Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Experiments in Fluids 36 (2004) 575585

DOI 10.1007/s00348-003-0699-5

Three-component velocity field measurements of propeller wake using


a stereoscopic PIV technique
Sang Joon Lee, Bu Geun Paik, Jong Hwan Yoon, Choung Mook Lee

575
Abstract A stereoscopic PIV (Particle Image Velocimetry) propeller loading may lead to problems such as noise, hull
technique was used to measure the three-dimensional flow vibration, and cavitation at high speed. The geometry of a
structure of the turbulent wake behind a marine propeller propeller should therefore be designed to minimize them.
with five blades. The out-of-plane velocity component was In general, modern propeller blades have a complicated
determined using two CCD cameras with an angular dis- geometry, making the wake behind the propellers com-
placement configuration. Four hundred instantaneous plicated too. Therefore, any serious attempt to optimize
velocity fields were measured for each of four different the geometrical shape of modern propellers will require a
blade phases, and ensemble averaged in order to find the reliable wake analysis based on detailed experimental
spatial evolution of the propeller wake in the region from measurements.
the trailing edge up to one propeller diameter (D) down- Marine propeller performance has been predicted using
stream. The influence of propeller loading conditions on the potential-based panel method (Lee 1987). Recently,
the wake structure was also investigated by measuring the Cho and Lee (2000) introduced the B-spline higher order
velocity fields at three advance ratios (J=0.59, 0.72 and panel method and showed that it yields more accurate
0.88). The phase-averaged velocity fields revealed that a numerical results. It is well-known that numerical mod-
viscous wake formed by the boundary layers developed eling of propeller wakes only yields a satisfactory predic-
along the blade surfaces. Tip vortices were generated tion of the formation of tip or trailing vortices if it includes
periodically and the slipstream contracted in the near- adequate wake sheet modeling, based either on experi-
wake region. The out-of-plane velocity component and mental data or on a special theory. In addition, it is very
strain rate had large values at the locations of the tip and important to predict the strength and trajectory of the tip
trailing vortices. As the flow moved downstream, the tur- vortices, which cause energy loss in propulsion, hull
bulence intensity, the strength of the tip vortices, and the vibration, and noise. In most previous numerical analyses
magnitude of the out-of-plane velocity component at of propeller wakes (Lee 1987; Cho and Lee 2000), the actual
trailing vortices all decreased due to effects such as viscous vortex sheet of the wake was assumed to be a thin filament
dissipation, turbulence diffusion, and blade-to-blade in order to predict the formation and trajectory of tip or
interaction. trailing vortices with or without an adequate wake sheet
model.
Keywords Stereoscopic PIV, Propeller wake, Tip vortex, Experimental characterizations of propeller wakes have
Wake sheet predominantly been carried out using point-wise experi-
mental techniques, such as hot film, Pitot tube and LDV
(Laser Doppler Velocimetry), which measure flow veloci-
1 ties at discrete points by scanning the flow field with an
Introduction array of measurement probes. For example, Stella et al.
As marine vehicles become larger and faster, the loading (1998) used LDV to measure the axial velocity component
on their propeller blades increases. This increased of a propeller wake. Chesnakas and Jessup (1998) also used
LDV to study the tip vortex flow, although they found it
impossible to simultaneously measure all velocity vectors
in a section. However, point-wise methods such as LDV
Received: 6 December 2002 / Accepted: 22 July 2003
Published online: 10 March 2004
require substantial time to acquire the phase-averaged
 Springer-Verlag 2004 velocity field data.
Particle Image Velocimetry (PIV) has also been used to
S. J. Lee (&), B. G. Paik, J. H. Yoon, C. M. Lee measure the flow around a marine propellers. Compared
Department of Mechanical Engineering, Pohang University with LDV, PIV has the advantages that it is less intrusive
of Science and Technlogy, San 31, Hyo-Ja Dong, and requires only a short measuring time to obtain the
790-784 Pohang, Korea
E-mail: sjlee@postech.ac.kr
phase-averaged velocity field over a large area, although
Tel.: +82-54-2792169 LDV has the advantage that it can obtain some flow
Fax: +82-54-2793199 information without the need for spatial reconstruction.
The present work was supported by the National Research Lab- Controni et al. (2000) used a PIV technique to investigate
oratory Program of the Ministry of Science and Technology the near-wake of a marine propeller at the longitudinal
(MOST) of Korea. plane in a cavitation tunnel. More recently, Lee et al.
(2002) used the adaptive hybrid PTV (Particle Tracking 2
Velocimetry) method to investigate the propeller wake at Basic principles of SPIV
