Journal of Ship Research, Vol. 64, No. 2, June 2020, pp.

118–126
http://dx.doi.org/10.5957/jsr.2020.64.2.118

Air Layer on Superhydrophobic Surface for Frictional Drag Reduction

Bradley C. Peifer, Christopher Callahan-Dudley, and Simo A. Mäkiharju
University of California, Berkeley, Berkeley, California

We examined the feasibility of combining a superhydrophobic surface (SHS) and air

layer drag reduction (ALDR) to achieve the frictional drag reduction (DR) shown
achievable with traditional ALDR, but at a reduced gas flux to increase the achievable
net energy savings. The effect of a commercial SHS coating on the gas flux required to
maintain a stable air layer (AL) for DR was investigated and compared with that of a
painted non-SHS at Reynolds numbers up to 5.1
106. Quantitative electrical im-
pedance measurements and more qualitative image analysis were used to charac-
terize surface coverage and to determine whether a stable AL was formed and
maintained over the length of the model. Analysis of video and still images for both the
SHS and painted surface gives clear indications that the SHS is able to maintain AL
consistency at significantly lower gas flux than required on the non-SHS painted
surface. Hydrophobicity of the surfaces was characterized through droplet contact
angle measurements, and roughness of all the flow surfaces was measured. The
results from these preliminary experiments seem to indicate that for conditions ex-
plored (up to Rex ¼ 5.1
106), there is a significant decrease in the amount of gas
required to establish a uniform AL (and hence presumably achieve ALDR) on the SHS
when compared with a hydraulically smooth painted non-SHS.

Keywords: frictional drag; air layer; superhydrophobic

1. Introduction SHS and ALDR to attempt to achieve benefits shown for traditional
ALDR, but at a reduced gas flux to increase the achievable net
To achieve air layer drag reduction (ALDR), a significant amount energy savings. As an added benefit, this will also reduce the size of
of energy is required to supply the air from compressors. Although gas compressors and piping needed, and hence shorten payback
net energy savings are achievable, large compressors and piping time.
lead to reduced savings and lengthen the payback time on traditional Frictional drag can account for approximately 60% of a cargo
ALDR systems. Superhydrophobic surface (SHS) alone has been ship’s propulsive power requirement; hence, techniques that can
shown to enable reduced frictional drag in the laboratory, but its significantly reduce a ship’s frictional resistance can have a sub-
benefits are lost if gas trapped in microscopic air pockets (plastrons) stantial impact, both economically and environmentally, on the cost
on the surface is lost because of diffusion or entrainment. Fur- of operation. We explore the possibility of combining two pre-
thermore, durability of a SHS and its effectiveness at high Reynolds viously studied frictional drag reduction (DR) techniques, namely,
numbers are also still a topic of research. We propose to combine a ALDR and SHS DR.
ALDR is based on the general concept of air lubrication, pro-
posed as early as during the 19th century (Latorre 1997). Other
related air lubrication techniques are bubble DR (Madavan et al.
Manuscript received by SNAME headquarters February 4, 2019; accepted 1985; Kodama et al. 2000) and partial cavity DR (Butuzov et al.
March 12, 2019. 1999; Lay et al. 2010; Mäkiharju et al. 2013). ALDR has been
Corresponding author: Simo A. Mäkiharju, makiharju@berkeley.edu
This manuscript is from the 32nd Symposium on Naval Hydrodynamics
studied in the laboratory (Elbing et al. 2013) and during sea trials
presented in Hamburg, Germany, on August 6, 2018. (Mizokami et al. 2010) by numerous investigators, and its energy
Co-author serving as reviewer: Simo A. Mäkiharju, University of California, economics have been analyzed based on the reported laboratory
Berkeley, Berkeley, California. makiharju@berkeley.edu results (Mäkiharju et al. 2012). When air is injected beneath a

