Experimental Thermal and Fluid Science 120 (2021) 110246

Contents lists available at ScienceDirect

Experimental Thermal and Fluid Science

journal homepage: www.elsevier.com/locate/etfs

Stereoscopic PIV measurements of flow over a riblet surface at high T

Reynolds number
G.M. Ozkana, , G.E. Elsingab, W.-P. Breugemb, D. Stübingc, K.J. Reynoldsd, J. Westerweelb
⁎

a
Dep. of Mechanical Engineering, Çukurova University, 01330, Saricam, Adana, Turkey
b
Laboratory for Aero & Hydrodynamics, Process and Energy Department, Delft University of Technology, Mekelweg 2, 2628 CD Delft, the Netherlands
c
Fraunhofer Institute for Manufacturing Technology and Advanced Materials IFAM, Bremen, Germany
d
AkzoNobel/International Paint Ltd, Gateshead, UK

ARTICLE INFO ABSTRACT

Keywords: The effect of drag reducing riblets on the flow structure was examined experimentally for a turbulent boundary
Turbulent boundary layer layer at Reθ = 9890 and riblet spacing s+ = 13.4. Trapezoidal riblets were used, which were attached to the
Drag reduction water tunnel wall as a coating. Force measurements were performed to quantify the amount of drag reduction.
Riblet coating Then, the mechanism underlying this reduction was investigated by stereo-PIV measurements in the cross-stream
Stereo-PIV
plane. To determine the effect of the drag reducing riblets, the results were compared with the smooth flat plate.
Time-averaged turbulent statistics such as turbulent kinetic energy and Reynolds shear stress were found to be
lower over the riblets compared to the flat surface. Two-point correlations of the fluctuating velocity components
were calculated to elucidate the average flow structure size and strength, where riblets significantly suppressed
the turbulent structures. Quadrant analysis of the Reynolds shear stress was performed to assess the change in
ejection and sweep events and the results were found to be in correspondence with previous works.

1. Introduction studied at NASA Langley Research Center [7] considering different

shapes including rectangular, triangular (trapezoidal, sawtooth or v-
The challenge of reducing skin friction has been quite attractive to groove) and scalloped geometries. Detailed numerical [8] and experi-
scientists, since any decrease in drag would result in such substantial mental [9] researches covering fairly large parameter ranges have re-
energy savings in modern technology and even a small amount of fuel vealed the effect of various riblet shapes on drag reduction. The ex-
savings can be crucially important for all kinds of vehicles in trans- periments in an oil channel of Bechert et al. [9] reported up to 9.9%
portation. For instance, typically about half of the total drag of an drag reduction for blade type riblets; however, even though the blade
aircraft at subsonic speeds is due to the skin friction, while it is up to type gives the best efficiency, it is not easy to implement in practice as
90% for underwater vehicles. Therefore, turbulent boundary layer they are mechanically weak and easily break off the surface. Therefore
theory and control techniques over surfaces have been studied in- the triangular riblets have been widely investigated [9–11] yielding
tensively over the past decades. These studies were initiated by Prandtl 4–8% drag reduction and have been adapted to real engineering ap-
[1] who is the inventor of boundary layer theory and the first re- plications concerning aeronautics and marine industry.
searcher on boundary layer control. There is abundant experimental The underlying mechanism of drag reduction by riblets has also
and numerical research showing a remarkable drag reduction on sur- been an important research topic in the scientific community. The main
faces classified with active and passive methods [2,3]. Nature serves as idea was that the riblets impede the cross-stream translation of the
a useful area for the inspiration of various flow control mechanisms, streamwise vortices in the viscous sublayer [12], which results in the
which is called “biomimetics”. The air and marine animals intrinsically attenuation of cross-stream velocity fluctuations by keeping the vortices
use individual flow control techniques for their survival which could above the riblet tips [13]. According to this physical explanation, al-
possess potential methods to be implemented on real engineering ap- though the surface area is larger for the riblet case, which might in-
plications [4–6]. For example, the structure of dermal denticles, i.e., the crease drag, the lower cross-stream fluctuations limit the generation of
tooth-like structure of shark skin can be mimicked by longitudinal shear stress and momentum transfer near the wall. Therefore, the total
grooves on a flat plate, called riblets. This surface modification was first drag acting on the riblet surface is reduced compared with the flat

⁎
Corresponding author.
E-mail address: [email protected] (G.M. Ozkan).