low Reynolds numbers up to the region of two propeller SPIV measurements were carried out using angular con-
diameters downstream in the longitudinal direction. They figurations in which the two optical axes were neither
measured the wake structure with good spatial resolution; parallel nor perpendicular to the measurement plane, since
the experimental results revealed a flow pattern similar to the angular displacement method, as shown in Fig. 1, re-
that reported by Controni et al. (2000), including blade-to- quires the image planes to be tilted at some angle with
blade interactions. respect to the object plane in order to focus the whole
Recently, stereoscopic PIV (SPIV) has been used to entire measuring volume. Variations in the refractive in-
investigate the 3-D flow structure. Usually, the SPIV dex result in aberrations and particle image distortion,
technique uses two cameras at different angles to acquire leading to non-uniformities in the magnification. Since the
576
two particle images of the same instantaneous particle tilting of image planes and camera lenses causes image
configuration synchronized with the pulsed laser light distortion and variations in the magnification, a sophisti-
sheet. Having simultaneously captured images of the flow cated calibration procedure and a small aperture are
from different angles at a range of times, intermediate required to obtain accurate flow data.
procedures can then be used to produce 3-D velocity fields When the ray tracing method is used in conjunction
from the 2-D particle displacement information contained with the angular configuration, the geometric relationship
in the flow images. This approach makes more accurate between the real particle displacements Dx = (Dx, Dy, Dz)
flow analysis possible than can be achieved using the and the projected displacements DX = (DX1, DY1, DX2, -
conventional 2-D PIV technique, which uses a single DY2) in the image-recording plane must be derived. The
camera and can therefore supply only in-plane velocity geometric relationship between the two displacements in
field information. In 2-D PIV measurements, information the image plane, where two cameras are installed sym-
about the out-of-plane motion is embedded in the in-plane metrically with respect to the Z-axis as shown in Fig. 1,
velocity field data, an approach that creates measurement can be expressed as follows:
errors. Therefore, previous measurements of 3-D flow
structure using 2-D PIV may contain small errors caused DX1 a1 Dxa2 Dz 1
by the out-of-plane velocity component. The SPIV tech-
DY1 = a3 Dy + a4 Dz 2
nique offers the opportunity to obtain flow information
without these errors. DX2 = a5 Dx + a6 Dz 3
Prasad and Adrian (1993) investigated the flow field
under a rotating disk using a two-camera SPIV technique DY2 = a7 Dy + a8 Dz 4
and analyzed the measurement errors due to the interfaces
between air, liquid and glass. Soloff and Adrian (1997) The four equations for three unknown particle dis-
developed a 3-D calibration method that does not require placements (Dx, Dy and Dz) are derived from the geo-
system geometry information during reconstruction. They metric relationship between the object plane and the two
measured calibration data at three different locations cameras. These equations can be solved by a least squares
parallel to the object plane and calculated a non-linear method.
mapping function providing the relationship between the In the present work, image distortion and optical
3-D object field and the 2-D image field for each camera. aberration were compensated for using the calibration
Particle displacements could then be measured using this procedure developed by Soloff and Adrian (1997). The
mapping function. Yoon and Lee (2002) measured the flow relation between the 3-D object volume and 2-D image
around an axial fan using a SPIV technique with a 3-D plane for stereo cameras can be written as
calibration method. By directly comparing the in-plane
velocity components and local turbulence intensity mea-
sured using the 2-D PIV technique with the SPIV results,
they established that the maximum differences due to the
out-of-plane motion were about 2.7% and 5.8% respec-
tively. Recently, Calcagno et al. (2002) used a SPIV tech-
nique to measure the complicated 3-D flow behind a
propeller in the transverse plane, thereby revealing the
spatial evolution of the 3-D propeller wake.
The main objectives of the present study are to illus-
trate the usefulness of the SPIV technique for analyzing
complex 3-D flows, and to investigate the flow character-
istics of the wake behind a marine propeller in detail. In
the present work, SPIV is used to investigate the near-wake
of a marine propeller in the longitudinal plane. Four
hundred instantaneous velocity fields were measured at
four different phases of the propeller blade and phase-
averaged to investigate the spatial evolution of the flow Fig. 1. Angular displacement configuration for SPIV
structure and turbulence statistics of the propeller wake. measurements
577

