Вы находитесь на странице: 1из 11

Ocean Engineering 101 (2015) 264274

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Investigation on drag performance of anti-fouling painted at plates


in a cavitation tunnel
Bu-Geun Paik a,b,n, Kyung-Youl Kim a, Sung-Rak Cho a, Jong-Woo Ahn a, Sang-Rae Cho c
a

Korea Research Institute of Ships & Ocean Engineering, KIOST, 171 Jang-dong, Yuseong 305-343, Daejeon, Republic of Korea
Ship and Ocean Plant Engineering, Korea University of Science & Technology(UST), Republic of Korea
c
Daewoo Shipbuilding & Marine Engineering Co. Ltd., Republic of Korea
b

art ic l e i nf o

a b s t r a c t

Article history:
Received 11 March 2014
Accepted 4 April 2015

The at plate coated with silicone-type Tinfree self-polishing co-polymer (SPC) or the conventional
metal-type Tin-free SPC is prepared to investigate the drag performance of the anti-fouling SPC. The
local skin friction of anti-fouling paints is evaluated by a at plate model test method in the cavitation
tunnel. The properties of the boundary layer and the drag performance are investigated by ow and
force measurement techniques. The silicone-type SPC paint shows better drag performance than the
metal-type paint in the high speed regime. The silicone-type SPC paints also show decreasing roughness
function (U ) with the increase of displacement thickness Reynolds number (Ren) and roughness
Reynolds number (ks ). Even in the same silicone-type SPC paints with similar roughness function, drag
performance appears differently. The different drag performance in the silicone-type SPC painted
surfaces is considered to be affected by different turbulent vortical structures caused by the surface
roughness. Y-directional peak position of streamwise turbulence intensity is utilized to estimate the
existence of vortical structure. To investigate the reason of the different drag performance in the
silicone-type SPC painted surfaces, the POD analysis, extracting the most energetic ow elds, is adopted
to nd the effects of cross-ow velocity component caused by the turbulent vortical structure.
& 2015 Elsevier Ltd. All rights reserved.

Keywords:
Anti-fouling paint
Local skin friction
Turbulent boundary layer
Proper Orthogonal Decomposition (POD)
Laser Doppler Velocimetry (LDV)
Particle Image Velocimetry(PIV)

1. Introduction
Reducing the energy consumption of marine vehicles without
losing speed has been attracting much attention because reduction
of hydrodynamic drag is essential in obtaining higher speed and
lower energy consumption in marine vehicles. The paint used to
coat the hull of seagoing marine vehicles generally must have anticorrosion and anti-fouling properties to provide protection against
seawater and marine organisms. Anti-fouling paints are crucial in
preventing marine fouling by slime, algae and barnacles, which
attach to the outside of the hull and the mechanical components
of marine vehicles (Atlar, 2008). This fouling not only detracts
from the beauty of the view in a ship but also increases the surface
roughness at macro-scale and augments the resistance of the
marine vehicle in the sea. The conventional anti-fouling paint used
to contain a toxic organ Tin compound and is now strictly
prohibited by International Maritime Organization (IMO), because
it causes genetic mutations in aquatic organisms when it is
released into seawater (Omae, 2006). On the other hand, antifouling paint (metal-type Tin-free SPC) with antifoulant Cu2O is

Corresponding author. Tel.: 82 42 866 3464; fax: 82 866 3449.


E-mail address: ppaik@kriso.re.kr (B.-G. Paik).

http://dx.doi.org/10.1016/j.oceaneng.2015.04.026
0029-8018/& 2015 Elsevier Ltd. All rights reserved.

now widely used because it has little effect on marine pollution,


compared to the conventional paint containing Tin. The antifoulant of anti-fouling paint is released to outside the paint surface by
hydrolysis, when it makes contacts with seawater. This sort of
paint is called self-polishing co-polymer(SPC) paint. As the antifoulant is released from the resin surface of SPC, the seawater ow
gradually abrades the surface layer of the water-soluble resin.
Candries and Atlar (2004) investigated the drag reduction performance of the metal-type Tin-free SPC paint and foul- release paint
in the recirculating water tunnel, and reported that the coating
affected the near-wall turbulence intensities.
Recently, resin and antifoulant have been developed to reduce local
skin friction and fouling. The lm of SPC paint releases antifoulant by
hydrolysis and forms the surface structure of micro-roughness. Thus, it
is very important to study the properties of the turbulent boundary
layer over the rough surface, which can inuence the drag reduction.
Several great works have focused on the rough wall boundary layer
investigation. Kendall and Koochesfahani (2006) proposed models for
the boundary layer prole for estimating wall friction, and showed the
usefulness to provide accurate estimates of wall shear based on sparse
data points. Schultz (2000) studied turbulent boundary layers on the
at surfaces coated by marine lamentous algae, and reported algaecovered surfaces increased the wall-normal turbulence intensity,
Reynolds stress, and skin friction. nal et al. (2012) also investigated

