Effects of a flexible leading edge on the aerodynamic performance of a vertical axis wind turbine model

Wind energy ranks as the fourth-largest source of power generation in the United States, contributing 9.8% of the nation’s electricity as of 2023.¹ Industrial wind turbines are generally classified according to the direction of their rotational axis into two types: horizontal axis wind turbines (HAWTs) and vertical axis wind turbines (VAWTs). HAWTs have a main rotor shaft and generator mounted horizontally at the tower hub, requiring orientation into the wind for optimal performance. By contrast, VAWTs feature a vertically oriented rotor axis, which allows them to operate effectively without alignment to the wind, making them advantageous in settings where wind direction is highly variable. Although HAWTs currently dominate the market owing to their efficiency in power generation, and have been the subject of many studies,^2–6 VAWTs show potential for future market resurgence, particularly for offshore applications.^7–9 VAWTs present several advantages over HAWTs at larger scales, including suitability for low-wind-speed environments, lower construction and installation costs owing to the use of ground-based gearboxes, reduced land footprint, greater resistance to extreme weather, higher scalability, and enhanced stability due to their lower center of gravity.

A significant technical challenge that limits the commercialization of VAWTs is their inherent dynamic stall,^10,11 which contributes to their lower efficiency compared with HAWTs.¹² Most VAWTs in current applications use rigid, fixed blades. Unlike HAWTs, whose blades generate a relatively stable torque regardless of angular position under uniform wind conditions, VAWT blades encounter cyclic lift fluctuations due to changes in the angle of attack along their circular path. This uneven variation in tangential force results in torque ripple in the output, placing substantial stress on mechanical components.¹³ Additionally, the staggered torque production—where not all blades produce torque simultaneously—further limits VAWT efficiency. This cyclic structural loading in VAWTs also accelerates wear, leading to early failure commonly observed in industrial use. All of these decreases in reliability and efficiency arise largely from dynamic stall occurring as the angle of attack changes rapidly during operation.¹⁴ 

Dynamic stall plays a crucial role in shaping the aerodynamic behavior of airfoils during rotation. This phenomenon occurs when oscillating airfoils undergo rapid, transient, and unsteady motion, leading the angle of attack to exceed the critical angle for static stall. Extensive wind tunnel studies have been conducted to analyze dynamic stall and its effects.^15,16 One of the primary effects of dynamic stall is the delayed onset of conventional flow separation, which generates additional lift by interacting with the leading-edge vortex.¹⁷ To enhance airfoil performance, various flow control techniques have been explored over the past several decades.^18–21 Broadly, these techniques can be categorized as either active or passive flow control methods. Passive flow control, unlike active methods, requires no additional energy input, thereby simplifying operational and maintenance demands. Typically, passive flow control is achieved through geometric modifications to the airfoil. Vortex generators, for example, are a widely used passive flow control method across multiple applications. These include traditional boundary-layer vortex generators and sub-boundary-layer vortex generators, also known as micro-vortex generators.^22–24 Studies of sub-boundary-layer vortex generators have shown that they significantly reduce flow separation, by an amount comparable to that achieved with their larger conventional counterparts, but without the added drag typically associated with the latter.²⁵ 

Although passive flow control mechanisms, such as vortex generators, have been widely studied for wind turbine blades, further research is needed on the dynamic stall of VAWT airfoils using passive flow control techniques. Such studies are essential to fully understand the complex phenomena involved and to improve VAWT performance through cost-effective methods. In recent years, novel passive methods, such as flexible surface designs, have been proposed to improve airfoil performance. He et al.²⁶ investigated the impact of a flexible trailing edge on the flow dynamics around an airfoil at high angles of attack using time-resolved particle image velocimetry. They found that the flexible plate exhibited periodic vibrations that generated and shed trailing-edge vortices, which influenced the formation of leading-edge vortices in the separated shear layer. In another experimental study, Siala et al.²⁷ investigated the energy harvesting capabilities of an oscillating airfoil with passive flexibility. They focused on the combined heaving and pitching motions of the airfoil, with the trailing edge being passively actuated using a torsion rod. Their results suggested that the pitching motion positively contributed to the mean power output, while the heaving motion had a smaller, negative impact. Brousseau et al.²⁸ performed a numerical study of the dynamics of a two-dimensional partly flexible plate immersed in a turbulent fluid flow. They found that at low pitching frequencies (drag mode), the flexible leading edge reduced the amplitude of hydrodynamic force fluctuations without significantly affecting the mean force value. At high pitching frequencies (propulsive mode), the flexible leading edge increased both the magnitudes and mean values of hydrodynamic forces. Serdar Genç et al.²⁹ investigated fluid–structure interaction phenomena on a NACA 4412 airfoil with partially flexible membrane material mounted on both its suction and pressure surfaces. Their results demonstrated that this partially flexible airfoil significantly improved aerodynamic performance, particularly at lower angles of attack, by doubling the lift coefficient and reducing the drag coefficient. The flexibility of the membrane caused the shear layer to approach, suppressed the laminar separation bubble, and shrunk the wake region, leading to higher lift and lower drag forces.