Fig. 2. Schematic diagram of the SPIV


system

X = F(x) 5 1.2 m 0.2 m 0.3 m (depth). Figure 3 shows the


Kp505 propeller for the KRISO 3600TEU container vessel
Here X = (X1, Y1, X2, Y2) is the projected image posi- tested in present study. The Kp505 propeller has five
tion and x = (x, y, z) is the real particle position in the blades with a design advance ratio J(=Va/(nD)) of 0.72.
object volume. The function F can be expressed in poly- Here, Va is the advance velocity, n is the rotational fre-
nomial form. Soloff and Adrian (1997) used a polynomial quency and D is the propeller diameter. For the Kp505
that was cubic in the x and y coordinates and quadratic in propeller, the propeller diameter is 54 mm and the hub
the z coordinate: ratio is 0.18. To examine the influence of propeller loading,
velocity field measurements were carried out at three ad-
F(x) = a0 a1 xa2 ya3 za4 x2 a5 xya6 xz vance ratios, of J=0.59, 0.72 and 0.88, corresponding to
a7 y2 a8 yza9 z2 a10 x3 heavy, design, and light loading respectively. The uniform
6 free stream velocity was fixed at 32.5 cm/s and the Rey-
a11 x2 ya12 x2 za13 xy2 a14 xyza15 xz2 nolds number based on the propeller diameter was about
3 2 2 18,000. The measurement plane was illuminated from the
a16 y a17 y za18 yz
bottom of the water channel with a thin laser light sheet
The particle displacement DX can be approximated of thickness 3 mm and the field of view was 66 cm.
as F(x)Dx. Here, F(x) can be obtained from the
measured calibration result F(x). Then the four equations
and three unknown particle displacements can be
calculated. Since the two CCD cameras were arranged in
a symmetric configuration, the displacement equations
for the Y direction in the first and second image planes
were averaged to improve the measurement accuracy
of DY1 and DY2.

3
Experimental apparatus and method
The SPIV system consists of a dual-head Nd:YAG laser,
two Kodak two-frame CCD cameras, a synchronizer, and
an IBM PC, as shown in Fig. 2. The Nd:YAG laser has a
pulse width of about 7 ns with a pulse energy of 125 mJ for
each head. The two CCD cameras, which are installed at an
angle of 50.8 to the Z-axis, each capture a particle image
of spatial resolution 10241024 pixels. The CCD cameras
and laser were synchronized with the angular position of
the propeller blade.
The circulating water channel in which the propeller Fig. 3. Geometry and specifications of the Kp505 propeller model
wake was measured had a test section size of tested in this study
578

Fig. 4. Typical instantaneous velocity fields subtracted by U0 at


/=0: a J=0.88; b J=0.72; c J=0.59

Silver-coated hollow glass beads with a mean diameter of


10 lm were used as seeding particles.
A servo-motor was used to drive the propeller and a
support strut was installed to prevent the transmission of
surges or vibrations of the propeller shaft (diameter 0.13D)
to the working fluid. The propeller, whose location was
defined to be X=0, was driven from downstream to ensure
that the wake generated by the supporting struts did not
affect the propeller wake. Since the propeller shaft aligned Fig. 5. Contour plots of phase-averaged axial velocity at /=0: a
with the X-axis (see Fig. 2), the center of the propeller J=0.88; b J=0.72; c J=0.59
wake could not be measured. The encoder mounted on the
servo-motor generated trigger signals to synchronize the
laser and CCD camera with an accuracy of 0.36 for each propeller model were measured at four different phases
selected angular position. The encoder signals were filtered (/=0, 18, 36, and 54), where a phase angle of /=0
with a low pass filter to obtain a clear trigger signal. The corresponds to the configuration with the blade in the
time interval Dt between two consecutive particle images upright position. The angular spacing between adjacent
was set to 300 ls, which corresponds to a propeller rota- measurement phases (18) corresponds to an elapsed time
tion of 0.9. The velocity fields of the wake behind the of about 6 ms.
camera lenses, thereby ensuring accurate flow field data.
Three-dimensional velocity components (axial velocity
component u, vertical velocity component v, and out-of-
plane velocity component w) were deduced from two sets
of consecutive particle images captured from the left and
right cameras of the SPIV system. In the experiments,
2424 pixel interrogation windows overlapping by 50%
were used to obtain instantaneous velocity fields from the
particle images captured by each CCD camera.
Four hundred instantaneous velocity fields were ob-
tained for each measurement phase. The time-averaged in-
579
plane and out-of-plane velocity distributions, vorticity
distribution, strain rate distribution, and turbulence
characteristics such as turbulence intensity, were obtained
by ensemble-averaging the instantaneous velocity fields in
order to investigate the spatial evolution of the wake
structure.

