DETAILED EVALUATION OF CFD PREDICTIONS AGAINST LDA MEASUREMENTS FOR FLOW ON AN AEROFOIL A. Benyahia...
-
Upload
adele-morrison -
Category
Documents
-
view
217 -
download
1
Transcript of DETAILED EVALUATION OF CFD PREDICTIONS AGAINST LDA MEASUREMENTS FOR FLOW ON AN AEROFOIL A. Benyahia...
DETAILED EVALUATION OF CFD PREDICTIONS AGAINST LDA MEASUREMENTS FOR FLOW ON AN
AEROFOIL
A. Benyahia 1, E. Berton 1, D. Favier 1, C. Maresca 1, K. J. Badcock 2, G.N.Barakos 2
1 L.A.B.M., Laboratoire d’Aérodynamique et de Biomécanique du Mouvement, UMSR 2164 of CNRS and University of Méditerranée,
163 Avenue de Luminy, case 918, 13288 Marseille Cedex 09, France
2 University of Glasgow, Department of Aerospace Engineering? Glasgow G12 8QQ, U.K
• Introduction and objectives
• Experimental methodology: LDV measurements
• Numerical methodology: PMB approach
• Results and discussion
• Conclusions and prospects
Presentation Outline
• The detailed knowledge of the boundary-layer response to unsteadiness induced by oscillating models is of major interest in a wide range of aeronautical applications.
• This work concerns an experimental and numerical investigation of the unsteady boundary-layer on a NACA0012 aerofoil oscillating in pitching motion.
• The experimental part of the present study is based on the Embedded Laser Doppler Velocimetry, developed at LABM for unsteady boundary-layer investigation.
• From the numerical point of view, simulations were performed using the PMB solver developed at GU based on a RANS approach of turbulence.
Introduction and objectives
• Introduction and objectives
• Experimental methodology: ELDV measurements
• Numerical methodology: PMB approach
• Results and discussion
• Conclusions and prospects
Presentation Outline
V
(t) = 0 + cos ( t )
(b) Pitching motion
Forced unsteadiness law
model
optical fibres
pitching fore-and-aft
particules seeding tube
0,5
mU
laser source
S2 Luminy wind-tunnel-Experimental Set-up
Embedded Laser Doppler Velocimetry Method (ELDV)
• Embedded optical head linked with the model
• Reference frame linked with the moving surface
• Focal length f = 400 mm• Survey along the chord
(x direction)• Survey along the normal
(y direction)Optical fibres
Beam-expander 45¡ mirror
Measuring volume
tn
0U°
vu
ymin = 0,2 mm
ymax = 145 mm
Optical head
Oscillating model
turntable
Transmiting block
Optical fibers
PM1
Oscilloscop
BSA (Burst Spectrum Analyser)
Postprocessing:
Labview Macintosh
sensor position
IEEE
SETUP SETTINGSMEASUREMENTS
7 8 9
4
1
5
2
6
3
0 +/-.ENTER
SHIFT
1
3.95
9.318
1.00
3.00
4.00
8
0
1560
64
10.67
BSA enhanced
DANTEC
Opticalhead
U V
A and B
TIME/DIV
and DELAY TIME
POWER
A TRIGGER
VOLTS:DIV
CH 1
465 B
OSCILLOSCOPEVOLTS:DIV
CH 2
TRIGGER
Lasersource
000000000
000000000
SETUP SETTINGSMEASUREMENTS
7 8 9
4
1
5
2
6
3
0 +/-.ENTER
SHIFT
1
3.95
9.318
1.00
3.00
4.00
8
0
1560
64
10.67
BSA enhanced
DANTEC
PM2
Color separator
ELDV Acquisition
data acquisition
t
cycle n
One period
cycle n-1
U
with 1 n Nc (Nc 150)
STEP 1
tT
Data plot on one period STEP 2
Uuniquefictif cycle
cycle unique fictif
tT
Statistic threatment on each intervalSTEP 3
une des 512 cases<U>
<U>
tT
STEP 4 Périodic signal/ Fourrier analyse
256 harmonics~U
u'(t) = U(t) - U(t)~
Unsteady measurements processing
• Introduction and objectives
• Experimental methodology: LDV measurements
• Numerical methodology: PMB approach
• Results and discussion
• Conclusions and prospects
Presentation Outline
Parallel Multi-Block (PMB) : RANS approach
for the turlence
PMB RANS approach principle
0
y
GG
x
FF
t
W ii
• Transport equations
TevuW ,,,
uhuvpuuF i ,,, 2 vhpvuvuG i ,,, 2
xyyxxxyxx qvuF ,,,0Re1 yyyxyyyxy qvuG ,,,0Re
1
ky
v
x
u
x
uTxx
3
2
3
22
ky
v
x
u
y
vTyy
3
2
3
22
y
v
x
uTxy
• The resolution method
Implicite scheme : 1,,,,
,,
,,1
,, 1
nkjikji
kji
nkji
nkji WR
Vt
WW
Turbulence models: 1 or 2 equations models. Spalart-Allmaras
(SA), k- et SST.
