|
Items in cart [0] |
Questions? Call 888-624-8373 |
| Copyright © 2009. National Academy of Sciences. All rights reserved. Terms of Use and Privacy Statement |
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 485
Experimental And Numerical Investigation Of The Flow
Around The Appendices Of A Whitbread 60 Sailing Yacht
P. Planquart, M. Riethmuller (don Karman Institute for Fluid Dynamics, Belgium)
ABSTRACT
In this paper, we present results obtained
experimentally and numerically during the study of
the flow around the appendices of a Whitbread 60
sailing yacht. Experiments are carried out in a wind
tunnel on a model of reduced scale (1/154. Force
measurements are made using a three component
balance. Two different bulbs are tested for different
angle of attack. Velocity measurements are made
using the P.I.V. technique and a hot wire anemometer.
Vortex shedding frequencies are computed from the
P.I.V. results and drag values are obtained from the
velocity measurements in a plane behind the bulb
(wake survey analysis). The experimental results are
used to validate the numerical simulations performed
using the commercial code FLUENT 5.2. Different
turbulence models are compared and a good
agreement between the experimental and numerical
results was found with the use of the standard k-£
turbulence model and wall-functions.
INTRODUCTION
Figure 1 - Scaled model
Two different geometries of bulb have been tested.
The first one corresponds to the shape of an existing
bulb (Swarbrick 1997) and the second one was
designed at the von Karman Institute using simple
geometrical shapes.
WIND TUNNEL TESTS
Description of the facility
The experimental investigation is carried out
using the scaled model (scale 1/15) represented in
figure 1. This model is placed in the wind tunnel L2B
of the von Karman Institute (Figure 24.
1_1
An experimental and numerical study is
actually performed, at the von Karman Institute, on
the investigation of the flow around the appendices
and the hull of a Withbread 60 (now called VOR 60 -
in reference to the Volvo Ocean Race).
The study combines experiments on a scaled model in
a wind tunnel and numerical simulations using the
commercial code FLUENT. Wind tunnel experiments
have already been used extensively in the past (Tinoco
et al 1993, Caponnetto 1993) for the study of IACC
appendages and have provided useful information
regarding the effect of bulb and wingless on the
performance (Claughton et al 1998) of IACC keels.
The aim of this research is to validate CFD tools for
the design of appendices for a racing sailing yacht.
The validation of the numerical simulations is made
using the following data:
1. Velocity measurements around the model (mean Figure 2 - L2 wind tunnel
and instantaneous).
2. Measurement of drag and lift of the appendices
using a three-component balance.
The section of the wind tunnel is 35:k35 cm2 and the
maximum velocity that can be reached in the test
section is 35 m/s.
t..~..f..i...~....~..-...:.
i. .;.
The Reynolds number, when performing wind tunnel
tests, is lower than in the real case, but the main
OCR for page 486
objective of the research is to have a database for
validation, and therefore the Reynolds number is not
the key parameter. Also, wind tunnel tests allow easy
optical access which is an important point when using
an advanced optical measurement technique like the
Particle Image Velocimetry (P.I.V.~.
Description of the two geomletries
1. The first bulb, that will be referred as the
"existing" bulb has been tested in three different
configurations represented in the Figure 4. A side
view of the "existing" bulb is represented in
Figure 3.
Figure 3 - side view of the "existing" bulb
............ _ ~~ ~
. . . ... .. ... .. . . .
......................................... . . . ...
Figure 4 - configuration for the "existing" bulb
2. The second bulb, designed at the von Karman
Institute, consists of a body of revolution created
by the combination of an ellipse and a cone. This
bulb will be referred as the "new" bulb.
The comparison between the shape of the "existing"
bulb and "new" bulb is shown in the following
~-
ngures.
1 2 ~
(3 5~& art
4 5 ~
~ ~ 'art)
7 8 1
- ~,~.
~~! ~ ~~
CROS S-SECTION
Figure 6 - comparison of the two bulbs
Flow visualization
Surface flow patterns observed using oil
visualization are footprints of the outer field (Delery
19924. When using the surface oil flow technique, we
must first coat the hull and the appendices with a
special prepared paint. The air flowing over the
surface will carry the oil with it and a streaky deposit
of the powder remains to mark the direction of the
flow. Using this technique, we are able to analyze the
local direction of the flow at the surface and by
implication, the general flow structure in the three-
dimensional field above it. This information is also
used for the validation of skin friction computation.
The technique allows determining region of separation
and attachment of the flow on an obstacle. This
technique has been applied to study the "existing"
bulb and one result is given in Figure 7 for an angle of
attack of 6°. Close to the tip of the bulb, painting is
accumulating, meaning that the "existing" bulb has
not an optimal aerodynamic shape.