B.-G. Paik et al. / Ocean Engineering 101 (2015) 264274

the turbulent boundary layer over at surfaces coated by nanostructured marine antifoulings, and showed nanostructured coatings provided good properties of local skin friction coefcients, roughness
functions and Reynolds stress.
The addition of a small amount of silicone-type (silyl-type)
material to SPC paint can reduce the surface energy and is known
to abrade the surface more evenly. It is necessary to validate the
performance of drag reduction in silicone-type Tin-free SPC paint,
comparing with metal-type Tin-free SPC paint. In the present
study, we conducted a dynamic similarity model test in a mediumsized cavitation tunnel with a ow speed of up to 9 m/s. Three
types of painted surface, one metal-type Tin-free SPC painted
surface and two silicone-type Tin-free SPC painted surfaces, were
assessed to give reliable experimental data for the paint researchers. The difference of two silicone-type SPC surfaces was only
caused by paint spraying skill of painters, and the ingredients of
paint and the stacking methods were similar to each other. The
painted at plates were deposited in the static seawater for one
month to give enough surface roughness after releasing antifoulant from the resin surface of SPC.
Flow measurement techniques of Laser Doppler Velocimetry
(LDV) were employed to investigate the properties of the turbulent
boundary layer over the painted surfaces. In each boundary layer,
the turbulence intensity distribution was measured to make sure
the existence of the coherent turbulent vortical structure on the
surfaces. For drag performance assessment, the roughness function was calculated in terms of local skin friction coefcient and
was investigated according to the displacement thickness and the
roughness Reynolds numbers. The conventional force measurement technique was also employed to assess the drag performance
globally. From these investigations, we were trying to conrm that
the silicone-type SPC paint has less local skin friction than metaltype SPC paint. Although the silicone-type SPC paints showed
similar roughness function each other with the ow speed
increase, the drag performance in the point of drag-increase
appeared differently. To gure out the reason of the different drag
performance in the high speed regime, the POD analysis was also
employed to nd the effects of cross-ow velocity component in
the turbulent boundary layer.
In Section 2, there are three sub-sections of 2-D at plate model
and paints, surface roughness and force measurements, and ow
velocity measurements of turbulent boundary layer. The following
Section 3 will present ve sub-sections such as the properties of
turbulent boundary layers, turbulence intensity in the boundary
layer, drag performance assessments using roughness function,
force measurements and POD method.

265

dimensions of the at plate were 950 L  340 W  10 H mm3. The


at plate was connected to 6-component force balance through
the strut of NACA0024 section. The leading edge of the at plate
had a bevel angle of 38.61 to avoid violent ow separation around
the edge. The trailing edge of the at plate had same bevel angle to
prevent ow separation increasing the drag. The experimental
conditions for the drag performance tests are presented in Table 1.
The self-polishing activity of the paint lm is activated when it
comes into contact with seawater. The seawater tank and plate
holders were prepared to simulate normal hydrolysis effects in
seawater, as in Fig. 1. The painted plate remained static in the
seawater for one month. A half of the lled seawater was relled
with fresh seawater once after 15 days.
Fig. 2 shows the photograph of the at plate installed in the cavitation
tunnel's test section. A 500 m-depth area for painting was prepared at
the bottom of the plate, as in Fig. 3. The thickness of one paint layer was
about 100250 m. Two or three layers were coated together with tie
coat. Two kinds of silicone-type Tin-free SPC painted surfaces (AF 1 and
AF 2) were used in the present study. A metal-type Tin-free SPC painted
surface (AF 3) was also used for comparison. The whole area of the plate,
except the painted area, was coated with polyamide to prevent seawater
corrosion during one month of seawater deposition. The coating thickness of the polyamide was approximately 100 m. Only the designated
bottom area of the plate was painted with the SPC paint. Thus, the
comparison between the smooth and polyamide or the smooth and
painted plate was not meaningful, and the polyamide plate having anticorrosion function was chosen as a reference in the drag
performance test.

2.2. Surface roughness and force measurements


As the surface roughness of the paint lm could affect the drag
performance with variation of the ow speed, surface roughness
was measured in advance before the seawater deposition. A portable
surface roughness tester (SJ-210 standard, Mitutoyo, Japan) was used

Table 1
The experimental conditions.
Plate type and number

Painted plate: 3
Polyamide plate: 1

Reynolds number
LDV/PIV measurement speed
Drag measurement speed
Roughness measurement number
Static pressure in test section
Density of water

0.5  1064.3  106


3, 5, 7 m/s
19 m/s, 1 m/s interval
2 (before and after sea water deposition)
Atmospheric pressure
1000 kg/m3

2. Experimental apparatus and method


2.1. 2-D at plate model and paints
The prole drag and LDV/PIV measurements on 2-D at plates
were carried out in the medium-sized cavitation tunnel of KRISO
(Korea Research Institute of Ships & Ocean Engineering, former
MOERI). The dimensions of the rectangular test section were
0.6 W  0.6 H  0.6 L m3. The ow speed at the test section was
varied from 3 to 9 m/s in the present study.
The at plate models were designed and manufactured to
estimate the local skin friction of the Tin-free SPC paints. In
addition to SPC paints, one more reference plate was manufactured to obtain standard data on non-SPC paint lm. The reference
plate was fully coated with polyamide to avoid seawater corrosion.
The material of the at plates was mild steel. The plates were heat
treated to maximize the atness of the plates and to minimize
their twisting and bending effects against uid ow. The

Fig. 1. Sea water deposition of painted plate.

266

B.-G. Paik et al. / Ocean Engineering 101 (2015) 264274

Fig. 5. Force balance for drag measurements.


Fig. 2. Photo of at plate model in cavitation tunnel.

Leading
edge

Trailing
Space for painting

edge

Fig. 3. Side view sketch of the at plate.

Fig. 6. LDV for ow velocity measurements.