https://doi.org/10.1016/j.expthermflusci.2020.110246
Received 7 April 2020; Received in revised form 10 July 2020; Accepted 1 September 2020
Available online 12 September 2020
0894-1777/ © 2020 Elsevier Inc. All rights reserved.
G.M. Ozkan, et al. Experimental Thermal and Fluid Science 120 (2021) 110246

surface. This claim has been supported by numerical research of Table 2

Goldstein et al. [11] showing that the generated vortices formed on the Summary of the turbulent boundary layer properties at streamwise location of
surface remain above the riblet tips by creating a low-velocity channel x = 1.8 m obtained from PIV measurements.
in the riblet valleys. Then, this low velocity reduced the velocity gra- Flat surface Riblet coating
dient compared with flat surface, and hence the wall shear stress and
the drag force. According to the mentioned studies, there are some Ue (m/s) 3.8 3.8
δ99 (mm) 26.4 23.2
transient secondary vortex formations inside the riblet valleys, although
θ (mm) 2.5 2.1
their contribution to drag is insignificant. Even though above physical uτ (m/s) 0.14 0.1385
explanations have been verified, drag reducing mechanism of the riblets ReL = Ue L/ʋ 7.207 × 106
is still under investigation by researchers using various modern tech- Reτ = uτ δ99/ʋ 3847 3344
Reθ = Ueθ/ʋ 9890 8300
niques to obtain well-resolved structures within the flat plate boundary
s+ = uτ s/ʋ – 13.2
layer, especially at high Reynolds numbers. Through these, hot-wire
anemometry measurements of Park and Wallace [14] close to the riblet
valleys proved that vertical flux of streamwise momentum is sig- as the riblet coating. A literature overview is presented in Table 1 to
nificantly reduced resulting in diminished wall shear stress near the emphasize the difference of current research from other experimental
bottom of the riblet valley with a 4% drag reduction calculated by in- investigations over riblet surfaces by means of riblet spacing and the
tegrating wall shear stress over the riblets. Laser Doppler velocimetry Reynolds number. Throughout the listed studies, current work focuses
(LDV) and hydrogen bubble flow visualization were used to observe the on comparably small riblet spacing and high Reynolds number flow (see
coherent structures in the near wall region [15]. Results revealed that Table 2).
the riblet surface enhances the thicknesses of viscous sublayer, buffer
region and integral constant C in the log-law while reducing the low-
speed streak spacing by 20% as an indication of lower drag. High-re- 2. Facility and technique
solution PIV experiments [16] including limited data for viscous and
buffer layers have revealed that the number distributions of the small- 2.1. Experimental set-up and material
scale spanwise vortices were significantly decreased in the riblet surface
and the turbulence intensity. Also, Reynolds stresses were reduced by Experiments were performed in a closed loop water tunnel (Fig. 2a)
the drag-reducing effect of the riblet grooves. at the Ship Hydromechanics group of the TU Delft. Dimensions of the
Recently, the actual denticles structure of sharkskin (from Isurus test section are (2 × 0.3 × 0.35) m3 in length, width, and height, re-
oxyrinchus) was compared with those of riblet surfaces in the turbulent spectively. The boundary layer under consideration develops over the
boundary layer [17]. They reported that the actual sharkskin denticles top wall of the tunnel at (nearly) zero pressure gradient conditions due
are not able to reduce the high shear rates close to the surface which to a sloping opposite wall. The flow speed can be increased up to 7 m/s,
resulted in larger drag forces by 44–50% compared with the flat sur- which enables to study realistic flow conditions for surface modification
face; however, 5% drag reduction was achieved by the use of riblets. with riblet, compliant coatings, etc. More details for the tunnel can be
Another study by planar, volumetric and long-range microscopic PTV found in Zverkhovskyi [24]. A zigzag strip was placed at the entrance of
[18] reported the reduction in turbulent fluctuations and attenuation of the tunnel to ensure the boundary layer flow is turbulent [25].
ejection and sweep motions close to the riblet surface as compared with The coating used in this study has a trapezoidal riblet shape where
the smooth wall. the spacing between the riblet tips, s, is 92 µm. The coating is manu-
In this research, a trapezoidal type riblet coating was investigated factured from a commercial fouling release product (Intersleek®
utilizing drag force and stereo-PIV measurements and analyzed for a 1100SR, International Paint / AkzoNobel) whose composition has been
further understanding of the relationship between drag reduction and modified by increasing the overall volume solids from 72% to 96% to
flow dynamics. Force measurements were compared with the study of enable preparation of a surface with defect free, high fidelity, trape-
Benschop et al. [19] which includes Taylor- Couette measurements over zoidal riblet shapes (see Fig. 1). Further detail of the coating, including
the same riblets. The novelty of this research is that the coating has a SEM (scanning electron microscope) pictures, can be found in Benschop
realistic riblet dimension (s = 92 µm) required for actual flow condi- et al. [19], their Fig. 2b.
tions found in most industrial applications in which the Reτ is on the The virtual origin of the coating was calculated according to Bechert
order of 103–105 [20]. Furthermore, this research might be one of the and Bartenwerfer [26]; i.e., the origin of the velocity profile lies below
rare PIV implementations performed in water at this high Reynolds the riblet tips where the related distance is described as the protrusion
number dealing with turbulent boundary layer over a flat plate, as well height which cannot exceed 22% of the lateral riblet spacing. The
coating was applied to a flat plate of length L = 2 m and width