SIDE VIEW
TOP VIEW
Figure 5 - comparison of the two bulbs
Figure 7 - Oil visualization on the "existing" bulb
A vortex tube attached at the tip of the bulb has also
been observed when the keel is in incidence. This
vortex tube has been put in evidence using a small
piece of wool that has been translated inside the flow
field. By recording the X-Y coordinates of the tube,
OCR for page 487
we can find the exact position of the vortex tube.
Figure 8 shows the rotating wool and we have plot on
Figure 9 the position of the vortex tube with respect to
the hull, keel and rudder. It is worth mentioning that
the vortex tube does not reach the rudder.
Figure 8 - vortex tube - visualization.
. ___.
~ .........
it.
Figure 9 - position of the vortex tube for an angle of attack of 6°
P.I.V. measurements
The principle of P.I.V. (Lourenzo 1999) is to
illuminate particles injected in the flow by a Laser
sheet and to observe the scattered light. In order to
perform velocity measurements, two Laser pulses,
separated by a short and know time interval At, are
emitted to provide two images recorded on the same
photographic plate or by a CCD camera providing the
image in a numerical form. By measuring the
displacement of the particles, the velocity components
contained in the plane of the image is deduced in a
straightforward manner. P.I.V. is very precious for
the study of unsteady phenomena since it allows to
freeze the velocity field at a given instant.
Mean and instantaneous velocity fields have been
recorded in the three planes represented in Figure 10.
At
.....
....
.....
.....
I...
Figure 10 - measurement plane for the PIV
The measurements have been taken for the three
different keel configurations of the "existing" bulb.
Using the instantaneous data from the Particle Image
Velocimetry measurements, we were able to compute
the vortex shedding frequencies that characterize the
detachment of vortices at the keel. Shedding
frequencies for the "existing" bulb are given in Table
1. These frequencies should be divided by 45 for the
extrapolation to the real case.
The frequencies were computed using the distance
measured between a negative and a positive peak of
vorticity. Knowing the mean velocity, we can
compute a first value for the shedding frequency by
taking the consecutive distance between two peaks.
However, a statistical approach has been used to
compute all the possible combination of peaks and
different frequencies have been obtained. This
approach has been used because two consecutive
peaks of vorticity, as shown on the right of Figure 11,
might not been the result of two independent vortex
shedding.
| Frequency 1
T negative 8070
T positive 9390
Z 7450
Frequency2 | Frequency 3 |
4040 2630
4000 2550
3980 2470
Table 1 - frequencies computed from the PIV results
Figure 11 shows a typical P.I.V. results measured
behind the keel in the plane A represented in Figure
10. On the left side, we see the instantaneous velocity
field and on the right side a contour map of vorticity
extracted from the instantaneous velocity field using
relation (14:
014. AL
~ ~ .
~ = _
By Ox
(1)
OCR for page 488
55
50
45
4a
~ 35
al 36
~ 25
:
15
1a
, ...
3 10 20 30 40
X: Largeur lmml
an
Figure 11 - Instantaneous velocity field and vorticity
Drag and lift measurements
Forces on the keel and the bulb (drag and lift) are
measured using an in-house three-component balance
represented in Figure 12. The keel and the bulb
mounted on the balance are represented on Figure 13.
Drag curves for the existing bulb as a function of the
angle of attack is represented in Figure 14 for the
"existing" bulb and the new one. For the "existing"
bulb, the best results are obtained with the
configuration T-negative. The worst results are
obtained with the configuration T-positive.
Comparing the new bulb with the existing one, we
observe a decrease in the drag for all angle of attack
with the new bulb.
Drag value as function of angle of attack
Z 07
g 06
05
04
03 _
existing bulb - T positive .......
~ existing bulb - T negative
0 ~ :: existing bulb - Z
· new bulb
—Poly. ( new bulb)
~ Poly. (existing bulb - T positive)
------- Poly. (existing bulb - Z)
~Poly. (existing bulb - T negative)
s ~~
angle of attack- °
Figure 14 -Drag value as function of attack
Figure 12 - Schemativ view of the balance
·~ ~
Figure 13 - picture of the experimental set-up (balance + model)
Lift value as function of angle of attack
,
angle of attack - °
Figure 15 - Lift value as function of angle of attack
Lift curves are represented in Figure 15. We observe
a rapid increase of the lift for small angles of attack,
but starting at 3° to 5° of angle of attack, depending
on the shape of the bulb and the configuration, the lift
does not increase anymore due to separation. The
flow around the existing keel in configuration T-
negative separates for the highest value of angle of
attack. The flow around the new bulb starts to
separate for 3° of angle of attack. These low values of
angle of separation can be explained by the high
aspect ratio of the keel (Marchaj 19914.
