t/c = 42 % @ LTT without BL fence:
• Drop in lift polar around 7.5 deg in CFD - Why?• Corner vortex is already present at 7.5 deg in the CFD, but is not detected
in the experiment• At 12.5 deg corner vortex is present on one side in the experiment• At 15 deg there is a the corner vortex at both wind tunnel end walls• The difference in the slope of the polar between 7.5 and 12.5 deg is
related to a earlier development of the corner vortex in CFD
Suzlon Energy Ltd.Wind tunnel CFD framework - Examples
24
0.0
0.5
1.0
1.5
2.0
2.5
0 2 4 6 8 10 12 14 16AoA
Cl
Exp. PP
CFD PP Trimmed QCR
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0 2 4 6 8 10 12 14 16AoA
Cd
Exp. PP
CFD PP Trimmed QCR
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 2 4 6 8 10 12 14 16
x/c
AoA
Transition SS/PS
Exp. SS Exp. PS
CFD SS Trimmed QCR CFD PS Trimmed QCR
• Airfoil tested @ DWG• Running AoA sweep in 3D CFD• Tuning of CFD model (mesh settings, model parameters, etc.)• Good agreement in terms of Lift and transition location
Suzlon Energy Ltd.Initial wind tunnel CFD models – First results
7
CFD model of t/c = 59% - AoA = 15 deg @ DWG CFD model of t/c = 42% - AoA = 10 deg @ LTT
• Large LE radius leads to corner separation due to wall interference
• Corner separation vortex covers large or total area of the airfoil SS
• High uncertainty in polars extracted from WT tests can be expected
top related