1 Introduction

Many researchers have studied using plasma actuators for the separation control of airfoils. Most of their studies were focused on using a dielectric barrier discharge (DBD) plasma actuator in various applications, such as separation control on an airfoil under laminar flow (Sosa and Artana 2006), actuator location optimization (Jolibois et al. 2008), airflow reattachment using a DBD actuator (Bernard et al. 2008), lift and drag improvement of airfoils (Bernard et al. 2009) and separation control using unsteady actuation (Amitay and Glezer 2002; Rethmel et al. 2011). Interest in DBD plasma actuator is growing because of their unique features such as no moving parts, quick response and extremely low mass. Other devices, e.g., mechanical slats and flaps, add more weight and produce noise on the aircraft.

A DBD plasma actuator consists of an exposed electrode and encapsulated electrode that are separated by a dielectric material. Plasma is created when sufficient voltage is applied between the exposed and encapsulated electrodes. The DBD plasma actuator is usually excited with an AC voltage to allow continuous plasma generation. Momentum is transferred from the plasma discharge to the ambient air through ion collision which creates an ionic wind or induced air flow. Forte et al. (2007) conducted Pitot tube measurements and obtained ionic wind velocities of up to 7 m/s at 0.5 mm from the wall for a single actuator and 8 m/s for multiple actuators. Meanwhile, Craig et al. (2010) developed a new configuration for plasma actuators. They claimed that the major problem with plasma actuators is the relatively low induced velocity of the jet. If the jet velocity can be increased, more momentum can be imparted to the flow. Therefore, increasing the jet velocity will make this device suitable for use over a wider range of Reynolds numbers. They studied the potential of multiple encapsulated electrodes (MEE) to increase the induced velocity. In their results, they were able to use MEE actuators to produce an induced velocity that was 36.5 % higher than the baseline case. However, a higher ionic velocity is not strictly related to improve flow control performance (Bernard et al. 2009). They found that unsteady actuation increases the aerodynamic performance of an airfoil compared to steady actuation. There is a strong suspicion that the vortex interaction has a larger influence on the improved results than the ionic velocity.

Balcon et al. (2009a) claimed that successful separation control at high angles of attack depends on vortex interaction due to the effects of an unsteady actuation or pulse modulation drive. This means that the DBD plasma actuator is triggered by on and off modes at a specific modulation frequency. For instance, a pulse modulation drive was successfully used to delay stall compared to the steady actuation case (Md Daud et al. 2014). This is because the pulse modulation drive may be able to cause a separated shear layer near the leading edge to roll up and develop. Asada et al. (2009) also reported that a smaller duty ratio provides stronger separation control capability despite a lower input energy. Jukes and Choi (2012) demonstrated that a DBD vortex generator develops streamwise vortices in their experiments. They successfully used their method to reduce the separation region. Therefore, the vortex interaction due to the actuation effect or actuator position is important for improving the flow control performance.

The electrical discharge of a DBD plasma actuator is powered by a sinusoidal waveform signal. However, Balcon et al. (2009b) modified the waveform signal. They studied the effect of positive and negative sawtooth signals on electric wind. Bernard and Moreau (2010) demonstrated the capabilities of plasma actuators by adjusting the voltage waveform into burst, superposition and ring modulation signals. In the present study, we applied a signal comprising pulse modulation and amplitude modulation (PM + AM) to a DBD plasma actuator. In this case, the DBD plasma actuator was placed near the leading edge of the NACA 0015 airfoil model. The objective of this study was to improve the lift coefficient compared to the result of a previous experiment (Md Daud et al. 2014). In this previous experiment, a pulse modulation drive with a constant voltage amplitude signal was applied under a high angle of attack condition. We compared the results to examine the flow pattern when both the pulse and amplitude are modulated.

2 Experimental setup

The experimental apparatus used in this study was the same as that used in our previous study (Md Daud et al. 2014). Figure 1 shows an overview of the experimental setup. The experiment was conducted in a low-speed, semi-closed wind tunnel with a square cross-section having dimension of 500 mm × 500 mm. An NACA 0015 airfoil with a chord length of 100 mm and span of 150 mm was tested at Re ≈ 67,000 in an airflow of 10 m/s. The DBD plasma actuator was installed at x/c = 0.025 of the chord length.

Fig. 1
figure 1

Experimental setup using airfoil model NACA 0015 in a low-speed, semi-closed wind tunnel with Re ≈ 67,000 and an airflow velocity of 10 m/s

Full size image

The DBD plasma actuator consisted of two copper-tape electrodes; each electrode was 50 μm thick and 5 mm wide. The two electrodes (exposed and encapsulated) were arranged in parallel with a 1 mm overlap. These electrodes were separated by a 100 μm thick Kapton film that acted as a dielectric.