4
Results and discussion
Fig. 6. Variation of axial velocity profile at several downstream Figure 4 shows typical instantaneous velocity fields (with
locations at J=0.72 the free stream velocity U0 subtracted off) at the phase
angle /=0. Here, the positive X- and Y-axes are nor-
malized by the propeller diameter D and lie parallel to the
The angular-type camera arrangement used in the main flow and vertically upward, respectively. For the
present SPIV system was carefully calibrated so as to advance ratios of J=0.59 and 0.72, wake sheets and tip
minimize errors due to any image distortion or non-uni- vortices are generated periodically from the blade tips in
form magnification created by the tilted image planes and the region around a depth of Y=0.5D. Under the light

Fig. 7. Spatial evolution of


phase-averaged spanwise vor-
ticity at J=0.72: a phase /=0;
b phase /=18; c phase /=36;
d phase /=54
580

Fig. 9. Open-water test results at Reynolds number = 1.3107

observed for the other loadings, apparently showing a


breaking condition. This unexpected phenomenon seems
to occur due to the Reynolds number scaling being too low
in the present experiment. The axial velocity component
has relatively small values near the blade tip (Y=0.5D)
compared to the corresponding values for J=0.59 and 0.72.
Figure 6 shows the axial velocity profiles at several
downstream locations for the advance ratio of J=0.72. The
velocity decreases outside the blade tip indicate the pres-
ence of a viscous wake. Since a laminar boundary layer
forms on the blade surface due to the low Reynolds
number, the retardative motion of the fluid particles in the
boundary layer are subject to stronger centrifugal forces
and push the viscous wake outside the tip radius (Y=-
0.5D). The viscous wake, which is obvious up to X=0.2D in
the downstream region but has disappeared by X=0.5D,
seems to be influenced by turbulent mixing because the
highly turbulent wake decreases the local velocity defect in
a short time period. The propeller wake containing trailing
vortices mixes with the free stream, and this turbulent
mixing decreases the local velocity defect rapidly as the
wake goes downstream.
An increase in axial velocity is observed within the
slipstream of the propeller wake at downstream locations
of X=0.05D and 0.2D. The viscous wake manifests in the
near-wake region due to the merging of two boundary
layers, developed on the pressure and suction sides of the
propeller blade respectively. As the propeller loading in-
Fig. 8. Contour plots of phase-averaged spanwise vorticity at creases (in other words, as J decreases), the magnitude of
/=0: a J=0.88; b J=0.72; c J=0.59
the axial velocity component increases within the slip-
stream of the propeller wake. The slipstream contracts
loading condition (J=0.88), no distinct vortices are toward the propeller shaft and this contraction continues
observed. up to 1D downstream. In a previous 2-D flow measure-
Contour plots of the phase-averaged axial velocity at ment of wake behind the same propeller model using a
/=0 are shown in Fig. 5. For J=0.59 and 0.72, the axial 2K 2K CCD camera and a hybrid PTV algorithm, Lee
velocity of the slipstream is larger than that of the free et al. (2002) observed fluctuations of the slipstream near
stream. For J=0.88, however, the shape of the slipstream is X=1D in the downstream region. In the present experi-
less clear and its axial velocity is much smaller than those ment, however, these oscillations of the slipstream were
581

