1. Introduction
Drag reduction (DR) in fluid dynamics is a critical focus of both aerodynamics and hydrodynamics, with broad applications from improving energy efficiency in transportation systems to advancing environmentally sustainable technologies. In this context, hydrophobic surfaces (HSs) have garnered significant attention over the past two decades as an effective passive flow control strategy for reducing drag in turbulent flows, particularly in underwater and surface vehicles (Rothstein Reference Rothstein2010; Temprano-Coleto et al. Reference Temprano-Coleto, Smith, Peaudecerf, Landel, Gibou and Luzzatto-Fegiz2023; Yu et al. Reference Yu, McClelland, del Campo Melchor, Lee and Lee2023; Liu et al. Reference Liu, You, Cao, Xu, Yang, Li, Lv, Sun and Duan2024; Xie et al. Reference Xie, Fairhall and García-Mayoral2024). Inspired by the natural examples, such as the lotus and salvinia, these surfaces are typically fabricated by combining microstructures and nanostructures with a low surface energy coating. These structures create an air–water slip interface, known as the plastron, which separates the wall from the water and serves as a key mechanism for underwater DR. Under ideal conditions, HSs have demonstrated DRs of up to 75 % (Park et al. Reference Park, Sun and Kim2014). However, the performance of these surfaces is highly sensitive to the integrity of the plastron. Once the plastron is compromised, the DR dissipates and may even lead to increased drag. The effectiveness of HSs in DR is closely related to the coverage and geometric characteristics of the plastron.
In laminar flow, without the significant influence of high shear entrainment and turbulent fluctuations, it has been experimentally confirmed that the plastron and the associated DR can be consistently maintained (Lee, Choi & Kim Reference Lee, Choi and Kim2016). This stability of DR in laminar flow is attributed to the uniform coverage of the air–water interface on HSs. Under such conditions, numerous experimental studies (Ou, Perot & Rothstein Reference Ou, Perot and Rothstein2004; Lee et al. Reference Lee, Choi and Kim2008) have demonstrated strong agreement between measured DRs and predictions from analytical models and numerical simulations that incorporate slip or partial-slip boundary conditions to represent the air–water interface. For HSs with certain structures, the DR in laminar flow can be well estimated by the slip length or the slip velocity as a function of the gas friction and the structural pitch (Lauga & Stone Reference Lauga and Stone2003). It has also been revealed that the slip length is sensitive to the deformation or displacement of the plastron under external influences such as static pressure, shear stress (Davis & Lauga Reference Davis and Lauga2009; Teo & Khoo Reference Teo and Khoo2010) and surfactant contamination (Tomlinson et al. Reference Tomlinson, Gibou, Luzzatto-Fegiz, Temprano-Coleto, Jensen and Landel2023). Specifically, deformation of the plastron reduces the effective slip length, thereby diminishing DR. These findings have significantly advanced our understanding of DR characteristics and mechanisms of the HSs in laminar flows, where their performance can be reliably estimated for surfaces with well-defined microstructures.
However, the majority of practical applications of HSs are faced with turbulent flow conditions other than the laminar flow. Despite numerous laboratory studies have demonstrated successful DR using superhydrophobic surfaces (SHSs), translating this technology to underwater vehicles operating at turbulent flows has proven challenging. The primary obstacle is the maintenance of the plastron under turbulent flow conditions, as well as the incomplete understanding of the turbulent DR characteristics and mechanisms of HSs. In turbulent flows, the increased shear stress (Yu et al. Reference Yu, McClelland, del Campo Melchor, Lee and Lee2023) and pressure fluctuations (Aljallis et al. Reference Aljallis, Sarshar, Datla, Sikka, Jones and Choi2013; Seo, Garcia-Mayoral & Mani Reference Seo, Garcia-Mayoral and Mani2015) induced by turbulence can lead to the deformation, displacement or even depletion of the plastron. These effects severely degrade the DR capabilities of HSs, sometimes resulting in increased drag (Aljallis et al. Reference Aljallis, Sarshar, Datla, Sikka, Jones and Choi2013; Gose et al. Reference Gose, Golovin, Boban, Mabry, Tuteja, Perlin and Ceccio2018). Up to now, there has been a discrepancy between the experimental and numerical results and no consensus has been reached. (Park, Choi & Kim Reference Park, Choi and Kim2021). Numerous numerical studies (Park, Park & Kim Reference Park, Park and Kim2013; Rastegari & Akhavan Reference Rastegari and Akhavan2015; Seo & Mani Reference Seo and Mani2016; Alamé & Mahesh Reference Alamé and Mahesh2019; Safari et al. Reference Safari, Saidi, Salavatidezfouli and Yao2023) have consistently reported that a substantial turbulent DR could be achieved with HSs and SHSs. For example, DRs up to 88.5 % have been reported in direct numerical simulations (DNS) with HSs featuring a void fraction of 63/64 at a friction Reynolds number (Re τ ) of 442 (Rastegari & Akhavan Reference Rastegari and Akhavan2019). These studies generally indicate that DR increases with the void fraction created by the microstructures. However, due to computational limitations, most DNS studies are performed at Re τ values below 1000, which are significantly lower than those encountered in practical applications. Additionally, numerical models often treat the plastron as a static, undeformed air–water interface using idealized slip or partial-slip boundary conditions. These assumptions diverge from experimental observations, where the plastron is often deformed or compromised under turbulent flow conditions. As such, trends observed in numerical studies have not yet to be verified by experimental data.
As for the turbulent experiments, to the best of the authors’ knowledge, experimental results of HSs both in turbulent channel flow and boundary layer flow show no clear trend. Some experiments (Daniello, Waterhouse & Rothstein Reference Daniello, Waterhouse and Rothstein2009; Park, Sun & Kim Reference Park, Sun and Kim2014; Li et al. Reference Li, Ji, Tan, Li, Xiang, Lv and Duan2020; Xu, Yu & Kim Reference Xu, Yu and Kim2021) reported DR percentages comparable to those from DNS, while others (Aljallis et al. Reference Aljallis, Sarshar, Datla, Sikka, Jones and Choi2013; Gose et al. Reference Gose, Golovin, Boban, Mabry, Tuteja, Perlin and Ceccio2018) observed negligible DR or even drag increases. Moreover, the relationship between DR and Reynolds number in experiments remains unclear. Some studies report a decrease in DR with increasing Reynolds number (Aljallis et al. Reference Aljallis, Sarshar, Datla, Sikka, Jones and Choi2013; Ling et al. Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016), while others observe an increase (Zhang et al. Reference Zhang, Tian, Yao, Hao and Jiang2015; Van Buren & Smits Reference Van Buren and Smits2017) or a non-monotonic trend where DR first increases and then decreases (Xu et al. Reference Xu, Yu and Kim2021). Though most of these previous experiments do not quantitively check or show the plastron coverage of the HSs during turbulent experiments, discrepancies across experimental studies are speculated to be due to variations in the coverage and shape of plastron among different experiments. For example, Xu et al. (Reference Xu, Yu and Kim2021) demonstrated that by manually eliminating the negative effects of partial wetted portions of HSs, DR could monotonically increase with Reτ (from 1858 to 5667). Moreover, as quantitatively evaluated by Ling et al. (Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016), the deterioration or depletion of the plastron can induce roughness effects, especially for the random HSs. Even a small increase in the dimensionless surface roughness (i.e. k + = 1.71) can significantly degrade DR performance, further complicating the interpretation of results.
Thus, the inconsistent experimental findings, largely attributed to the deformation, partial loss or complete failure of the plastron, have led to a fragmented understanding of turbulent DR characteristics and mechanisms over HSs. To date, the characteristics and mechanisms of turbulent DR over HSs remain unclear. A key limitation has been the inability to accurately monitor and actively restore the plastron during turbulent flow tests, yet few experiments have systematically controlled these parameters. To address these challenges, the objective of this study is to evaluate the turbulent DR characteristics and mechanisms of HSs under controlled plastron conditions. Specifically, we employ air-fed hydrophobic surfaces (AFHS) with longitudinal grooves, enabling active maintenance and reversible restoration of the plastron during turbulent flow tests. The experimental set-up combines wall shear stress sensor, planar particle image velocimetry (PIV) and interfacial visualization techniques to simultaneously measure the skin friction drag, turbulent boundary layer (TBL) velocity profiles and the plastron coverage. These methodologies allow for a comprehensive investigation of the relationship between plastron integrity, turbulent flow characteristics and DR, providing valuable insights into the turbulent DR mechanisms of HSs with controlled plastron coverage and shape. The rest of the paper is structured as follows: § 2 describes the experimental set-up and methodologies; § 3 presents the results, including plastron sustainability, skin friction measurements, and analysis of DR mechanisms; § 4 summarizes the conclusions.
2. Experimental set-up and methods
2.1. Air-fed HSs with longitudinal grooves
In this experiment, the conventional HS and the AFHS with longitudinal grooves are fabricated and tested. The schematics of the HS and AFHS are presented in figure 1. The primary distinction between the HS and AFHS lies in their substrates: the HS features a flat plate substrate, whereas the AFHS incorporates an array of air-fed channels embedded within its substrate. By connecting this substrate to the air chamber, the air in microgrooves of the AFHS, i.e. the plastron, can be actively supplied and refreshed. To precisely control the plastron, the pressure difference between the air chamber and the test section is monitored and maintained using a pressure difference transducer and a high-precision pressure regulator, ensuring stable and controlled plastron conditions during testing. The longitudinal microgrooves on the HSs are manufactured using femtosecond laser micromachining (Universal laser systems PLS6MW) by adhering a layer of polytetrafluoroethylene (PTFE) film onto a smooth aluminium substrate (6061-T6). The AFHSs are fabricated through a two-step process. First, air-fed channels were created on the aluminium substrate using microelectromechanical systems techniques. Subsequently, a PTFE film was adhered to the substrate, and longitudinal grooves with the desired dimensions were machined into the PTFE film using femtosecond laser micromachining. Figure 2 provides a detailed view of the AFHS, along with the corresponding three-dimensional (3-D) microscopic morphology of the grooves measured using light scanning interferometry (Veeco NT9100). The nominal dimensions of both the HSs and AFHSs are 50 mm in width and 100 mm in length. As shown in figure 2(a), the designed air-fed channel pairs span the entire width of the sample, ensuring uniform air supply while maintaining sufficient structural strength. The 3-D surface profiles of the grooves, obtained through light scanning interferometry, indicate that the overall dimensions of the grooves closely match their design specifications. However, small bumps are observed along the edges of the grooves, likely caused by the thermal effects of femtosecond laser processing. These submicron and nanoscale bumps are deemed too small to significantly impact drag performance or the flow field over the surfaces.
Schematic of (a) conventional HS and (b) AFHS with longitudinal grooves.

