|
Items in cart [1] |
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 916
24th Symposium on Naval Hydrodynamics
Fukuoka, JAPAN, ~13 July 2002
Phase-Averaged PTV Measurements of Propeller Wake
Sang Joon Lee, Bu Geun Paik and Choung Mook Lee
(Pohang University of Science arid Technology, Korea)
ABSTRACT
The objective of present paper is to apply an
adaptive hybrid 2-frame PTV (Particle Tracking
Velocimet~y) technique for measuring the flow
characteristics of turbulent wake behind a marine
propeller with 5 blades. Compared to the conventional
PIV method, he hybrid PTV technique increases the
spatial resolution
and measurement accuracy
significantly, while reducing the computation time. For
each of four different blade phases of 0°, 18°, 36° and
54°, four hundred instantaneous velocity fields were
measured. They were ensemble averaged to investigate
the spatial evolution of the propeller wake in the region
ranged from the trailing edge to two propeller
diameter(D) downstream location. The phas~averaged
mean velocity fields show that the trailing vorticity and
the viscous wake are formed by the boundary layers
developed on the blade surfaces. The vorticity contours
at each phase angle show that the tip vortices are
produced periodically. The slipstream contraction
occurs in the near-wake region up to about x = 0.5D
downstream. Thereafter the unstable oscillation occurs
due to the separation of tip vortex from the wake sheet
behind the maximum contraction point. As the tip
vortex evolves downstream, its strength is reduced due
to turbulent diffusion, viscous dissipation and active
mixing between tip vortices and adjacent wake flow.
The technique presented here can be readily extended
to investigate the nominal and effective wake
distribution as well as the details of the flow field of
fore and aft of a rotating propeller behind a ship model.
INTRODUCTION
A marine propeller is the main source of noise. hull
vibration, and cavitation at high speed. In order to
increase the propulsion efficiency, the geometry of
propeller should be optimally designed. To achieve this
objective, an accurate wake analysis based on detailed
experimental results is required. In general, a modern
propeller blade has a complicated geometry and makes
the wake behind a propeller more complex.
Kerwin (1978) predicted the steady and unsteady
marine propeller performance by numerical lifting
surface theory. However, the lifting surface theory
cannot give accurate prediction of the flow around the
leading edge and the blade tips. Lee (1987) and Kim et
al. (1993) used the potentiaLbased panel method to
improve the prediction of the steady performance of a
propeller. A further improvement was made by Cho
and Lee (2000) by applying the gasoline higher order
panel method to obtain more accurate numerical results.
It is well known that any numerical attempts based on
either potential or viscous flow model would not yield
a satisfactory prediction of the formation and trajectory
of tip or trailing vortices without an adequate wake
sheet modeling. In the conventional numerical methods,
the velocity field around a propeller blade is calculated
from the position of wake sheet assumed in the
previous iteration step, and a new position of wake
sheet satisfying the boundary condition is determined
and the process is repeated until a numerical
convergence is obtained. To avoid such a tedious
repetition of calculation, some numerical investigations
used a linear or non-linear wake sheet model which
was obtained from the experimental data .
Hoshino (1989) computed the flow field around a
propeller using the wake contraction and trailing vortex
sheet model. However, he did not consider the viscosity,
hub effects and velocity field influenced by the wake.
Although the actual trailing vortex sheet in wake has a
finite thickness, it has been assumed as a thin filament
in the numerical analysis resulting in a decrease in the
reliability of the prediction of the trailing and tip
vortices. It is very important to accurately predict the
tip vortex which can be the major source of energy loss
in propulsion, hull vibration and noise. The position of
tip vortex could be estimated using conventional
numerical methods; however, the prediction of the
strength or thickness of tip vortex requires much more
accurate information which can be obtained only from
reliable experiment focused on the wake and tip vortex
behavior.
The wake of a propeller has been measured with
point-wise experimental techniques such as Pitot tube
and LDV(Laser Doppler Velocimetry). Therefore, most
previous investigations neasured flow velocities at
OCR for page 917
discrete points by scanning the flow field with an array
of measurement probes. Stella et al. (1998) measured
the axial velocity component of a propeller wake, and
Chesnakas and Jessup (1998) investigated the tip
vortex flow using LDV. Unfortunately, these point-wise
methods take substantially long time to get the phase-
averaged velocity field information.