118 JUNE 2020 0022-4502/20/6402-0118$00.00/0 JOURNAL OF SHIP RESEARCH

horizontal plate at increasing rates, a transition in the topology can shown that a “water-repellent surface,” whose properties were not
take place from a bubbly flow to a continuous air layer (AL). The further described, could enable what appeared to be a continuous
first indication of bubble coalescence and persistent DR of >20% gas layer to persist with nominal thickness of .5–1.0 mm up to
defines the beginning of the transitional ALDR regime. In this speeds O (4 m/sec). However, continuity of the gas layer, the critical
region, the hull is covered by interwoven regions of bubbly flow and flux to maintain a stable continuous gas layer, and details of the
segments of AL. As the gas flux is further increased, the fraction of physical mechanisms enabling the water-repellent surface to sustain
the surface covered by fragmented AL increases, until finally a the gas layer (with flux yielding only a .5- or 1-mm nominal
continuous layer covers the entire surface, with frictional DR ex- thickness) were not further investigated. Nevertheless, this earlier
ceeding 80%. The gas flux required to transition from bubble to article reports on three separate experiments and presents a sig-
transitional to ALDR, along with the percentage DR versus air flux, nificant amount of tantalizing data, which seem to further support
is discussed in Elbing et al. (2013). As ALDR has been proven to be the present hypothesis.
technically feasible, its energy economics have also been consid- The unbroken AL on SHSs, albeit thinner than that required on
ered (Mäkiharju et al. 2012). Whereas gross reduction in propulsive conventional surfaces, will presumably still have a significantly
power requirements of approximately 20% has been reported, the higher local DR (Elbing et al. 2013) than transitional AL. This
net energy savings have typically been in the range of 4–8%, due to would reduce the air compressor power consumption, as well as the
significant energy input for the air compressors. size and cost of the compressor required, and, hence, lead to a
The use of a SHS as a passive DR method has gained much significant improvement over the net savings predicted in
attention for its potential to reduce the frictional drag in a wide Mäkiharju et al. (2012). As a combined SHS and ALDR technique
variety of applications, from ships to pipelines (Watanabe et al. may prove more economical than either technique alone, the ob-
1999). SHS DR is achieved owing to the SHS’s ability to trap a thin jectives of the present research was to find out whether the critical
layer (or pockets) of gas and enable apparent slip on the surface air flux required to achieve ALDR is modified by the presence of
(Tretheway & Meinhart 2002). However, for turbulent flows, a SHS SHS.
has been applied with varied success (Rothstein 2010). According
to recent work by Seo and Mani (2016), the main challenge for
2. Experimental approach
developing SHS-based DR that will work for turbulent flows is the
limitations on robustness of the air–water interface. 2.1. Model design
However, it is conceivable that surfaces such as those using the
Salvinia effect (Barthlott et al. 2010) may reduce entrainment from To explore the validity of the assumptions of combined SHS–
SHSs. In general, the DR is lost if the gas layer/pockets are removed ALDR performance, we gathered data using a 3.3-m-long flat plate
due to entrainment or by diffusion, if the gas–liquid interface is (Fig. 1), where the first 1.0 m is for boundary layer development,
immobilized because of dirt, if the coating itself is damaged, or due followed by a gas injection slot at the base of a backward-facing step
to a combination of these. In addition, DR may not be achieved if the (similar to Elbing et al. 2013), and the last 1.8 m is either a painted
surface contains a significant number of features protruding through surface or a SHS. The experiments were conducted at UC Ber-
the viscous sublayer (e.g., kþ > 5, where kþ is the ratio of the keley’s Richmond Field Station Towing Tank, which is 61 m long,
average surface roughness to the flow’s viscous length scale) (Gose 2.5 m wide, and 1.4 m deep. It has two carriages, manned and
et al. 2018). unmanned, that enable towing of models up to 1.7 and 10 m/sec,
On a full-scale ship, the AL used for ALDR is typically O (1 cm) respectively.
 Hence, the Reynolds number based on length
thick layer flowing along with the liquid. In comparison, the layer of UL
ReL ¼ range over which the experiments can be conducted
air trapped on a SHS is typically O (10 micron) thick and comprises ν
discontinuous air pockets attached to the surface. at this facility with a 3-m-long model extends up to ReL ¼ 5.1
106
Our hypothesis is that owing to the multiphase flow physics for the carriage used. For these initial experiments, the SHS was
involved, it may be possible to maintain a stable AL over a SHS achieved by application of the NeverWet commercial coating
with a lesser gas flux than over a conventional surface. Intriguingly, system, which is known to have limitations especially regarding
as discovered after present experiments, Fukuda et al. (2000) had durability as discussed in Wojdyla et al. (2017).

Fig. 1 Model with flow from the left to right, as indicated by the arrow. Dimensions are in meters. Gas is injected through a 3-mm-tall slot on a 6-mm-
tall backward-facing step. All lower surfaces, except for the boundary layer trip, are hydraulically smooth with k1 < 5

JUNE 2020 JOURNAL OF SHIP RESEARCH 119

Fig. 2 Assembled model with the leading edge on the left side of the picture. The vertical hydrofoils were installed to minimize vibrations resulting
from vortex shedding of the support struts