(a) The picture of the AFHS and (b) the 3-D microscopic morphology of the grooves measured by the light scanning interferometry. The nominal dimensions of the AFHS as well as the HSs are 50 mm in width and 100 mm in length.

Table 1 summarizes the dimensions and contact angles of the tested samples, including the smooth surface, the HS and the AFHS with air void fraction (φ
groove
) varying from 0.2 to 0.4. Since longitudinal grooves were used in the hydrophobic surface, both the apparent streamwise and spanwise contact angles are measured using an optical contact angle meter (OCA 20, Dataphysics). As shown in table 1, the streamwise contact angles of the surfaces with longitudinal grooves are slightly larger than the spanwise contact angles. The smooth surface exhibits a contact angle of 109.3 ± 0.5°, whereas the HSs demonstrate significantly higher contact angles, approximately 150°. Within the current range of the φ
groove
, a high fraction of air contributes to achieving a high apparent contact angle, and no significant differences in contact angles are observed between surfaces with varying φ
groove
. Within the tested turbulent Reynolds numbers, the wall unit (
$l_{v}$
= ν/uτ
, uτ
is the wall friction velocity, and ν is the kinematic viscosity) varies from 13.23 to 8.02 μm, and the corresponding normalized structural pitch (W
+
= W/
$l_{v}$
) ranges from 9.07 to 14.96.
Dimensions and contact angles of the tested samples. Within the tested turbulent Reynolds numbers, the wall unit (
$l_{v}$
) varies from 13.23 to 8.02 μm and the corresponding normalized structural pitch (W
+
) ranges from 9.07 to 14.96.

2.2. Experimental facility and measurements
Figure 3 illustrates the water tunnel and the corresponding test section used in the experiments. The experiments were conducted in the water tunnel at the Marine Design and Research Institute of China (MARIC), which was originally designed and manufactured by the West Japan Fluid Engineering Laboratory. The dimensions of the rectangular test section are 1.0 m × 0.12 m × 0.18 m (length, height and width) with Plexiglass tunnel walls for flow visualization. The operating speed of the tunnel is adjustable between 0 and 3.5 m s−1, with a low incoming turbulence level.
Schematics of (a) the water tunnel and measurement systems, (b) the test section of the water tunnel, the assembled shear stress sensor and the tested sample. The dimensions of the water tunnel rectangular test section are 1.0 m × 0.12 m × 0.18 m (length, height and width) and the operating speed of the tunnel is adjustable between 0–3.5 m s−1. The tested sample, embedded in the assembled shear stress sensor, is mounted on the ceiling of the test section, with a distance of 682.5 mm between the turbulence generator and the centre of the sample.