The aim of the investigation described in this paper was to explore the effects of a novel passive flow control mechanism, specifically a flexible leading edge (LE), on the dynamic stall behavior and aerodynamic performance of a VAWT blade with a NACA 0018 airfoil. Two experiments were conducted to achieve this objective. First, wind tunnel tests were performed on 2D airfoil sections with and without the flexible LE, under both steady flow and dynamic pitching conditions. Second, a scaled VAWT model equipped with both rigid and flexible LE blades was constructed and tested under an open flow condition using an variable-speed fan to compare aerodynamic performance. The effects of the flexible LE were evaluated through surface pressure measurements and electrical output analysis. This work provides preliminary insights into a novel and cost-effective passive flow control strategy for improving dynamic loading and overall power efficiency in VAWT rotors.

Figure 1 shows schematics of the flexible LE airfoil model and the mechanisms used to generate pitching motion in the wind tunnel study. The cross-section of the airfoil was based on the NACA 0018 profile, as shown in Fig. 1(a). A baseline airfoil was 3D-printed using ABS filaments, resulting in a fully rigid surface. A second flexible LE model was created by replacing the bottom 30% of the LE surface with flexible silicone rubber. The silicone rubber sheet has a thickness of 1/16 in. (1.6 mm) and a durometer grade of 30 A. The chosen silicone rubber exhibited high elasticity, allowing it to deform adequately in response to changes in surface pressure. As noted by Meththananda et al.,³⁰ there is a reasonably well-defined relationship between Shore hardness and Young’s modulus within the hardness range studied here (30.2–62.9). Hence, the selected rubber material had an approximate Young’s modulus of 1.1 MPa according to an empirical equation provided in Ref. 30. Both models were hollowed out internally to connect pressure tubes for surface pressure measurements. Figure 1(a) also shows the locations of pressure taps, covering the surfaces from the leading edge to 70% of the chord, except for the flexible rubber sheet region. The flexible portion was able to change its camber in response to loads produced by air pressure during the pitching cycle.

Experiments were conducted under both static (fixed angle of attack) and dynamic (pitching motion) scenarios. For the dynamic case, the sinusoidal pitching mechanism shown in Fig. 1(b) was developed to provide sinusoidal pitching motion of the airfoil with adjustable frequency and amplitude. Specifically, the small disk, directly driven by a stepper motor, was connected to the big disk through a connecting shaft. This connection transformed the motor’s 1D rotation into oscillating motion and allowed the amplitude to be adjusted by altering the connecting shaft length. The center of the big disk was directly connected to the airfoil at 25% chord length from the LE, which was fixed inside the wind tunnel section.

Both static and dynamic testing of the 2D airfoil section were conducted in an open-circuit subsonic wind tunnel in the NDSU Advanced Fluid Dynamics Laboratory. The airfoil and the pitching motion mechanism were mounted in the 1 × 1 ft² (30.5 × 30.5 cm²) test section and its static angle of attack (AOA) could be precisely controlled by programming the stepper motor. For both tests, the wind speed was set at 5 and 10 m/s, giving Reynolds numbers of Re = 46 000 and 92 000, respectively. A Dantec MiniCTA hot-wire anemometer was used to measure the flow velocity in the wind tunnel at a frequency response of 5 kHz. The hot-wire probe was calibrated in a velocity calibrator and was able to provide an average accuracy of ±1% of reading for wind speeds up to 30 m/s. Before the airfoil was mounted, the uniformity and the turbulence intensity of the wind tunnel flow were measured using a hot wire across the test section. The wind tunnel had a low turbulence intensity of ∼1.3%. Surface pressure measurements were performed using a 16-channel high-speed pressure scanner (Scannivalve DSA Model 3217). The pressure scanner featured 16 temperature-compensated piezoresistive pressure transducers, a pneumatic calibration valve, RAM, a 16-bit A/D converter, and a microprocessor, all housed in a compact, self-contained module. This design provided a network-ready Ethernet pressure scanning module. The pressure scanner had ±0.05% full scale accuracy and featured automatic zero-offset correction, ensuring highly accurate and repeatable results. During this test, 16 pressure channels were linked to surface ports via plastic tubing positioned along the center of the airfoil span, as shown in Fig. 1(a). The 16 standard inputs of the pressure scanner sensors worked at 500 Hz sampling frequency. On the airfoil surface, the pressure ports were evenly distributed on the upper and lower surfaces of the baseline airfoil. For the flexible LE airfoil, the flexible portion (35% chord from the LE) was not connected with pressure tubes, to prevent the tubing affecting the deformation of the surface. Therefore, for the flexible LE model, only eight pressure taps were available on the flexible side.