The leading edge was a 8:1 machined PVC ellipse to avoid flow Wilkerson modular filter for removal of oil, particulates, and
separation above and below the model. On the under side of the moisture contamination down to .5 microns before flow mea-
leading edge, a 5-cm-wide strip spanning the width of the model surement. The gas flow rate was measured with an Alicat M-series,
with randomly distributed 140 micron roughness elements was used 0-1000 SLPM flow meter (Alicat Scientific, Tucson, Arizona), with
to trip the boundary layer. The painted plate downstream of the manufacturer-specified accuracy of .2% of full scale þ.8% of
injector had an arithmetic mean roughness of 8  2 microns and reading. One control valve at the receiver and another control valve
the root mean square roughness was 10  2 microns. For SHS, the after the flow meter were used to adjust the amount of air injected
arithmetic mean roughness was <1 microns. Assuming a turbulent through the piping. The air was equally divided between six 1/4 ”
liquid boundary layer with a zero pressure gradient and using the NPT injection pipes via two separate round manifolds. A diagram of
formula of Monkewitz et al. (2007) to estimate boundary layer the compressed air piping and valves is shown in Fig. 3. The six
quantities and shear stress, the painted plate’s roughness corre- pipes threaded into a .9-m
.3-m
1.6-cm PVC injection plate
sponds to kþ < 1, such that the surface can be considered hy- machined with a cavity for even air flow through the injection slot at
draulically smooth. (Note: Were this not the case, as discussed in the backward-facing step. A .9-m
.3-m
.3-cm aluminum plate
Elbing et al. (2013), the roughness alone would increase the critical was screwed onto the injection plate and sealed with an O-ring to
gas flux to achieve ALDR.) prevent air leakage. The gas discharged from the back of the plate
The trailing edge of the model was a polyvinyl chloride (PVC) via a .8-m
.32-cm injection slot on a 6-mm backward-facing step, as
plate machined into a cone to fair the aft edge of the 1.3-cm-thick shown by Fig. 4. Albeit, for practical application, it should be noted
aluminum plate. The model was attached to the carriage using four
vertical supports that were clamped to the carriage frame. A Na-
tional Advisory Committee for Aeronautics (NACA) 0015 hy-
drofoil with a chord length of 25.4 cm was 3D printed with
polylactic acid (PLA) filament and painted to seal each hydrofoil
and prevent water intrusion. The hydrofoils were placed over the
vertical supports to avoid vortex-induced vibrations at the struts
(Fig. 2). Two-direction portable bubble levels were attached at four
locations on the test section and used to ensure the model was at zero
trim and list for the duration of the testing. The bottom of the
injection plate was maintained at .2 m (6 in) draft below the free
surface for all experiments. Fig. 3 Compressed air system diagram for the experimental setup
Compressed air was stored in a 227-L (60-gal) receiver and used. With strictly enforced symmetry at the pipe tee connection and
passed through a SMC 7- to 125-psi (SMC Corporation, Chiyoda, identical piping downstream of the tee connection, air was assumed to
Tokyo, Japan) nonrelieving pressure regulator with manufacturer- be uniformly distributed to the six injection pipes leading into the gas
specified accuracy of 2.5%. The compressed air passed through a injection plate

120 JUNE 2020 JOURNAL OF SHIP RESEARCH

Fig. 4 Diagram showing the route of compressed air through the injection plate. The dashed line indicates the air cavity machined from the PVC plate.
Gas was injected to the model via a 3-mm-tall injection slot on a 6-mm-tall backward-facing step. To contain air under the test surface, 10.2-cm PVC
strakes (not pictured) were placed along the port and starboard sides of the test surface, with the strakes extending 5.1 cm below the bottom of the test
plate. Drawing is not to scale

that gas would likely be injected through discrete holes (e.g., Adobe Photoshop image adjustment and measurement tools, the
Mäkiharju et al. 2017; Mäkiharju & Ceccio 2018), and in this case, static contact angle was recorded for each surface. For the painted
presence of SHS may further promote initial formation of ALDR. To aluminum plate, the contact angle measured 74 3° and the average
contain the air under and prevent it from leaking out the sides of the contact angle for the superhydrophobic (SH) coatings measured 156
test section, PVC strakes were installed on the side of the plate, 2°, as shown in Table 1. The experimental contact angle results were
extending 5.1 cm below the bottom of the test section. in close agreement with the coating company’s published contact
The carriage speed (i.e., flow speed in frame of the plate) was angle of 160–175°. Figure 5 shows the gradient image results on the
recorded based on the carriage wheel rotation rate using a Monarch painted non-SH aluminum plate and four different surface prepa-
ROS-W remote optical sensor (Monarch Instruments, Amherst, rations using the SHS coating. The results showed that the effec-
New Hampshire) through a F2A1X frequency to analog converter tiveness of the NeverWet commercial coating system is not
(Monarch Instruments, Amherst, New Hampshire), with manufacturer- measurably dependent on the roughness of the surface being applied
specified accuracy of .005%. All output signals were measured at 250 to, for surfaces examined. It should be noted that after testing, the
kHz with a National Instrument USB-6351 data acquisition board surface was found to remain hydrophobic, albeit degraded, in 5 days
(DAQ) (National Instruments, Austin, Texas) and recorded in Lab- of testing.
View. Each test run was conducted at the maximum carriage speed of
1.7 m/sec corresponding to ReL ¼ 5.1
106. Two underwater cameras Table 1 Static contact angle for the MIC-6 aluminum plate using
(GoPro 5, San Mateo, California) were used to visualize the AL from various surface preparation techniques after application of two base
directly below and at an oblique angle during the experiments. During coats and four top coats of NeverWet. All measurements pre-
the experiment, the water temperature in the towing tank was measured sumably had an uncertainty of ±3° based on repeated contact angle
and found to be constant at 19  1°C. measurements