Fig. 4. Portable surface roughness tester SJ-210 for roughness measurements.

to measure the surface roughness. The measuring ranges in horizontal and vertical directions were 17.5 mm and 360 m, respectively. The cut-off length of a Gaussian lter was 2.5 mm. It has
stylus features of displaying sectional calculation results. Although
we tried to assess proles with a resolution of 0.02 m, the probe
volume of the roughness measurement device might be larger than
the actual measurement resolution of it. Fig. 4 shows the roughness
measurements of the lm surface at the micro-scale.
A 6-component balance was used to measure the drag, as
shown in Fig. 5. The strut and model holding jig were rmly
connected to the top of the tunnel test section and the
6-component balance. The X-axis ran along the tunnel centerline,
pointing downstream, and the Z-axis was set horizontally toward
the port (left) direction looking from downstream. The Y-axis was
arranged along the Y-directionally upward. The origin was positioned at the leading edge of the at plate, and distances in the X-,
Y-, and Z-axes were normalized by the plate length. The capacity of
the force balance was 7500 N in the X- and Z-directions, and
7800 N in the Y-direction. The total uncertainty in the force
measurements was 0.3% in each direction. The systematic and the
random errors were 0.125% and 0.275%, respectively.
2.3. Flow velocity measurements of turbulent boundary layer
To measure the boundary layer over the surface of the paint, 1-D
LDV (FlowExplorer 300, Dantec Dynamics) was employed and a 3-D

auto traverse was used to move the probe, which had a resolution of
10 m. Titanium dioxide (TiO2) seeding particles with a mean
diameter of 3 m moved through the measurement volume. The
mean velocity was obtained from sampling of 20,000 Doppler signals
at each spatial position. Fig. 6 shows the LDV for the ow velocity
measurements.
The uncertainty analysis of the calibration for LDV was performed
by a ywheel method. The measurement uncertainty on the ywheel
velocity reference was 0.018%. There were several sources of uncertainties such as linearity and error weight sensitivity. All uncertainties
added together and reached a value of 0.056%. For the uncertainty
analysis of turbulence statistics, the 95% condence limits for the mean
and RMS values were computed, based on the following expressions
(Benedict and Gould, 1996).
s
2
U
Mean :
1:96 
N

RMS :

s
2
U
1:96 
2N

Here, U is the measured ow velocity and N is the total sample


number. The calculated uncertainties of employed LDV measurements were approximately 1.43% in mean value and approximately 1% in RMS value.
In addition to the LDV measurements, PIV velocity eld measurements were utilized to investigate the drag performance further. The
two-frame PIV system consists of a dual-head Nd:YAG laser (200 mJ
per pulse), CCD camera, a synchronizer, and a frame grabber. The
measurement region beneath the bottom of the at plate was set to an

B.-G. Paik et al. / Ocean Engineering 101 (2015) 264274

267

Table 3
Properties of the boundary layer.
U (m/s)

Re (x L/2)

item

AF 1

AF 2

AF 3

Polyamide

1.43  106

(mm)
n(mm)
(mm)
H
Re

10.652
1.712
1.213
1.411
3639

9.635
1.662
1.175
1.414
3525

10.438
1.715
1.208
1.419
3624

8.701
1.488
1.072
1.388
3216

2.39  106

(mm)
n(mm)
(mm)
H
Re

8.831
1.511
1.096
1.391
5430

9.025
1.575
1.136
1.386
5680

9.082
1.506
1.077
1.398
5385

7.799
1.272
0.926
1.373
4630

3.35  106

(mm)
n(mm)
(mm)
H
Re

8.687
1.492
1.076
1.361
7672

8.864
1.462
1.083
1.349
7581

8.843
1.495
1.083
1.380
7581

7.591
1.191
0.879
1.355
6153

Flow

Laser

1M CCD camera

Fig. 7. Experimental set-up of PIV system.

Table 2
Measurement uncertainties of displacement vectors measured by the PIV system.

X-direction
Z-direction

Velocity (%)

Standard deviation (%)

7 1.34
7 1.25

7 1.28
7 1.16

XZ measurement plane. The coordinate system and the XZ measurement plane used in this study are shown in Fig. 7. The eld of view was
1.7  1.7 cm2 and the measurement plane was located at y E100. The
CCD camera has a resolution of 1024  1024 pixels, and it was located
outside the bottom wall of the tunnel to capture pairs of particle
images separated by short time intervals using the frame-straddling
method. TiO2 particles were also used as PIV tracers. The velocity elds
were extracted using the PIV algorithm of the cross-correlation method
based on the fast Fourier transformation (FFT). In this experiment,
interrogation windows of 32  32 pixels, overlapped by 50%, were used
to obtain 800 instantaneous velocity elds.
The analysis of measurement uncertainty was carried out for
actual particle images for which the ow eld was known in
advance. The quiescent ow was tested for the uncertainty
evaluation of the present PIV system following the procedure
recommended by Raffel et al. (1998) to estimate the measurement
errors of PIV systems under the same condition as explained in the
experimental apparatus and method. The standard errors encountered in measuring the displacement vector were calculated and
the results are summarized in Table 2. The size of the seeding
particles in the ow images must exceed a certain minimum size
for correct velocity vectors to be extracted. In the present work,
the average particle diameter was over 3 pixels, and hence the
measurement uncertainty with respect to the particle diameter
was sufciently small to prevent the peak-locking effect.

