Journal of Colloid and Interface Science 393 (2013) 384–396

Contents lists available at SciVerse ScienceDirect

Journal of Colloid and Interface Science

www.elsevier.com/locate/jcis

Shark skin inspired low-drag microstructured surfaces in closed channel flow

Gregory D. Bixler, Bharat Bhushan ⇑
Nanoprobe Laboratory for Bio & Nanotechnology and Biomimetics (NLB2), The Ohio State University, 201 W. 19th Avenue, Columbus, OH 43210-1142, USA

a r t i c l e i n f o a b s t r a c t

Article history: Living nature is the inspiration for many innovations and continues to serve as an invaluable resource to
Received 4 October 2012 solve technical challenges. Skin from fast swimming sharks intrigue researchers since its low-drag riblet
Accepted 25 October 2012 structure is applicable to many engineering applications. In this study, riblet-lined closed channel (rect-
Available online 28 November 2012
angular duct) internal flow was examined since its effect is less understood than with open channel
external flow. With one experimental setup and two fluids, this study examines various dimensional
Keywords: aspects of microstructured riblets. Experimental parameters include riblet geometry, fluid velocity (lam-
Riblets
inar and turbulent flow), fluid viscosity, riblet combinations, channel size, wettability, and scalability. For
Shark skin
Biomimetics
direct comparison, the sample flow channel was fabricated to accommodate multiple samples with water
Low-drag and air in various flow conditions, where drag is characterized by measuring pressure drop. Results are
Antifouling discussed and conceptual models are shown suggesting the interaction between vortices and the riblet
Closed channel surfaces.
Ó 2012 Elsevier Inc. All rights reserved.

1. Introduction oleophobic (oil-fearing) surface [22]. Fast swimming shark skin is

low drag and believed to be antifouling due to its riblet microtex-
Inspired by designs found throughout living nature, researchers ture, flexion of scales, and a mucous layer. The skin contains scales
are reverse engineering the world’s flora and fauna to solve techni- called dermal denticles (similar to teeth), which are covered by
cal challenges. Much attention is given to structures and materials specially sized and spaced riblets oriented parallel to the swim-
since living nature efficiently uses resources and incorporates inge- ming direction. Riblets are believed to lift and pin the naturally
nious designs to solve problems. Therefore by using lessons from occurring fluid vortices in the viscous sublayer. Lifting reduces
living nature, bioinspired designs are serving as the basis for many the total shear stress since vortices contact just the small riblet
new innovations. Examples include boats inspired by low drag dol- tips, as opposed to the total surface area. Pinning reduces the cross
phins and sharks, ‘‘self-cleaning’’ windows inspired by the super- stream motion of fluid and ejection of vortices from the viscous
hydrophobic lotus leaf, and wall-climbing robots inspired by sublayer, which reduces energy losses. Lower drag increases fluid
adhesive gecko feet [1–7]. flow at the skin, reduces microorganism settlement time, promotes
Two engineering challenges solved in living nature but plaguing washing, and allows for faster predatory swimming [10,11,23–33].
a variety of industries are fluid drag (skin friction) and biological Turbulent vortices are also present in other fluids, therefore the
fouling (biofouling). Drag is the resistant force fluid imposes on shark inspired riblets are expected to provide a similar drag-reduc-
an object in either open or closed channel flow. Biofouling is the ing benefit with air.
accumulation of unwanted biological matter, with biofilms created Experimental results indicate that shark skin-inspired surfaces
by microorganisms and macroscale biofouling created by macroor- effectively reduce drag in open and closed channel flow, although
ganisms. In addition to biofouling, inorganic fouling can occur as a limited data is available with closed channel [11,17,24–26,28,34–
result of deposits from corrosion, crystallization, suspended parti- 42]. Flow in closed channel differs from open channel due to the
cles, oil, and ice. A low drag surface often equates less fouling and surrounding walls interacting with the fluid. In open channel flow,
better energy conservation, which is important for many industries the boundary layer continues growing over the length of the sur-
with example applications such as hospital catheters, ship hulls, face (until plateauing), whereas in closed channel the boundary
and pipelines [8–18]. layer is limited by the channel size. Furthermore, in closed channel
Certain flora and fauna benefit from low drag and effectively re- flow, turbulent vortices are trapped by the enclosure, unlike in
sist biofouling, unlike their contemporaries [19–21]. For example, open channel flow [8]. These differences warrant investigation into
fish scales remain clean in oil-polluted water due to their super- drag-reducing properties of riblet surfaces for open and closed
channel flow.
⇑ Corresponding author. Both drag (for open channel) and the pressure drop (for closed
E-mail address: [email protected] (B. Bhushan). channel) measurements characterize the riblet drag reduction

0021-9797/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcis.2012.10.061
 G.D. Bixler, B. Bhushan / Journal of Colloid and Interface Science 393 (2013) 384–396 385