Drag has also been computed by measuring the
velocity field in two planes located before and behind
the keel as represented in Figure 16. By computing
OCR for page 489
the balance of momentum between these two planes,
we can deduce the drag using relation (24. The
velocity measurements have been performed using a
hot wire anemometer mounted on a displacement
table. Mean values have been recorded, but also
R.M.S. value in order to compute the second term of
the right side of equation (24.
P Zulu (1-U— )lydz+ pil~u'~-u' Hyde
(2)
Figure 16 - plane of measurement
A typical velocity map (mean value) in the
downstream plane for configuration T-negative of the
existing bulb is shown Figure 17. This map has been
recorded 5 cm behind the tip of the bulb. The free
stream velocity is around 34 m/s and the minimum
velocity measured behind the bulb is equal to 30 m/s.
The R.M.S value of the velocity, in the same plane
and for the same configuration, is represented in
Figure 18.
Velocity magnitude ~ ale of attack 6°
Ann
Figure 17 - velocity magnitude - configuration T-negative
R. M S of the veloGi~ - angle of attack 6°
-300 -200 -1 00 o 1 00 Coo 300 400
Width [mml
Figure 18 - R.M.S. value - configuration T-negative
The following table shows the value of drag measured
by the balance compared with the value computed
using the velocity map (wake survey analysis).
configuration
DRAG
T - negative
T - negative
T - positive
T - positive
Z
Z
New bulb
New bulb
Angle of
attack
O
6
O
6
O
6
O
6
O
Balance
N
0.54
0.436
0.636
0.466
0.551
0.462
0.49
0.414
Wake survey
analysis
N
0.543
0.444
0.63
0.456
0.543
0.438
0.506
0.398
Table 2 - comparison of drag values for two angle
of attack
NUMERICAL SIMULATION
Description of the simulations
The numerical simulations are performed
using the commercial code FLUENT™ 5.2, which
solves the Reynolds Averaged Navier-Stokes
equations using a finite volume technique. The
following turbulence models have been tested in order
to determine their influence on the computation of
drag and lift: k-£ standard model, k-£ realizable model
and Spalart-Allmaras model. Wall functions are used
with the k-£ turbulence models. The Spalart-Allmaras
turbulence model is a relatively simple one-equation
model that solves a modeled transport equation for the
kinematic eddy (turbulent) viscosity and is well suited
for the computational of external flow around
aerodynamic shape (Spalart et al, 19944.
OCR for page 490
The geometry of the "existing" bulb has been created
using the same CAD file as for the construction of the
experimental model in order to avoid geometrical
differences. The computational domain is shown in
Figure 19. For an angle of attack of 0°, a symmetry
plane is used to reduce the number of cells. The grid
is generated using the commercial grid generator
GAMBIT.
S5O
it.\
\
Figure 19 - computational domain for the numerical simulations
A typical surface grid on the keel and the bulb for the
T-negative configuration is shown in Figure 20.
Triangular cells are used for the discretisation of the
surface of the bulb. Special attention has been taken
for the discretisation near the wall when generating
the volumetric grid. Tetrahedral cells are used for the
volumetric grid. Due to the limited computational
resources available for this study, the total amount of
discretisation cells was always below 500.000. All the
simulations are performed using a second-order
discretisation scheme (second-order upwind).
Figure 20 - surface grid of the keel and the bulb
Results - Angle of attack = 0°
The influence of the turbulence model on the
value of drag for the configuration T-negative of the
"existing" bulb is shown in the Table 3. A value of
0.44 N has been measured experimentally. A good
agreement is found when using a standard k-£
turbulence model (within 2% of the experimental
value).
Drag
Angle of attack = 0° N
k-£ standard turbulence model 0.430
k-£ realizable turbulence model 0.321
Spalart-Allmaras turbulence model 0.363
Table 3 - Drag for the T-negative configuration
Using the same turbulence model, the values of drag
for the three configurations of the "existing" bulb are
compared in Table 4. The relative variation of the
drag agrees well with the experimental observations,
but only the k-£ turbulence model gives quantitative
data close to the experimental values (within 2 %.
Configuration
T-negative
T-positive
z
Numerical
Simulation
k-£ std.
Drag - N
0.430
0.478
0.462
Numerical
. Simulation
. Spalart-Al.
Drag - N
. 0.363
0.380
0.376
Experl.
Results
Drag - N
0.436
0.466
0.462
Table 4 - Drag value for the different configurations
Results - Angle of attack = 6°
One simulation (configuration T positive) has
been performed for an angle of attack of 6°, using the
standard k-£ turbulence model and wall functions.
Values of drag and lift are given in Table 5. We
observe that the agreement with the experimental
values is not as good as for an angle of attack of 0°. It
should be mentioned that the discretisation is worst
than in the previous cases, because the computational
domain has doubled (no symmetry plane). An
analysis of the sensitivity of the results with the grid
size must still be performed for this case.
Configuration
T-positive
T-positive
.