3. Results and discussion


3.1. The properties of turbulent boundary layers
The properties of the boundary layer are presented in Table 3.
In Table 3, is the boundary layer thickness. The displacement
thickness n is dened by following equation:

1
0


1


U
dy
U0

where U0 denotes the freestream ow velocity. In addition, the

momentum thickness was obtained from the following equation.



Z 1 
U
U
dy
2

1
U0
U0
0
The ratio of displacement thickness to momentum thickness is
called the shape factor, which is used to determine the nature of
the boundary layer ow.
H

The properties of the turbulent boundary layer can be checked by


the shape factor or the momentum thickness Reynolds number. A
turbulent boundary layer can be formed when H has a range of
1.31.4. A laminar boundary layer will appear when H is 2.42.6.
Based on the boundary layer property of the shape factor H, a
turbulent boundary was well generated in the whole velocity
region. The Reynolds number, which is based on the momentum
thickness, has often been utilized to conrm if the simulated
velocity prole of a boundary layer was asymptotically similar to a
full-scale turbulent boundary layer. The denition of the Reynolds
number is as follows:
Re

U0

Here, is the kinematic viscosity. David et al. (1999) reported


that the velocity prole of a full-scale turbulent boundary layer
could be simulated asymptotically when Re 46000, and sufciently developed turbulent boundary layer was formed in the
ow range larger than 3 m/s in the present study.
Fig. 8 shows the typical boundary layer proles measured by
LDV. The velocity measurements were conducted at a high spatial
resolution of 10 m near the surface of the plate wall. The entire
boundary layer was plotted in Fig. 8(a). The Y-axis represents the
non-dimensionalized vertical distance divided by the boundary
layer thickness. Fig. 8(b) shows an enlarged plot of the region
below (Y )/ 0.3. The value of error in origin was ranged from
10 to 30 m in the painted and polyamide surfaces. The X-axis
represents the non-dimensionalized velocity divided by the free
stream speed U0. The non-dimensionalized velocity proles of the
AF paints was similar each other and similar to polyamide, as
shown in Fig. 8.
The distance Y between the wall and any uid position can be
non-dimensionalized as y .
y

YU n

268

B.-G. Paik et al. / Ocean Engineering 101 (2015) 264274

0.2

0.4

0.6

0.8

2
polyamide(7m/s)
AF 1(7m/s)
AF 2(7m/s)
AF 3(7m/s)

1.5

(Y+)/

As reported elsewhere, the turbulent boundary layer has a coherent


turbulence structure (Smith, 1984) as well as a specic velocity
distribution. The coherent turbulence structure has a longitudinal
vortex structure whose shape is similar to a hairpin in the region of
y o200, and it generally interacts with the surface of the plate wall.
The local skin friction increases when the longitudinal vortex
approaches the wall and strongly interacts with the wall, and vice
versa when the longitudinal vortex goes away from the wall to weaken
the interaction with the wall. Typically, riblet-patterned swimsuits
adopt this sort of mechanism, with the tip of the riblet preventing any
interaction between the vortex and the wall, or breaking the vortex
structure to decrease the interaction (Lee and Lee, 2001). In addition to
the slip or nonslip on the surface, the interaction between the longitudinal vortex and the wall is important because the turbulence
structure caused by the paint lm affects the local skin friction.

1.5

0.5

0.5

0
0.2

0.4

0.6

0.8

3.2. Turbulence intensity in the boundary layer

U/U0

0.2

0.4

0.6

0.8

The vortex is a swirling ow that increases the instability and


the uctuation component of the ow around it, augmenting the
turbulence intensity. The turbulence intensity can be dened as
follows:
p
u02
Tu
9
U0

1
0.3

0.3

(Y+)/

polyamide(7m/s)
AF 1(7m/s)
AF 2(7m/s)
AF 3(7m/s)

0.2

0.2

0.1

0.1

0.2

0.4

0.6

0.8

U/U0
Fig. 8. Velocity proles of boundary layer measured at U0 7 m/s, (a) wide view
and (b) narrow view.

Here, Un means the friction velocity on the wall and is dened


by following equation.
Un

is the wall shear stress and is the uid density.


The velocity distribution of the turbulent boundary layer
follows the law of the wall. The logarithmic region of the boundary
layer is given by
U

ln Y B

The Von Krmn constant and constant B were 0.41 and


5.0 for a smooth or rough wall, respectively (Schlichting and
Gersten, 2000). The boundary layer velocity U should be nondimensionalized as follows:
U

U
n
U

According to the law of the wall, the viscous sublayer, buffer


layer, and inner layer are classied by the regions of y o5,
5 oy o30, and 30oy o  100, respectively. The region of
y 4  100 is generally called the outer layer.