Fig. 11. Phase-to-phase variation of out-of-plane velocity


component at J=0.72

tangential velocity between the upper and lower surfaces


of a propeller blade. The trailing vortices, referred to as the
potential wake, originate from the trailing edge of the
propeller blade. The tip vortex appears as concentric
circles, whereas the trailing vortices have a curled shape
extending toward the propeller shaft from the tip.
Figure 8 shows a contour plot of the phase-averaged
spanwise vorticity at /=0 for three advance ratios. The
vorticity in the blade tip region increases significantly as
the propeller loading increases. This indicates that, com-
pared with the case of a light loading, more flow energy is
lost at the heavy loading due to the generation of stronger
tip vortices. For J=0.59 and 0.72, the tip vortex is skewed
upward in the downstream region up to X=0.5D. However,
the tip vortices are less skewed for the light loading of
J=0.88. This skewing of the tip vortex seems to be caused
by the strong interaction between the tip vortex and wake
sheet. As the wake flow goes downstream, the skewing
diminishes because of a weakening of the interaction be-
tween the tip vortex and wake sheet. For J=0.59 and 0.72,
the path of the tip vortices contracts toward the propeller
shaft downstream of X=0.5D. This larger contraction of
the tip vortex trajectory over the region up to X=0.5D
seems to result from blade-to-blade interaction. Specifi-
cally, the blade wake within the slipstream is faster than
the projection of the tip vortices in the axial direction,
leading to an interaction between the blade wake of the
Fig. 10. Spatial distributions of out-of-plane velocity (w) at /=0: present blade and the tip vortex generated from the pre-
a J=0.88; b J=0.72; c J=0.59 vious blade. The trailing vortices are composed of two
vortex sheets with opposite sign due to the boundary
not clearly resolved due to the lower spatial resolution of layers that develop along the upper and lower blade sur-
the 1 K 1 K CCD cameras and the conventional SPIV faces. For J=0.88, the trailing vortices of positive sign ap-
algorithm used to extract the velocity vectors. pear clearly at regular intervals and are stronger than those
Figure 7 shows the spatial evolution of the phase- of negative sign. However, as the loading increases, the
averaged spanwise vorticity (s-1) at J=0.72 for four con- trailing vortices of negative sign increase in magnitude,
secutive phase angles. These plots show that as the flow whereas those of positive sign decrease.
moves downstream, regularly-spaced tip vortices evolve. For J=0.88, the axial velocity within the slipstream is
These tip vortices are generated by the difference in smaller than that of the free stream, compared to that for
582

Fig. 12. Phase-averaged in-plane strain rate xy distributions at


/=0: a J=0.88; b J=0.72; c J=0.59 Fig. 13. Phase-averaged out-of-plane normal strain rate (zz)
distributions at /=0: a J=0.88; b J=0.72; c J=0.59