As shown in figure 3(b), turbulence generator and particle injectors are positioned upstream of the sample to control the TBL and visualize its profiles. The tested sample, embedded in the assembled shear stress sensor, is mounted on the ceiling of the test section with a distance of 682.5 mm between the turbulence generator and the centre of the sample. Static pressures at the sample location and inside the air chamber are measured using pressure transducers (Meacon, MIK-P3000), while the corresponding pressure difference is recorded with a pressure difference transducer (Meacon, MIK-P2051). The pressure transducers are all purged and calibrated before recording data for each experimental condition, ensuring accurate measurements. The calibration of the pressure transducers is linear, with an R-squared value of approximately 0.99, and the uncertainty in pressure and pressure difference measurements is less than 0.075 %, as specified by the vendor. Meanwhile, a dissolved oxygen meter (Meacon, MIK-DM3000) is used to monitor the dissolved oxygen levels in the water tunnel throughout the experiment to ensure consistent test conditions.
As the AFHS incorporates an array of air-fed channels embedded within its substrate, the air in the microgrooves of the AFHS can be actively supplied and refreshed by connecting this substrate to the air chamber. To precisely control the plastron, a high-precision pressure regulator (Bellofram, TYPE 70) is employed together with the pressure difference transducer (Meacon, MIK-P2051) to control the pressure difference (ΔP) between the test section and the air chamber. The pressure difference between the air chamber and the test section can be monitored and maintained by the feedback control using the pressure difference transducer and the pressure regulator. In this way, the pressure inside the grooves counters external static pressure. Meanwhile, by monitoring the pressure difference transducer and simultaneously adjusting the pressure regulator, the plastron shape can also be regulated between concave, flat and convex with the ΔP gradually increasing from negative to positive.
(a) Schematic and (b) photograph of the assembled shear stress sensor and the tested sample for skin friction drag measurement. The sensor comprises a mounting base, air chamber, spring element, optical encoder and optical scale. The tested sample is mounted on the sensor, aligned flush with the front and rear walls, while maintaining a 50 μm gap around it to ensure proper isolation and minimize interference.

Calibration of shear stress sensor for skin friction drag measurement. (a) The comparison between the measured skin friction drags and the semiempirical equation (Schlichting & Gersten Reference Schlichting and Gersten2016). (b) The comparison of the measured skin friction coefficient against theoretical predictions across different Reynolds numbers. The calibration process involved repeated measurements of skin friction drag over a smooth plate under different incoming flow velocities.

To simultaneously measure the skin friction drag, the plastron and the wall-bounded turbulent, an in-house-designed shear stress sensor, macroscopic and microscopic interfacial visualization systems and PIV are utilized. Figure 4 shows the schematic and a photograph of the assembled shear stress sensor along with the tested sample. This high-sensitivity shear stress sensor is specifically designed and fabricated in-house to measure skin friction drag over the samples. The design of this sensor is initially based on the work in the literature (Xu et al. Reference Xu, Arihara, Tong, Yu, Ujiie and Kim2020), and is further enhanced to improve the stiffness in both the spanwise and gravitational directions, as well as to integrate with the air supply system. The sensor primarily comprises a mounting base, air chamber, spring element, optical encoder and optical scale. The tested sample is mounted on the sensor, aligned flush with the front and rear walls, while maintaining a 50 μm gap around it to ensure proper isolation and minimize interference. When a streamwise force is exerted on the surface, the spring beam undergoes displacement, which is detected by the optical encoder. The corresponding force is then calculated using the calibrated stiffness of the sensor. Figure 5 shows the in situ calibrated results of the shear stress sensor. To validate its accuracy, the sensor was used to measure the skin friction drag over a smooth plate at various incoming flow velocities. Measurements were repeated three times, and all results aligned well with the semiempirical equation (Schlichting & Gersten Reference Schlichting and Gersten2016), confirming the sensor’s reliability. Further comparison of skin friction coefficients at different Reynolds numbers is presented in figure 5(b). In laminar flow, the measurement uncertainty is relatively high due to the small magnitude of the skin friction drag. However, in turbulent flow, the experimental data closely match the theoretical predictions, demonstrating that the shear stress sensor can accurately measure skin friction drag over the tested samples under turbulent flow conditions.
Figure 6 illustrates the macroscopic and microscopic interfacial visualization systems used for monitoring the plastron on the AFHS during the test. For macroscopic visualization, a single-lens reflex (SLR) camera (Nikon, D7500) is employed to quantitatively characterize the overall coverage of the plastron by leveraging the reflection contrast between the air–water interface and the solid wall. For microscopic visualization, a high-resolution Grasshopper3 camera (Flir, GS3-U3-51S5M, 2448 × 2048 pixels, 3.45 μm pixel–1) is used in conjunction with a Cassegrain optical telescope system (Seika, Maksutov–Cassegrain system) to capture the detailed shape of the plastron. As shown in figure 7, a shiny interface is observed in the macroscopic images when the tested sample is covered by a flat plastron, whereas the interface appears dark when the plastron exhibits a concave or convex curvature. For microscopic visualization, as shown in figure 7, the behaviour is reversed. When the plastron is flat, the microscopic image appears dark. Conversely, when the plastron interface becomes concave or convex, a bright reflective region is observed on the right or left side of the groove, respectively. By combining macroscopic and microscopic interfacial visualization, the overall coverage and the interfacial shape of the plastron (flat, concave or convex) can be quantitatively characterized throughout the tests. This dual visualization approach provides valuable insights into the interfacial dynamics and structural behaviour of the plastron under various flow conditions.
(a) Schematic and (b) actual set-up of the macroscopic and microscopic interfacial visualization systems. For macroscopic visualization, a collimated light source and a SLR camera (Nikon D7500, denoted as Camera 1 in the image) are utilized to quantitatively characterize the overall plastron coverage. This was achieved by measuring the reflection contrast between the air–water interface and the solid wall. A high resolution Grasshopper3 camera (Flir, GS3-U3-51S5M) is integrated with a 45° reflector and a Cassegrain optical telescope system (Seika, Maksutov–Cassegrain system) to capture the fine details of the plastron’s shape.

The overall and local views of the plastron captured by the macroscopic and microscopic interfacial reflection systems. (a) The flat plastron on the microgroove appears shiny in the macroscopic image but exhibits a dark groove in the microscopic image. (b) The concave plastron demonstrate a dark macroscopic image which reveals a bright reflective region on the right side of the groove in the microscopic image. (c) The convex plastron shows a dark macroscopic image but displays a bright reflective region on the left side of the groove in the microscopic image.