Here, u0 is U  Uand Uis the time-averaged value of the total


velocities. If the longitudinal vortex is close to the plate wall, high
turbulence intensity will appear near the wall. The turbulence
intensity also increases as the vortex strength increases (Paik et al.
2004). Fig. 9 shows the turbulence intensity measured at each
painted plate. The Y-axis was non-dimensionalized by the boundary layer thickness. The peak value and peak position in the
turbulence intensity varied according to the ow speed and the
surface condition of the paint lm. The locations of the maximum
turbulence intensity were far from (Y )/ 0 at a ow speed of
3 m/s in the AF paints. In the 3 m/s case, all AF surfaces showed
similar turbulence intensity. The Y-directional peak position of the
maximum turbulence intensity in AF 1 was slightly closer to the
wall than that in AF 2. The peak locations at 7 m/s were clearly
different each other. The Y-directional peak position of AF 1 was
farther away from the wall than that of AF 2. Thus, the turbulent
vortical structure in the AF 2 boundary layer seemed to show a
stronger interaction with the wall than AF 1, as the ow speed
increased. Although the Y-directional peak position of AF 3 was
similar to AF 1 at 7 m/s, AF 3 showed largest turbulence intensity
which could be related the swirling strength of vortical structure.
The polyamide showed much larger turbulence intensity than AF
surfaces and most closer to the plate surface in both ow cases.
The inuence of the coherent turbulence structure was indirectly
investigated by measuring the turbulence intensity in the boundary layer, although the surface's topological structure was not
revealed totally in the paint lm. Later studies could examine
systematically how the paint components, their combination, and
painting methods affect the coherent turbulence structure in
similar types of paints.
3.3. Drag performance assessment using roughness function
The wall shear stress has to be considered to calculate the
local skin friction coefcient Cf.
Cf

0:5U 20

10

To obtain , the friction velocity value should be known rst


from Eq. (6). The friction velocity can be obtained directly by using
gauges to read the shear stress of the wall, but we could not

B.-G. Paik et al. / Ocean Engineering 101 (2015) 264274

0.1

0.05

0.1

polyamide(3m/s)
AF 1(3m/s)
AF 2(3m/s)
AF 3(3m/s)

0.09
0.08

101

0.2
0.1

102

103

30

30

25

25

20

20

15

15

0.09
0.08
0.07

0.06

0.06

0.05

0.05

0.04

0.04

0.03

0.03

0.02

0.02

0.01

0.01

0.07

(Y+)/

0.15

269

viscous sublayer
logarithmic form
polyamide(3m/s)
AF 1(3m/s)
AF 2(3m/s)
AF 3(3m/s)

10

0
-0.01

0.05

0.1

0.15

-0.01
0.2

10

u'/U 0

10

(y+)
0.1

0.05

0.1

polyamide(7m/s)
AF 1(7m/s)
AF 2(7m/s)
AF 3(7m/s)

0.09
0.08

(Y+)/

0.15

0.2
0.1

10

10

Fig. 10. Velocity proles of polyamide and AF surfaces using the tting method.

0.09
0.08

Table 4
Local skin friction coefcients (  103).

0.07

0.07

0.06

0.06

Type

3 m/s

5 m/s

7 m/s

0.05

0.05

0.04

0.04

Smooth
Polyamide

2.990
3.362

2.827
3.042

2.592
2.780

0.03

0.03

0.02

0.02

0.01

0.01

-0.01
0

0.05

0.1

0.15

-0.01
0.2

u'/U 0
Fig. 9. Comparison of turbulence intensity distributions in turbulent boundary
layers, (a) 3 m/s case (b) 7 m/s case.

perform direct measurement of it because the ush-mounted


shear stress sensor was not employed in this study. Instead, the
indirect tting method (Clauser, 1956) was utilized in the present
study, as the friction velocity varies depending on the nature of the
paint lm and the ow velocity. Frankly, the Clauser method is not
appropriate for rough surfaces because the optimization of error in
origin is not possible. But in the present study, the Clauser method
was considered for relative comparison and estimation of rough
surfaces, assuming that the error in origin was approximately
found from mean velocity and turbulence intensity proles.
The iterative methods of tting process depend on the logarithmic region of the boundary layer over the surface wall, and
then the calculation of friction velocity can be followed (nal et al.,
2012). Fig. 10 shows the velocity proles of the polyamide and AF
surfaces, obtained through the tting process at a ow velocity of
3 m/s. The source of the irregularity at the region of y 4 30 might
come from thin strut installed at upstream contraction part of the
tunnel. Actually, this medium-sized cavitation tunnel has been
used for propeller cavitation observation and the thin strut
supporting propeller shaft made some weak wake in the test
section. The at plate was installed at the position of small wake
region as much as possible; however, slight perturbation caused
by strut wake still existed in the results of the boundary layer
measurements. In the case of polyamide plate, the increase of the

local skin friction was 12.3 7 0.5%, compared to the smooth plate.
The drag increase in the polyamide coated plate was meaningful
because of providing reference value in drag performance to other
surfaces. The local skin friction coefcients of smooth and
polyamide plates are displayed in Table 4 for reference. The
tting method was not easy to use at the velocity range higher
than 3 m/s because of the contraction of the boundary layer and
the augmentation of the turbulent uctuating velocity
components.
The roughness function U (Hama, 1954) is related to the
increase of the local skin friction and the function of local skin
friction coefcient Cf for smooth and coated surface at the same
displacement thickness Reynolds number Ren.
s!
s!
2
2

U

11
cf
cf
Smooth

Rough

The roughness functions of four kinds of surfaces were plotted


with respect to Re as in Fig. 11. As the Reynolds number increases,
the polyamide and AF 3 surfaces showed saturated trend; however, abruptly decreasing trend appeared in other silicone-type
SPC paints. It means that the local skin friction of silicone-type SPC
surfaces decreased more than those of the polyamide and metaltype surfaces with increasing Reynolds number, which were
similar to the results reported by nal et al. (2012) who investigated the turbulent boundary layer of silicone-containing foulrelease coating.
Table 5 shows the roughness results measured at 6 points from
leading edge to trailing edge before the seawater deposition and
after the drag performance tests. Ra, Rq, and Rt are average
roughness, root mean square roughness and maximum roughness
height, respectively. Rsk represents the skewness of the surface
elevation probability density function and a measure of the

270

B.-G. Paik et al. / Ocean Engineering 101 (2015) 264274

7
Polyamide
AF 1
AF 2
AF 3

1.8
1.6

1.2

4
+

1.4

Polyamide
AF 1
AF 2
AF 3
Colebrook (1939)

0.8
0.6

0.4

1
0.2
4000

5000

6000

7000

8000

9000

10000

11000

Re *

0 -1
10

10 0

10 1
+
ks

Fig. 11. Roughness functions with the displacement thickness Reynolds number
(uncertainty in U : 72.3%).