A Darrieus VAWT model, scaled at 1:75 from the 34-m Sandia VAWT design, was created with a DC electrical generator mounted at the top for experimental testing. This generator was connected to a data acquisition (DAQ) system for power performance data collection. A schematic and the dimensions of this VAWT design are shown in Fig. 3. The model, designed as an H-rotor, had three blades, an 18-in. (45.7 cm) rotor diameter, and a height-to-diameter aspect ratio of 1.0. The rigid portions of the blades were 3D-printed using ABS material and were based on the 2D section of the NACA 0018 airfoil, with each blade having a chord length of 7 in. (17.8 cm). Similar to the wind tunnel test model, the leading edge of the VAWT blades was replaced with a silicone rubber sheet, covering 30% of the outward-facing leading-edge surface, as shown in Fig. 3. This flexible portion was designed to act as a passively adaptive surface, deforming in response to the cyclic dynamic loading experienced by the rotor blades during rotation. Owing to the large size of the modified H-rotor and limitations of the wind tunnel capacity, it was tested with an industrial variable-speed fan in an open flow environment.

Figure 5 shows the velocity fluctuations downstream of the centerline of the pitching airfoil at a distance of one chord length. Figures 5(a) and 5(b) demonstrate that at low Re, the pitching motion of the airfoil induces clear periodic velocity fluctuations. For example, in the k = 0.07 case, the dominant frequency is 1.63 Hz (period 0.61 s), exactly twice the pitching frequency of 0.82 Hz (period 1.21 s). This result suggests that two primary vortices shed from the leading edge—one from the upstroke and one from the downstroke of the pitching motion—are causing the fluctuations in the center of the wake flow. Owing to the presence of the large-scale vortices, the turbulence intensity increases significantly to 0.44. The velocity characteristics of cases with other values of k and Re were similar and therefore are not discussed here. As Re increases to 96 000, the velocity fluctuations become quite chaotic as vortex mixing intensifies in the wake. The normalized turbulence intensity is relatively low, around 0.27–0.29, owing to the breakdown of the large vortex from the pitching airfoil.

Figure 6 presents the surface pressure coefficient distributions on rigid and flexible LE airfoils at various AOAs for Re = 46 000. It is clear from Fig. 6(a) that as the AOA increases above 13°, aerodynamic stall occurs, indicated by the flattened pressure distribution on the suction side. For the flexible LE airfoil, when the flexible LE is on the pressure side, Fig. 6(b) shows that stall over the suction surface is slightly delayed to 14°, while the suction peak pressure is decreased. This reduction in suction peak pressure is induced by the concave deformation of the flexible LE. However, the delay of static stall can be considered a favorable result. Figure 6(c) shows the pressure distribution on the suction side when the flexible LE is on that side. It is evident that the flexible LE dramatically delays static stall to 16°. When Re is increased to 92 000, the performance difference between the flexible LE airfoil and the baseline airfoil becomes more significant. It can be seen from Fig. 7(a) that the baseline rigid airfoil experiences stall at an AOA of 15°, whereas the flexible LE airfoils with the flexible LE on the pressure side and the suction side do not experience stall until the AOA reaches 17° [Fig. 7(b)] and 18° [Fig. 7(c)], respectively.