Table 1
Literature overview of relevant researches on drag reducing riblet surfaces.
Experiments on continuous riblet Reynolds number Experimental Facility Type of wall-bounded Riblet spacing (mm) & s+ Experimental Technique
surfaces turbulence interval

Vukoslavcevic et al. [21] Reθ = 1000 Wind tunnel Flat plate s = 10 & s+= 35 Hot wire
Park & Wallace [14] Reθ = 1200, 2100, 2500 Wind tunnel Flat plate s = 10 & s+= 18, 28, 46 Hot-wire
Bechert et al. [9] 10,000 ≤ Rech ≤ 33000 Oil channel Channel flow Effect of riblet spacing was Shear stress balance
investigated
Lee & Lee [13] Reθ = 2340, Reθ = 4950 Wind tunnel Flat plate s = 3 &s+=25.2, 40.6 PIV and PTV
Lee & Choi, [16] Reθ = 3790 Wind tunnel Flat plate s = 0.3 & s+=10.4 PIV
Li et al. [22] Reθ = 1200, Reθ = 2080 Wind tunnel Flat plate s = 0.75 & s+=24,45 PIV and µPIV
Hou et al. [18] Reθ = 2384 Water channel Channel flow s = 0.75 & s+=11 2D-3D PIV, PTV, µPTV and
tomo-PIV
Rowin et al. [23] Rech = 4360 Water channel Channel flow s = 0.75 s+ = 8.6, 17.3, and Planar PIV
34.6
Benschop et al. [19] 104 ≤ Res ≤ 105 Taylor- Couette set-up Taylor- Couette flow s = 0.092 & 1 ≤ s+≤ 20 Torque-meter
Current work Reθ = 9890 Water tunnel Flat plate s = 0.092 & s+=13.4 Force balance, Stereo-PIV

2
G.M. Ozkan, et al. Experimental Thermal and Fluid Science 120 (2021) 110246

Fig. 1. Schematic presentation of the riblet shape with geometric details. For
more details, see Benschop et al. [19].

w = 0.298 m, which were mounted in the test section replacing the top
wall. The results for the riblet coated plate were compared with those
for a hydraulically smooth flat plate in order to examine the drag re-
duction mechanism. The drag reducing range, determined with respect
Fig. 2b. Schematic presentation of the cross-view of experimental set-up.
to dimensionless riblet spacing, s+ = su / (where = 0.949 × 10-6
m2/s) was 2 < s+ < 14. The friction velocitiy, uτ, were evaluated
from force measurements using u = w / where w is the averaged attached, satisfying the attached flow on the test plate.
wall shear stress (N/m2) on the plate and is the water density in kg/ The non-dimensional drag coefficient is defined as,
m3.
CD = 2Fd/ Ue2 A (2.1)

2.2. Force measurement where A is the surface area of the plate, Ue is the free-stream velocity of
the flow, and ρ is the density (kg/m3) calculated at the average tem-
The force measurements were performed using an external moving perature during the measurements. To validate the results with avail-
frame that is specially designed for having one degree of freedom in the able theory, the drag coefficient was also calculated using the re-
direction of the free stream flow. Any plate with specific dimensions lationship:
can be attached on this moving frame and carried by two leaf springs CDtheo = 0.031/ ReL1/7, where ReL = Ue L/ (2.2)
enabling direct measurement of the drag force, FD. The test plate was
2
precisely positioned using alignment screws on fixed frame, resulting in (υ is the kinematic viscosity in m /s) derived by one-seventh power law
an angle of attack to the flow of O(0.01°). A picture is presented in [27]. The results are presented in Fig. 3 for a wide range of Reynolds
Fig. 2a showing the location of force balance, including a pulley me- number. It is clear that the measurements for low ReL demonstrate
chanism for calibration. The force sensor (ZEMIC L6D Class-3, with a strong deviations from the theoretical curve according to non-turbulent
precision of ± 0.023% of the full-scale) used in the experiments was transition around Re 106; therefore ReL > 2 × 106 looks to be reliable
calibrated using this pulley mechanism applying various weights ac- for processing and will be used for further analysis. Water level of the
cording to the measurement range. The sampling frequency of the force tunnel was kept constant as hw = 0.3 m for all velocities to eliminate
sensor was 1 kHz, and the signals were analyzed by calculating the possible bias caused by the water height. Since the force measurements
sample mean for the data recorded. A cross-view of the set-up is pre- on the test plate started with U∞ = 0.4 m/s, measurements were cor-
sented in Fig. 2b to show the details of the drag measurement me- rected by fitting a second-order polynomial to the force data with re-
chanism. The leaf springs are attached to a fixed frame on the top, and spect to velocity, and then an offset (about 1% of the measured force)
the test plates on the bottom. There exist 1 mm gaps at both sides of the was applied to the data, considering zero force at zero velocity. For
plate, which may cause water leaks and hence the error in the mea- ReL > 2 × 106, the uncertainty in the measurements was estimated to
surement. In order to prevent water leaks, thin, flexible strips were be less than 0.1% at 95% confidence level.