Recently, velocity field measurement techniques
have been applied to measure the flow around a marine
propeller. The PIV(Particle Image Velocimetry)
technique does not interfere with the flow structure and
also does not require much time, compared to the LDV
method, to measure the velocity field in a reasonably
large area. Controni et al. (2000) investigated the near-
wake of a marine propeller in a cavitation tunnel using
PIV method. Their results show a good spatial
resolution compared with conventional LDV data.
In the present study, an adaptive hybrid PTV(Particle
Tracking Velocimetry) method(Kim and Lee (2002)) is
used to investigate the near-wake of a marine propeller
in detail. To demonstrate the effectiveness of the
method, a simple case of measuring the wake of a
propeller model in an open-water channel was chosen.
Once the reliability of the method is affirmed, it s
anticipated to apply the method to more complex flow
problems in the ship stern region. This velocity field
measurement technique requires a preliminary PIV
routine to determine the local match parameters needed
for the 2-frame PTV algorithm(Baek and Lee, 1996).
The original 2-frame PTV based on match probability
determines the match parameters globally for the whole
flow field. The adaptive hybrid scheme by utilizing the
advantages of both PIV and PTV techniques recovers
significantly more velocity vectors while reducing
computation time and errors. In addition, the adaptive
hybrid PTV shows the superior spatial resolution,
compared with the conventional PIV technique.
The several hundreds instantaneous velocity fields
were measured at four different phases of the propeller
blade and they were phas~averaged to investigate the
spatial evolution of the vertical structure and
turbulence statistics for the propeller wake.
DIAL APPARATUS ANI) MliTHOD
The hybrid 2-frame PTV system consists of a
Nd:YAG laser, a high-resolution CCD camera, a
synchronizer, motor controller and an IBM PC as
shown in figure 1. The dual-head Nd:YAG laser has a
pulse width of about 7 ns with a pulse energy of 125
mJ for each head. The CCD camera can capture a
couple of particle images at a time and has the spatial
resolution of 2048 x 2048 pixels. A thin laser light
. . . . ... . .
sneer was used tO Illuminate the measurement planes
and the scattered particle images were captured by the
CCD camera for velocity field measurements. The
Circulating Water
Channel
Sen,~motor and Encoder
2.2D I_
11 1
1\ L85D
~ ~ ~ X
1 _-
/ Kodak 2K'dK
~ / ~~ ~
Optics:
r
1 1
IBMPC |
Grabber |
Dual-Head I ~ |-Controller l
Nd:Yag ~ - Low pass filter
Laser - Synchronizer
Figure 1: Schematic diagram of experimental set-us
_ __
CCD came ra and laser were synchronized with angular
position of the propeller blade. Particle centroids were
detected from the captured particle images in the
preprocessing of the hybrid PTV. The post-processing
routine includes particle tracking, data validation and
interpolation to get the instantaneous velocity field on
the regularly spaced grid points at each section. Details
are described in Kim and Lee (2002) about the velocity
field measurement technique.
The circulating water channel where the propeller
wake was measured has a test section size of 120~ x
30W x 2(F in centimeter. Figure 2 shows the propeller
KP505 for the 3600TEU container vessel tested in
present study. The KP505 propeller of 54mm in
diameter has 5 blades with a design advance coefficient
J of 0.72. The diameter of the propeller shaft is 7 mm.
Velocity field measurements were carried out at three
advance coefficients of J = 0.59, 0.72 and 0.88 to
examine the influence of propeller loading. The free
r/R StOE ELEVATION PROJECTED BLADE £XPANOEO BLADE P/O
. to
moo _
U.'DU - '
eco
=
.
you
coo
, O. .~
, u an d~
.. I, o,
_
~C-
1 \ ': ~
Ll ~ ~
~_O ~ FR PRINCTPAI PART lCULARS
DIAMETER(MM) psalm SCAI~.E RATIO 4 000
(P/O} mCon : O .950 PP0P TYPE ~ FED
Ar/Ao O. BOO SECTION NACA66
Mug RAT lo o. HO DRAWING SCALE 2,rJ00
NO OF B`,ADES 5 COMN4FNTS
Figure 2: Propeller geometry
OCR for page 918
stream velocity was 32.5 cm/s and the Reynolds
number based on the propeller diameter is about 18000.
no
.n ~
n R
.1 ~ _
no
-0.6
-na
. ._
0 0.2
_ ~ . .