J=0.59. The breaking condition of negative thrust seems to to achieve J=0.88. Apparently, such a low revolution speed
be brought on by viscous effects due to the low Reynolds was insufficient to generate the designed thrust magnitude,
number of 1.8104. In fact, the advance ratio J can be which was reflected in the results of U/U0 in the slipstream
chosen by holding either the advance velocity or the at J=0.88, as shown in Fig. 5.
propeller revolution speed constant while the varying the Figure 9 shows the open-water test results of the Kp505
other variable. Given the limited choice of low advance propeller at a Reynolds number of 1.3107, where the
velocity with the present experimental setup, the revolu- boundary layer flow over the blade surface is turbulent.
tion speed of the model propeller was significantly lowered The thrust and torque acting on the propeller in
component (w) distribution was measured by SPIV. As
shown in Fig. 10, the out-of-plane velocity component has
large values along the trailing vortices. As the loading is
increased, the value of w becomes larger within the slip-
stream. This indicates that the magnitude of the out-of-
plane motion is greatest near the trailing vortices in the
slipstream region. For light loading (J=0.88), the value of w
near the trailing vortices is smaller than for the other
loadings, indicating smaller out-of-plane motions. The
present observation of non-axial flow motion of the wake
indicates that not all of the propulsion energy is used
583
effectively. Therefore, measurements of the out-of-plane
velocity component together with the strength of the tip
vortices, such as those presented here, could be used to
guide the design of more effective marine propellers.
Figure 11 shows the phase-to-phase variation of the
out-of-plane velocity component at the locations of the tip
vortices for J=0.72. The out-of-plane velocity of the tip
vortices decreases gradually beyond the downstream
location of X=0.2D. These results show that the magnitude
of the out-of-plane velocity of the tip vortices is about
813% of the free stream velocity in the near-wake region.
This means that, if the same propeller wake were measured
with using a 2-D PIV system, the in-plane velocity field
would be slightly overestimated due to perspective error
caused by the out-of-plane velocity component.
The in-plane strain rate xy was calculated for the three
loading conditions to investigate the behavior of the
spanwise vortices near the tip vortices. The diagonal and
off-diagonal terms in the strain rate tensor
ij (= (Ui,j + Uj,i)/2) are related to extensional strains and
shearing strains, respectively. The in-plane strain rate
xy (= u/y + v/X) can be calculated using the phase-
averaged in-plane velocity components from which the
perspective errors due to out-of-plane motion have been
removed. Figure 12 shows the spatial distribution of the
phase-averaged in-plane strain rate at each loading condi-
tion. Under the design (J=0.72) and heavy (J=0.59) loading
Fig. 14. Phase-to-phase variation of turbulence intensity in at
J=0.72: a axial turbulence intensity; b vertical turbulence intensity
conditions, the in-plane strain rate shows strong positive
values at the locations of the tip vortices. On the other hand,
weak in-plane strains with negative values are observed
open-water are normalized by the fluid density, revolution beneath the tip vortices. At the light loading condition
speed and propeller diameter to give the parameters KT (J=0.88), weak in-plane strains with positive values appear
and KQ, respectively. The values of KT and KQ decrease in the region of the negative vortices, whereas a region of
monotonically with increasing advance ratio. At J=0.88, KT strong negative in-plane strains is observed outside the
has the small value of 0.09. As the Reynolds number negative vortices shown in Fig. 8a. The negative in-plane
increases above 1.0105, the skin friction slowly decreases strains are in a staggered arrangement with the positive in-
up to a Reynolds number of 106 and then the flow becomes plane strains. From these results, we can conjecture that at
predominantly turbulent. For small Reynolds numbers the light loading condition, both the positive vortices and
(<1.0105), the variation in skin friction is roughly negative vortices are associated with the tip vortices.
inversely proportional to the square root of the Reynolds In addition to the in-plain strain rate xy, the out-of-
number. This implies that the skin friction is higher at plane normal strain rate zz was also estimated using the
smaller Reynolds numbers, leading to a greater tendency following continuity equation:
toward flow separation under those conditions. Therefore,
the significant reduction of thrust at J=0.88 observed in the @w @u @v
present experiments, which leads to the breaking condi- ezz   exx  eyy 7
@z @x @y
tion, can be attributed to laminar boundary layer separa-
tion resulting from the low Reynolds number of about 104 The normal strain rate zz corresponds to the out-of-
used in this study. plane velocity gradient in the direction normal to the
To investigate the out-of-plane motion of the flow measurement cross section, and therefore provides a
structure at the three loadings, the out-of-plane velocity measure of the out-of-plane motion around the tip and
584