The planar PIV system is also applied for the TBL velocity profile measurement. Figure 8 shows the planar PIV system used in the experiment as well as the velocity contour measured by the planar PIV system. A high-frequency Nd–YAG Laser (CNI, AO-X-532, 532 nm) and a high-resolution charge coupled device (CCD) camera (ImperX, ICL-B3340, 3312 × 2488 pixels, 5.5 μm pixel–1) equipped with a microlens (Nikon, Micro-Nikkor 60 mm f/2.8G ED) are employed. The field of view (FOV) of the camera is centred on the test sample, with a physical streamwise length of 16.15 mm and a wall-normal height of 12.14 mm. The thickness of the laser sheet within the FOV is maintained at approximately 1 mm. The frame-straddling method is applied by synchronously triggering the laser and the CCD camera using a high-frequency synchronizer (Siglent Ltd, SGD1032X). For each experimental case, more than 300 image pairs are recorded, with the time interval between two adjacent frames varying from 25 to 50 μs, depending on the incoming flow velocities. Polymethylmethacrylate microspheres with a diameter of approximately 10 µm are used as seeding particles. To enhance the particle density within the TBL, a particle spreader with 27 evenly distributed 0.15 mm holes, which is located 211 mm upstream of the tested sample, is applied to slowly seed particles using a microsyringe pump, following the methodology applied in a previous study (Ling et al. Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016).
(a) The planar PIV system for the TBL velocity profile measurement. A high-frequency Nd-YAG Laser (CNI, AO-X-532, 532 nm) and a high-resolution CCD camera (ImperX, ICL-B3340, 3312 × 2488 pixels, 5.5 μm pixel–1) equipped with a microlens (Nikon, Micro-Nikkor 60 mm f/2.8G ED) are employed. (b) The velocity contour measured by the planar PIV system at ReL = 1.85 × 106.

Figure 9(a) presents the wall-normal mean velocity profiles on a smooth wall as measured by the planar PIV. The data closely align with the classical laws of the TBL, capturing both the viscous sublayer and the log-law region. High-resolution near-wall velocity measurements are achieved, successfully resolving the viscous sublayer velocities for y
+
< 3. Figure 9(b) further shows the streamwise and wall-normal velocity fluctuations. The 〈u
′
u
′〉 profiles generally agree with those reported in the literature for similar Reτ
values (Ling et al. Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016). The peak values of 〈u
′
u
′〉 occur in the range of y
+
= 12–20 and increases with Reτ
in the log-law region. The peak values are slightly lower than those reported by Ling et al. (Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016) for the corresponding Reτ
. For 〈
$v^{\prime}v^{\prime}$
〉, the fluctuations gradually increase with y
+
before plateauing at approximately 2.0 around y
+
= 100, which is consistent with theoretical expectations. The experimental conditions of this study involve mean flow velocities in the test section of the water tunnel ranging from 1.86 to 3.35 m s–1, corresponding to Reynolds numbers based on length, ReL
, from 1.21 × 106 to 2.17 × 106. Based on the velocity profile measurements, the friction velocity (uτ
) is estimated to range from 0.0756 to 0.125 m s–1, with the corresponding friction Reynolds number (Reτ
) varying from 885 to 1723.
Calibration of PIV measurement. (a) The normalized mean velocity profiles, and (b) the streamwise and wall-normal velocity fluctuations of the smooth surface. The classical law of the wall is shown for comparison and the results of Ling et al. (Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016) measured by the in-line digital holographic microscopy at similar friction Reynolds numbers are also included.

3. Results and discussion
3.1. Plastron coverage and sustainability
Controlling the plastron shape and its coverage in turbulent flows has long been a challenge in achieving DR over HSs. In this experiment, as demonstrated in figure 7, dynamic control of the plastron curvature and coverage is achieved using microchannel pressure regulation within the AFHS, while the plastron curvature and coverage are quantitatively characterized using interfacial reflection systems. Figure 10 shows the plastron shape, specifically the curvature of the air–liquid interface, as captured by the microscopic interfacial reflection system. By monitoring the pressure difference transducer and regulating the high-precision pressure regulator (as mentioned in § 2.2), the pressure difference between the air chamber and the external flow can be precisely adjusted, transitioning from negative to positive pressure differences. As we can see in figure 10, within the region marked by two red dashed lines, the reflective shiny band shifts from the right side (marked as R in figure 10) to the left side of the groove (marked as L in figure 10). This transition signifies a change in the plastron shape, dynamically switching between concave, flat and convex configurations. That is the plastron curvature of the AFHS is dynamically controlled. Additionally, the corresponding tangential angle and penetration height of the plastron can be estimated using the analytical solution of the Young–Laplace equation.
Control of plastron shape on the AFHS by pressure regulation. From (a) to (e), with ΔP gradually increases from negative to positive, the plastron shape changes from concave to convex, i.e. in the region marked by two red dashed lines, a reflective shiny band shifts from one side to another side. The notations R and L indicate the right and left side of the groove. The
$\theta$
and
$h$
are the tangential angle and penetration height of the plastron estimated by the Young–Laplace equation, respectively.

Figure 11 further compares microscopic interfacial images of HS and AFHS under a range of ReL . For the HS, as can be seen in figure 11(a–e), the intensity of reflective shiny band increases with the increment of the ReL . This indicates that the plastron shape becomes increasingly concave as water gradually penetrates deeper into the grooves. In contrast, for the AFHS, the image intensity within the grooves remains nearly constant across the tested ReL range. This observation suggests that the plastron shape on the AFHS remains flat, irrespective of the ReL , due to the dynamic pressure regulation and real-time microscopic inspection. By maintaining a flat plastron during the tests, the influence of interfacial curvature on experimental outcomes is effectively eliminated. To the best of authors’ knowledge, this is the first instance in which the plastron shape on HSs has been dynamically controlled during a TBL DR experiment. For conventional HSs, controlling the plastron shape is nearly impossible, as the air–water interfacial curvature depends solely on the static pressure of the flow.
The comparison of microscopic interfacial images between HS (a)–(e) and AFHS ( f)–(j) under different ReL . Notations ‘R’ and ‘L’ indicate the right and left side of the groove. The red dashed line pairs are added for comparison of the reflective intensity.

To further evaluate plastron sustainability, macroscopic interfacial reflection images of HS and AFHS are compared across a range of ReL . As shown in figure 12, the overall plastron coverage is captured using macroscopic interfacial reflection imaging, and it is evident that the sustainability of the plastron on the AFHS is significantly better than that on the HS. As the ReL increases, part of the HS gradually darkens, indicating a transition from the Cassie–Baxter state to the Wenzel state. In contrast, the AFHS remains relatively unchanged, maintaining the Cassie–Baxter state. The overall plastron coverage (φ) is quantitatively assessed through image processing and extraction in figure 13. The image processing and extraction are processed by an in-house MATLAB code, incorporating procedures such as thresholding, median filtering and area calculation. A watershed segmentation algorithm, in combination with edge detection function, is employed to perform binary thresholding and extract the plastron-covered area. The macroscopic images captured under calm water conditions are used as the reference for determining the full plastron coverage. As shown in figure 13, in laminar flow, the difference in plastron coverage between the HS and AFHS is relatively small. However, in turbulent flow, the plastron coverage on the HS decreases significantly, whereas the coverage on the AFHS remains relatively unchanged as ReL increases. This demonstrates that the AFHS enhances plastron sustainability and coverage in turbulent flow conditions. Thus, by controlling the plastron shape and enhancing plastron sustainability, it is possible to reveal the DR and turbulence statistics in TBL over HSs.
Comparison of macroscopic plastron images of (a) HS and (b) AFHS across a range of ReL . The flow direction is from the top to the bottom. The bright region indicates that the tested sample is covered with a layer of plastron, and the dark region indicates the deterioration or depletion of the plastron.