efficiency. Previous experiments have utilized a variety of riblet shape, with a base width of 100–300 lm, summit radius of about
geometries, materials, fluids, and flow conditions (laminar and tur- 15 lm, height of 200–500 lm, and spacing of 100–300 lm [41].
bulent flow). Geometries include blade, sawtooth, bullnose, and These values served as the basis for new laser-etched blade riblet
scalloped designs with continuous and segmented (aligned and samples with an expanded range of dimensions. Early in the exper-
staggered) configurations in air, water, and oil. Wind tunnel stud- imental design process, the fluids (water and air) and rectangular
ies have included continuous metal [38] and aligned segmented channel size were chosen, since this drives the riblet sizes and con-
polymer [24] blade as well as continuous metal and polymer saw- sequently the fabrication technique.
tooth, bullnose, and scalloped riblets [34,35]. Closed water channel Sample riblet dimensions with various thicknesses (t), valley
studies have included continuous polymer [42] and aligned seg- widths (vw), spacing (s), gaps (g), lengths (L), and heights (h) were
mented polymer [41] blade as well as continuous polymer saw- selected. Published blade riblet studies suggest that h/s equaling
tooth riblets [39]. Conversely, open water channel studies have 0.5, small t/s, and with a dimensionless riblet spacing (defined later
included continuous polymer sawtooth riblets [39]. Open oil chan- in Eq. (4)) near 15 optimizes drag reduction [26]; also see review
nel studies have included continuous metal [26] and staggered paper [31]. Fabrication limitations and material properties deter-
segmented metal blade riblets [36]. Additionally, riblets have been mined the minimum riblet dimensions for the segmented and con-
studied during airplane flight tests [37,40]. Closed water channel tinuous samples. Blade dimensions were incrementally varied to
experiments with polymer riblets show pressure drop reduction change the h/s and t/s values to fully understand their relationship
of 23% [41]; whereas open channel oil experiments with metal rib- in closed channel.
lets show drag reduction of nearly 10% [26]. Shark skin replicas A wide range of blade h/s and t/s values (0.16 < h/s < 0.62 and
have also been evaluated, and have reduced pressure drop up to 0.18 < t/s < 0.38) were studied; with g and L held constant, and h,
30% in closed channel water flow [41,43]. Experimental and vw, and t varied individually; see Table 1 for dimensions. The top
numerical data help explain the correlation between turbulence image of Fig. 1 shows the segmented blade configuration. Samples
and drag reduction of real and simulated shark skin [32,44]. are referred to as their ‘‘description’’ and also their h/s and t/s val-
Since limited closed channel drag data indicates that fluid dy- ues when necessary, so for example the segmented blade riblet
namic properties differ from open channel flow, relevant parame- sample with h/s equaling 0.31 and t/s equaling 0.31 is called ‘‘Base-
ters and scaling effects for closed channels warrant investigation line (0.31, 0.31)’’ or more simply ‘‘baseline’’.
and are the subject of this study. Unknown parameters include rib- Continuous sawtooth riblets (with h = s = t) were obtained from
let geometry, fluid velocity, fluid viscosity, riblet combinations, 3 M Corp. (Minneapolis, Minnesota) and inserted into the sample
channel size, wettability, and scalability. Relative drag was deter- flow channel; see Table 1 for dimensions. The bottom image of
mined by measuring pressure drop between the inlet and outlet Fig. 1 shows the sawtooth pattern configuration. The 3 M experi-
of a closed channel, with lower pressure drop indicating lower mental vinyl riblet sheets are reported to be either coextruded
drag. Blade riblets were fabricated using laser etching, which al- through a die capable of a patterned surface or more likely em-
lows sizes to be incrementally and systematically varied. Addition- bossed with a negative master mold using heat and/or pressure;
ally, experimental sawtooth riblets obtained from a vendor were both with an adhesive backing applied [45].
evaluated, as well as blade riblet samples with a hydrophobic Studies suggest that superhydrophobic surfaces exhibit lower
nanostructured coating. Each sample was evaluated in water and drag [4,16,41,50–53] and self-cleaning [4,16,54] properties, which
air with flows ranging from laminar to turbulent regimes. Further- are believed to promote antifouling; therefore an applicable coating
more, experiments were conducted with various orientations such was applied to select samples. For the nanostructured coating, silica
as riblets on top, bottom, and both top and bottom; as well as with particles were selected as they are known to provide high durability
different channel heights. Numerical and conceptual models sug- and transparency, if desired [55,56]. Replicas were dip-coated with
gest explanations relating the pressure drop (drag) reduction to a solution consisting of 50 nm (±15 nm) hydrophobized silica nano-
the interactions of vortices, as well as scaling effects. particles (by Evonik-Degussa Corporation, Parsippany, New Jersey)
combined with methylphenyl silicone resin dissolved in tetrahy-
drofuran and isopropyl alcohol. Uncoated samples of each were also
2. Experimental details fabricated in order to study the effects of superhydrophobicity and
low adhesion. One sample included completely coating the riblets
2.1. Fabrication of samples and a second included coating only the riblet valleys.

Feasible riblet fabrication techniques include using metal shims 2.2. Model for riblet geometry
[26], machined acrylic [42], machined aluminum [35], extruded/
coextruded polymer [45], embossed polymer [45], soft lithography Optimizing riblet geometries is twofold; one is to lift and pin
[41], photolithography [46], wet and dry etching [46], grinding vortices, and the other is to minimize drag (skin friction) due to
[47], rolling [48], and laser etching [49]. Each method has advanta- the riblets themselves. Since riblets protrude into the flow channel,
ges and disadvantages related to precision, accuracy, and costs. The the increased surface area equates to increased drag. In order to
leading choices for riblet fabrication included photolithography optimize drag reduction, the riblets should lift and pin the vortices
and micromachining, but these were not economically feasible as well as allow efficient flow between the riblets. Such efficient
for the desired sample quantity, so milling was attempted. Com- flow is found by considering each valley between the riblets as
puter numerically controlled milling proved ineffective due to an open channel and minimizing its wetted perimeter, since in-
unacceptably large minimum achievable tolerances (±60 lm). creased wetted perimeter leads to increased drag. Such calcula-
In this study, laser etching was attempted using a Universal La- tions help explain the experimental results of the various
ser Systems M-300, which produced acceptable tolerances samples, in order to understand the tradeoffs necessary for riblet
(±20 lm). Cast acrylic was chosen since it is laser compatible and design.
optically clear (which aids alignment during experimentation). With a constant channel cross sectional area and Reynolds
Varied parameters include laser speed, number of passes, intensity, number (Re), the goal is to minimize the wetted perimeter by
pulse rate, raster versus vector mode, and distance to the sample. determining the optimal h/vw ratio. To start, it is helpful to first
The baseline size for riblets was derived from the dogfish shark consider open channel laminar flow (Re < 2000), where the friction
Squalus acanthias, whose riblets have a triangular cross section factor (f) is [8]:
386 G.D. Bixler, B. Bhushan / Journal of Colloid and Interface Science 393 (2013) 384–396

Table 1
Summary of samples for experiments with water and air.

Sample set # Material and fabrication processes Geometry Descriptiona Dimensions (lm) h/s t/s
1 Laser etched acrylic Segmented blade Effect of dimensions
L = 850 lm Baseline h = 127 0.31 0.31
g = 200 lm vw = 280
t = 127
Shallow ½h, vw, t 0.16 0.31
Deep 2 h, vw, t 0.62 0.31
Narrow h, 3=4 vw, t 0.38 0.38
Thin h, vw, ½t 0.37 0.18
Single versus. double sided
Shallow ½h, vw, t 0.16 0.31
Effect of coatings
Coated ½h, vw, t 0.16 0.31
Valleys coated ½h, vw, t 0.16 0.31
2 Continuous blade Baseline h, vw, t 0.31 0.31
L = 101 mm

3 Embossed polymer sheets supplied by 3M Corp Continuous sawtooth 80 lm h = t = s = 80 1 1

L = 101 mm 100 lm h = t = s = 100 1 1
vw = 0 150 lm h = t = s = 150 1 1
a
Samples are referred to as ‘‘Description (h/s, t/s)’’ throughout article.