Grit Direction of application Static contact angle (°)

2.2. SHS
Painted plate
To compare the effectiveness of the commercial NeverWet
coating, individual test areas of a second MIC-6 aluminum plate None N/A 74
(TCI Aluminum, Hayward, California) were sanded with varying
grits and directions to examine the resulting water-repelling SH plate
properties based on different initial plate surface roughnesses. 60 Lateral 159
The aluminum plate was cleaned and etched before sanding and 60 Orbital 157
application of the NeverWet base coat. Following the manufac- 150 Lateral 159
turer’s instructions, two base coats of the coating were applied to the 150 Orbital 157
test surfaces, followed by four additional top coats. The base and top 220 Lateral 155
coat layers were allowed to dry for a minimum of 30 minutes 220 Orbital 158
between applications. The final application was allowed to cure 400 Lateral 157
overnight before testing. 400 Orbital 158
Water droplets measuring .1 mL and colored with a red dye were 800 Orbital 156
1000 Orbital 158
deposited onto each test section. A Nikon DSLR (Nikon, Minato,
2000 Orbital 158
Tokyo, Japan) mounted on a fixed tripod with a 105-mm Nikon AF- None N/A 156
Nikkor macro lens was used to photograph the static bubbles. Using

JUNE 2020 JOURNAL OF SHIP RESEARCH 121

 machined to a depth of 3.2 mm (.125 in) that spanned the width of
the model and was filled with 3M Scotch-Weld DP-460 (3M,
Maplewood, Minnesota) nonconductive epoxy with a listed volume
resistivity of 2.4
1014 (Ohms/cm). Two 1.59-mm (.063-in) square
slots were then machined into the nonconductive potting epoxy
spanning the width of the model at an on-center distance of 4.0 mm
(.157 in) from each other. Each slot held a probe element consisting
of a 1.0-mm (.041-in)-diameter stainless steel wire bedded in AI
Technology Inc. EG8020 (AI Technology Inc., Princeton Junction,
New Jersey) conductive epoxy with an electrical resistivity of .0004
(Ohms/cm). New probe locations were maintained from the early
model design at 54.6 cm (21.5 in) aft of the air injection slot for the
bow location and a distance of 156.2 cm (61.5 in) aft of the injection
slot for the stern location.
The circuit design for the third configuration incorporated two
Stanford Research SR850 DSP lock-in amplifiers (Stanford Research
Systems, Sunnyvale, California), one for each probe, to allow for the
isolation of probe signals, reducing cross-talk and minimizing ex-
ternal noise. The signals were generated, as before, via a Tektronix
AFG1022 (Tektronix, Beaverton, Oregon) function generator,
producing a 1 VRMS sine wave at 5 kHz for the bow probe and 8 kHz
for the stern. Frequency selection was made to allow impedance to
Fig. 5 Contact angle of water droplets on the MIC-6 aluminum plate. dominate the measurements as the primary area of interest is higher
The top image shows a droplet on a painted non-SH aluminum plate. The air–water void fractions as described in Ceccio and George (1996).
lower four images are after two base coats and four top coats of the The lock-in amplifiers then read the voltage drop across the probe
NeverWet SHS coating. Top—painted; middle left—60 grit in the lateral wires and sent the signal to case ground with a minimal 10-Ohm
direction; middle right—150 grit in the orbital direction; bottom left—2000 resistance. The lock-in amplifier then outputs a DC voltage equiv-
grit in the orbital direction; bottom right—no surface preparation alent to 10 times the magnitude of Volts Alternating Current (VAC)
read across the probe, which is recorded in LabView using the DAQ.
2.3. Electrical impedance–based void fraction measurements In all three cases, the probe wires were connected to a shielded
wire pair using screw terminals located outside the flow under the
Electrical impedance–based measurement of the void fraction of model. Exposed wiring was waterproofed and sealed with silicone.
the AL was made using three different probe configurations. The Screw terminals or cable ties, taking all load forces imposed on the
first probe configuration was located at the bow of the test plate and signal wires during carriage running, avoided imparting any po-
consisted of parallel wire probes suspended off the model surface, tential motion or oscillation to the probe wires. The signal wires
matching one of the methods discussed briefly in Ceccio and were made as short as practical to reduce stray current and foil-
George (1996). This configuration was located at the bow, 56.4 cm shielded to minimize electromagnetic interference. The internal
(21.5 in) aft of the air injection slot. Two 304 stainless steel wires of drain wire, as with all instrumentation, was grounded to the building
.05 mm (.002 in) diameter were stretched across the model’s common ground and carried to the tow tank carriage by an umbilical
91.4 cm (36 in) span. Then, .8-mm-thick tabs were laser cut from cable bundle that moved alongside the carriage.
nonconductive Delrin sheets and installed across the model span at Calibration of the probes was necessary to relate the measured
even intervals to ensure the wires were equally spaced 4 mm apart signal to void fraction. The model was oriented such that the spanwise
and remained 2.3 mm off the model surface. length was perpendicular to the water surface and submerged in-
The stern electrical impedance wire probe configuration used an crementally by known distances. Voltage readings were taken from
electrode pair placed directly on the surface, reminiscent of the 0% to 100% submersion. The voltage drop measured was given ample
surface-embedded electrode configuration described in George time (minutes) to reach a steady state before recording, as the process
et al. (2000). It was located near the model’s stern, at a distance of of lowering the model introduced unwanted motion to the model and
156.2 cm (61.5 in) aft of the injection slot. In this case, two 304 fluid surface disturbance. To ensure the accuracy of calibration, the
stainless steel wires of .05 mm (.002 in) diameter were affixed voltage measurements were taken with both probes ON, bow OFF and
directly to the model surface using a double-sided adhesive strip stern ON, and bow ON stern OFF. Although the effect of one probe on
approximately 10 mm (.39 in) wide. The two spanwise electrodes the other was measurable (for the first two configurations, we did not
were separated 4 mm apart. have access to lock-in amplifiers as used in similar measurements by
A third probe configuration was used in later experiments and Elbing et al. [2008] and Mäkiharju et al. [2013]), it was found to be at
replaced the first two probes near the bow and stern at the same least two orders of magnitude lower than the variation in voltage
streamwise locations. In this article, we mainly discuss the results of readings, and as such was not significant. Modifications made to the
the third (improved) probe configuration and compare the results circuit in the third probe design resulted in a decreasing calibration
with those of the first two presented in Peifer et al. (2018). curve. The calibration curve was monotonic, allowing for easy cor-
The third probe design was adapted from a flush-mounted relation of voltage to void fraction as shown in Fig. 6.
electrode pair design as mentioned in Ceccio and George (1996). Probes were calibrated for both the nonconductive painted
The probe design consisted of a 12.7-mm (.5-in)-wide groove surface and the SHS in advance of data collection, and recalibrated