Fig. 2a. General view of the tunnel and force measurement system.

3
G.M. Ozkan, et al. Experimental Thermal and Fluid Science 120 (2021) 110246

Fig. 3. Drag coefficients calculated by experiments and theory (one-seventh power law).

2.3. Velocity measurement pulsed Nd:YLF laser having 527 nm wavelength and 10 mJ/pulse
output with a repetition rate of 10 Hz was used for the illumination. The
All three components of the velocity were measured in a cross- sheet was formed with two lenses and directed such as to obtain for-
stream plane using stereo-PIV as shown in Fig. 4. The experiments were ward scattering in which the tracer particles scatter more light ac-
performed considering the case with highest drag reduction (s+ cording to Mie theory [29]. The thickness of the laser sheet and time
=13.2). Two high-speed CMOS cameras (pco.dimax, 4 M, between images were adjusted to 1 mm and 25 µs, respectively, en-
2016 × 2016 pix resolution, 12-bit dynamic range, 1279 fps) at 45° suring that there is only limited out of plane motion. The thin light
viewing angles were used to capture the particle images (Fig. 4). The sheet perpendicular to the mean flow resulted in a low signal-to-noise
cameras were mounted with 105 mm Nikon lenses at f /5.6 satisfying ratio [31] . The calibration was done using a plexiglass plate with “+”
the Scheimpflug condition [28]. Furthermore, water filled prisms were signs printed in increments of 5 mm. A traversing mechanism was used
used to minimize the optical deformation of the images due to re- for displacing the plate in steps of 0.5 mm to calibrate the stereo PIV
fractive changes at air-water interface. The flow was seeded with system. Furthermore, the self-calibration procedure was performed
neutrally buoyant hollow glass spheres with 10 µm diameter. A double- [30] in order to minimize the errors caused by misregistration [31].
1570 images were captured and used for self-calibration for all data
sets. Davis v.8.0 was used for the calibration and cross-correlation
analysis. A final interrogation window size of 24 × 24 pixels was
employed with a 50% overlap. The final field of view has dimensions of
(y, z) = (41.8 × 54.2) mm2 corresponding to an average magnification
of M = 0.41, yielding a resolution of 37.6 px/mm in the object plane.
The spatial resolution of the measurements was, therefore,
Δy = Δz = 0.3188 mm. The number of spurious vectors was less than
5%. The correlation noise of the y- and z- components of the vector field
were σcorr ≈ 0.1 px, in agreement with Westerweel [32]. Hence, con-
sidering the maximum value of the free-stream flow, the uncertainty in
the velocity was estimated to be less than 2%. The velocity gradients,
hence the vorticity were estimated by means of flow circulation by first
applying Gaussian kernel to attenuate the noise level [33]. According to
Foucaut and Stanislas [34], the uncertainty in the vorticity estimation
was calculated to be less than 5%.

2.4. Two-point correlations

In order to quantify the statistical behavior of turbulent structures

within the available flow domain, two-point spatial correlations of the
fluctuating velocity components are widely utilized [35]. For example,
two-point spatial correlation of the streamwise velocity fluctuations, u,
are used to gain information about the hairpin packets [36] in the log
region. For the available flow field and coordinate system, it is defined
as:
Fig. 4. Schematic presentation of stereo-PIV configuration.

4
G.M. Ozkan, et al. Experimental Thermal and Fluid Science 120 (2021) 110246

Fig. 6. Dimensionless mean velocity profile for flat and riblet surfaces with
Fig. 5. Drag reducing effect of the riblet coating, together with Bechert et al. Spalding’s fit [37] and hotwire data of Carlier & Stanislas [38].
[9] and Benschop et al. [19].