p
u02
Fig. 15. Spatial distribution ofpaxial
turbulence intensity U0 and
v02
vertical turbulence intensity U0 at /=0: a J=0.88; b J=0.72;
c J=0.59
trailing vortices. The zz distributions at /=0, depicted in the strength and skew of the tip vortices decreased. The in-
Fig. 13, show that the out-of-plane motion is concentrated plane strain rate information suggested that the positive
within the slipstream region and that the strength of the vortices outside the tip vortices at light loading are asso-
out-of-plane motion becomes stronger as the loading ciated with the tip vortices.
increases. However, the out-of-plane velocity component In the near wake region, the axial turbulence intensity
cannot be recovered from the normal strain rate zz, which decreased gradually whereas the vertical turbulence
only indicates the existence of complicated 3-D flow intensity remained almost constant as the wake moved
structure. downstream due to the rotation of the propeller. For the
The spatial distributions of the turbulence intensity design and heavy loading conditions, the axial turbulence
were obtained by ensemble averaging the fluctuation of intensity was concentrated on the wake region, and the
velocity fields. Figure 14 shows the variation of the axial energy of the wake flow was diffused as the flow moved
585
and vertical turbulence intensities near the tip vortices and downstream.
trailing vortices for J=0.72. As the wake goes downstream Strong out-of-plane motion was observed within the
in the near-wake region of 0<X<0.5D, the axial turbulence propeller slipstream. As the propeller loading increased,
intensity decreases. However, in the same region, the the out-of-plane motion in the wake region was enhanced.
vertical turbulence intensity stays almost constant at The out-of-plane velocity beyond the location of the tip
44.5%. Beyond the near-wake region, the axial and ver- vortices was about 813% of the free stream velocity in
tical turbulence intensities gradually converge to the free the near-wake region. This large out-of-plane velocity
stream value (2%) due to viscous dissipation and tur- component may be the main source of the perspective
bulent diffusion. errors that lead to overestimation of the in-plane velocity
p p
Figure 15 shows the turbulence intensity distributions field for 2-D PIV. Therefore, the present results indicate
02 02
of the axial and vertical velocity components ( Uu0 ; Um0 ) that a SPIV technique that measures the three velocity
normalized by the free stream velocity U0. As the propeller components will be very useful for the accurate charac-
loading increases, the turbulence intensity for the axial terization of propeller wakes with strong out-of-plane
velocity component increases markedly. For the case of motions. Given that an accurate analysis of the wake
J=0.72 and 0.59, the axial and vertical turbulence intensi- characteristics is helpful to optimizing the design of ship
ties display local high values along the traces of the tip and propellers, consideration of SPIV results in the propeller
trailing vortices, compared with the other regions of the design process should lead to the production of more
wake. The flow interaction between the tip vortices and the efficient propellers.
wake sheet leads to the presence of finite turbulence
intensities of the axial and vertical velocity components in
the far downstream region. The magnitude of the axial
turbulence intensity is larger than that of the vertical References
turbulence intensity within the propeller wake region Calcagno G, Di Felice F, Felli M, Pereira F (2002) Propeller wake
tested in this study. analysis behind a ship by stereo PIV. Proc of 24th Symposium on
The vertical turbulence intensity distribution matches Naval Hydrodynamics, Fukuoka, 3:112127
Chesnaks C, Jessup S (1998) Experimental characterisation of pro-
well with the vorticity contours for all three loading cases peller tip flow. Proc of 22nd Symposium on Naval Hydrody-
(see Figs. 8 and 14), indicating that the turbulence inten- namics, Washington D.C., pp 156169
sity produced from the rotating propeller shaft makes only Cho CH, Lee CS (2000) Numerical experimentation of a 2-D B-spline
a small contribution to the overall turbulence intensity, higher order panel method. J Soc Naval Architects Korea 37(3):27
36
and therefore that the propeller shaft has only a small Cotroni A, Di Felice F, Romano GP, Elefante M (2000) Investigation of
effect on the wake structure. the near wake of a propeller using particle image velocimetry. Exp
Fluids 29:S227236
5 Lee JT (1987) A potential-based panel method for the analysis of
Conclusions marine propellors in steady flow, Ph. D. Thesis, Dept. Ocean
Engineering, MIT, Mass.
The complicated 3-D flow structure of the turbulent wake Lee SJ, Paik BG, Lee CM (2002) Phase-averaged PTV measurements of
behind a marine propeller was investigated experimentally propeller wake. Proc of 24th Symposium on Naval Hydrody-
using a SPIV technique with an angular configuration. namics, Fukuoka, 4:1825
Tip vortices were formed from the propeller and moved Prasad AK, Adrian RJ (1993) Stereoscopic particle image velocimetry
applied to liquid flows. Exp Fluids 15:4960
downstream as the propeller rotated. A viscous wake dis- Soloff SM, Adrian RJ (1997) Distortion compensation for generalized
playing velocity deficits was clearly observed in the region stereoscopic particle image velocimetry. Meas Sci Technol
near the blade tip (0<X<0.2D). As the flow moved down- 15:14411454
stream, the traces of the tip vortices contracted toward the Stella A, Guj G, Di Felice F, Elefante M (1998) Propeller wake evolu-
tion analysis by LDV. Proc of 22nd Symposium on Naval
propeller shaft up to X=0.5D and thereafter contracted Hydrodynamics, Washington D.C., pp 171-188
further due to a blade-to-blade interaction. As the wake Yoon JH, Lee SJ (2002) Direct comparison of 2-D PIV and 3-D ste-
moved downstream and the propeller loading decreased, reoscopic PIV measurements. Meas Sci Technol 13:16311642

You might also like