122 JUNE 2020 JOURNAL OF SHIP RESEARCH

Fig. 6 Calibration curve for the bow and stern impedance probes (with the third configuration) generated by submerging the model coated in the SHS
(rotated 90° to be port side down) a known distance in the water and measuring the VDC value at steady submersion. The blue filled circle indicates data
for the bow, whereas the red asterisk indicates data for the stern

after any changes to the model. During actual data collection, and lowest flow rate measured corresponding to a tAL of 1.0 mm, the AL
with the assumption of a uniform bubbly mixture underneath the on the SHS extends to nearly the entire 1.8-m length of the alu-
plate, the recorded voltage drop was correlated with the calibration minum plate from the injection slot. The conical perturbation on the
curve established for the surface type to estimate the void fraction AL surface, appearing more significant because of the shallow
at a given carriage velocity and air injection rate. angle of lighting, is due to the pressure perturbation originating
from the leading edges of the strakes. We can note that for the
painted surface at tAL ≈ 1 mm, the AL (just a partial cavity forming
3. Results off the backward-facing step) covers only approximately 20% of the
length of the plate, and is spread nearly uniform across the span of
3.1. Critical gas flux results the plate. For increasing gas flux rates, the SHS plate continues to
Using underwater cameras mounted below and at an oblique maintain a steady AL coverage through the entire plate, whereas
angle, videos of the flow under the plate were taken and compared at small patches of air–water interactions occur on the painted plate.
similar gas injection flux rates at the draft pressure for the SHS and Over the SHS, a patch-free AL forms over the entire plate when tAL
painted plates. To ensure a similar aspect ratio between the two ∼ 2.8 mm. On the painted plate, an AL forms when tAL ∼ 6.4 mm,
surfaces, each photo was captured as the injection slot aligned with which is similar to the value predicted of tAL ∼ 6.5  .5 mm by
the right side of the frame. Figure 7 displays the results for varying extrapolating based on previous studies summarized in Mäkiharju
nominal AL thickness, tAL, increasing from the top to bottom, with et al. (2012).
the painted surface on the left and the SHS on the right. The nominal
AL thickness was defined, as in Elbing et al. (2008), as
3.2. Void fraction results
Q
tAL ¼ : (1) Time-averaged void fraction data collected at the bow and stern
U‘ S
locations, 54.6 cm and 156.2 cm aft of the injection slot, re-
Here, Q is the volumetric flux at draft pressure, U‘ is the free-stream spectively, for the painted and SHS plates are shown in Fig. 8. The
velocity, and S is the model span. time-averaged data were collected when the carriage reached top
It is immediately clear, from Fig. 7, that the flux required to speed and the air injection reached a steady-state value. Early
form a stable AL at draft pressure is significantly less for a SHS impedance data collected were unreliable for the bow probe location
than a non-SHS hydraulically smooth painted surface. Near the and have been omitted. With the redesigned flush-mounted probe