viscous sublayer [10] resulting in a decreased friction velocity and skin

< u (y , z ) u (y + y , z + z ) >
Ruu (y , z , y , z ) = friction. This upward shift in the velocity profile has been observed
2
u (2.3) frequently for drag reducing riblets [13,17,18,39,40] and described as a
where σu is the standard deviation of streamwise velocity fluctuations possible consequence of modified near- wall turbulent structures due to
and, Δy and Δz are the wall-normal and spanwise separation, respec- the energy balance between reduced turbulent kinetic energy produc-
tively between the data points in measurement plane. The brackets tion and viscous dissipation [41].
represent the ensemble average of the data over the recorded image The effect of the riblets is to reduce the velocity fluctuations espe-
frames. The PIV velocity field has inhomogeneous velocity components cially in the streamwise and wall-normal directions as reported pre-
since the current study focuses on the cross-stream plane where the wall viously [16–18]. This indicates a decreased momentum transfer and
and riblets introduce non-uniformity. Thereby, equation (2.3) was di- turbulent mixing in the turbulent boundary layer. The present results
rectly computed at the selected reference points, (y, z). shown in Fig. 7 agree well with those earlier findings; here the free-
stream and friction velocities used for normalization were considered
individually for flat and riblet surfaces. The turbulent kinetic energy
3. Results
and Reynolds shear stress profiles for the riblet surface are slightly
below the respective profiles for the flat surface. The profiles match
The drag reducing effect of coating is expressed as a drag change in
with each other outside of the boundary layer (y > δ). Discrepancies in
percentage calculated by DC = 100 × (CDriblet CDflat )/CDflat , which is
turbulent kinetic energy and Reynolds shear stress tend to get smaller
presented and compared with Bechert et al. [9] and Benschop et al.
when friction velocity is used for normalization; however, the riblet
[19] in Fig. 5. Here, the low Reynolds number data of current study
profile is still below the profile for the flat plate. Physically, decreased
(Re < 2 × 106) are not included due to inaccurate drag measurements
statistics of fluctuating velocity is key evidence for reduced turbulent
of non-turbulent transition (Re 106), which was shown in Fig. 3. It can
mixing and less efficient transport of high-momentum outer fluid to-
be seen that the drag reduction behavior is identical to that observed by
ward the surface. Besides, the lower Reynolds shear stress implies that
Bechert et al. [9], and a maximum reduction of 5.9% is obtained. The
sweep and ejection motions, which significantly contribute to the skin
results are also in good agreement with the Taylor-Couette measure-
friction, are attenuated.
ments of the same riblet coating presented in Benschop et al. [19], al-
One of the main characteristics of turbulence are vortical structures,
beit their flow geometry is different. The agreement between the results
hence the effect of riblet surface on streamwise vorticity fluctuations,
of current study and Benschop et al. [19] shows that less time con-
< x+ > = < ( x' ) 2 > / u is presented in Fig. 8. Related research has
suming and lower-cost Taylor-Couette flow measurements may be used
revealed that up to a certain wall-normal distance, i.e. y+ = 15 [18] or
as an alternative to turbulent boundary layer measurements over a flat
y+ = 130 [42], the vorticity fluctuations are reduced above riblets
plate.
compared to the flat surface. These studies reported that < x+> tends to
Once the drag reducing effect of riblet coatings has been verified,
have the same magnitude over flat and riblet surfaces beyond the
PIV measurements were performed at the speed corresponding to the
mentioned wall-normal distances. However, these investigations have
highest drag reduction, i.e. s+ =13.2 . In Fig. 6, semi log profiles of
considered channel flow, and therefore they have limited data for larger
mean velocity as a function of wall-normal distance normalized by local
wall-normal distances. The present vorticity profiles (Fig. 8) are in good
friction velocities are presented and compared with the generalized
agreement with these earlier works covering the near wall region
formula of Spalding [37] and hot-wire data of Carlier & Stanislas [38].
[10,18,42], while presenting additional information for the whole
As evident from the figure, the mean velocity profile of the flat surface
boundary layer. The vorticity fluctuations seem to be slightly lower
fits well with the literature [37] showing a visible logarithmic region in
over riblets up to y+ = 500 and at this point the values are identical for
the overlap layer evaluated using von Kármán and integral constants as
both surfaces. However, an obvious increase in the deviation occurs for
k = 0.4 and B = 5, respectively. The mean velocity profile for the flat
y+ ≥ 500. Compared with the flat surface, a reduction of about 6%
surface also agrees well with results from Carlier & Stanislas [38],
occurs in vorticity fluctuations within the log layer of the riblet surface.
however with slight deviations. The discrepancy of data points close to
The two-point correlations give information about the typical size
the wall is due to the limited spatial resolution of the PIV measure-
and strength of turbulent flow features. Fig. 9 presents the auto corre-
ments. As compared with the flat surface, an upward shift is obtained
lation of the streamwise velocity fluctuations which can be used to
for the coating, which is associated with an apparent thickening of the

5
G.M. Ozkan, et al. Experimental Thermal and Fluid Science 120 (2021) 110246

Fig. 7. Comparison of time-averaged Turbulent Kinetic Energy, < TKE > and Reynolds Shear Stress, < u′v′ > normalized by free-stream velocity, Ue (top) and
local friction velocity, uτ (bottom).