The lift force was measured directly with a load cell (LMC-3501-50N, Nissho Electric Works, Japan) having a rated load of 50 N and accuracy of ±0.2 %. The sampling frequency was set to 1 kHz with a low-pass filter of 10 Hz over approximately 4 s. The velocity was measured at the mid-span of the airfoil with an I-type hot-wire probe and constant temperature anemometer. The sampling frequency was 50 kHz, the measurement time was approximately 4 s and electromagnetic noise was removed with a 2 kHz low-pass filter. The thermal effect caused by the plasma generation seemed to be small, because the measurements were conducted at a distance from the plasma actuator. Therefore, the values measured by the hot-wire probe were obtained without any compensation. The output signal from the anemometer was converted with an A-D converter (WE7000, Yokogawa, Japan), after which a customized LabVIEW program processed the data.

The flow was visualized with a high-speed camera (GX 8, nac Image Technology, Japan) and continuous laser sheet (SVL-532-1000-45L, Sanctity Laser, China). The laser sheet intersected the airfoil at its mid-span, and the axis of the high-speed camera was oriented perpendicular to that of the laser sheet. The camera frame rate was 2000 fps, and its resolution was 1024 × 768 pixels. In the experiment, smoke was produced by the smoke wire method, and the airflow speed was set to 5 m/s. The quantity of nickel powder used as seeding particles was so small that its effect on the flow was not detected even near electrodes. In the results, the Reynolds number for the flow visualization was half that used in our other experiments. However, the trend of the aerodynamic performance did not vary significantly, especially for St = 0.6 and St = 4.0.

2.1 Pulse modulation

In our previous experiment (Md Daud et al. 2014), a high-voltage AC current was supplied to the exposed electrode while the encapsulated electrode was grounded. The base waveform was an 8 kHz sinusoidal wave, generated by a digital function generator (DF1906, NF Corporation, Japan). The signal was amplified by a high-voltage amplifier (HEOPT-10B10-LI, Matsusada Precision Inc., Japan) to provide a peak-to-peak voltage of 6 kV. Unsteady actuation was performed through the application of a low modulation frequency to the base wave. Figure 2 shows the ON–OFF controlled waveforms for the PM case at St = 0.6 and St = 4.0.

Fig. 2
figure 2

Waveforms for the PM case (f B = 8 kHz, Duty 10 %)

Full size image

2.2 Amplitude modulation

In this experiment, the PM + AM case was designed to have the same power consumption as the PM case. The power consumption was measured with the electric current method proposed by Ashpis et al. (2012). The time series of the voltage and current signals was recorded with a GW Instek GDS-2062 oscilloscope (8 bits, 60 MHz bandwidth, 1 GS/s/channel @ 2 active channels). 12,500 samples of data were recorded at 0.25 MHz sampling frequency for St = 0.6 and 1.25 MHz sampling frequency for St = 4.0. These were the suitable sampling frequency to obtain low amplitude current spikes. A resistor with a known resistance, R = 100 Ω, was placed in series between the exposed electrode and ground to calculate the current. The applied voltage was calculated from the monitored output of high-speed and high-voltage amplifier. The PM + AM case was designed using DF0106 software (NF Corporation, Japan). The waveform of the PM + AM case was set with the same modulation frequency f M and same base frequency f B as that of the PM case. In the results, the T on and T periods were the same for the PM + AM and PM cases. Thus, the power consumptions for the PM and PM + AM cases were about 0.46 W when St = 4.0 and 0.36 W when St = 0.6. From our result, we concluded that the power consumption is only proportional to V p–p if St is less than 2.0.

When St = 0.6, there were 13 cycles of the base wave within one ON time period, while only two cycles of base waves existed when St = 4.0; this is shown in Fig. 2. We set a maximum voltage of 7 kV and minimum voltage of 5 kV for the St = 0.6 case, while a maximum voltage of 7 kV and minimum voltage of 2 kV were used for the St = 4.0 case. This is a suitable maximum voltage because a maximum voltage of greater than 7 kV will cause the Kapton film to become damaged more easily. In addition, obtaining the same power consumption as in the PM case is difficult, if the maximum voltage is lower than 7 kV. Figure 3 shows waveforms for the PM + AM case with decreasing and increasing patterns. The results obtained for each pattern were compared (not shown), and the lift coefficient C l of the decreasing pattern was found to be slightly better than that of the increasing pattern, although the difference was very small. Therefore, we adopted a decreasing amplitude modulation in this study. It is also important to highlight that in the case of St = 4.0 which has two cycles, it is not proper to express it as amplitude modulation. This is an extreme and rare case in the case of amplitude modulation.