These results clearly suggest that the flexible LE improves the aerodynamic performance of the NACA 0018 airfoil by delaying stall, regardless of whether the LE surface is present on the pressure side or the suction side. The effect is more significant at higher Reynolds numbers, particularly at Re = 92 000. The beneficial effects of the flexible LE surface are primarily due to its ability to generate dynamic camber in response to varying AOAs. When the flexible LE is positioned on the pressure side of the airfoil, the high surface pressure near the stagnation point induces an inward deflection. This inward movement increases the camber of the airfoil, enhancing lift generation and stabilizing the airflow. Conversely, when the flexible LE is on the suction side, the negative suction pressure causes an outward deflection of the membrane. This outward movement helps sustain a favorable pressure gradient even at high AOAs, effectively delaying flow separation. By continuously adjusting its shape, the flexible LE maintains smoother airflow over the airfoil surface, mitigating the onset of dynamic stall. These observed phenomena are consistent with findings from previous studies, which have demonstrated that an entirely flexible airfoil can significantly improve aerodynamic performance by adjusting its shape to surface pressure distributions.

For the dynamic pitching analysis, pressure data were collected at 500 Hz by the DSA pressure scanner for airfoils pitching within −35 to +35° at two reduced frequencies k = 0.07 and 0.1. To illustrate the transient process, the pressure coefficient C_p variations were plotted for individual pressure taps during a complete pitching cycle. Figures 8–10 provide an in-depth view of the pitching-up and pitching-down processes in relation to the flexible LE side. In the context of these figures, “pitching up” refers to rotation of the airfoil where the angle of attack (AOA) increases, causing the flexible LE to interact with the flow more directly on the pressure side. Conversely, “pitching down” indicates a decrease in AOA, with the flexible LE side experiencing the suction side pressure as it moves away from the flow.

Figure 8 compares the pressure variation loops of the baseline rigid airfoil and the flexible LE airfoil for k = 0.07 at three different measurement locations (X/C = 0.03, 0.34, and 0.69) on the rigid side of the airfoil. The pressure tap located at X/C = 0.03 is very close to the LE of the airfoil and is typically where the suction peak is observed. The tap location X/C = 0.34 represents the end of the LE and is also the mirrored location of the end of the flexible LE surface on the other hand. The tap location X/C = 0.69 represents the characteristic pressure variation near the trailing edge. The comparison clearly shows that the presence of the flexible LE surface significantly alters the pressure variation on the opposite side of the rigid surface during dynamic pitching. Figures 8(a) and 8(b) suggest that although the maximum suction pressure is about the same, the flexible LE airfoil experiences an early ramp-up in suction pressure when the airfoil is pitching up (with the flexible surface on the bottom) for both locations X/C = 0.03 and 0.34. At X/C = 0.03, the suction pressure is significantly greater than for the baseline airfoil during the 5°–15° pitching-up process. The decrease in suction pressure after 15° is a consequence of the moving LE vortex, which can be explained by the increase in suction pressure at X/C = 0.34 after 15°. At X/C = 0.69, both airfoils exhibit similar pressure coefficient variations and exhibit a late increase in suction when they are pitched up near 30° [Fig. 8(c)]. Overall, it is clear that the presence of the flexible LE leads to a prolonged period of high suction over the rigid side of the airfoil during the pitching-up process near the LE.

Figure 9 shows a similar trend, but with a greater magnitude of difference when the airfoils are pitching at k = 0.1. In addition to the early and prolonged suction pressure, it should be noted that the flexible LE also modifies the pitching-down process by creating a quicker recovery to positive pressures when the airfoil is pitching from +35° to −35°. Hence, the increased area of the pressure variation loop indicates improved dynamic performance of the flexible LE airfoil model.

The improved suction pressure is more remarkable on the flexible LE side. Figure 10 shows the pressure coefficient loops for surface tap locations X/C = 0.34 [Figs. 10(a) and 10(c)] and X/C = 0.69 [Figs. 10(b) and 10(d)] behind the flexible LE surface. To be consistent with the previous discussion, we define the “pitching-up” and “pitching-down” processes with respect to the flexible LE side, i.e., with a flipped AOA, which, owing to pitching symmetry, does not affect the interpretation of the results. Figures 10(a) and 10(c) suggest that the flexible LE modifies the aerodynamics of this side and significantly increases the suction pressure near the LE for the entire pitching-up process. Figures 10(b) and 10(d) show that the impact of the flexible LE reaches X/C = 0.69. For both k = 0.07 and k = 0.1, an increase in suction pressure is observed as early as 20° during the pitching-up process for flexible LE airfoils.