Comparison of the calculated plastron coverage (φ) between HS and AFHS at different ReL . The overall coverage of the plastron is quantitively extracted from the macroscopic plastron images by image processing and extraction methods.

3.2. Skin friction drag and DR
Time traces of skin-friction drag over the smooth plate, the HS and AFHS, measured by the shear stress sensor are presented in figure 14. The results clearly indicate that the skin-friction drag of the HS and AFHS is significantly smaller than that of the smooth plate, demonstrating the drag-reduction effect induced by hydrophobicity. Notably, the AFHS exhibits a lower skin-friction drag compared with the HS. This reduction is attributed to the higher plastron coverage of the AFHS, which is maintained through controlled air supplementation. By regulating both the coverage and curvature of the plastron, the AFHS achieves superior drag-reduction performance compared with the HS. Although variations of skin-friction drag are observed in figure 14, the mean values remained stable throughout the experiment.
Time traces of skin-friction drag measured by the shear stress sensor over the smooth plate, the conventional HS and the AFHS at ReL = 2.17 × 106.

The DR can be further calculated based on the mean value of skin friction drag. In figure 15, the DR of HS and AFHS from ReL = 1.21 × 106 to ReL = 2.17 × 106 are summarized. As we can see, DR of different HSs varies between 14.8 % to 35.8 %, demonstrating substantial drag-reduction performance. Though the DR is impressive, the DRs are slightly lower than the nominal designed air fraction (φ groove ) of the HSs (20–40 %). This aligns with previous findings (Daniello et al. Reference Daniello, Waterhouse and Rothstein2009; Rothstein Reference Rothstein2010; Park et al. Reference Park, Park and Kim2013), which show that DR often falls short of the designed air fraction, approaching a theoretical maximum of DR ∼ φ groove . At the same designed air fraction (i.e. φ groove = 0.4), the AFHS and AFHS-H exhibit higher DR than the HS, which is attributed the superior plastron coverage of the AFHS. Despite the same air fraction, the DR of the AFHS (H = 130 μm) is higher than that of the AFHS-H (H = 180 μm). Though it is not fully understood yet, the deeper groove may cause the plastron from one groove interacting with the nearby groove, leading to the coalescence and/or break from their contact point. As a result, the HSs with deep grooves hinder plastron sustainability, warranting further investigation into the influence of the groove depth on DR.
(a) The DR of the HS and AFHS at same designed air fraction (φ groove = 0.4) under different Re L. The HS (H = 130 μm) is the conventional hydrophobic surface, the AFHS (H = 130 μm) and AFHS-H (H = 180 μm) are the AFHS. (b) The DR of AFHS at the same groove depth but different designed air fraction (AFHS, φ groove = 0.4; AFHS-D1, φ groove = 0.3; AFHS-D2, ϕ groove = 0.2).

For all the AFHS, surfaces with higher designed air fractions generally exhibit larger percentages of DR. However, trends in DR with increasing ReL differ across surfaces with high air fractions (φ groove = 0.4) and those with lower air fractions (φ groove = 0.3 or 0.2). As we can see, DRs of HS, AFHS and AFHS-H (φ groove = 0.4) first slightly decrease and then increase as ReL rises. In contrast, for AFHS-D1 (φ groove = 0.3) and AFHS-D2 (φ groove = 0.2) DR consistently increases with ReL . For surfaces with φ groove = 0.4, the initial decrease in DR from ReL = 1.21 × 106 to 1.52 × 106 is attributed to a partial wetting transition. Despite the use of AFHS with longitudinal grooves to prevent plastron loss during tests at varying Reynolds numbers, small portions of the plastron may still disappear, as evidenced by interfacial visualization. This partial plastron loss can impact the drag-reduction performance at certain flow conditions.
The DR normalized by the plastron coverage (φ) versus the ReL . The fitted line is obtained by the least square fitting method based on data from different HSs tests.

In order to evaluate the DR performance of HSs with increasing Reynolds number while minimizing the influence of surface wetting transition, the DR normalized by the plastron coverage (φ) is calculated and presented in figure 16. This approach enables the relationship between DR and Reynolds number to be determined without interference from variations in plastron coverage, as both parameters were simultaneously measured. In figure 15, the DR and their trends with increasing ReL vary across different HSs. However, in figure 16, the normalized drag reduction (DR/φ) calculated from different HSs collapses onto a single trend. Within the tested ReL range, DR/φ generally increases monotonically with the Reynolds number. This indicates that if plastron curvature and coverage remain constant, the DR percentage of HSs increases as ReL rises. This result is consistent with the longitudinal grooved SHSs experiments in external flow (Xu et al. Reference Xu, Yu and Kim2021), where corrected DR data (accounting for non-wetted areas) similarly showed enhanced performance at higher Reynolds numbers. It is noteworthy that a similar trend of increasing DR with Reynolds number has also been reported in low-speed internal flow experiments on random SHSs, even though the coverage of the plastron was not explicitly measured (Zhang et al. Reference Zhang, Tian, Yao, Hao and Jiang2015). For the experiments conducted in this study and those of Xu et al. (Reference Xu, Yu and Kim2021), DRs at different Reynolds numbers are tested using fixed physical surface microtextures. As Reynolds numbers increase, the size of the surface microtextures in wall units also increases. According to the DNS study (Rastegari & Akhavan Reference Rastegari and Akhavan2019), in the absence of the wetting transition or roughness effects, the increase in microtexture size in wall units enhances the magnitude of DR. This effect is evident in surfaces like the AFHS-D1 and AFHS-D2. However, once the wetting transition happens, the increase in drag caused by the degradation of the plastron overshadows the drag-reduction enhancement caused by larger wall-unit-scale microtextures, leading to an overall drag increase. Therefore, by eliminating the influence of plastron coverage, curvature and surface roughness in this study, the results reveal that DR on HSs is indeed enhanced with increasing Reynolds numbers, which provides robust confirmation of DNS studies (Rastegari & Akhavan Reference Rastegari and Akhavan2019).
3.3. Turbulence statistics and DR mechanism
Figure 17 presents the mean velocity profiles, the viscous and Reynolds shear stress over the smooth wall and the AFHS at ReL = 1.85 × 106 (i.e. Reτ = 1427). Here the velocities are scaled by their own inner units, while the viscous and Reynolds shear stress are scaled by the total wall stress of the smooth wall for the quantitative comparison. As we can see, the velocity profile over the AFHS is noticeably higher than that of the smooth wall, indicating the presence of a significant slip velocity. As the longitudinal grooved AFHSs are applied together with controlled plastron coverage and shape, the upward shift of the velocity both in the viscous sublayer and log-law region is in correspondence. This observation aligns with prior findings from 3-D particle tracking velocimetry (PIV) measurements in low Reτ turbulent channel flow over random SHSs (Abu Rowin & Ghaemi Reference Abu Rowin and Ghaemi2019). These measurements revealed a similar slip velocity behaviour in both the viscous sublayer and the log-law region when normalized wall roughness (i.e. krms +) is small. Similar trend is also reported in the DNS of low Reτ (i.e. Reτ = 222–442) turbulent channel flow over SH longitudinal microgrooves and SH aligned microposts (Rastegari & Akhavan Reference Rastegari and Akhavan2019). When the surface fully covers the surface without visible deformation, the upward shift of the velocity both in the viscous sublayer and log-law region is in correspondence. However, when interfacial deformation, such as pinned deformation is introduced, slip velocities in both regions are notably reduced, particularly in the log-law region (Rastegari & Akhavan Reference Rastegari and Akhavan2019). The deterioration of slip velocity in log-law region has also been observed in high Reτ (i.e. Reτ = 1328–1389) turbulent channel flow over random SHSs with small normalized wall roughness (i.e. krms + < 1) (Ling et al. Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016). It is revealed that the velocity shift in the log region is much smaller than that in the viscous sublayer when wall roughness effects are negligible. Such deterioration of the slip velocity is inapparent at low Reτ cases, but becomes predominant at high Reτ cases. Though the reason of the deterioration of the slip velocity in log-law region at the high Reτ is not explicitly explained in the literature (Ling et al. Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016), it is hypothesized to result from a reduction in plastron coverage and increased interfacial deformation at high Reτ .
(a) Mean velocity profiles (b) viscous shear stress and Reynolds shear stress at ReL = 1.85 × 106 in TBL flow for the smooth wall and the AFHS, respectively. The velocities are scaled by their own inner units, while the viscous and Reynolds shear stress are scaled by the total wall stress of the smooth wall for the quantitative comparison of shear stresses.