decreasing the size of turbulent eddies within the log region of a tur-
bulent boundary layer.
The quadrant analysis applied to Reynolds shear stress is a useful
tool to evaluate information on the sign of velocity fluctuations, hence
the special events occurring in a turbulent boundary layer. The quad-
rants are related to velocity information according to: Q1(u > 0,
v > 0), Q2(u < 0,v > 0), Q3(u < 0,v < 0), Q4(u > 0,v < 0)
which are also known as sweep (Q4)- ejection (Q2) motions and inward
(Q1)-outward (Q3) interactions [44]. The products of the fluctuations
classify the events. Fig. 10 presents the conditional averaged and nor-
malized Reynolds shear stress distributions in wall-normal direction for
all quadrants, agreeing well with relevant literature [45]. The gradient-
type motions; Q2 and Q4 events have the most significant contribution
to Reynolds shear stress [46], i.e. the turbulence production. However,
in general, Q2 has a larger contribution compared to Q4 [45], since the
obtained data is in the log layer. Moreover, all quadrants have slightly
decreased magnitudes over riblet surface in correspondence with re-
lated studies [10,18,42]. Inherently, the deviation in magnitudes for Q2
and Q4 events is larger compared with Q1 and Q4, and this deviation is
valid throughout the outer region. Therefore it should be noted that the
Fig. 8. Streamwise root mean square vorticity fluctuation normalized using
attenuation effect of riblet surface on turbulence production lasts even
local friction velocities of flat and riblet surfaces.
for the wake region.

characterize the high and low speed regions. The flow field was
cropped, and only positive correlation contours are presented to make it 4. Conclusions
clear for comparison. Firstly, the correlations clearly show the growth
with increasing the wall-normal reference position, which is consistent In this study, the difference between flow structures over flat and
with Townsend’s attached eddy hypothesis in which the size of an eddy riblet surfaces in the cross-stream plane is analyzed and interpreted in
along the spanwise and wall-normal directions is proportional to its order to understand the underlying mechanism of drag reduction. In
distance from the wall [43]. Next, the structure size and shape over this sense, the Reynolds number considered was much higher than
riblet and flat surfaces look almost identical at y+= 232. However, the before, together with a comparably small-sized riblet spacing, both
structure size over riblet surface tends to decrease in width. This ten- corresponding to the actual flow conditions encountered in most in-
dency yields an absolute difference in structure size where it becomes dustrial applications. The drag reducing effect of a trapezoidal riblet
narrower over the riblet surface compared to the smooth wall at wall coating is first verified by force measurements agreeing well with the
normal distances of y+= 511 and y+ = 976. The decrease in vorticity literature [9,19]. According to the agreement in force measurements of
fluctuations depicted in Fig. 8 for y+ ≥ 500 could also be associated Benschop et al. [19], it was revealed that the Taylor-Couette flow fa-
and in accordance with the two-point correlations obtained at y+= cility may be used as an alternative set-up over more time- consuming
511. Therefore it can be concluded that the riblets are effective on and costly measurements over a flat surface.

6
G.M. Ozkan, et al. Experimental Thermal and Fluid Science 120 (2021) 110246

Fig. 9. The correlation of streamwise velocity fluctuations, Ruu of flat and riblet surfaces normalized with local friction velocities at wall-normal positions of (a) y+=
232, (b) y+= 511 and (c) y+= 976. The contour levels range from 0.3 to 1 with a spacing of 0.1.

Next, the stereo-PIV measurements were performed at the max- turbulent structure over the riblet surface is smaller for y+ ≥ 511
imum drag reducing case. The data was post-processed with a focus on following the results of vorticity fluctuations. It was concluded that this
time-averaged turbulent statistics. For the mean velocity profile, an is an indication of possible suppression on quasi-streamwise vortices
upward shift was obtained for the riblet surface due to the decrease in and decrease in transport of vorticity, hence the turbulence production.
friction velocity. Time-averaged Reynolds shear stress and turbulent
kinetic energy over riblet surface were attenuated compared with the 5. Compliance with Ethical Standards
flat surface as an indication of reduced turbulent transport towards the
wall. This is further elaborated through time-averaged vorticity fluc- Funding
tuations, two-point spatial correlations, and quadrant analysis.
Over the riblet surface, it was found that the evaluated vorticity The manufacturing of riblet coating used in this research has re-
fluctuations were slightly reduced close to the wall, however, for y+ ≥ ceived funding from the European Union Seventh Framework
500, the reduction found was obtained greater, i.e., a maximum of 6%. Programme in the SEAFRONT project [grant agreement number
The special events, especially the sweep and ejection within the 614034]. The first author is funded by The Scientific and Technological
boundary layer, is known to be the primary source of the transport of Research Council of Turkey (TUBITAK) under application number of
Reynolds shear stresses. It was evaluated that over riblets the magni- 1059B191600985.
tudes of sweep and ejection events were slightly dropped throughout
the boundary layer. Furthermore, two-point spatial correlations aided CRediT authorship contribution statement
to further understanding providing signatures of hairpin vortices. Two-
point correlations of streamwise velocity fluctuations showed that the G.M. Ozkan: Conceptualization, Formal analysis, Investigation,