JUNE 2020 JOURNAL OF SHIP RESEARCH 123

 model, the spanwise void fraction was estimated at both locations
by examining the area of each measurement location over the video
time frame, allowing for a better comparison with the void fraction
data collected. As expected, these void fraction visual estimations
increase as the gas flow rates increase. For the area void fraction
estimations, a snapshot of a 101.6-cm
91.4-cm area near the
trailing edge was evaluated as it passed over the camera. The area
image was formatted into a binary image and processed in Adobe
Photoshop to find the number of pixels for the area covered in air
versus water. The results of this processing are shown in Fig. 9 and
are compared with the improved electrical impedance measure-
ments at the stern location. The measurement uncertainty for all
visual observations was estimated to be 5%.

3.3. Discussion of scaling

Although a SHS enabling the formation of a continuous gas layer
at a reduced gas flux may intuitively make sense, it also seems
curious that properties of a surface covered by a nominally con-
tinuous macroscopic (O BL thick) gas layer could affect the critical
gas flux required to sustain the said layer (once it has formed).
Indeed, similarly, it is interesting that surface roughness that is
orders of magnitude less than the AL thickness would affect the
critical gas flux, as was shown in Elbing et al. (2013).
However, let us start by assuming that a thick continuous gas
layer has already been formed. The 1) instabilities at the interface
(Kelvin–Helmholtz) and 2) turbulent fluctuations of the liquid flow
can still be expected to cause the liquid to periodically come in
contact with the solid surface by pushing the liquid through the AL.
Hence, we will have attempts by the liquid to wet the surface, and
the frequency of such events will depend on both outer flow
quantities and the thickness of the gas layer. Once the liquid
contacts the surface, we may expect the surface’s roughness and SH
properties to control the rate of wetting and rate of receding. We also
expect these to be influenced by the hydrostatic pressure difference
related to the gas layer thickness. Adopting this simplified repre-
sentation of the flow leads us to search for a scaling for the critical
gas flux that is not only a function of the outer flow quantities but
also depends on surface tension, surface roughness, and properties
Fig. 7 Results of spanwise uniform gas injection on the painted alu-
minum surface (left) and SHS (right) at ReL ¼ 5.1
106 and with gas flux of the solid surface (which as a first approximation could be as-
corresponding to tAL ¼ 1.0 (A), 1.5 (B), 1.8 (C), 2.1 (D), 2.8 (E), 3.9 (F), 4.9 sumed to be represented by the contact angle). Although this is a
(G), 6.1 (H), 6.4 (I), and 7.5 mm (J) for images from the top to bottom. Each topic of the authors’ ongoing work, we could expect to find a
photo was taken when the injection slot was aligned with the right side of dependence on a dimensionless number capturing the ratio of in-
the image. The camera field of view is approximately 180 cm
59 cm for ertial forces to forces effecting rate at which a contact line recedes
each tAL, with the exception of (C), where the field of view is approxi- on the surface. Hence, such a scaling might include ratios of forces
mately 116 cm
46 cm. The flow is from the right to left related to
1) Perturbation: liquid’s density ρw and free stream velocity
U‘
configuration, the impedance data collected from both the bow and 2) Restoring: density ρw, gravity g, and AL thickness tAL
stern probes show a clear difference in the void fraction measured at 3) Helping or hindering AL restoration: a combination of
given nominal AL thicknesses for a SHS coating versus a painted cosine of the contact angle, cos θ, and surface tension, γ.
surface. A nominal AL of tAL > 3.8 mm shows a consistent re-
duction of gas flux requirement for a stable AL spanning the test These, through dimensional analysis, yield a group we could take as
surface for the SHS. The data collected for tAL < 3.8 mm were the Froude number, FrAL ¼ gtAL/U‘, the already dimensionless
limited; however, with the incorporation of lock-in amplifiers and cos θ, and a new group gγ/ρwU‘. The latter could be interpreted as
redesigned probes, it is clear that the flux requirement at tAL < the ratio of surface tension, gravity, and inertial forces squared, or
3.8 mm is consistently lower for a SHS than a hydraulically smooth interpreted as a combination of Weber and Froude numbers with a
painted surface. In addition to electrical impedance measurements, (to be determined) characteristic length scale, l, which cancels in the
image analysis of the spanwise and area void fraction was com- combined group. However, it is not clear presently if these truly are
pleted. Using an underwater camera looking directly underneath the the proper and sufficient quantities to consider, and one could