Fig. 1. Blade and sawtooth riblet geometries are shown on the top and bottom, respectively. Flow direction in top views is left to right and end view is into the page. Blade
riblets are both segmented (top) and continuous (not shown); whereas sawtooth are all continuous (bottom). The riblet height (h), thickness (t), valley width (vw), spacing (s),
and length (L) describe the geometry. Optical images show actual samples.

k P ¼ v w þ 2h ð3Þ
f ¼ ð1Þ
Re
where k is the dimensionless friction coefficient (56 < k < 96 for Using the above equations and measured h and vw values from each
smooth and 130 < k < 240 for rough sections). sample in Table 1, Fig. 2 shows calculated friction factors and corre-
In open channel turbulent flow (Re > 4000), the friction factor sponding wetted perimeters. These were calculated for laminar
(f) is [8]: (Re = 1000) and turbulent (Re = 10,000) flows, with the e value of
, 0.0058 for a typical pipe roughness [8]. Lower friction factors corre-
 2
e 5:74 spond with lower drag and vice versa. Fig. 2 also shows wetted
f ¼ 0:25 log10 þ ð2Þ
3:7D Re0:9 perimeter compared to h/vw values. Values of h/vw near zero trans-
late into a flat shallow channel, which increases hydraulic diameter,
where the open channel hydraulic diameter is D ¼ v4h v w , h is the rib- surface area, wetted perimeter, and friction factor. It is believed
wþ2h
let height, vw is the distance between riblets, and e is the equivalent when h/vw equals 0.5, the wetted perimeter is minimized indicat-
sand roughness height (defined as height of sand grains attached to ing the optimal ratio for open channel drag reduction. Shown for
the inside of a pipe that would produce the same pressure drop per each sample is the wetted perimeter as well as laminar and turbu-
unit length as a test pipe of the same diameter). Also Re ¼ VD v where lent friction factors, where the laminar factor holds constant and
V is the average flow velocity and v is the kinematic viscosity of the the turbulent factor varies between samples. The effect of riblet
fluid. wetted perimeter and friction factors in open channel is applicable
Riblet open channel wetted perimeter is P defined as [8]: to studying riblet effects in both open and closed channels.
 G.D. Bixler, B. Bhushan / Journal of Colloid and Interface Science 393 (2013) 384–396 387

2.3. Pressure drop measurements

The general size of the closed channel was based on hospital

catheter tubes (3–5 mm diameter) commonly used in the health-
care industry to transport aqueous fluids. With a smaller cathe-
ter-sized closed channel, considerations include the difficulty of
applying the riblets inside of a channel. For this reason, the rectan-
gular channel sandwich design was chosen, where riblets are ap-
plied to one or both sides and then sandwiched together to
create a rectangular closed channel. For one sided experiments,
the top side is a milled channel and the bottom side contains the
riblet sample.
Fig. 3 illustrates the final experimental apparatus to measure
pressure drop for water and air flow experiments. To achieve de-
sired Reynolds numbers, experiments were conducted with an ele-
vated container, syringe pump (New Era Pump Systems NE-300),
and laboratory air. The two sample flow channel halves were care-
fully aligned, sealed with gaskets, clamped, and then purged of air
bubbles. As shown, entrance and exit flat sections were seamlessly
incorporated (with a roughness Ra = 3.28 lm). Each sample was
measured with an optical microscope and calipers to ensure accu-
rate flow rate and theoretical pressure drop calculations. With the
elevated container experiments, water was pumped from the res-
ervoir to the elevated container (via the fill line), which then flo-
wed down the supply line. To ensure a constant flow rate, the
control valve and overflow line regulated the water level, and the
flow rate (thus Reynolds number) was varied by changing the con-
tainer elevation. The syringe pump delivered a specified volumet-
ric flow rate, and velocity was calculated. Laboratory air connected
to an adjustable Omega FL-1478-G rotameter allowed for incre-
mental variation of the flow velocity. The laboratory air flow veloc-
Fig. 2. Top plot shows modeling results for blade riblet samples with the calculated
laminar (Re = 1000) and turbulent (Re = 10,000) flow friction factors, along with the ity was calculated based on the rotameter value and the channel
riblet valley wetted perimeter (where wetted perimeter = vw + 2 h). Larger riblets cross sectional area.
translate into a higher wetted perimeter and thus higher friction. Tradeoffs occur To maintain kinematic viscosity, water temperature was moni-
when designing riblets in various flow conditions. With a constant flow channel tored with a CND DTQ450X thermometer, and held constant
area, the bottom plot shows the ideal height (h) and valley width (vw) ratios for
turbulent open channel flow (Re = 10,000), where the lowest wetted perimeter is
(18.5 < temp. < 19 °C). The pressure drop between the inlet and
desirable. outlet was measured with an Omega PX26-005DV differential
manometer (potted in room temperature vulcanized silicone). Data
was collected at 10 Hz for 30 s with a Vishay 2311 Laboratory
Amplifier and a Measurement Computing USB-1208LS DAQ card.
Dimensionless riblet parameters denoted by the + symbol allow The system was calibrated prior to use with an Ametek RK-
comparison of riblet geometries in different flow conditions. The 1600W6 pneumatic pressure system.
dimensionless units for riblet spacing (s+) and height (h+) are calcu-
lated with [36]: 2.3.1. Model for predicting closed channel pressure drop
Comparing predicted pressure drop to the measured values en-
sV s sured that the system was behaving properly by detecting leaks and
sþ ¼ ð4Þ
v misalignments. This was done by comparing the milled channel
sample to the predicted values. Additionally, the milled channel
sample serves as the reference sample when reporting pressure
þ hV s
h ¼ ð5Þ drop percentage values. Predicting pressure drop of a flat hydro-
v philic rectangular closed channel requires use of the incompressible
flow equations for straight uniform pipes. Since the Mach number is
In turbulent flow conditions, the diameter of vortices is equal to the less than 0.3 for all experiments, incompressible flow equations
s+ value of 35–50 [36,57] and is believed to be approximately equal may be used [8]. Each riblet sample was measured using calipers
to 2–3 the optimized s+ [36]. Also Vs is the wall stress velocity and for the height and optical microscopy for the pitch, and the channel
so is the wall shear stress, which is calculated by [36]: dimensions were measured with calipers. The predicted pressure
 1=2 drop was calculated using the effective cross sectional flow area (to-
so tal channel cross sectional area minus riblet cross sectional area).
Vs ¼ ð6Þ
q This allows for direct sample comparison by accounting for the flow
reduction due to the physical protrusion of riblets into the channel.
Pressure drop (Dp) between two points in a flat hydrophilic
so ¼ 0:039v 1=4 qV 7=4 D1=4 ð7Þ straight uniform closed channel with incompressible and fully
developed flow is found with the Darcy-Weisbach formula [8]:
where V is the mean flow velocity and D is the closed channel
hydraulic diameter (see following Eq. (9)). The dimensionless values
qV 2 fL
Dp ¼ ð8Þ
will aid in understanding and comparing experimental results. 2D
388 G.D. Bixler, B. Bhushan / Journal of Colloid and Interface Science 393 (2013) 384–396