Figure 17(b) compares the viscous and Reynolds shear stresses over the smooth wall and AFHS. In the near wall region (y + < 30), the viscous stress of the AFHS is consistently lower than that of the smooth wall. In the outer region, however, the viscous stresses of the AFHS and smooth wall almost collapse. The Reynolds shear stress of the AFHS is slightly higher than that of the smooth wall in the ranges y + < 10 and 40 < y + < 200. Moreover, the maximum value of the Reynolds shear stress for the AFHS is smaller than that for the smooth wall, with the location of this maximum shifted closer to the wall. These trends are in accordance with the observations reported in Abu Rowin & Ghaemi (Reference Abu Rowin and Ghaemi2019) , who found that the viscous stress over a SHS is lower than that over a smooth wall for y + < 30, while the Reynolds shear stress over the SHS is larger than that over the smooth surface at y + < 8 and 20 < y + < 30. It should be noted that the data over y + = 30 is not provided in their study.
The increment of the near-wall Reynolds shear stress , as reported by Abu Rowin & Ghaemi (Reference Abu Rowin and Ghaemi2019), is attributed to the larger streamwise velocity fluctuations. In the present experiment, both streamwise and wall-normal velocity fluctuations, i.e. 〈u
′
u
′〉
+
and 〈
$v^{\prime}v^{\prime}$
〉
+
, are obtained in the near-wall region using PIV. Figure 18 shows the velocity fluctuations in streamwise and wall-normal directions for the smooth wall and the AFHS at ReL
= 1.85 × 106. Within the vicinity of the wall (y
+
$\leqslant$
10), the 〈u
′
u
′〉
+
of the AFHS is larger than that of the smooth wall, where the difference in the magnitude of 〈u
′
u
′〉
+
between the AFHS and the smooth wall diminishes with increasing distance from the wall. In the outer region (y
+
> 10), 〈u
′
u
′〉
+
for the AFHS becomes smaller than or converges with that of the smooth wall. Although the peak value of 〈u
′
u
′〉
+
for the AFHS occurs closer to the wall than that for the smooth wall, the overall distribution of 〈u
′
u
′〉
+
for the AFHS resembles the smooth wall profile. It increases with distance from the wall, peaks around y
+
= 10 and then decreases. It appears that the profile of 〈u
′
u
′〉
+
of AFHS shifts towards the wall while compared with that of the smooth wall. The enhancement of the streamwise velocity fluctuations over the HSs has also been observed in the literature through the high-resolution near-wall measurement (Ling et al. Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016; Abu Rowin & Ghaemi Reference Abu Rowin and Ghaemi2019). The increasement magnitude of the streamwise velocity fluctuations in this study is smaller than those reported in the literature, likely because random SHSs were used in those experiments. For the 〈
$v^{\prime}v^{\prime}$
〉
+
, the wall-normal velocity fluctuation of AFHS is larger than that of smooth wall within the region of y
+
$\leqslant$
20, whereas values for the AFHS become smaller than those for the smooth wall in the outer region.
Comparison of velocity fluctuations between the smooth wall and the AFHS in (a) streamwise direction, and (b) wall-normal direction at ReL = 1.85 × 106. Results are scaled by the smooth wall inner units.

To further investigate the differences in velocity fluctuations between the smooth wall and the AFHS, the vortex swirling method (Zhou et al. Reference Zhou, Adrian, Balachandar and Kendall1999) is employed to visualize the near-wall swirling strength. This method calculates the imaginary part of the complex eigenvalue (λci
) of the velocity gradient tensor, which quantifies localized spiralling motion while excluding regions with vorticity but no swirling, such as boundary layers. This provides a more detailed view of near-wall vortex dynamics. Figures 19(a) and 19(b) show the local swirling strength (
$\skew2\bar{\lambda}_{ci} $
) contours for the smooth wall and the AFHS at ReL
= 1.85 × 106. Both surfaces exhibit high swirling strength stripes oriented in the streamwise direction, indicating the presence of coherent vortex structures. These stripes are characteristic of TBLs, where streamwise-aligned vortices dominate the near-wall region. While the overall structure is similar, the wall-normal position of the maximum swirling strength differs between the two surfaces. By extracting the streamwise averaged swirling strength,
$\skew2\bar{\lambda}_{ci}$
, the wall-normal position of the maximum
$\skew2\bar{\lambda}_{ci}$
of the smooth wall and the AFHS can be compared. The streamwise-averaged swirling strength reveals that the maximum
$\skew2\bar{\lambda}_{ci}$
occurs at y
+
= 10.56 for the smooth wall and at y
+
= 9.50 for the AFHS. This result demonstrates a 10 % downward shift in the location of maximum swirling strength for the AFHS. The reduction in wall-normal position suggests that near-wall vortex cores are drawn closer to the surface in the presence of slip velocity. By bringing vortex cores closer to the wall, the AFHS alters the energy cascade within the boundary layer. This redistribution of energy could enhance DR by reducing the turbulence intensity farther from the wall.
The comparison of time-averaged local swirling strength contours of (a) the smooth wall and (b) the AFHS at ReL
= 1.85 × 106. The local swirling strength (Zhou et al. Reference Zhou, Adrian, Balachandar and Kendall1999),
$\skew2\bar{\lambda}_{ci}$
, is applied to visualize the near-wall vortical structures by extracting the imaginary part of the complex eigenvalue of the velocity gradient tensor. The x
+
and y
+
are the normalized streamwise and wall-normal distance.