0.4 0.6 0.8 1
X/D
(a)J=0.88
XID
(b) J = 0.72
-1 .2
2
0 0.2 0.4 0.6 0.8 1
X /D
(b)J=0.59
Figure 3: Instantaneous velocity field subtracted by
Uo at o= 0°
The laser light sheet was illuminated from the bottom
of water channel and the field of view was 11.8 x 11.8
in centimeter.
Silver coated hollow glass beads with mean diameter
of 10 ,um were used as seeding particle. A servo-motor
attached with an encoder was used to derive the
propeller, and the support strut was installed to prevent
transfer of the surge or vibration of the propeller shaft
with 0.1 3D diameter to the working fluid. The
propeller was driven from
downstream to avoid the effect of wake generated by
the supporting struts. The center of propeller wake
could not be measured because the propeller shaft was
located at the downstream. The encoder mounted on
the servo-motor generates trigger signals to
synchronize laser and CCD camera with an accuracy of
0.36° for each selected angular position. The encoder
signals were filtered with a low pass filter to get a clear
trigger signal. The time interval between two
consecutive particle images As set to 300 ,us, for
which the propeller rotates 0.9°. The velocity field
measurements were carried out at four different
phases(~= 0°, 18°, 36°, 54°) with the angular interval
between two adjacent measurement phases of 18°, and
the corresponding elapse time was about 6 ms. A total
of 400 instantaneous velocity fields was obtained for
each measurement phase using the adaptive hybrid 2-
frame PTV method. The turbulence characteristics such
as turbulence intensity and Reynolds shear stress
distribution were obtained by ensemble-averaging the
instantaneous velocity fields.
RESllTIS AND DISCUSSION
The abscissa and ordinate of all results are
represented in the plane of X and Y axes, normalized
by the propeller diameter D. The positive X-axis is in
the direction of the main flow and the positive Y-axis
is directed vertically upward. The propeller is placed at
X = 0 and the propeller shaft is placed at Y = 0. Figure
3 shows the instantaneous velocity field subtracted by
Uo=32.5 cm/s at the phase angle of o= 0°. The periodic
wake sheets and tip vortices in the clockwise rotation
are shed successively from the blade tips with a regular
interval are clearly seen at J = 0.59 and 0.72. In figure
3(a) counter-clockwise rotating vortices can be
observed. These vortices do not appear to be associated
with the tip vortices but are diagnoised to be generated
from the shear interaction between the slipstream of
propeller and the free stream.
The contour plots of phase-averaged axial velocity at
0 = 0~ are shown in figure 4. The axial velocity
component has relatively small values in the blade tip
and propeller shaft. The viscous wake indicating the
defects of axial velocity component appears in the
OCR for page 919
near-wake region due to the merging of two boundary
not
06
(c)J=o.59
Figure 4: Contour of phase-averaged axial velocity
layers developed on both sides of propeller blade
414
Q4
ED
(a)J=0.88
o~i Q5 1
1 125 1.5 1.75
In
(b)J=0.72
025 05 Q75 1 125 15 1.75
(c) J = 0.59
Figure 5: Contour of phas~averaged vorticity at ¢
OCR for page 920
no
-0.6 _
___ ~
-0.8 _
~ _
-1.2
Figure 6: Trajectories of tip vortices for four blade
angles at J = 0.72
-0.2 t
-0.4
-0.6(
-0.8
-1 .2
C.
~~+ . 47~-~
Figure 7: Variation of trajectory of tip vortices at
three propeller loadings at o= 0°
surfaces. As the advance coefficient(J) decreases
(increasing propeller loading), the magnitude of the
axial velocity component is increased within the
slipstream of the propeller wake. The slipstream
fluctuates unstably in the downstream region behind X
= 0.5D for the advance coefficient of J = 0.59 and 0.72.
This fluctuation seems to be originated mainly from the
separation of tip vortex from the wake region. For the
case of light loading(J = 0.88), the slipstream oscillates
just from the trailing edge of propeller blade. On the
average, the axial velocity has minimum values around
0.7R of the propeller blade span where R is the
propeller radius. This agrees well with the propeller
design which assigns the maximum loading around
0.7R.