Fig. 3. (a) Schematic of the apparatus that provided pressure drop measurements for samples in closed channel, high velocity water flow. The flow channel was connected to
a manometer to measure pressure drop. Water was pumped into the elevated container and then traveled to the flow channel. Adjusting the bottle height allowed for
variations in flow rates and thus Reynolds numbers, ranging from laminar to turbulent flows. Interface views are shown highlighting the top channel and bottom riblet
channels that are sandwiched together during experimentation. (b) Schematic of the apparatus that provided pressure drop measurements for samples in closed channel air
and low velocity water flow. Air experimentation was conducted with laboratory air regulated by a rotameter connected to the flow channel. Lower velocity water
experimentation was conducted with a syringe pump connected to the flow channel.

where q is the fluid density and L is the length between two points. long. Fig. 4a and b shows results from syringe pump laminar low
Flow velocity (V) is determined by dividing the volumetric flow rate velocity flow (0 < Re < 500) and Fig. 5a and b elevated container
by the channel cross sectional area. In air experiments, the rotame- laminar and turbulent high velocity flow (0 < Re < 6000). Figs. 4a
ter values were used with manufacturer provided charts to deter- and 5a show results from a series of blade experiments comparing
mine the flow velocity. The rectangular closed channel hydraulic the effect of nanoscale roughness, h/s and t/s, continuous versus
diameter is [8]: segmented, and coated versus uncoated. Figs. 4b and 5b show re-
sults from sawtooth experiments comparing the effect of h. One
2ab
D¼ ð9Þ plot is shown with the predicted flat channel line using Eq. (8)
aþb
and all others the milled channel control sample for comparison,
where a is the width and b is the height. The friction factor for rect- each with parabolic fit trend lines connected to the origin. Percent-
angular closed channel flow is [8]: age pressure drop difference is calculated from the milled sample.
   Calculations use mass density (q) equaling 1000 kg m3 and kine-
64 2 11 b b
f ¼ þ 2 ð10Þ matic viscosity (v) equaling 1.034  106 m2 s1 [58].
Re 3 24 a a
The top row of Fig. 4a shows baseline compared with the flat la-
where ba 6 1. ser and milled channel samples, which are all within one standard
Friction factor values for laminar and turbulent flow can also be deviation of each other. The predicted flat channel line is higher
found with the Moody chart to conveniently account for various than the samples, and the milled channel outperforms the laser
surface roughness values [8]. channel, indicating that surface nanoscale roughness contributes
to pressure drop. Additionally, the variables vw and t affect pres-
3. Results and discussion sure drop, when s remains constant by varying vw and/or t. With
constant h and s values, the narrow sample shows a lower pressure
3.1. Pressure drop with water flow drop as compared to thin sample.
The middle row of Fig. 4a shows segmented compared to con-
To understand the effects of riblets with water flow, a series of tinuous blade samples. The baseline and continuous sample pres-
experiments were conducted to study various parameters. With sure drops are similar, and are within one standard deviation of
the top and bottom assembled the rectangular channel measured each other. The continuous sample line is slightly higher than the
0.7 mm high by 3.2 mm wide; with the sample section 101 mm baseline sample line, indicating that such riblets offer negligible
 G.D. Bixler, B. Bhushan / Journal of Colloid and Interface Science 393 (2013) 384–396 389

Fig. 4. Pressure drop in laminar water flow with (a) blade and (b) sawtooth riblets (low velocity). The h/s and t/s variations in (a) are shown with the solid and dashed underlined
notations, respectively. See Table 1 for sample descriptions. Closed channel laminar flow (0 < Re < 500) was achieved with a syringe pump. For comparison, the predicted
pressure drop for a flat closed channel is shown, calculated using the Darcy-Weisbach formula (Eq. 8). Also shown is the milled channel line, which serves as the reference
sample when reporting pressure drop percentage values. For blade riblets, parameters affecting pressure drop include the nanoscale roughness, geometry, h/s and t/s values,
continuous and segmented, and coatings. Higher pressure drop translates into higher drag; therefore lower pressure drop is desirable. The top row of (a) indicates higher
nanoscale roughness increases pressure drop and shallower riblets reduce pressure drop. The bottom row of (a) indicates continuous and segmented blade riblets are similar in
the left figure, and that hydrophobic nanostructured coating reduces pressure drop in water flow in the right figure. Also (b) indicates that sawtooth riblets perform similarly,
suggesting that the effect of equilateral sawtooth riblet height (h) is negligible. Error bars show ±1 standard deviation and found to be ±2%, which is hardly visible in the plot.

improvement. Coated and uncoated sample results indicate that heights are different for each sample, indicating that pressure drop
hydrophobic nanostructured coating offers improvement when ap- is not influenced by h. It should be noted that the sawtooth sam-
plied to riblets. The shallow sample was chosen to receive the coat- ples differ from the blade samples, since for sawtooth h = t = s,
ing due to positive preliminary results, and is compared with the which is not the case with blade riblets. Each sample is within
coated samples of the same dimensions. Fig. 4b shows sawtooth one standard deviation of each other indicating negligible
sample results comparing the three samples, indicating pressure differences.
drop reduction and performance similar to the shallow blade sam- When comparing the pressure drop performance between the
ple. Interestingly, the pressure drops are similar even though the syringe pump in Fig. 4a and b to the elevated bottle experiments
390 G.D. Bixler, B. Bhushan / Journal of Colloid and Interface Science 393 (2013) 384–396

Fig. 5. Pressure drop in laminar and turbulent water flow with (a) blade and (b) sawtooth riblets (high velocity). The h/s and t/s variations in (a) are shown with the solid and
dashed underlined notations, respectively. See Table 1 for sample descriptions. Closed channel flow (0 < Re < 6000 where Re > 4000 is turbulent) was achieved with an
elevated container. For comparison, the predicted pressure drop for a flat closed channel is shown, calculated using the Darcy-Weisbach formula (Eq. 8). Also shown is the
milled channel line, which serves as the reference sample when reporting pressure drop percentage values. Riblet geometries and pressure drop trends are similar to low
velocity water flow in Fig. 4. Error bars show ±1 standard deviation and found to be ±1%, which is hardly visible in the plot.