Fig. 12. Roughness functions with the roughness Reynolds number (uncertainty in
U : 72.3%).
Table 5
Roughness measurement results of four coatings.

Polyamide
AF 1
AF 2
AF 3

before
after
before
after
before
after
before
after

Ra (m)

Rq (m)

Rt (m)

Rsk

3.57
3.57
2.88
2.94
1.76
1.85
3.05
3.32

4.25
4.25
3.31
3.39
2.05
2.06
3.57
3.72

16.20
16.20
33.13
35.84
15.28
19.28
14.42
14.66

 0.02
 0.02
 0.11
 0.12
 0.25
 0.25
 0.10
 0.11

12

This correlation has been known to work well for sand grain and
packed spheres covered with grit. In addition, Eq. (12) was not
validated clearly for the coating with negative skewness of the
surface elevation. Thus, the Rt was measured at six positions
several hundred times and averaged to estimate the roughness
height(Schultz, 2000) in this study. The roughness Reynolds
number was given by ksUn/. The variation of the roughness
function according to ks is shown in Fig. 12, which includes the
correlation curve of the well-known formula (Colebrook, 1939).
The silicone-type surfaces showed highly decreasing trend with
increasing ks , which was the opposite of the tendency in the
correlation curve. This is also similar to the results reported by
nal et al. (2012). The metal-type and polyamide surfaces showed
rather slowly decreasing trend with roughness Reynolds number.
The noticeable thing is that AF 1 and AF 2 have similar U values
at different ks . In other words, the different surface structures
existed in the silicone-type SPC surfaces even with similar Cf.

10

Polyamide
AF 1
AF 2
AF 3

0.020

average of the rst derivative of the surface. There was a little


difference in the roughness before and after the drag measurements. Although the roughness Rt of AF 1 was much larger than
that of Polyamide or AF 2 and AF 3, the other surface roughness
parameters can affect the relation with the local skin friction.
The roughness function can be presented in terms of roughness
Reynolds number ks . The equivalent roughness height ks value
can be obtained by following equation (Flack and Schultz, 2010)
as:
ks 4:43Rq 1 Rsk 1:37

0.020

0.018

0.018

0.015

0.015

0.013

0.013

CD

type

10

Velocity(m/s)
Fig. 13. Drag measurement results on four at plates.

3.4. Drag performance assessment using force measurements


The prole drag is the summation of the local skin friction, the
form drag caused by the plate thickness, and the pressure drag
caused by the pressure separation around the strut. The value of
the form and pressure drags of each surface would be expected to
be the same because the drag measurements of each plate were
conducted under the same experimental conditions. Therefore, the
prole drag difference between the polyamide and the AF paint
denoted the difference in their mean skin friction values. Based on
this estimation method, the mean skin friction was calculated in
the present study. Fig. 13 shows the prole drag measurement
results of the polyamide and AF paints. The drag coefcient CD
decreased in accordance with an increase in the ow speed
(Reynolds number). The drag coefcient was expressed as follows:
CD

D
1
2

AU 0 2

13

B.-G. Paik et al. / Ocean Engineering 101 (2015) 264274

Here, the measured drag D was obtained by 6-component force


balance, and was the density of the fresh water, and A represents
the painted area of the plate. The uncertainty in CD is shown in
Table 6. The drag coefcients of AF 1 and AF 3 were larger than
that of the polyamide at 3 m/s. They were still larger than that of
the polyamide at 9 m/s. To show the measured results more
quantitatively, each value was shown in Table 7. In Eq. (14), R
meant the drag reduction rate of the painted plates with respect to
the polyamide plate. The drag reduction was dened as follows:

R%

C D_paint  C D_poly
C f _poly

14

x100

CD was calculated with the drag D measured by the force balance.


The CD_paint  CD_poly was the prole drag difference between the
painted and the polyamide plates, and it was equivalent to the
difference in the mean skin friction because the form and the
pressure drags were the same at the same experimental conditions. The local skin friction coefcient of polyamide surface was
obtained from the friction velocity at each ow speed, which were
already tted by the Clauser method.
AF 1 showed a drag reduction rate of 0.95% at 5 m/s, but it
showed a drag-increase of 2.50% with respect to the polyamide at
9 m/s. AF 2 showed a drag reduction of 3.32% at 5 m/s and a dragincrease of about 4.31% at 9 m/s. The drag reduction effect of AF
2 was apparent at a medium speed of 5 m/s, whereas the same
effect was not apparent with AF 1 at 5 m/s. The metal-type SPC
paint AF 3 showed a signicant increase of drag, compared to
silicone-type SPC paints. The low surface energy of the siliconetype SPC was found to be more effective than the metal-type in
terms of drag reduction.
The dominant factor affecting drag performance is known to be
the surface roughness. The vortical structure in the boundary layer
is also inuenced by the roughness of the surface. Considering
Table 7 and Fig. 9, the turbulent vortical structure is considered to
be another factor in drag performance. Although the shape and the
structure of the roughness were not totally understood, the surface roughness affected the turbulence structure over the surface.
As AF 1 with relatively high roughness Reynolds number showed
less drag-increase than AF 2 from 5 m/s to 9 m/s, the ow
structure over the paint surface should be investigated more to
know the relation between the drag performance and the turbulence structure.