Fig. 3
figure 3

Waveforms for the PM + AM case (f B = 8 kHz, Duty 10 %) a decreasing pattern b increasing pattern

Full size image

3 Results and discussion

3.1 Effect of PM and PM + AM on lift coefficient

Figure 4 shows the lift coefficient C l plotted against the angle of attack α. In the PM case, C l increased compared with the Duty 100 % case in the previous study (Md Daud et al. 2014). However, the PM + AM case showed larger values of C l compared to the PM case for both St = 0.6 and St = 4.0. Here, St is the non-dimensional pulse modulation frequency. For Duty 100 %, the airfoil stalled at α = 15°, while C l continued to increase to a high attack angle when PM or PM + AM was applied. In this experiment, the duty ratio was set to 10 % based on the results obtained by Asada et al. (2009). They found that the most effective duty ratio setting for the pulse modulation drive is 10 %.

Fig. 4
figure 4

The effect of PM and PM + AM cases on lift coefficient C l for Duty 10 %

Full size image

The PM and PM + AM cases produced similar trends for C l when St was the same. When St = 4.0, C l decreased sharply after the maximum point, while C l decreased gradually when St = 0.6. In the previous study, we examined the St = 0.6 and 4.0 cases under the stall control condition (α = 16°) and at a high angle of attack (α = 18°). Therefore, we applied the same condition to the PM + AM case. In this paper, the stall control condition is defined as the parameter set for which no stall occurs and C l increases beyond the stalled point of the base case, (α = 16°).

Figure 5a shows the lift coefficient C l plotted against St at α = 16°. Figure 5b shows the ratio of C l for PM + AM to that for PM as a function of St. At the stall control condition (α = 16°), C l gradually increased with an increase in St and the PM + AM case showed little improvement in C l compared to the PM case for each duty ratio. C l was a few percentage points greater for the PM + AM case than for the PM case for all St. On the other hand, at a high angle of attack (α = 18°) as shown in Fig. 6, C l reached a local maximum and then rapidly decreases as St increased.

Fig. 5
figure 5

Effect of modulation frequency on C l at α = 16° (stall control condition)

Full size image
Fig. 6
figure 6

Effect of modulation frequency on C l at α = 18° (high angle of attack condition)

Full size image

At α = 18°, C l was again greater for the PM + AM case than for the PM case for all St, but the rate of increase was greater than that at α = 16°. The rate of increase in C l was less than 5 % at α = 16°, while it was greater than 5 % at α = 18°. It reached 25 % when Duty 50 % and St = 4.0. PM + AM was clearly more effective than PM at improving the lift coefficient at high angles of attack.

3.2 Flow around airfoil

We examined the flow around the airfoil by focusing on the effect of PM + AM on C l at a high angle of attack (α = 18°). Figure 7 shows the mean velocity contour map when St = 0.6 and St = 4.0 for both of the PM and PM + AM cases. The mean velocities near the airfoil surface for the PM and PM + AM cases were almost the same when St = 0.6. However, the low velocity region at the suction side was slightly narrower for the PM + AM case. This situation occurred in the same manner when St = 4.0, where the low velocity region for the PM case was slightly wider compared to the PM + AM case. This indicates that the flow can be expected to attach slightly closer to the airfoil surface in the PM + AM case compared to the PM case when St = 0.6 and St = 4.0. The flow around the airfoil was also analyzed based on the results of the captured images for the velocity fluctuation and flow.

Fig. 7
figure 7

Time–mean velocity contour map at α = 18° for Duty 10 %

Full size image

Figure 8 shows the turbulence intensity contour map for the PM and PM + AM cases when St = 0.6 and St = 4.0. When St = 0.6, a high level of turbulence intensity for the PM case was only detected near the leading edge and mid-point area of the airfoil surface. However, for the PM + AM case, a high level of turbulence was detected across the entire airfoil surface from the leading edge to the trailing edge. This was due to the strong vortex shedding near the leading edge, which grew to a large vortex and flowed downstream along the airfoil surface. In other words, when St = 0.6, the PM + AM case is predicted to produce larger and stronger vortices near the leading edge and entire airfoil surface compared to the PM case. This larger vortex may be caused by the amplitude modulated actuation of the actuator. For St = 4.0, however, the region with a high level of fluctuating velocities was limited to near the leading edge. The turbulence intensity near the leading edge was slightly lower for the PM case than for the PM + AM case. The vortex produced from the leading edge should be much stronger for the PM + AM case than for the PM case.