in Fig. 5a and b, the results follow similar trends. At higher Rey- pressure drop is reduced by 19% with the shallow sample (h/
nolds numbers, drag reduction generally improves, as observed s = 0.16 at Re = 4700) and increased by 26% with the deep sample
throughout the results. Since turbulent flow contains natural vorti- (h/s = 0.62 at Re = 4700). This indicates a relationship between
ces, the riblets offer more benefit in turbulent versus laminar flow. pressure drop and riblet height when the spacing remains con-
For instance, the shallow sample reduces pressure by 6% at stant, where larger height increases pressure drop, likely due to in-
Re = 550 and 19% at Re = 4700, and it is expected this drag reducing creased riblet valley wetted perimeter and friction (see also Fig. 2).
benefit will plateau around s+  15. This is further enhanced by
adding the nanostructured coating, where the valleys coated sam- 3.2. Pressure drop with air flow
ple shows a pressure drop reduction of 29% at Re = 550 and 34% at
Re = 4700. When considering the effects of h/s and t/s, pressure To understand the effects of riblets with air flow, a series of
drop reduction seems most dependent on h/s. For instance, experiments were conducted to study various parameters. When
 G.D. Bixler, B. Bhushan / Journal of Colloid and Interface Science 393 (2013) 384–396 391

the top and bottom were assembled, the rectangular channel mea- nected to the origin. Percentage pressure drop difference is calcu-
sured 0.7 mm high by 3.2 mm wide, with the sample section lated from the milled channel sample. Calculations use mass
101 mm long. Fig. 6a and b shows results from laminar through density (q) equaling 1.2 kg m3 and kinematic viscosity (v) equal-
turbulent air flow (0 < Re < 6000). Fig. 6a shows results from a ser- ing 1.51  105 m2 s1 [58].
ies of blade experiments comparing the effect of nanoscale rough- Fig. 6a and b shows similar trends to the water experimenta-
ness, h/s and t/s, continuous versus segmented, and coated versus tion, indicating that riblet efficiency is independent of fluid viscos-
uncoated. Fig. 6b shows results from sawtooth experiments com- ity, with a few noteworthy differences. The air data contains
paring the effect of h. One plot is shown with the predicted flat slightly higher pressure drops, perhaps due to the lower viscosity
channel line using Eq. (8) and all others the milled channel control and the tendency for the vortices to translate more easily. For in-
sample for comparison, each with parabolic fit trend lines con- stance, pressure drop reduction with the shallow sample is 19%

Fig. 6. Pressure drop in laminar and turbulent air flow with (a) blade and (b) sawtooth riblets. The h/s and t/s variations in (a) are shown with the solid and dashed underlined
notations, respectively. See Table 1 for sample descriptions. Closed channel flow (0 < Re < 6000 where Re > 4000 is turbulent) was achieved with a laboratory air supply. For
comparison, the predicted pressure drop for a flat closed channel is shown, calculated using the Darcy-Weisbach formula (Eq. 8). Also shown is the milled channel line, which
serves as the reference sample when reporting pressure drop percentage values. Pressure drop reductions follow similar trends but are slightly lower than found with water
flow. Error bars show ±1 standard deviation and found to be ±1%, which is hardly visible in the plot.
392 G.D. Bixler, B. Bhushan / Journal of Colloid and Interface Science 393 (2013) 384–396

in water and 13% in air, at Re = 4700 and 4900 respectively. With experiments were conducted with the channel height doubled. Re-
air, the achievable Reynolds number range was higher as compared sults indicate that pressure drop with the riblets on the bottom are
to the water, and the higher Reynolds numbers show continued unaffected by channel height, but pressure drop improves with
pressure drop reduction. However, it is expected above a certain topside riblets and 2 channel height. This improvement in the
Reynolds number, that the pressure drop reduction will plateau. top side riblets with 2 channel height is believed to have a con-
When comparing continuous blade samples in Figs. 5a and 6a, a nection with the vortices interaction, as described later when pre-
lower pressure drop is shown with air than with water flow, sug- senting the conceptual model. With 2 channel height, results are
gesting continuous blades offer more benefit with air as compared within one standard deviation.
to water. The continuous blades may reduce cross stream move- The bottom row of Fig. 7 shows results from riblets on both the
ment with air more efficiently than with segmented blades. As top and bottom sides of the channel for two channel heights in
the lower viscosity air vortices rotate in the channel, the gap area water flow. Compared to the milled channel line, results indicate
between segmented blades perhaps disrupts and unpins the vorti- that the pressure drop increases with the 1 channel height and
ces. The shallow and baseline samples are nearly identical, which drops with a 2 channel height. It is believed that the vortices from
indicates the minimum h/s ratio. A lower value would increase the 1 channel height interact and increase the drag, but the larger
pressure drop since the riblets essentially disappear. The 100 lm gap distance in the 2 channel height reduces this effect.
sawtooth sample shows the highest pressure drop reduction of The top row of Fig. 8 shows results from riblets on the top only
26% at Re = 4900. and on the bottom only for two channel heights in air flow. The
bottom row of Fig. 8 shows results from riblets on both the top
3.3. Effects of riblet combinations and channel dimensions and bottom sides of the channel for two channel heights in water
flow. When comparing the water and air experimental data, simi-
To understand the effects of riblets in various combinations, a lar trends are shown when considering riblets on the top and bot-
series of water and air flow blade experiments were conducted. tom, but trends differ when considering riblets on two sides of the
Figs. 7 and 8 show results from experiments including 1 channel channel. Water data deviates more than the air data, where the 1
height, 2 channel height, riblets on top side, and riblets on top channel height water results are higher above the milled channel
and bottom. When the top and bottom were assembled, the rectan- line as compared to the same experiment with air.
gular channel measured 0.7 mm high (1.4 mm for 2 channel
height) by 3.2 mm wide; with the sample section 101 mm long. 3.4. Scalability of riblets
The top row of Fig. 7 shows results from riblets on the top only
and on the bottom only for two channel heights in water flow. Re- Effectively scaling riblets is important since dimensions are
sults indicate that riblets located on the bottom side perform dif- usually determined by the particular application and available fab-
ferently from the same riblets located on the top side. The rication processes. For instance, millimeter sized riblets are proba-
pressure drop increases by 20% when the riblets are transferred bly more feasible for pipeline or ship applications as compared to
from the bottom to the top, although both combinations show micrometer sized riblets. Therefore the scalability effects
pressure drop benefit compared to the milled channel line. Since are important to understand in order to optimize riblets of differ-
the channel height contributes to pressure drop, the same ent dimensional magnitudes for particular flow conditions.

Fig. 7. Effects of channel height and blade riblet location on pressure drop in laminar and turbulent water flow. Also shown is the milled channel line, which serves as the
reference sample when reporting pressure drop percentage values. Compared are the 1 and 2 channel height as well as riblets on top versus bottom, where riblets on
bottom are believed to provide more pressure drop reduction. The bottom row shows riblets on both the top and bottom, and that the 2 channel height provides more
benefit. Error bars show ±1 standard deviation and found to be ±1%, which is hardly visible in the plot.
 G.D. Bixler, B. Bhushan / Journal of Colloid and Interface Science 393 (2013) 384–396 393

Fig. 8. Effects of channel height and blade riblet location on pressure drop in laminar and turbulent air flow. Also shown is the milled channel line, which serves as the
reference sample when reporting pressure drop percentage values. Compared are the 1 and 2 channel height as well as riblets on the top versus bottom, where riblets on
the bottom are believed to provide more pressure drop reduction. The bottom row shows riblets on both the top and bottom, and that the 2 height provides more benefit.
Error bars show ±1 standard deviation and found to be ±1%, which is hardly visible in the plot.