7
G.M. Ozkan, et al. Experimental Thermal and Fluid Science 120 (2021) 110246

[12] B. Dean, B. Bhushan, Shark-skin surfaces for fluid-drag reduction in turbulent flow:
a review, Philos. Trans. Roy. Soc. 368 (2010) 4775–4806.
[13] S.J. Lee, S.H. Lee, Flow field analysis of a turbulent boundary layer over a riblet
surface, Exp. Fluids 30 (2) (2001) 153–166.
[14] S.R. Park, J.M. Wallace, Flow alteration and drag reduction by riblets in a turbulent
boundary-layer, AIAA J. 32 (1) (1994) 31–38.
[15] J.J. Wang, et al., Experimental study on the turbulent boundary layer flow over
riblets surface, Fluid Dyn. Res. 27 (4) (2000) 217–229.
[16] S.J. Lee, Y.S. Choi, Decrement of spanwise vortices by a drag-reducing riblet sur-
face, J. Turbul. 9 (23) (2008) 1–15.
[17] A. Boomsma, F. Sotiropoulos, Direct numerical simulation of sharkskin denticles in
turbulent channel flow, Phys. Fluids 28 (3) (2016).
[18] J.F. Hou, et al., Three-dimensional measurement of turbulent flow over a riblet
surface, Exp. Therm Fluid Sci. 85 (2017) 229–239.
[19] H.O.G. Benschop, A.J. Guerin, A. Brinkmann, M.L. Dale, A.A. Finnie, W.-
P. Breugem, A.S. Clare, Stübing, D.C. Price, K.J. Reynolds, Drag-reducing riblets
with fouling-release properties: development and testing, Biofouling 34 (5) (2018)
532–544.
[20] A.J. Smits, I. Marusic, Wall-bounded turbulence, Phys. Today 66 (9) (2013) 25.
[21] P. Vukoslavcevic, J.M. Wallace, J.-L. Balint, Viscous drag reduction using stream-
wise- aligned riblets, AIAA J. Tech. Notes 30 (1992) 4.
[22] W. Li, W. Jessen, D. Roggenkampa, M. Klaas, W. Silex, M. Schiek, W. Schröder,
Turbulent drag reduction by spanwise traveling ribbed surface waves, Eur. J. Mech.
B/Fluids 53 (2015) 101–112.
[23] W.A. Rowin, J. Hou, S. Ghaemi, Turbulent channel flow over riblets with super-
hydrophobic coating, Exp. Therm Fluid Sci. 94 (2018) 192–204.
Fig. 10. Conditionally averaged Reynolds shear stresses normalized with local [24] O. Zverkhovskyi, Ship Drag Reduction by Air Cavities, Ph.D. Thesis, TU Delft, 2014.
friction velocities for flat and riblet surfaces. [25] G. Elsinga, J. Westerweel, Tomographic-PIV measurement of the flow around a
zigzag boundary layer trip, Exp. Fluids 52 (4) (2012) 865–876.
[26] D.W. Bechert, M. Bartenwerfer, The viscous flow on surfaces with longitudinal ribs,
Writing - original draft, Visualization, Data curation. G.E. Elsinga: J. Fluid Mech. 206 (1989) 105–129.
Writing - review & editing, Supervision, Methodology. W.-P. Breugem: [27] M.F. White, Fluid Mechanics, McGraw Hill, 2003.
[28] A. Prasad, Stereoscopic particle image velocimetry, Exp. Fluids 29 (2000) 103–116.
Writing - review & editing, Project administration, Funding acquisition. [29] M. Born, E. Wolf, Principles of Optics, Pergamon Press, Oxford, 1975.
D. Stübing: Resources, Methodology. K.J. Reynolds: Resources, [30] B. Wieneke, Stereo-PIV using self-calibration on particle images, Exp. Fluids 39
Methodology. J. Westerweel: Supervision, Conceptualization, Writing (2005) 267–280.
[31] V. Doorne, J. Westerweel, Measurement of laminar, transitional and turbulent pipe
- review & editing, Methodology. flow using Stereoscopic-PIV, Exp. Fluids 42 (2007) 259–279.
[32] J. Westerweel, Theoretical analysis of the measurement precision in particle image
Declaration of Competing Interest velocimetry, Exp. Fluids 29 (Supplemantary) (2000) S3–S12.
[33] C.C. Landreth, R.J. Adrian, Impingement of a Reynolds number turbulent circular
jet onto flat plate at normal incidence, Exp. Fluids 9 (1990) 74–84.
The authors declare that they have no known competing financial [34] J.M. Foucaut, M. Stanislas, Some considerations on the accuracy and frequency