Numerical
Simulation
Drag - N
0.54
Lift - N
2.89
Experimental
Results
Drag - N
0.63
Lift - N |
3.5 1
Table 5 - lift/drag - angle of attack=6°
OCR for page 491
Contours of static pressure on the bulb and on the keel
are shown in Figure 21. The contour of velocity
magnitude in a vertical plane located 5 cm behind the
tip of the bulb is represented in Figure 22. The deficit
of velocity due to the bulb is clearly shown but not the
one due to the keel. The velocity map confirms the
lower value of drag obtained by numerical
simulations.
4.80e+02
........ 3.40e+02
2.00e+02
e.OOe+01
-8. OOe+O 1
_ -2.20e+02
-3.60e+02
_ 1111 -5. 00e+02
3.35~+01
31e+O1
3.~+01
S.23e+01
3.lOe+01
3.1 5e+01
3.1 1e+01
3.07~+01
S.03e+01
2.~9e+01
2. gEe+0 1
Figure 21 - contours of pressure
Figure 22 - velocity magnitude 5 cm behind the tip of the bulb
Validation
The numerical results for drag and lift
generated by the keel are validated with the data
obtained using the balance, only when the simulations
are performed for no incidence. In these cases, a very
good agreement is found when using the k-£ standard
turbulence model with wall functions. The use of wall
function substantially saves computational resources,
because the viscosity-affected is not resolved and is
well suited for this type of computation. Results with
the Spalart-Allmaras, which is specifically designed
for aerospace applications, did not agree much with
the experimental data, but this is mainly due to the
insufficient grid resolution near the wall. Also this
model is not performing very well when there is a
change in length scale, such as when the flow changes
abruptly from a wall-bounded flow to a free shear
flow.
Simulations performed with an angle of attack of 6°
have not been validated by the experimental results.
Grid is now being modified in order to check the
sensitivity of the results with the grid size.
Contours of skin friction on the surface of the
"existing" bulb are validated with the visualization
using the surface oil technique. Near the tip of the
bulb, the skin friction reaches zero (Figure 23), and
this explains the accumulation of oil observed
experimentally. As mentioned already before, the
shape of the "existing" bulb is not an optimal
aerodynamic shape.
CONCLUSIONS
Figure 23 - contours of skin friction
Results of a detailed experimental
investigation of the aerodynamic performances of two
bulbs of different geometry have been presented.
Experimental results include visualization of
separation zone, the determination of vortex shedding
frequencies and the measurement of values of drag
and lift for different angles of attack. Two different
methods are used for the determination of drag and
similar results are obtained. One geometry has also
been investigated using numerical simulations with
the commercial code FLUENT 5.2. A very good
validation is found when computing the drag value for
no incidence. This good agreement has only been
found when using the standard k-£ turbulence model
with wall functions. For an angle of attack of 6°, the
comparison between the experimental and numerical
results is not so good. This might be due to the poor
grid resolution. Having now more computational
resources, the numerical simulations will continue on
more refined grids. The "new" bulb will also be
investigated numerically.
OCR for page 492
ACKNOWLEDGEMENTS
The athors would I kc to 6 mk the following
persons for thei help du ing the experiments or the
m merical simoktiom: F. Helleput, R D Vogh 1, T.
And 6 md D Dupont
REFERENCES
J. SwarLnek: D cwing of WOR 60
commumication, l997
E.N. Tinoco:' IACC Appendages studies', The
Elevendh Chescpeake Sailmg Yaht Symposium,
Jarulary 1993
M. Caponnetto: 'A review on 11 Moro di Venezic
Desig ', The Eleventh Chescpeake Sciling Yaht
Symposium, Jmuary 1993
A.R. Claughton, R.A. Shenoi, J.F. Welheome:
'Sciling Yaht desing: Theory', Addsion Wesley
Longm m Limited, 333 pages, l 998
J.M. D61ery: 'Physics of Vortical Flows', Jou md of
Al maf, Vol. 29, N°S, sept-oct 1992
S.F. Hoerner: 'Fluid-Dynamic Drag, P'aoticcl
i fommation on aerodynamic d cg md hyd odynamic
resistmce', publish dbythe Author, 1958
P.R. Spalart, S.RT. AUmaras: 'A one-equ~tion
tubuence model for aerodynamic flows', Lc
Recherche A6ro pati~le, n°l, 5-21,1994
L.M. Lourenio: 'Particulate imcge ad tmcking
velocimeby', von Kum m Institue L ctme Series,
cou se note 148,1999
C.A. Marehaj: 'Aero-hyd odynamics of sciling',
AdkrdColesNcutical,1991
FLUENT 5 User's Guide: Flu nt is c regi tered
trad mark of Flu nt Inco porcted
GAMBIT User's Guide: Gcmbit is c ~egistered
trad mark of Flu nt Inco porcted
Representative terms from entire chapter:
turbulence model