Understanding scalability is conducted by comparing the pressure on their s values to provide a wide range of s+ values, with spacing
drops of several samples at various dimensionless spacing (s+) val- values shown in Table 1. To study the effects of fluid viscosity, flow
ues (calculated from Eq. (4)). The s+ values were varied by changing conditions ranged from laminar to turbulent flow for low to high
the spacing (s) values, and both blade and sawtooth samples were values of s+ with both air and water.
chosen for evaluation. For given s+ values, the pressure drop can be Scalability study results are presented in Fig. 9. The top row
compared amongst different samples. Riblets were selected based compares the s+ values to pressure drop for water experimentation,

Fig. 9. Scalability effects of riblets in water and air flow. Shown is the pressure drop as a function of s+ for various samples. The top row shows water and the bottom row air
experimentation with the blade and sawtooth riblets. Results from the studied samples indicate that lower h/s for blade and higher h and t for equilateral sawtooth riblets are
desirable. However the s+ value must be such that the vortices remain lifted and pinned. Error bars show ±1 standard deviation and found to be ±0.3%, which is hardly visible
in the plot.
394 G.D. Bixler, B. Bhushan / Journal of Colloid and Interface Science 393 (2013) 384–396

with blade riblets (left) and sawtooth (right). As the s+ values in- angle when viewed in the streamwise direction, since over time
crease for each blade sample, the pressure drops indicate that for the droplet slowly flattens. When viewing the droplet from the
higher s+ values, lower h/s values are desirable (until expected pla- cross-stream direction, the droplets are pinned between the saw-
teauing). The equilateral sawtooth riblets also provide a compari- tooth riblets and exhibit a higher contact angle. This effect was
son between s+ and pressure drop, where h = t with a constant h/ not present with the segmented blade geometries but was evident
s and t/s between samples. As the s+ values increase for each saw- with the continuous baseline sample.
tooth sample, the pressure drops indicate that for higher s+ values, When comparing pressure drop results with wettability, there
higher h values are desirable (until expected plateauing). Similar is not a direct correlation, since the deep sample exhibits a high
results are shown with the air experimentation in the bottom row. contact angle but also higher pressure drop. When combined with
An overall trend suggests that water and air flow data are sim- appropriately sized riblets, a hydrophobic surface provides benefit,
ilar when comparing samples with varying h/s and t/s ratios. Re- as shown with the coated sample, which provides the greatest
sults from the studied samples indicate that lower h/s for blade pressure drop reduction.
and higher h and t for equilateral sawtooth riblets is desirable.
However the s+ value must be such that the vortices remain lifted
3.6. Models for vortices behavior
and pinned. The s+ value may be changed by altering the spacing,
which is accomplished by changing the vw and/or t. If vw is too
Experimental results indicate a pressure drop reduction and
small, then the vortices are believed to intermingle and increase
thus drag reduction for a variety of riblets and combinations. The
the pressure drop. The goal is to minimize pressure drop and select
the optimum s+ value for use in different flow conditions. Combin-
ing this information with previous data, producing optimal and
scalable drag-reducing blade or sawtooth riblets for any fluid is be-
lieved to require low h/s, t/s, t, vw, and s+  15.

3.5. Wettability of samples

To understand the effects of wettability combined with riblets,

the apparent contact angles were measured for each sample. This
was conducted with a Rame-Hart automated goniometer model
290-F4, where 5 lL water droplets were deposited onto the riblet
surface and measured. Results are summarized in Fig. 10.
The blade samples show contact angle measurements ranging
from hydrophilic to hydrophobic. For comparison, flat milled and
laser channel samples were measured at 91° and 108° respectively.
The shallow sample shows a lower contact angle at 83°, presum-
ably due to the Wenzel wetting when water penetrates between
the individual riblets. As expected, the coated and valleys coated
samples exhibit a higher contact angle than the uncoated sample,
showing the hydrophobic effectiveness of the nanostructured
coating.
The sawtooth riblets show higher contact angles than the blade,
with measurements ranging from 135° to 145° for the 80 lm and
150 lm samples, respectively. The droplets show a lower contact

Fig. 10. Apparent contact angle measurements for blade and sawtooth riblets in Fig. 11. Scale conceptual models of vortices interacting with various riblet
flat, segmented, and continuous configurations. Measurements are conducted with geometries. Drag reduction is achieved by lifting and pinning the naturally
a goniometer and water droplet. The highest contact angle is shown with the occurring fluid vortices. This results in less shear stress, momentum transfer, and
sawtooth samples followed by the hydrophobic coated samples. Error bars show ±1 vortices ejection from the viscous sublayer. A minimum riblet height is necessary to
standard deviation and found to be ±1%. lift and pin the vortices.
 G.D. Bixler, B. Bhushan / Journal of Colloid and Interface Science 393 (2013) 384–396 395

Fig. 12. Scale conceptual models of vortices’ behaviors with the flat, shallow sample geometry, one or two sided, and 2 channel height combinations. Vortices are believed
to rotate clockwise on a flat plate and counterclockwise on riblets. When considering the two-sided nature of a closed channel, the rotational direction accounts for drag
increase or reduction. A 1 channel height is believed to allow vortices from the top and bottom to interact and either increase the rotational speed or cancel each other. Cross
stream movement occurs when vortices are not properly pinned, for instance with a flat plate. Increasing the channel height moves the top and bottom side vortices apart and
the vortices are believed to not interact in the same manner.