Fig. 8
figure 8

Turbulence intensity contour map at α = 18° for Duty 10 %

Full size image

Figures 9 and 10 use smoke wire to visualize St = 0.6 and St = 4.0, respectively, at a high angle of attack. Here, T is the period of one ON–OFF cycle. Flow separation occurred for all cases under these conditions. For St = 0.6, the separated shear layer rolled up and developed into a small-scale vortex at the leading edge during the T/3 period. The vortex scale became larger during the 2T/3 and 3T/3 periods in the middle area of the airfoil surface for the PM case (Fig. 9a). For the PM + AM case (Fig. 9b), the separated shear layer rolled up more definitely and developed into a vortex near the leading edge, which caused the vortex structure to become larger during the T/3 period. The growing vortex rolled downstream over the entire airfoil surface during the 2T/3 and 3T/3 periods. As a result, the lift coefficient was higher for the PM + AM case than for the PM case because of the growing vortex, which forced the flow closer to the wing surface.

Fig. 9
figure 9

Captured images of airflow over an airfoil at α = 18° when St = 0.6 (Duty 10 %)

Full size image
Fig. 10
figure 10

Captured images of airflow over an airfoil at α = 18° when St = 4.0 (Duty 10 %)

Full size image

For St = 4.0, the separation area appeared to be broader for both the PM and PM + AM cases. For the PM case, a small vortex structure was detected near the leading edge during the T/3 and 2T/3 periods. Then, the vortex dispersed downstream during the 3T/3 period. On the other hand, for the PM + AM case, the separated shear layer near the leading edge rolled up and grew to a slightly larger-scale vortex than that in the PM case during the T/3 period. The large vortex rolled near the wing surface over the mid-area of airfoil during the 3T/3 period.

To distinguish the results of the PM and PM + AM cases shown in Figs. 9 and 10 (especially for St = 4.0), the flow on the leading edge of the airfoil was visualized. Figures 11, 12 and 13 show images of the leading edge of the airfoil visualized with the smoke method. As shown in Fig. 11, when St = 0.6, the vortex shedding at the leading edge increased in size and flowed downstream (PM and PM + AM). After 3T/6, the vortex was no longer seen in these images. However, during the T/6 periods, the starting vortex was slightly larger for the PM + AM case than for the PM case (Fig. 11). This was due to the unsteady amplitude modulated from the PM + AM signal, which created a massive change in the separated shear layer near the leading edge. This stimulated the vortex strength which forced the flow closer to the wing surface. This agreed with the results presented in Fig. 8 (PM + AM) when St = 0.6. In addition, the movement of the vortex was much faster when the PM + AM case was applied to the actuator. This result is consistent with that shown in Fig. 9 during the 3T/3 period. The vortex covered up to 90 % of the wing surface for the PM + AM case, but only 70 % for the PM case during this period.

Fig. 11
figure 11

Captured images of airflow over the leading edge of an airfoil at α = 18° when St = 0.6 (Duty 10 %)

Full size image
Fig. 12
figure 12

Captured images of airflow over the leading edge of an airfoil for the PM case when St = 4.0 (α = 18°, Duty 10 %)

Full size image
Fig. 13
figure 13

Captured images of airflow over the leading edge of an airfoil for the PM + AM case when St = 4.0 (α = 18°, Duty 10 %)

Full size image

When St = 4.0 with a shorter period than in the St = 0.6 case, almost all vortex structures were captured in these images (Figs. 12, 13). The vortex size in both cases (PM and PM + AM) was the same during the T/6 and 2T/6 periods. However, for the PM + AM case, the unsteady amplitude modulated signal caused a rapid change in the separated shear layer. As a result, the vortex size became slightly larger as the starting vortex combined with the secondary vortex during the 3T/6 period (Fig. 13). Therefore, the combined vortices flowed slightly closer to the wing surface compared to the PM case. This result supports the findings shown in Fig. 8b, where the vortex near the leading edge was stronger rather than in the PM case. Therefore, the lift coefficient was higher for the PM + AM case than for the PM case. Overall, the lift variation is seemed to be influenced mainly by the amplitude modulation when St is low and it is influenced not only by the amplitude modulation but also by power consumption when St becomes high.

4 Conclusion

An experimental study was performed to investigate the impact on the lift coefficient C l of an amplitude modulated signal, applied to a DBD plasma actuator mounted on the leading edge of an NACA 0015 airfoil. The results are summarized as follows:

  1. 1.

    The lift coefficient C l showed similar trends for both the PM and PM + AM cases at the same St values. When St = 4.0, C l decreased rapidly after reaching a maximum. However, when St = 0.6, C l decreased gradually.

  2. 2.

    The PM + AM case improves the lift coefficient more effectively than the PM case at a high angle of attack (α = 18°), with minimal improvement under the stall control condition (α = 16°).

  3. 3.

    An unsteady amplitude modulated signal stimulates vortex growth more for the PM + AM case than for the PM case; the former forces the flow closer to the wing which improves the lift coefficient.