124 JUNE 2020 JOURNAL OF SHIP RESEARCH

Fig. 8 Bow (top) and stern (bottom) void fraction measurements for locations 56.4 cm and 156.2 cm aft of the injection slot, respectively, for varying
AL thicknesses. The blue filled circle corresponds to the visual observations of the void fraction on the painted plate, whereas the red filled circle
corresponds to the SH coated plate. The blue square box marker indicates the data collected via the electrical impedance probe on the painted plate
and the red square box indicates data collected on the SHS coated plate. The third configuration impedance probe data are included in both figures
marked by green asterisk. Because of difficulties in impedance measurements at the bow location on the painted plate, the data are omitted. The
uncertainty of visual coverage void fraction is estimated to be 5%

expect that surface roughness should also be explicitly included. In

addition, versions of Capillary number may also enter the scaling.
Rather than speculating further, we note that determining a proper
scaling for the critical gas flux for SHS–ALDR is a topic of the
authors’ ongoing work and will be further discussed in a subsequent
publication.

4. Conclusion

On the hydraulically smooth non-SHS painted MIC-6 aluminum

surface, the gas flux required to form an AL is in fair agreement with
what was expected from previous studies, with the minimum re-
quired nominal layer thickness tAL ∼ 6  1 mm. However, on a SHS
plate, a significantly lower gas flux, tAL ∼ 3  1 mm, sufficed to
achieve an AL that persisted for the length of the model. Hence,
these preliminary experiments, especially when taken together with
the results of Fukuda et al. (2000), appear to confirm the authors’
hypothesis that a stable AL, presumably leading to an ALDR similar
to that reported in Elbing et al. (2013), can be formed on a SHS at a
significantly reduced gas flux. Thus, SHS–ALDR may offer a DR
Fig. 9 Area void fraction calculation based on image processing of a technique with improved net energy saving potential and reduced
101.6-cm
91.4-cm area near the trailing edge of the model for the same
compressor and piping size requirements, while also reducing the
tAL described in Fig. 7. The blue filled circle corresponds to the visual
observations of the void fraction on the painted plate, whereas the red
payback period on a retrofitted existing vessel. However, further
filled circle corresponds to the SH coated plate. New electrical imped- study is needed at higher Reynolds numbers, on longer models, and
ance measurements from the improved probe at the stern location, using more advanced instrumentation than available during the
shown as green asterisk, are compared with the existing visual area void present study. Hence, the authors plan to significantly expand on the
fraction data work reported.