Table 6
Measurement uncertainties in drag coefcient CD (99.7% of the readings lie within
7 3 of mean value, is the standard deviation).
Item

A(%)

U0 (%)

D (%)

Overall (%)

3 m/s
5 m/s
7 m/s
9 m/s

7 0.5
7 0.5
7 0.5
7 0.5

7 0.65
7 0.54
7 0.66
7 0.73

7 0.23
7 0.25
7 0.29
7 0.35

1.215
1.164
1.233
1.287

271

3.5. Drag performance assessment using POD method


To elucidate why AF 1 showed smaller drag-increase than AF
2 in the high-speed regime, the cross-ow produced by coherent
vortical structure was investigated by using the standard deviation
of the mean velocity component and proper orthogonal decomposition (POD) analysis. Fig. 14 shows the results of the standard
deviation () of the mean velocity component in the Z-direction.
In the 3 m/s case, the AF 1 plate showed higher than the AF
2 plate in the overall region. It meant that X-directional uid
motion was less inuenced by coherent turbulence structure in
the AF 2 plate than in the AF 1 plate. This low in the AF 2 plate
was related to the velocity augmentation of the buffer and inner
layers in the turbulent boundary layer. If the ow speed increased
to 7 m/s, the value of AF 2 increased more than at a ow speed
of 3 m/s, and the value showed a similar distribution to AF 1.
Thus, it is difcult to ascertain why the drag-increase of AF 2 was
larger than that of AF 1 from 5 m/s to 9 m/s.
In the next stage, POD analysis was conducted to investigate
the most energetic component of an innite-dimensional process
with only a few modes (Lumley et al., 1996). The present study
employed the snapshot method (Sirovich, 1987) represented with a
nite set of snapshots' sampled at discrete times. Members of the
data set {ui} can be represented as the sum of the coefcients of
projection aj and basis functions j as follows:
ux

N
X

aj j x;

15

j1

Here N is the total number of modes. To derive aj and j, we


should choose to maximize the averaged projection of u onto
in a Hilbert space, suitably normalized as follows:
D
 E
u; 2
;
16
max
2

where j U j denotes the modulus and U is the norm: f f ; f 1=2 .


Maximizing this inner product leads to the solution of the
following eigenvalue problem:
n

D Sij x; x0 j x0 dx0 i x
n

17
n
i

The are eigenvalues and the are eigenfunctions for each


POD mode n. The maximum in Eq. (16) corresponds to the largest
eigenvalue 1 of Eq. (17).
The normalized eigenvalues were extracted from the averaged
two-point correlation tensor, and the rst eigenmode contained
about 15% of the turbulent kinetic energy in the unsteady
turbulent boundary layer as shown in Fig. 15. This implies that
the dominant rst eigenmode contains a large amount of turbulent kinetic energy, and that the ow pattern can be represented
by the rst eigenmode. At a ow speed of 3 m/s, the cross-ow
velocity component by vortical structure in the turbulent boundary layer for AF 1 was larger than AF 2. However, if the ow speed
increased to 7 m/s, AF 1 showed very little cross-ow and AF
2 showed a similar cross-ow to that observed at 3 m/s. As seen in
Fig. 16, positive X-directional motion was increased due to fewer

Table 7
Results of skin friction reduction rate.
U0

3 m/s

type

AF 1

AF 2

AF 3

AF 1

AF 2

AF 3

AF 1

AF 2

AF 3

AF 1

AF 2

AF 3

CD (  10  2)
R (%)

1.586
 0.66

1.588
 0.22

1.603
3.08

1.423
 0.95

1.442
 3.32

1.461
6.24

1.351
0.34

1.360
1.69

1.398
7.22

1.325
2.50

1.340
4.31

1.362
9.50

5 m/s

7 m/s

9 m/s

272

B.-G. Paik et al. / Ocean Engineering 101 (2015) 264274

Fig. 14. Results of standard deviation in Z-direction, (a) AF 1, 3 m/s, (b) AF 2, 3 m/s, (c) AF 1, 7 m/s and (d) AF 2, 7 m/s.

Energy fraction

0.16

50

100

150

200
0.16

0.15

0.15

0.14

0.14

0.13

0.13

0.12

0.12

0.11

0.11

0.1

0.1

0.09

0.09

0.08

0.08

0.07

0.07

0.06

0.06

0.05

0.05

0.04

0.04

0.03

0.03

0.02

0.02

0.01

0.01

0
-0.01

small-scale eddy motions over the AF 1 plate at 7 m/s. The impact


of the coherent turbulence structure on the local skin friction (i.e.,
whether it was weakened or pulled away from the wall by the
micro-scale topology) has not been conrmed in detail. However,
the inuence of coherent turbulence structure was decreased
more in the AF 1 plate than in the AF 2 plate, and affected the
drag-increase tendency. Accordingly the X-directional momentum
was augmented in the turbulent boundary layer of AF 1.
We recommend that these experimental results be reected in
the current anti-fouling paint design, which is based on surface
roughness. It could be modied to control the turbulence structure
in the boundary layer. Since the drag performance tests using only
the force measurements are not sufcient to estimate or investigate the interaction between the uid ow and wall surface,
various hydrodynamic approaches are required to investigate the
boundary layer in detail.