The wall-normal profiles of the streamwise averaged swirling strength,
$\skew2\bar{\lambda}_{ci}$
, are further extracted and shown in figure 20(a). For the AFHS, the swirling strength is enhanced within the near-wall region (y
+ < 10) compared with the smooth wall, while it is suppressed in the outer near-wall region (10 < y
+ < 100). This behaviour is consistent with the DNS results by Park et al. (Reference Park, Park and Kim2013) which showed that near-wall turbulence structures are weakened due to the presence of the SHSs. Figure 20(b) further compares the profiles of
$\skew2\bar{\lambda}_{ci}$
in the wall-normal direction at different ReL
. For both the smooth wall and the AFHS, the overall magnitude of the
$\skew2\bar{\lambda}_{ci}$
increases with the ReL
, while the wall-normal position of the peak
$\skew2\bar{\lambda}_{ci}$
gradually shifts closer to the wall with the increment of ReL
. Except the case of the lowest ReL
, all
$\skew2\bar{\lambda}_{ci}$
of the AFHS within y
+
$\leqslant$
10 is enhanced, while it is suppressed in the other region comparing with the smooth wall. The aforementioned trends indicate that as ReL
increases, the slip condition imposed by the AFHS has a greater influence on the redistribution of energy and turbulence structures within the boundary layer. Specifically, the downward shift in the position of peak
$\skew2\bar{\lambda}_{ci}$
reflects a reorganization of the near-wall turbulence, with coherent vortices drawn closer to the wall.
(a) The wall-normal profile of the streamwise averaged swirling strength,
$\skew2\bar{\lambda}_{ci}$
for the smooth wall and the AFHS at ReL
= 1.85 × 106. (b) The comparison of wall-normal
$\skew2\bar{\lambda}_{ci}$
profiles of the smooth wall and the AFHS at different ReL
.

In addition to analysing the influence of the ReL
on the wall-normal
$\skew2\bar{\lambda}_{ci}$
profiles, we evaluate the effects of the ReL
on near-wall mean velocity profiles and the corresponding slip velocities. The slip velocity is extracted from the mean velocity profile in the linear viscous sublayer measured by the planar PIV. As we can see in figure 21(a), there is a distinct upward shift in
$\bar{u}^{+}$
, i.e. the normalized slip velocity, for all ReL
cases of the AFHS when compared with the velocity law of smooth wall. This upward shift confirms the presence of slip velocity, consistent with findings from previous DNS studies (Min & Kim Reference Min and Kim2004) and experiments (Ling et al. Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016; Abu Rowin & Ghaemi Reference Abu Rowin and Ghaemi2020).
(a) Mean velocity profiles of the AFHS at different ReL
. The classical law of the wall is shown for comparison. (b) The normalized mean slip velocity,
$\bar{u}_s^{+}$
, at different ReL
. The slip velocity is extracted from the mean velocity profile in the linear viscous sublayer measured by the planar PIV.