Figure 5 shows the contours of phase-averaged
vorticity at ~ = 0° for three advance coefficients. Tip
vortex generated from the pressure difference between
upper and lower surfaces of a propeller blade is rolled
up near the blade tip and forms vortex sheets. Tip
vortices evolve downstream with a regular spacing
periodically. The trailing vortices called as the potential
wake are originated from the trailing edge of the
propeller blade. Tip vortex has a concentric circles
shape, while, the trailing vortices have the curled shape
~ 1
{2
~4
{16
o
. ~ _ ~
0:~; 05
Y./D
(a)J=0.88
of _
X,D
(c) J = 0.59
L" Tu
20 Q1 5,
19 Q144
18 Q1~
17 Q133
16 Q128
15 Q122
14 Q116
1 3 Q1 11
12 Q1Di
11 Q1Q)
1 0 Q094
9 Q088
8 QOe3
7 Q077
6 Q072
5 QOe6
4 Q063
3 QO$
2 Q04
1 Q043
_ Let Tu
20 Q1~
19 Q1 44
Q1 39
17 Q1 33
16 Q1 28
15 Q1 i2
14 Q1 16
13 Q1 11
12 Q1 Q5
11 Q1 Q1
10 Q094
9 Q088
8 Q083
7 Q077
6 Q072
5 GOES
4 COO
3 QO$
2 Q04
1 Q043
Lad Tu
20 Q1 ~
19 Q144
18 Q13)
17 Q133
16 Q128
15 Q122
14 Q116
1 3 Q1 11
12 Q1Q5
11 Q1Q,
1 0 Q094
9 Q088
8 Q083
7 Q077
6 Q072
5 QOe6
4 QOeO
3 QO$
2 Q04
_ 1 Q043
Figure 8: Spatial distribution of axial turbulence
intensity at o= 0°
OCR for page 921
~2
~4
Q6
Q2
~4
~6 l
. ~ 1' ... I .'t. 1. t.'I I' . 51 ~ .. ~ It . t. I..
0:~ 05 075 1 1.~ 15 1J~
YJD
(a)J =0.88
B _
Ski 05 075 1 1.;5
ID
(b)J=0.72
15 1.75
Led ~
20 OlCo
19 aos6
1 8 GOSH
1 7 aom
1 6 Q084
1 s aoeo
1 4 0076
13 ao72
12 ao6s
11 0 -
10 Doe
g Dose
8 QOq'
7 Q04}
6 Go44
5 Q040
4 O
3 ao32
2 Got
_ 1 Goal
L - l 1;,
20 Q1CD
19 QO$
1 8 Doe.
1 7 QO`B
1 6 Q084
1 5 QO8O
1 4 QO~i
13 Q02
1 2 QOEB
1 1 Q064
1 0 QOeD
9 QO$
8 QO~
7 Q04}
6 Q044
5 QO~
4 QO:B
3 Qo32
2 Goal
_ 1 Qo24
't I''''I ' ''' I ''' 'I''' 'I It ~ ·1 · ·\ · 1.
OF QS 0.75 1 1Z; 15 1.75
ED
(c)J=o.59
Figure 9: Spatial distribution of vertical turbulence
intensity at o= 0°
toward the propeller shaft from the tip. The vorticity in
the tip region increases significantly as the propeller
loading increases. The peak values of the first tip
vorticity are -7.1, -3.9 and-2.3 sec~i for J = 0.59, 0.72
and 0.88, respectively. This indicates the fact that the
propeller of heavy loading loses more energy due to tip
vorices than that of light loading. For the small loading
at J = 0.88, the vortices generated by the shear
interaction between the slipstream of the propeller and
free stream show relatively greater intensity of 2.7 sec~i
than those of tip vortices whose maximum intensity is
-2.3 sect, which indicates that J=0.88 is not a good
operational condition for KP505 propeller. From this, it
can be judged that it is desirable to operate the
propeller at the design loading to get the optimized
propulsion efficiency. The trailing vortices are
composed of two vorticity layers with a positive or
negative sign, which are developed on the blade
surfaces and the vortex exists between these two layers.