0
0

50

100

150

-0.01
200

Mode No.
Fig. 15. Typical modal energy distribution of ow eld over the painted plate.

4. Conclusions
The drag performance of Tin-free SPC paints was studied using
an experimental method. In addition to measuring the drag

B.-G. Paik et al. / Ocean Engineering 101 (2015) 264274

273

Fig. 16. Distribution of Z-directional velocity in the rst eigen mode, (a) AF 1, 3 m/s, (b) AF 2, 3 m/s, (c) AF 1, 7 m/s and (d) AF 2, 7 m/s.

directly, the properties of the turbulent boundary layer over the


paint surfaces were measured by the LDV technique and analyzed.
The PIV measurements were also conducted to adopt POD method
for understanding different drag performance with ow speed.
Silicone SPC paints have similar U values even at different ks .
The ow velocity and force measurement data indicated that paint
lm with low roughness is effective in reducing drag at speeds less
than 5 m/s. The ow structure at speeds over 5 m/s showed that the
amount of drag-increase was found to be reduced when the magnitude of cross-ow velocity component decreased at the rst eigenmode extracted by POD analysis. The drag-increase was also reduced
when the Y-directional peak position of streamwise turbulence
intensity moved farther away from the wall. Although the turbulence
intensity and velocity elds were measured locally at the developed
turbulent boundary layer region, they were enough to support the
results of global drag performance. These experimental results in the
development of low frictional SPC paints indicate that the uiddynamical approach focusing on the turbulence structure is necessary
in addition to the conventional method that concentrates on the force
measurements only.

Acknowledgment
This research was sponsored by Project PES1810 and Project
PNS2260. We sincerely appreciate the supports for this research works.

References
Atlar, M., 2008. An update on marine antifoulings. 1420. 25th ITTC Group
Discussions 3 - Global Warming and Impact on ITTC Activities, Fukuoka, Japan,
pp. 563603.
Benedict, L.H., Gould, R.D., 1996. Towards better uncertainty estimates for turbulence statistics. Exp. Fluids 22, 129136.
Candries, M., Atlar, M., 2004. Experimental investigation of the turbulent boundary
layer of surfaces coated with marine antifoulings. J. Fluids Eng. 127 (2),
219232.
Clauser, F.H., 1956. The turbulent boundary layer. Adv. Appl. Mech. 4, 151.
Colebrook, C.F., 1939. Turbulent ows in pipes with particular reference to the
transition region between the smooth and rough pope ows. J. Inst. Civil Eng.
11, 133155.
David, B.D., Donald, R.W., John, K.E., 1999. The effect of Reynolds number on
boundary layer turbulence. Exp. Therm. Fluid Sci. 18, 341346.
Flack, K.A., Schultz, M.P., 2010. Review of hydraulic roughness scales in the fully
rough regime. J. Fluids Eng. 132, 041203.
Hama, F.R., 1954. Boundary-layer characteristics for smooth and rough surfaces.
Trans. SNAME 62, 333351.
Kendall, A., Koochesfahani, M., 2006. A method for estimating wall friction in
turbulent boundary layers. 25th AIAA Aerodynamic Measurement Technology
and Ground Testing Conference, 58 June, Sanfrancisco, California, USA.
Lee, S.J., Lee, S.H., 2001. Flow eld analysis of a turbulent boundary layer over a
riblet surface. Exp. Fluids vol. 30 (2), 153166.
Lumley, J.L., Holmes, P., Berkooz, G., 1996. Turbulence, coherent structures,
dynamical systems and symmetry. Cambridge University Press, Cambridge.
Omae, I., 2006. Chemistry and fate of organotin antifouling biodices in the
environment. The Handbook of Environmental Chemistry vol. 5, 1750.
Paik, B.G., Lee, C.M., Lee, S.J., 2004. PIV analysis of ow around a container ship
model with a rotating propeller. Exp. Fluids vol. 36 (6), 833846.
Raffel, M., Willert, C., Kompenhans, J., 1998. Particle image velocimetry. Springer,
ISBN: 3-540-63683-8.

274

B.-G. Paik et al. / Ocean Engineering 101 (2015) 264274

Sirovich, L., 1987. Turbulence and the dynamics of coherent structures Part 1:
Coherent structures. Q. Appl. Math. 45, 561572.
Smith, C.R., 1984. A synthesized model of the near-wall behavior in turbulent
boundary layers. In: Zakin J., Patterson G. (eds.). In: Proceedings of 8th
Symposium on Turbulence, University. Missouri-Rolla, Rolla, Missouri,
pp. 299325.
Schlichting, H., Gersten, K., 2000. Boundary layer theory. Springer, ISBN: 3-54066270-7, pp. 522524.

Schultz, M.P., 2000. Turbulent boundary layers on surfaces covered with laments
algae. J. Fluids Eng. vol. 122, 357363.
nal, U.O., nal, B., Atlar, M., 2012. Turbulent boundary layer measurements over
at surfaces coated by nanostructured marine antifoulings. Exp. Fluids vol. 52,
14311448.

Вам также может понравиться