In this study, it should be noted that the upward shift in
$\bar{u}$
+
is observed not only near the wall but also in the logarithmic region of the velocity profile. This is consistent with the DNS studies (Rastegari & Akhavan Reference Rastegari and Akhavan2018, Reference Rastegari and Akhavan2019) on the superhydrophobic longitudinal microgrooves, where the upward shift in the log region was found to be only slightly smaller than that in the viscous sublayer. By contrast, experimental studies of random SHSs by Ling et al. (Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016) and Abu Rowin & Ghaemi (Reference Abu Rowin and Ghaemi2020) have reported much smaller shifts in the log region compared with the viscous sublayer, or even overlapping semilogarithmic velocity profiles at different ReL
. The discrepancy between the current results and those of random SHSs can be attributed to the suppression of spanwise slip in the longitudinal microgroove HSs used in this study. Ling et al. (Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016) has attributed the smaller slip velocity in the log region observed for random SHSs to the spanwise slip induced by surface roughness. Such roughness disrupts the coherent slip effects and reduces the slip velocity in the logarithmic region. In contrast, the longitudinal grooves used in the current study minimize spanwise slip, thereby preserving the slip velocity in both the viscous sublayer and the logarithmic region.
Figure 21(b) further shows the relationship between the normalized slip velocity, i.e.
$\bar{u}_{s}^{+}$
, and ReL
of the AFHS. As we can see, the
$\bar{u}_{s}^{+}$
increases almost linearly with the ReL
within the measurement region (i.e. ReL
= 1.17 × 106–1.21 × 106, Reτ
= 885–1723). This is consistent with the DNS results of regular SHSs (Martell, Rothstein & Perot Reference Martell, Rothstein and Perot2010), which also reported an increase in u
s
+ with the increase of Reτ
from 180 to 590. More quantitative relationships between
$\bar{u}_{s}^{+}$
and ReL
have been reported in DNS studies (Lee, Jelly & Zaki Reference Lee, Jelly and Zaki2015; Rastegari & Akhavan Reference Rastegari and Akhavan2019) by modelling SHSs with slip-free walls and non-dimensionalized geometries. Rastegari & Akhavan (Reference Rastegari and Akhavan2019) associate the
$\bar{u}_{s}^{+}$
both in laminar and logarithmic regions to the non-dimensional size of the microcavities and solid fractions relative to the wall units. They mainly highlight the dependence of the normalized slip velocity on the non-dimensional size of surface microtextures and solid fractions in wall units, while showing minimal dependence on the Reynolds number itself. As further revealed by Xu et al. (Reference Xu, Yu and Kim2021), as long as the plastron sustained, the DR (the slip length in identical) increases almost linearly with the dimensionless structural pitch (W
+
= 4.0–16.0). However, these findings differ significantly from experimental measurements of random HSs, such as those by Abu Rowin & Ghaemi (Reference Abu Rowin and Ghaemi2020), where
$\bar{u}_{s}^{+}$
shows little or no variation with increasing Reynolds number (Re = 6200–9400) in small channel flows. Moreover, in these experiments, the dimensional slip length decreases with increasing Reynolds number. Abu Rowin & Ghaemi (Reference Abu Rowin and Ghaemi2020) attributed this discrepancy to air plastron depletion, which reduces or eliminates the slip effect at higher Reynolds numbers. Unlike DNS simulations, where the air plastron is assumed to remain constant as a slip-free boundary, experimental conditions often lead to dynamic thinning or collapse of the plastron with increasing Reynolds number.
Interestingly, the current experimental results align more closely with DNS studies than with previous experiments on random SHSs. This consistency can be attributed to the active pressure regulation applied in our experiments, which maintains plastron coverage and ensures its stability across different Reynolds numbers. By using AFHSs with longitudinal microgrooves, the plastron behaviour on the grooves approximates the idealized slip boundary conditions assumed in DNS. This allows the AFHS to sustain a predominant streamwise slip effect, resulting in a linear increase in
$\bar{u}_{s}^{+}$
with increasing ReL
. In contrast, random SHSs lack this level of plastron stability due to surface roughness and irregularities, which induce spanwise slip and lead to plastron thinning or depletion at high Reynolds numbers. This spanwise slip disrupts the coherent slip effects observed in DNS and contributes to the discrepancies between experiments and simulations. These findings clarify inconsistencies between previous experimental and numerical studies and underscore the importance of microstructures design and pressure regulation in achieving idealized slip conditions in turbulent conditions.
Figure 22 shows the relationship between the DR and the normalized slip velocity (
$\bar{u}_{s}^{+}$
) based on both the current experimental data and results from previous studies in the literature. To the authors’ knowledge, this study is the first experimental study that simultaneously measures DR and the
$\bar{u}_{s}^{+}$
directly while eliminating the influence of plastron degradation across varying Reynolds numbers by applying the AFHS with longitudinal microgrooves and pressure regulation. As shown in figure 22, the experimental data for longitudinal grooves aligns closely with theoretical predictions for cases where λ
y
= 0 (i.e. no spanwise slip), as well as DNS results (Park et al. Reference Park, Park and Kim2013; Lee et al. Reference Lee, Jelly and Zaki2015; Rastegari & Akhavan Reference Rastegari and Akhavan2019) for regular patterned SHSs reported in the literature. The DNS studies span a range of Reynolds numbers and demonstrate that
$\bar{u}_{s}^{+}$
is a strong predictor of DR when spanwise slip effects are negligible. For a given
$\bar{u}_{s}^{+}$
, cases with λ
z
= 0 consistently exhibit higher DR than cases with λ
z
= λ
x
, highlighting the negative contribution of the spanwise slip. As also recently reported by Yu et al. (Reference Yu, del Campo Melchor, Li, Jeon, Eldredge and Kim2024), the linear relationship between the normalized slip length and the DR is confirmed in practical application tests of longitudinal microtrench surfaces underneath a motorboat on the sea. This explains why HSs with streamwise grooves, which are dominated by streamwise slip, achieve significantly greater DR compared with HSs with random roughness or spanwise grooves. For HSs with longitudinal grooves, where streamwise slip dominates, DR is primarily determined by the magnitude of
$\bar{u}_{s}^{+}$
. However, the magnitude of DR decreases slightly with increasing Reynolds number. This behaviour is consistent with previous DNS studies (Park et al. Reference Park, Park and Kim2013; Lee et al. Reference Lee, Jelly and Zaki2015; Rastegari & Akhavan Reference Rastegari and Akhavan2019) and is speculated to be attributed the effects of the increased turbulence production and plastron deformation at higher Reynolds number. At higher Reynolds numbers, turbulence production intensifies, leading to higher energy dissipation rates within the boundary layer. While the slip velocity reduces frictional resistance at the wall, the enhanced turbulence production in the outer regions diminishes the net DR effect. Although the plastron is actively regulated in this study, higher Reynolds numbers may still introduce subtle interfacial deformation, reducing the efficiency of the slip boundary condition. This effect is much smaller than in random HS experiments, where plastron degradation is more pronounced.
Magnitude of DR versus the normalized slip velocity both measured in this experiment and those in the literature. Drag reduction is measured by the in-house made shear stress sensor and the normalized slip velocity is extracted from PIV data for varying Reτ = 885–1723. The experimental results by Ling et al. (Reference Ling, Srinivasan, Golovin, McKinley, Tuteja and Katz2016)and Abu Rowin & Ghaemi (Reference Abu Rowin and Ghaemi2020), and the DNS results by Martell et al. (Reference Martell, Rothstein and Perot2010), Park et al. (Reference Park, Park and Kim2013), Lee et al. (Reference Lee, Jelly and Zaki2015) and Rastegari & Akhavan (Reference Rastegari and Akhavan2019) are also shown. The theoretical prediction (lines) on the basis of Fukagata, Kasagi & Koumoutsakos (Reference Fukagata, Kasagi and Koumoutsakos2006) and Busse & Sandham (Reference Busse and Sandham2012) are included for comparison, where the λ x and λ z indicate components of the so-called slip lengths in streamwise and spanwise directions, respectively.

4. Conclusions
The current experimental investigation presents the first simultaneous experimental measurement of turbulent DR, plastron coverage and turbulence statistics under sustained plastron conditions in TBL. It has revealed the turbulent DR characteristics and mechanisms over AFHS with longitudinal grooves in TBL. The AFHS, designed with longitudinal grooves and integrated air supply channels, demonstrates the ability to actively control and sustain plastron coverage at friction Reynolds numbers up to 1723. This capability enables a detailed study of DR characteristics under turbulent conditions while minimizing the influence of plastron degradation or depletion, which often compromises results in similar studies. Using wall shear stress sensors, planar PIV and interfacial visualization systems, the skin friction drag, velocity profiles and plastron coverage are simultaneously measured. The results show DR percentages ranging from 14.8–35.8 % depending on the hydrophobic groove void fraction from 0.2 to 0.4. At the same void fraction, the AFHS exhibits higher DR than the conventional HSs. Furthermore, the experimental data confirm a consistent increase in DR with increasing Reynolds numbers, provided the plastron remains intact. These findings bridge the gap between experimental observations and DNS, which predict similar trends for HSs and SHSs with sustained plastron coverage. Turbulence statistics measured via PIV reveal a significant reduction in near-wall viscous shear stress, accompanied by corresponding slip velocities both in the viscous sublayer and the logarithmic region. The Reynolds shear stress and streamwise velocity fluctuations over the AFHS are found to be larger than those over a smooth wall. The near-wall vortex cores are revealed to be shifted 10 % towards the wall, as indicated by near-wall swirling strength maps. Additionally, the relationship of DR and the normalized slip velocity is in accordance with the DNS and the streamwise slip theoretical prediction. Previous discrepancy of the DR trends in different experiments are primarily due to the deterioration or depletion of the plastron during the turbulent tests. By maintaining plastron stability through active pressure regulation, the AFHS effectively eliminates the discrepancies observed in previous experimental studies, where plastron deterioration was a limiting factor. The sustained plastron coverage in AFHS replicates idealized slip boundary conditions, as assumed in computational studies, leading to enhanced DR and turbulence modulation. This observation aligns with the trends predicted by DNS, thereby bridging the gap between experimental and computational fluid dynamics in the context of hydrophobic surface DR.
Funding
This work is financially supported by the National Natural Science Foundation of China (no. 52201384), the Fundamental Research Funds for the Central Universities (no. 226-2022-00208) and the State Key Laboratory of Clean Energy Utilization.
Declaration of interests
The authors report no conflict of interest.
Data availability statement
The data that support the findings of this study are available upon reasonable request.