The tip vortex has a strong asymmetry shape in the
initial wake region up to X = 0.25D at J=0.59, 0.72.
However, for the advance coefficient of J=0.88, the tip
vortex shows a weak asymmetry. The asymmetry of tip
vortex was caused by the interaction between tip vortex
and wake sheet. As the flow goes downstream at J =
0.59 and 0.72, the asymmetry turns to symmetry as the
tip vortices are separated from the wake.
Figure 6 shows the trajectories of tip vortex cores for
four phases at the advance coefficient of J = 0.72. Each
tip vortex has a nearly constant trajectory in the initial
region O < X < 0.5D and then suddenly contracted from
the slipstream at about X=0.8D. After the large
contraction, tip vortices start to oscillate.
The traces of tip vortex cores for 0 = 00 at three
propeller loadings are shown in figure 7. The tip
vortices are formed at regular intervals for all propeller
loadings tested in this study. Up to the downstream
location of X=0.8D, as the propeller loading
decreases(J increases) the tip vortices moves slightly
lord Tv outward due to shortage of a axial momentum. In the
:D Q1~
it aOg~ region of X > 0.8D, the fluctuation of the trajectory of
,7 QO~ tip vortices at J = 0.88 appears slightly smaller than
345 X76 that at the other loadings. The separation of the tip
'2 Co678 vortex from the wake, viscous dissipation and turbulent
to Q064 diffusion seem to result in the decay of the trailing and
8 Y52 tip vortices.
7 oo4 Turbulence intensity distribution was obtained from
5 ~O30 the statistical analysis of the fluctuation velocity fields.
2 ~O238 Figure 8 and 9 show the turbulence intensity
distributions of axial and vertical velocity components
( ~IUO,~IUo ), respectively. The turbulence
intensity for the axial velocity component is increased
significantly, as the propeller loading increases. The
axial turbulence intensity has local maximum values
along the trace of tip vortices. This indicates that the
turbulece intensity for axial velocity component is
stronger near the tip vortices than the other wake
region. The turbulence intensity of vertical velocity
component also has large values near the tip vortices as
shown in figure 9. This may have resulted from the
active interaction between the tip vortices and the wake
sheet. However, the magnitudes of the vertical
OCR for page 922
~4
ma,
_
no . . ~ . . . . I . . ,
o
of 61
O:~i 05 075 1 1.Zi
Y0D
(a)J=0.88
e
42
~4
It
· 1~1
- of as 075 1 1.~5 15 1.75
Y7D
(b) J= 0.72
_
4B '. ~ 1, .., 1.... 1.... 1 .~..1 · ..
025 05 O~ 1 1~ 15 1.
ID
(C)J=0.59
Figure 10: Spatial distribution of Reynolds shear
stress at ¢= 0°
turbulence intensity at the design and higher propeller
loadings do not vary much. The interaction between the
tip vortices and the wake sheet transports the
turbulence intensity of axial and vertical velocity
components to the far downstream region.
The propeller wake displays isotropic turbulent
structure up to the range of X = 2D with large
fluctuations of axial and vertical velocity components.
Kiya and Sasaki(1983) mentioned that the maximum
turbulence intensity occurs usually at the core of shear
layer for a given flow. The shear layer start to oscillate
from the downstream location of X = 0.7D due to the
separation of the tip vortex from the wake sheet. The
effect of propeller shaft on the wake structure seems to
be small because the vertical turbulence intensity
distribution is well matched with the vorticity contour
and the turbulence intensity produced from the rotating
propeller shaft is not so high.
The Reynolds shear stress(u'v'/UO2 ) has large
values within the shear layer developed from the
propeller blade surface as shown in figure 10 within the
wake region observed in the present experiment, the
Reynolds shear stress becomes larger as the wake goes
downstream and the advance coefficient J increases.
The separation of tip vortices from the slipstream
expands the region having large values of Reynolds
shear stress. The tip vortex is mixed with near-wake
flow for the cases of J = 0.59 and 0.72. Due to active
mixing between tip vortices and wake flow, the
turbulent shear stress increases at the downstream
location.
CONCLUSION
The propeller wake in an open water condition was
investigated using an adaptive hybrid PTV technique,
and instantaneous velocity fields were measured at four
different blade phases of 0°, 18°, 36° and 54°.