most benefit is shown with the shallow samples in turbulent flow. Since closed channel flow operates differently than open chan-
In order to fully understand the mechanisms behind the pressure nel, a possible conceptual closed channel flow theoretical model
drop reduction, careful consideration is given to the vortices and was developed, as shown in Fig. 12. The illustration highlights
riblet interaction. With experimental data from both water and the assumed interaction between the vortices and the flat plate
air, we believe several theories are possible to imagine when as well as shallow riblets on top, bottom, both sides, and both sizes
describing flow behaviors in various conditions. with 2 channel height. This model, developed from the experi-
Understanding the vortices and dimensional values requires mental data and flow visualization (where turbulent vortex diam-
conceptual visualization based on the experimental data. Fig. 11 eter is approximately 35–50 dimensionless wall units) [29], shows
illustrates one possible explanation for how vortices may interact how the top and bottom vortices may interact. Understanding this
with the riblet geometries, flat plate, and each other. Shown are interaction will aid in the design of efficient closed channel flow
vortices in open channel flow and the effects of contact, surface systems.
area, and entanglement on drag. The riblets and vortices are be- During closed channel flow when neighboring wall effects play
lieved to be scalable; therefore s+ equals approximately the diam- a role, vortices from the top and bottom mix and perhaps their
eter of vortices or less. In general, minimizing the higher velocity rotation is either additive or subtractive. In this model, vortices
vortices’ contact is desired, along with minimizing wetted perime- seem to flow clockwise on flat plates and counterclockwise on rib-
ter for the lower velocity flow between riblets. let surfaces. This difference in direction affects how the vortices
As shown, the arrows indicate assumed correlations between mix by determining if they reduce or increase total drag. Vortices
vortices’ contact, surface area, and entanglement as related to total that rotate in opposite directions will cancel each other and reduce
drag. Entanglement is perhaps worse with the flat plate, and with total drag, as shown with riblets on one side. Conversely, vortices
riblets it increases as the spacing reduces, as shown with narrow that rotate in the same direction will add to one another, and in-
and thin samples. The height effect is evident when comparing crease total drag. This is shown with flat plates on the top and bot-
the three riblet heights, since the deep sample shows the highest tom, as well as with riblets on both sides (1 channel height).
drag due to its increased riblet valley wetted perimeter. The h/s ra- When considering riblets on both sides (2 channel height), the
tios were varied with increasing or decreasing height, which also vortices are believed to be lifted and pinned as expected. Further-
affects the wetted perimeter of the sample. This increase in wetted more since the channel height is ample, the top and bottom do not
perimeter increases friction (also see Fig. 2) and thus increases mix, thus reducing drag. In this case, channel height H equals at
pressure drop, as shown with the deep riblets. The best performer least 3 turbulent vortex diameter, or 150 wall units, or 10 s+
is the shallow sample, since it is believed that the vortices are still when s+ = 15, which is believed necessary for maximum drag
lifted and pinned, but the entanglement and wetted perimeter are reduction.
minimized. Furthermore, the values t and t/s should be minimized
in order to reduce the contact with the higher speed vortices, 3.7. Conclusions
which are believed to interact with riblet tips. Narrow riblets en-
sure that such interaction is minimized, although a minimum The optimization of shark skin-inspired riblets to reduce drag
thickness is necessary for structural integrity. was experimentally studied in closed channel flow. Drag was
396 G.D. Bixler, B. Bhushan / Journal of Colloid and Interface Science 393 (2013) 384–396