response of some derivative filters applied to particle image velocimetry vector
interests or personal relationships that could have appeared to influ- fields, Meas. Sci. Technol. 13 (2002) 1058–1071.
ence the work reported in this paper. [35] J.A. Sillero, J. Jiménez, R.D. Moser, Two-point statistics for turbulent boundary
layers and channels at Reynolds numbers up to δ+ ≈ 2000, Phys. Fluids 26 (2014)
105109.
References
[36] R.J. Adrian, Hairpin vortex organization in wall turbulence, Phys. Fluids 19 (2007)
041301.
[1] L. Prandtl, Über Flüssigkeitsbewegung bei sehr kleiner Reibung, in: Proc. Third Int. [37] D.B. Spalding, A single formula for the law of the wall, J. Appl. Mech. 28 (1961)
Math. Cong., Heidelberg, Germany, 1904, pp. 484–491. 455–457.
[2] M. Gad-el-Hak, Flow Control: Active, Passive and Reactive Flow Management, [38] J. Carlier, M. Stanislas, Experimental study of eddy structures in a turbulent
Cambridge University Press, 2000. boundary layer using particle image velocimetry, J. Fluid Mech. 535 (2005)
[3] M. Perlin, D.R. Dowling, S. Ceccio, Freeman scholar review: passive and active skin- 143–188.
friction drag reduction in turbulent boundary layers, J. Fluids Eng.-Trans. Asme 138 [39] O.A. El-Samni, H.H. Chun, H.S. Yoon, Drag reduction of turbulent flow over thin
(9) (2016). rectangular riblets, Int. J. Eng. Sci. 45 (2007) 436–454.
[4] D.M. Bushnell, K.J. Moore, Drag reduction in nature, Annu. Rev. Fluid Mech. 23 [40] A. Rastegari, R. Akhavan, The common mechanism of turbulent skin-friction drag
(1991) 65–79. reduction with superhydrophobic longitudinal microgrooves and riblets, J. Fluid
[5] D.W. Bechert, M. Bruse, W. Hage, R. Meyer, Fluid mechanics of biological surfaces Mech. 838 (2018) 68–104.
and their technological application, Naturwissenschaften 87 (2000) 157–171. [41] E. Coustols, Riblets: Main known and unknown features, in: K.-S. Choi, K.K. Prasad,
[6] B. Bhushan, Biomimetics: lessons from nature - an overview, Philos. Trans. Roy. T.V. Truong (eds.), Emerging Techniques in Drag Reduction, Mechanical
Soc. A-Math. Phys. Eng. Sci. 367 (1893) (2009) 1445–1486. Engineering Publications Limited, Bury St. Edmunds, London, UK, 1995, pp. 3–43.
[7] M.J. Walsh, Riblets as viscous drag reduction technique, AIAA J. 21 (1983) [42] M. Sasamori, H. Mamori, K. Iwamoto, A. Murata, Experimental study on drag re-
485–486. duction effect due to sinusoidal riblets in turbulent channel flow, Exp. Fluids 55
[8] P. Luchini, F. Manzo, A. Pozzi, Resistance of a grooved surface to parallel flow and (2014) 1828.
cross-flow, J. Fluid Mech. 228 (1991) 87–109. [43] A.A. Townsend, The Structure of Turbulent Shear Flow, second ed., Cambridge
[9] D.W. Bechert, M. Bruse, W. Hage, J.G.T. Van der Hoeven, G. Hoppe, Experiments on University Press, Cambridge, 1976.
drag-reducing surfaces and their optimization with an adjustable geometry, J. Fluid [44] J.M. Wallace, R.S. Brodkey, H. Eckelmann, The wall region in turbulent shear flow,
Mech. 338 (1997) 59–87. J. Fluid Mech. 54 (1972) 39–48.
[10] H. Choi, P. Moin, J. Kim, Direct numerical simulation of turbulent flow over riblets, [45] J.M. Wallace, Quadrant analysis in turbulence research: History and evolution,
J. Fluid Mech. 255 (1993) 503–539. Annu. Rev. Fluid Mech. 48 (2016) 131–158.
[11] D. Goldstein, R. Handler, L. Sirovich, Direct numerical-simulation of turbulent-flow [46] W.W. Willmarth, S.S. Lu, Structure of the Reynolds stress near the wall, J. Fluid
over a modeled Riblet covered surface, J. Fluid Mech. 302 (1995) 333–376. Mech. 55 (1972) 65–92.