The vis cous wake indicating the loss of axial
velocity component is related to the merging of
boundary layers of a propeller blade. The slipstream
starts to oscillate after the rapid contraction at about
X=0.7D at the design and higher loading conditions.
Periodic tip vortices are formed due to pressure
difference between two surfaces of propeller blade and
go downstream with a regular interval. The magnitude
of counter-clockwise rotating vortex increased as the
propeller loading decreases. The vorticity value and the
asymmetry of tip vortices are increased as the increase
of propeller loading. The broadening and contraction of
the gap between trailing vortices and tip vortex as the
propeller wake goes downstream were attributed to the
following factors; the separation of tip vortices from
the wake, the interaction between tip vortex and wake
sheet, turbulent diffusion and viscous dissipation.
With increasing propeller loading, the magnitude of
axial and vertical velocity component is increased and
the turbulence structure becomes an isotropic one.
After the region of slipstream contraction, the shear
layer oscillates unstably. This results from the
separation of tip vortex and the active interaction with
the near-wake flow.
OCR for page 923
ACKNOliVLEDGMENT
The present work is supported by National Research
Laboratory Program of Ministry of Science and
Technology(MOST) of Korea.
RICE
Back, S.J. and Lee, S.J., "A New Two-Frame Particle
Tracking Algorithm Using Match Probability,"
Excrements in Fluids, Vol.22, 1996, pp. 23-32
Chesnaks C., Jessup S., "Experimental Characterisation
of Propeller Tip Flow," 22nd SYmoosium on Naval
Hydrodynamics, Washington D.C.~ 1998* oD.156-169.
Cho, C.H. and
--on or ~
Lee, C.S., "Numerical
Experimentationof a 2-D Splice Higher Order Panel
Method,' Journal of the SocietY of Naval Architects of
Korea, Vol.37, No.3, 2000, pp.27-36.
Cotroni, A., Di, Felice F., Romano, G.P. and
Elefante,M., "Investigation of the Near Wake of a
Propeller Using Particle Image Velocimetry,"
Experiments in Fluids, Vol.29, 2()00, pp.s227-236.
Hoshino, T., "Hydrodynamic Analysis of Propellers in
Steady Flow Using a Surface Panel Method," Journal
of the Societv of Naval Architects of Janan, Vol.166,
1989, pp. 79-92.
Kerwin, J.E. and Lee, C.S., "Prediction of Steady and
Unsteady Marine Propeller Performance by Numerical
Lifting Surface Theory," Trans. S NAME Vol.86, 1978,
pp.218-253.
Kim H.B. and Lee, S.J., "Performance Improvement
of Two-frame Particle Tracking Velocimetr~y Using a
Hybrid Adaptive Scheme," Measunnent Science &
Technology, Vol.13, 2002, ppS7~582.
Kim, Y.G., Lee, J.T., Lee, C.S., and Sub, J.C.,
"Prediction of Steady Performance of a Propeller by
Using a Potential-Based Panel Method," Trans. of the
SocietY of Naval Architects of Korea, Vol. 30 No. 1
, ,
1993, pp.73-86.
Kiya, M. and Sasaki, K., "Structure of a Turbulent
Separation Bubble," Journal of Fluid Mechanics,
Vol.137, 1983,pp.83-113.-
Lee, J.T., "A Potential-based Panel Method for the
Analysis of Marine Propellers in Steady Flow," Ph.D.
Thesis, Department of Oc can Engineering, M.I.T.,
Cambridge, Mass., 1987.
Stella, A., Guj, G., Di, Felice F. and Elefante, M.,
"Propeller Wake Evolution Analysis by LDV," 22nd
Symposium on Naval Hydrodynamics, Washington
D.C., 1998, pp. 171
OCR for page 924
discrete points by scanning the flow field with an array
of measurement probes. Stella et al. (1998) measured
the axial velocity component of a propeller wake, and
Chesnakas and Jessup (1998) investigated the tip
vortex flow using LDV. Unfortunately, these point-wise
methods take substantially long time to get the phase-
averaged velocity field information.