' 'ppp 'uuu 'vvv
PMB RANS approach principle
• Le modèle SA1
~ fT
31
3
3
1
c
f
~
2
12
21
~~~~.
1~~~
d
fccScDt
Dbb
222
~~
f
dSS
12 1
1
f
f
6
1
63
6
631
cg
cgf rrcrg 6
2 22~~
dSr
135.01 bc 32 622.02 bc 41.0 762.21 c 3.02 c 23 c 1.71 c
PMB RANS approach principle
• Le modèle k- k
T
kPkkVt
kkT
**.Re
1.
2.Re
1. kPV
t T
95 403 1009* 21 21*
PMB RANS approach principle
• Le modèle de transport du tenseur visqueux (SST)
12;1max aF
kT
2
22
500;
09.02maxtanh
yy
kF
kPkkVt
kkT
**.Re
1.
kFkPVt T
21
21 12.
Re
1.
4
22
21
4;
500;
09.0maxmintanh
yCD
k
yy
kF
k
202 10;2
max
kCDk
31.01 a 09.0* 41.0
PMB RANS approach principle
Results
• Introduction and objectives
• Experimental methodology: LDV measurements
• Numerical methodology: PMB approach
• Results and discussion
• Conclusions and prospects
Presentation Outline
2 meshes results
Fine mesh, 27501 pointsCoarse mesh, 6975 points
Lift coefficient
Fixed incidence case
Turbulence Reynolds Number
k- model with imposed transition at s/c=0.2
Différents model with or without transition
Turbulence Reynolds Number
Fully turbulent Transition fixed at s/c=0.2 Transition increasing linéairly between s/c=0.1
and s/c=0.3
Velocity Profiles
Velocity Profiles
SST model for differents values of k
SST models for differents transition localisations and
k=0.001
Turbulence Reynolds Number
15 deg: SST model with transition fixed at s/c=0.05
15 deg: SST model fully turbulent
Turbulence Reynolds Number
Vortex shedding with a characteristic dimensionless frequency of k= 0.51
12°: k- model fully turbulent
12°: SST model fully turbulent
12°: SSTmodel with transition location at s/c=0.2
Turbulence Reynolds NumberPitching motion, (t)=0+cos(t)0=6deg, 6 deg, k=c/2U=0.188
Lift coefficient Pitching motion, (t)=0+cos(t)0=6deg, 6 deg, k=c/2U=0.188
Hysteresis loops
Velocity Profiles
SST model with transition located at s/c=0.2 an k-model fully turbulent
• Introduction and objectives
• Experimental methodology: LDV measurements
• Numerical methodology: PMB approach
• Results and discussion
• Conclusions
Presentation Outline
• Using such ELDV measurement methods, the boundary-layer behaviour can be fully investigated and characterized in a moving frame of reference.
• Analysis of the effects of forced unsteadiness (due to the pitching motion) on B.L.
• Dependence on turbulence model and transition
• Better agreement with experiment for transitionnal models for static incidence, before stall. Fully turbulent dodels are more adapted for oscillation case
Conclusions