In summary, the presence of the flexible LE provides earlier and prolonged high suction pressure during the pitching cycle on both sides of the airfoil. This effect could be attributed to the fact that the deformation of the flexible surface under the action of surface pressure changes disrupts the symmetry of the airfoil by creating camber. Figure 11 shows a sequence of images of the flexible LE airfoil pitching for k = 0.07 and Re = 92 000 in the wind tunnel section, captured by a FLIR Blackfly camera at 200 f/s. The images clearly show that the flexible LE surface deforms considerably when subjected to changing negative and positive surface pressures (negative and positive AOA, respectively) during a pitching cycle, which alters the overall surface pressure distribution around the airfoil.

The improvement in aerodynamic performance of the flexible LE VAWT model can be attributed to the ability of the adaptive surface to modify flow separation and dynamic stall. At low tip-speed ratios, VAWT blades experience significant flow separation and dynamic stall, owing to lower relative wind speeds, which increase the AOA and lead to flow separation on the blade surfaces. This results in large fluctuations in lift and drag forces, reducing overall efficiency. The flexible LE acts as an adaptive surface that deforms in response to changing aerodynamic loads, helping to maintain a more favorable camber and enhancing the lift generated by the blade. It should be noted that the current results for power enhancement by the flexible LE are limited in terms of the parameters examined. The placement of the flexible LE surface, whether on the inner or outer side, could affect the results. Additionally, parameters such as the size ratio of the flexible LE to blade size, the elastic modulus of the materials, the radius-to-height ratio, and the tip-speed ratio could also influence outcomes. Nevertheless, this work demonstrates that it is feasible to increase power output by simply installing a flexible LE on a scaled model. However, a more comprehensive experimental study is needed to cover all factors that could contribute to performance enhancement.

This experimental study has investigated the effects of a flexible leading edge (LE) on the aerodynamic performance of a NACA 0018 airfoil on a scaled vertical axis wind turbine (VAWT) model under steady flow conditions and dynamic pitching motion. The 2D airfoil section was analyzed using a high-frequency surface pressure scanner with a specially designed pitching motion mechanism installed in a wind tunnel. Additionally, a three-bladed H-rotor VAWT model was constructed to assess the effectiveness of the flexible LE blades in rotational situations.

Under steady flow conditions, the flexible LE airfoil showed significant improvements in aerodynamic performance by delaying the static stall angle compared with the baseline airfoil. For the dynamic pitching scenario, the airfoil’s performance was evaluated by plotting its pressure cycle during the pitching motion. The flexible LE structure induced an earlier increase and prolonged period of high suction pressure during the pitching-up process, contributing to overall higher lift generation. Images of the flexible LE airfoil during pitching motion at Re = 92 000 revealed deformation of the LE surface, demonstrating its potential to achieve high camber.

The scaled VAWT model with flexible LE blades was evaluated by comparing its performance with a conventional rigid blade VAWT. Both the rotational speed and electrical voltage data suggest that the VAWT model with flexible LE blades performs better, since the output voltage exceeded that of the rigid blade VAWT as wind speed increased. Overall, the flexible LE structure has been demonstrated to be an efficient and cost-effective passive method to improve VAWT aerodynamic efficiency.

The flexible LE, as a novel passive approach to enhancing aerodynamic efficiency, could have a significant impact in applications where dynamic stall and flow separation are common challenges, such as in wind turbines, rotorcraft, and UAVs. Unlike traditional fixed or rigid airfoil designs, a flexible LE can adapt to varying aerodynamic loads, improving lift, delaying stall, and increasing power output without the need for active control systems. This adaptability could lead to significant performance improvements, especially under unsteady and variable flow conditions, and it offers a lightweight, low-maintenance alternative to more complex active flow control technologies. Future studies on optimized design and large-scale tests of the flexible LE are strongly recommended to further elucidate its scale-up potential.

The authors have no conflicts to disclose.

Jiaming Zhao: Data curation (lead); Formal analysis (lead); Investigation (lead); Writing – original draft (equal); Writing – review & editing (equal). Yan Zhang: Conceptualization (lead); Project administration (lead); Supervision (lead); Writing – original draft (equal); Writing – review & editing (equal).

All data analyzed during this study are included in this published article. Raw datasets are available from the corresponding author on reasonbale request.