Recently, velocity field measurement techniques
have been applied to measure the flow around a marine
propeller. The PIV(Particle Image Velocimetry)
technique does not interfere with the flow structure and
also does not require much time, compared to the LDV
method, to measure the velocity field in a reasonably
large area. Controni et al. (2000) investigated the near-
wake of a marine propeller in a cavitation tunnel using
PIV method. Their results show a good spatial
resolution compared with conventional LDV data.
In the present study, an adaptive hybrid PTV(Particle
Tracking Velocimetry) method(Kim and Lee (2002)) is
used to investigate the near-wake of a marine propeller
in detail. To demonstrate the effectiveness of the
method, a simple case of measuring the wake of a
propeller model in an open-water channel was chosen.
Once the reliability of the method is affirmed, it s
anticipated to apply the method to more complex flow
problems in the ship stern region. This velocity field
measurement technique requires a preliminary PIV
routine to determine the local match parameters needed
for the 2-frame PTV algorithm(Baek and Lee, 1996).
The original 2-frame PTV based on match probability
determines the match parameters globally for the whole
flow field. The adaptive hybrid scheme by utilizing the
advantages of both PIV and PTV techniques recovers
significantly more velocity vectors while reducing
computation time and errors. In addition, the adaptive
hybrid PTV shows the superior spatial resolution,
compared with the conventional PIV technique.
The several hundreds instantaneous velocity fields
were measured at four different phases of the propeller
blade and they were phas~averaged to investigate the
spatial evolution of the vertical structure and
turbulence statistics for the propeller wake.
DIAL APPARATUS ANI) MliTHOD
The hybrid 2-frame PTV system consists of a
Nd:YAG laser, a high-resolution CCD camera, a
synchronizer, motor controller and an IBM PC as
shown in figure 1. The dual-head Nd:YAG laser has a
pulse width of about 7 ns with a pulse energy of 125
mJ for each head. The CCD camera can capture a
couple of particle images at a time and has the spatial
resolution of 2048 x 2048 pixels. A thin laser light
. . . . ... . .
sneer was used tO Illuminate the measurement planes
and the scattered particle images were captured by the
CCD camera for velocity field measurements. The
Circulating Water
Channel
Sen,~motor and Encoder
2.2D I_
11 1
1\ L85D
~ ~ ~ X
1 _-
/ Kodak 2K'dK
~ / ~~ ~
Optics:
r
1 1
IBMPC |
Grabber |
Dual-Head I ~ |-Controller l
Nd:Yag ~ - Low pass filter
Laser - Synchronizer
Figure 1: Schematic diagram of experimental set-us
_ __
CCD came ra and laser were synchronized with angular
position of the propeller blade. Particle centroids were
detected from the captured particle images in the
preprocessing of the hybrid PTV. The post-processing
routine includes particle tracking, data validation and
interpolation to get the instantaneous velocity field on
the regularly spaced grid points at each section. Details
are described in Kim and Lee (2002) about the velocity
field measurement technique.
The circulating water channel where the propeller
wake was measured has a test section size of 120~ x
30W x 2(F in centimeter. Figure 2 shows the propeller
KP505 for the 3600TEU container vessel tested in
present study. The KP505 propeller of 54mm in
diameter has 5 blades with a design advance coefficient
J of 0.72. The diameter of the propeller shaft is 7 mm.
Velocity field measurements were carried out at three
advance coefficients of J = 0.59, 0.72 and 0.88 to
examine the influence of propeller loading. The free
r/R StOE ELEVATION PROJECTED BLADE £XPANOEO BLADE P/O
. to
moo _
U.'DU - '
eco
=
.
you
coo
, O. .~
, u an d~
.. I, o,
_
~C-
1 \ ': ~
Ll ~ ~
~_O ~ FR PRINCTPAI PART lCULARS
DIAMETER(MM) psalm SCAI~.E RATIO 4 000
(P/O} mCon : O .950 PP0P TYPE ~ FED
Ar/Ao O. BOO SECTION NACA66
Mug RAT lo o. HO DRAWING SCALE 2,rJ00
NO OF B`,ADES 5 COMN4FNTS
Figure 2: Propeller geometry
OCR for page 925
DISCUSSION
F. Di Felice
INSEAN (Italian Ship Model Basin), Italy
Authors performed the analysis of a propeller
wake at very low Reynolds number by using
PTV and phase averaged technique. Analysis
extends far downstream the contraction
including the vortex breakdown region. The
images are analyzed using an Hybrid algorithm
in which, in a first pass, classical PIV cross-
correlation algorithm is applied to obtain the
main parameters to be used for the PTV analysis.