determined by measuring pressure drop using water and air in [2] B. Bhushan, Phil. Trans. R. Soc. 367 (2009) 1445–1486.
[3] B. Bhushan, Springer Handbook of Nanotechnology, third ed., Springer, New
laminar through turbulent regimes. Experimental parameters in-
York, 2010.
cluded riblet geometry, fluid velocity, fluid viscosity, riblet combi- [4] B. Bhushan, Biomimetics: bioinspired Hierarchical-Structured Surfaces for
nations, channel size, wettability, and scalability. Blade style riblets Green Science and Technology, Springer-Verlag, Heidelberg, Germany, 2012.
were laser etched in acrylic with a wide range of height (h) divided [5] R. Allen (Ed.), Bulletproof Feathers How Science Uses Nature’s Secrets to
Design Cutting Edge Technology, Ivy Press, London, 2010.
by spacing (s) (or h/s) and thickness (t) divided by spacing (or t/s) [6] R.E. Armstrong, M.D. Drapeau, C.A. Loeb, J.J. Valdes (Eds.), Bio-Inspired
ratios, and polymer sawtooth riblets were obtained from the 3 M Innovation and National Security, National Defense University Press,
Corp. Washington, 2010.
[7] Y. Bar-Cohen, Biomimetics: nature Based Innovation, CRC Press, Boca Raton,
With water and air, riblets show an improved pressure drop Florida, 2011.
reduction in turbulent versus laminar flow, presumably due to [8] R.D. Blevins, Applied Fluid Dynamics Handbook, Van Nostrand-Reinhold, New
the lifting and pinning of vortices. Results also indicate pressure York, 1984.
[9] L.F. Melo, T.R. Bott, C.A. Bernardo (Eds.), Fouling Science and Technology,
drop reduction improves in most cases with increasing Reynolds Kluwer Academic Publishers, Dordrecht, The Netherlands, 1988.
number (until expected plateauing). In turbulent water flow, the [10] P. Lowrey, J. Harasha, Energy 16 (1991) 631–642.
valleys coated sample provides maximum pressure drop reduction [11] D.W. Bechert, M. Bruse, W. Hage, R. Meyer, in: Paper # AIAA-1997-1960,
presented at AIAA 28th Fluid Dynamics Conference, AIAA, New York, 1997.
of 34%, indicating that properly sized riblets combined with an [12] M. Fingerman, R. Nagabhushanam, M.F. Thompson (Eds.), Recent Advances in
appropriate nanostructured coating is optimal. Furthermore, drag Marine Biotechnology, Science Publishers, Inc., Enfield, New Hampshire, 1999.
reduction is greater in water versus air flow, where in turbulent [13] J. Walker, S. Surman, J. Jass (Eds.), Industrial biofouling detection, Prevention
and Control, Wiley, New York, 2000.
air flow the 100 lm sawtooth sample provides the maximum pres-
[14] A.I. Railkin, Marine Biofouling Colonization Processes and Defenses, CRC Press,
sure drop reduction of 26%. Boca Raton, Florida, 2004.
Several design principles for riblet geometry optimization are [15] C. Hellio, D. Yebra (Eds.), Advances in Marine Antifouling Coatings and
derived or confirmed from the experimental results. These include Technologies, CRC Press, Boca Raton, Florida, 2009.
[16] B. Bhushan, Y.C. Jung, Prog. Mater. Sci. 56 (2011) 1–108.
low h, h/s, t, t/s, and vw; with the previously published optimum [17] Z. Deyuan, L. Yuehao, C. Huawei, J. Xinggang, Pipeline Gas J. 1 (March) (2011)
dimensionless riblet spacing value (s+) near 15. In addition when 59–60.
studying segmented and continuous riblets, it appears that seg- [18] G.D. Bixler, B. Bhushan, Phil. Trans. R. Soc. A 370 (2012) 2381–2417.
[19] W. Barthlott, C. Neinhuis, Planta 202 (1997) 1–8.
mented for water and continuous for air are desirable. Furthermore [20] A. Kesel, R. Liedert, Comp. Biochem. Physiol. A 146 (2007) S130.
in water flow, riblets with the hydrophobic nanostructured coating [21] E. Ralston, G. Swain, Bioinsp. Biomim. 4 (2009) 1–9.
reduce pressure drop more than the uncoated, due to the water [22] M. Liu, S. Wang, Z. Wei, Y. Song, L. Jiang, Adv. Mater. 21 (2009) 665–669.
[23] W. Reif, Courier Forschungsinstitut Senckenberg 78 (1985) 1–255.
repellency effect. [24] D.W. Bechert, G. Hoppe, W.E. Reif, in: Paper # AIAA-85-0546, Presented at
As far as scalability in water and air, results from the studied AIAA Shear Flow Control Conference, AIAA, New York, 1985.
samples indicate that lower h/s for blade riblets and lower h for [25] D.W. Bechert, M. Bartenwerfer, G. Hoppe, W.-E. Reif, Paper # ICAS-86-1.8.3,
vol. 2 (A86-48-97624-01), pp. 1044–1067, Proc. 15th ICAS Congress, AIAA,
equilateral sawtooth riblets are desirable. However the s+ value New York, 1986. .
must be such that the vortices remain lifted and pinned. This al- [26] D.W. Bechert, M. Bruse, W. Hage, J.G.T. van der Hoeven, G. Hoppe, J. Fluid
lows one to design riblets for various applications where the size Mech. 338 (1997) 59–87.
[27] D.W. Bechert, M. Bruse, W. Hage, R. Meyer, Naturwissenschaften 87 (2000)
may be a limiting factor.
157–171.
Placement of riblets either on the top or bottom or both sides [28] P. Luchini, F. Manzo, A. Pozzi, J. Fluid Mech. 228 (1991) 87–109.
inside the channel is found to affect the pressure drop reduction. [29] S.J. Lee, S.H. Lee, Exp. Fluids 30 (2001) 153–166.
Riblets on the bottom appear to perform better than on the top. [30] K. Krieger, Science 305 (2004) 636–637.
[31] B. Dean, B. Bhushan, Phil. Trans. R. Soc. A 368 (2010) 4775–4806.
With appropriately sized channel height, riblets on both sides per- [32] Z. Deyuan, L. Yuehao, L. Xiang, C. Huawei, J. Hydrodyn. 23 (2011) 204–211.
form better than on just one side. Vortices from the top and bottom [33] J. Oeffner, G.V. Lauder, J. Exp. Biol. 215 (2012) 785–795.
sides of the channel are believed to interact, increasing total shear [34] M.J. Walsh, in: Paper # AIAA-82-0169, Presented at AIAA 20th Aerospace
Sciences Meeting, AIAA, New York, 1982.
stress and thus drag. This is reduced when the spacing value (s) [35] M.J. Walsh, A.M. Lindemann, in: Paper # AIAA-84-0347, Presented at AIAA
equals the vortex diameter and the channel height (H) equals twice 22nd Aerospace Sciences Meeting, AIAA, New York, 1984.
the vortex diameter, which helps minimize the negative vortices’ [36] D.W. Bechert, M. Bruse, W. Hage, Exp. Fluids 28 (2000) 403–412.
[37] S.P. Wilkinson, J.B. Anders, B.S. Lazos, D.M. Bushnell, Inter. J. Heat Fluid Flow 9
interaction. (1988) 266–277.
Producing optimal and scalable drag-reducing blade riblets for [38] S.P. Wilkinson, B.S. Lazos, in: H.W. Liepman, R. Narasimha (Eds.), Turbulence
any fluid is believed to require the following design principles: Management and Relaminarisation, Springer-Verlag, Berlin, 1988, pp. 121–
131.
low h, h/s, t/s, t, and vw; with s+  15 and H  10  s+. These design [39] J.J. Rohr, G.W. Andersen, L.W. Reidy, E.W. Hendricks, Exp. Fluids 13 (1992)
principles are applicable for closed channel flow; thus providing 361–368.
new insight for various medical, marine, and industrial [40] P.R. Viswanath, Prog. Aerospace Sci. 38 (2002) 571–600.
[41] Y.C. Jung, B. Bhushan, J. Phys.: Condens. Matter. 22 (2010) 1–9.
applications.
[42] B. Dean, B. Bhushan, Appl. Surface Sci. 258 (2012) 3936–3947.
[43] G.D. Bixler, B. Bhushan, Soft Matter, http://dx.doi.org/10.1039/C2SM26655E.
[44] M.S. Khan, in: Paper # AIAA-86-1127, Presented at AIAA 24th Aerospace
Acknowledgments
Sciences Meeting, AIAA, New York, 1986.
[45] F. J. Marentic, T. L. Morris, United States Patent No. 5 133 516, 1992.
The financial support of this research was provided by a grant [46] A.B. Brennan, R.H. Baney, M.I. Carman, T.G. Estes, A.W. Feinberg, L.H. Wilson,
from the National Science Foundation, Arlington, VA (Grant # J.F. Schumacher, United States Patent No. 7 650 848, 2010.
[47] B. Denkena, J. Kohler, B. Wang, CIRP J. Manuf. Sci. Technol. 3 (2010) 14–26.
CMMI-1000108). The authors want to thank Mr. Matt Bernhardt [48] G. Hirt, M. Thome, Prod. Eng. Res. Devel. 1 (2007) 351–356.
(Knowlton School of Architecture, The Ohio State University) for la- [49] P.H. Dickinson, G.M. Proudley, United States Patent No. 4 994 639, 1991.
ser etching support. The authors also want to thank Dr. G. Memo [50] J. Ou, B. Perot, J.P. Rothstein, Phys. Fluids 16 (2004) 4635–4643.
[51] R.J. Daniello, N.E. Waterhouse, J.P. Rothstein, Phys. Fluids 21 (2009) 085103.
Izzi and Dr. Thomas Herdtle (3 M Corp. Renewable Energy Division [52] M.B. Martell, J.B. Perot, J.P. Rothstein, J. Fluid Mech. 620 (2009) 31–41.
Laboratory) for the sawtooth riblet samples as well as critical re- [53] M.B. Martell, J.P. Rothstein, J.B. Perot, Phys. Fluids 22 (2010) 065102.
view of the manuscript. [54] J. Genzer, K. Efimenko, Biofouling 22 (2006) 339–360.
[55] D. Ebert, B. Bhushan, J. Colloid Interface Sci. 384 (2012) 182–188.
[56] D. Ebert, B. Bhushan, Langmuir 28 (2012) 11391–11399.
References [57] S.J. Kline, W.C. Reynolds, F.A. Schraub, P.W. Runstadler, J. Fluid Mech. 30 (1967)
741–773.
[58] S. Goldstein, Modern Developments in Fluid Dynamics, Clarendon, Oxford,
[1] M.W. Collins, C.A. Brebbia (Eds.), Design and Nature II Comparing Design in
1938.
Nature with Science and Engineering, WIT Press, Southampton, UK, 2004.