JUNE 2020 JOURNAL OF SHIP RESEARCH 125

 Acknowledgments KODAMA, Y., KAKUGAWA, A., TAKAHASHI, T., AND KAWASHIMA, H. 2000 Experi-
mental study on microbubbles and their applicability to ships for skin friction
The authors would like to thank additional students who assisted reduction, International Journal of Heat and Fluid Flow, 21(5), 582–588.
LATORRE, R. 1997 Ship hull drag reduction using bottom air injection, Ocean
on the project, especially Anoushka Giri, who contributed signif- Engineering, 24(2), 161–175.
icantly to the model construction, and Anand Sagar, who assisted in LAY, K. A., YAKUSHIJI, R., MÄKIHARJU, S., PERLIN, M., AND CECCIO, S. L. 2010 Partial
running numerous towing tank tests. In addition, Daniel Grieb and cavity drag reduction at high Reynolds numbers, Journal of Ship Research, 54(2),
Monica Li provided useful contributions to improve the final model 109–119.
MADAVAN, N. K., DEUTSCH, S., AND MERKLE, C. L. 1985 Measurements of local
design and data collection and processing. Others who have con- skin friction in a microbubble modified turbulent boundary layer, Journal
tributed to this research include Brett Bussell, Felipe Cuellar, of Fluid Mechanics, 156, 237–256.
Ramtin Hosseini, Tai Nguyen, and Shoba Varma. MÄKIHARJU, S. A. AND CECCIO, S. L. 2018 On multi-point gas injection to form
an air layer for frictional drag reduction, Ocean Engineering, 147, 206–214.
MÄKIHARJU, S. A., ELBING, B. R., WIGGINS, A., SCHINASI, S., VANDEN-BROECK,
J.-M., DOWLING, D. R., PERLIN, M., AND CECCIO, S. L. 2013 On the scaling of
References air entrainment from a ventilated partial cavity, Journal of Fluid Mechanics,
732, 47–76.
BARTHLOTT, W., SCHIMMEL, T., WIERSCH, S., KOCH, K., BREDE, M., BARC- MÄKIHARJU, S. A., LEE, I. R., GRZEGORZ, P. F., MAKI, K. J., AND CECCIO, S. L.
ZEWSKI, M., WALHEIM , S., WEIS, A., K ALTENMAIER, A., LEDER, A., AND BOHN, 2017 The topology of gas jets injected beneath a surface and subject to liquid
H. F. 2010 The salvinia paradox: Superhydrophobic surfaces with hy- cross-flow, Journal of Fluid Mechanics, 818, 141–183.
drophilic pins for air retention under water, Advanced Materials, 22(21), MÄKIHARJU, S. A., PERLIN, M., AND CECCIO, S. L. 2012 On the energy economics
2325–2328. of air lubrication drag reduction, International Journal of Naval Architecture
BUTUZOV, A., SVERCHKOV, A., POUSTOSHNY, A., AND CHALOV, S. 1999 State of and Ocean Engineering, 4(4), 412–422.
art in investigations and development for the ship on the air cavities, Pro- MIZOKAMI, S., KAWAKITA, C., KODAN, Y., TAKANO, S., HIGASA, S., AND SHIGENAGA,
ceedings, International Workshop on Ship Hydrodynamics, Wuhan, China, R. 2010 Experimental study of air lubrication method and verification of
1–14. effects on actual hull by means of sea trial, Mitsubishi Heavy Industries
CECCIO, S. L. AND GEORGE, D. L. 1996 A review of electrical impedance Technical Review, 473(3), 41–47.
techniques for the measurement of multiphase flows, Journal of Fluids MONKEWITZ, P. A., CHAUHAN, K. A., AND NAGIB, H. M. 2007 Self-consistent
Engineering, 118, 391–399. high-Reynolds-number asymptotics for zero-pressure-gradient turbulent
ELBING, B. R., MÄKIHARJU, S., WIGGINS, A., DOWLING, D. R., PERLIN, M., AND boundary layers, Physics of Fluids, 19(11), 115101.
CECCIO, S. L. 2013 On the scaling of air layer drag reduction, Journal PEIFER, B. C., CALLAHAN-DUDLEY, C., AND MÄKIHARJU, S. A. 2018 Air layer on
of Fluid Mechanics, 717, 484–513. superhydrophobic surface for frictional drag reduction, Proceedings, 32nd
ELBING, B. R., WINKEL, E. S., LAY, K. A., CECCIO, S. L., DOWLING, D. R., AND Symposium on Naval Hydrodynamics, August 5–10, Hamburg, Germany.
PERLIN, M. 2008 Bubble-induced skin-friction drag reduction and the abrupt ROTHSTEIN, J. P. 2010 Slip on superhydrophobic surfaces, Annual Review of
transition to air-layer drag reduction, Journal of Fluid Mechanics, 612, Fluid Mechanics, 42, 89–109.
201–236. SEO, J. AND MANI, A. 2016 On the scaling of the slip velocity in turbulent flows
FUKUDA, K., TOKUNAGA, J., NOBUNAGA, T., NAKATANI, T., IWASAKI, T., AND KUNITAKE, over superhydrophobic surfaces, Physics of Fluids, 28(2), 025110-1–12.
Y. 2000 Frictional drag reduction with air lubricant over a super-water-repellent TRETHEWAY, D. C. AND MEINHART, C. D. 2002 Apparent fluid slip at hydro-
surface, Journal of Marine Science and Technology, 5(3), 123–130. phobic microchannel walls, Physics of Fluids, 14(3), L9–L12.
GEORGE, D. L., IYER, C. O., AND CECCIO, S. L. 2000 Measurement of the bubbly WATANABE, K., UDAGAWA, Y., AND UDAGAWA, H. 1999 Drag reduction of
flow beneath partial attached cavities using electrical impedance probes, Newtonian fluid in a circular pipe with a highly water-repellent wall, Journal
Journal of Fluids Engineering, 122, 151–155. of Fluid Mechanics, 381, 225–238.
GOSE, J., GOLOVIN, K., BOBAN, M., MABRY, J., TUTEJA, A., PERLIN, M., AND CECCIO, WOJDYLA, A., TAYLOR, A., DURAND, G., AND BOYD, I. W. 2017 New assessment
S. 2018 Characterization of superhydrophobic surfaces for drag reduction in criteria for durability evaluation of highly repellent surfaces, Wear, 390–391,
turbulent flow, Journal of Fluid Mechanics, 845, 560–580. 49–60.

126 JUNE 2020 JOURNAL OF SHIP RESEARCH