This approach allows an higher spatial resolution
as shown also by Keane et al (1995) because a
velocity vector is obtained for each particle
simultaneously detected on both PIV images,
even if randomly spaced data are obtained.
Authors show instantaneous velocity field
obtained interpolating the randomly spaced data
over a regular grid. The choice of the sampling
frequency is very critical because a dense re-
sampling grid will introduce too much
interpolated data while on the other hand a low
density grid will loose many information. Could
the authors explain how they managed such
problem especially looking at the statistical
computation, which can be affected by the
presence of interpolated data?
Looking at the result, figure 6 and 7 show the
location of the tip vortices. A separation of the
tip vortex trajectories downstream x/D= 1 is
apparent. This is a typical behavior of the
propeller vortex breakdown process that is
caused by the interaction of the tip vortex with
the wake of the previous blade. In fact, the actual
blade wake, traveling at higher speed with
respect the tip vortex, overtakes and interacts
with the previous blade tip vortex as observed
also at higher Reynolds number (Di Felice et al,
2000~. Did the authors observed such type of
blade to blade interaction?
REFERENCES
Measurement Science and Technology, vol 6,
1995, pp 754-768.
Symphosium on Naval Hydrodinamics, Vat de
Ruil (F), 2000.
AUTHORS' REPLY
(1) Using the grey level intensity of particle
image, the global thresholding algorithm was
applied to extract particle centroids. Thereafter
the particle tracking routine is applied. After
validating all PIV data, error vectors in the
velocity field are replaced with the vectors
estimated using the AGW(Adaptive Gaussian
Window) method for neighborhood correct
vectors (Spedding and Rignot(1993), Agui and
Jimenez(1987~. The AGW method is a simple
convolution of the known data (uk,vk) with an
adaptive Gaussian window ok. The weighting
coefficients ok are adjusted so that their sum is
always equal to unity, independent of the particle
location. As the velocity vectors extracted from
the present hybrid PTV algorithm has random
particle locations, the interpolation scheme is
required to obtain velocity vectors on regular
spaced grid points and future calculation of
turbulence statistics.
In the interpolation procedure, the number of
grids has to be determined in advance. The
proper number of grids for interpolation depends
on the number of particles and the spatial
resolution of camera.
(2) The separation of tip vortex from the wake
sheet results from the interaction between blade
wake and tip vortex. The previous experiment
(Di Felice et al, 2000), the blade wake overtakes
and interacts with the previous blade tip vortex at
high Reynolds number. On the other hand, we
measured the propeller wake at relatively small
Reynolds number. For low Reynolds number
flow, the viscous effect is usually a little
overestimated than the actual one. However, we
also observed the interaction between tip vortex
and blade wake in the separation and the
oscillation of tip vortices. The contour plots of
axial velocity (figure 4) show that higher axial
velocity at the inner radius makes the blade wake
bend and go faster than the tip vortex. The
Keane R D, Adrian R J. Zhang Y. "Super- vorticity evolution (figure 5) also shows that the
resolution Particle Image Velocimetry", blade wake goes downstream faster than the tip
vortex. Therefore, the blade wake travelling at
higher speed with respect to the tip vortex
overtakes and interacts with the previous blade
tip vortex in our experiment.
Di Felice F. Romano G P. Elefante M "Propeller
wake Analysis by means of PIV", 23th
OCR for page 926
REFERENCES
Agui J. C. and Jimenez J., "On the Performance
of Particle Tracking", Journal of Fluid
Mechanics, vol. 185, 1987, pp.447-468.
Di Felice F., Romano G. P., Elefante M.,
"Propeller wake Analysis by means of PIV",
23th Symposium on Naval Hydrodynamics, Val
de Ruil (F), 2000.
Spedding G.R. and Rignot E.J.M., "Performance
Analysis of Grid Interpolation Techinques for
Fluid Flows", Exp. Fluids, vol 15, 1993, pp. 417-
430.
Representative terms from entire chapter:
tip vortices
