Effects of Input Voltage and Freestream Velocity on Active Flow Control of Passage Vortex in a Linear Turbine Cascade Using Dielectric Barrier Discharge Plasma Actuator

: Passage vortex exists as one of the typical secondary flows in turbomachines and generates a significant total pressure loss and degrades the aerodynamic performance. Herein, a dielectric barrier discharge (DBD) plasma actuator was utilized for an active flow control of the passage vortex in a linear turbine cascade. The plasma actuator was installed on the endwall, 10 mm upstream from the leading edge of the turbine cascade. The freestream velocity at the outlet of the linear turbine cascade was set to range from U FS , out = 2.4 m/s to 25.2 m/s, which corresponded to the Reynolds number ranging from Re out = 1.0 × 10 4 to 9.9 × 10 4 . The two-dimensional velocity field at the outlet of the linear turbine cascade was experimentally analyzed by particle image velocimetry (PIV). At lower freestream velocity conditions, the passage vortex was almost negligible as a result of the plasma actuator operation ( U PA , max / U FS , out = 1.17). Although the effect of the jet induced by the plasma actuator weakened as the freestream velocity increased, the magnitude of the peak vorticity was reduced under all freestream velocity conditions. Even at the highest freestream velocity condition of U FS , out = 25.2 m/s, the peak value of the vorticity was reduced approximately 17% by the plasma actuator operation at V AC = 15 kV p-p ( U PA , max / U FS , out = 0.18).


Introduction
Axial-flow turbines are a main component of many modern turbomachines, and are commonly utilized in a vast array of industrial applications including aircraft propulsion jet engines, and electricity power generating gas turbines. In a turbine passage, the secondary flow, which is defined as the cross-flow deviated from the design freestream flow, gives rise to secondary vortices [1]. As shown in Figure 1, a passage vortex is a typical vortex in the secondary flow field among the turbine blades. The inlet boundary layer hits the leading edge of the blade, which leads to the formation of the horseshoe vortex. The pressure gradient within the turbine passage moves the pressure side leg of the horseshoe vortex towards the suction side of the neighboring blade, and causes the growth of a passage vortex. With the migration downstream, the rotation of the passage vortex is strengthened, and the vortex lifts from the endwall.
Endwall secondary losses occur because of the vortex formations due to the interaction between the endwall boundary layers and the blades in the turbine cascade passage. According to Sharma and Butler [2], the loss owing to the secondary flows in the endwall region accounts for almost 30-50% of the aerodynamic total pressure loss in a turbine passage. A typical value of the endwall secondary total pressure loss of turbine blades ranges from 0.01 to 0.04, which depends on the low-turning or high-turning blade designs [3]. Therefore, the secondary flow field in a turbine passage has been an important research subject for several decades, as shown in Langston [4], Sieverding [5], Wang et al. [6], and Coull [7]. The Reynolds number is an important dimensionless aerodynamic parameter, which represents the ratio of internal force to viscous force within a fluid. In general, the Reynolds number is defined as follows: where ρ is the density of the fluid (SI units: kg/m 3 ), U is the reference velocity of the fluid (m/s), L is the reference length (m), and μ is the dynamic viscosity of the fluid (Pa·s or kg/m·s). Turbines for aircraft propulsion systems operate over a very wide range of Reynolds number conditions. The operational Reynolds number is high when taking-off and landing, however, it can drop below 2.5 × 10 4 while cruising at high-altitudes due to low air density [8]. Although the secondary losses are largely insensitive to the variation of high values of the Reynolds number [9], the secondary losses increase significantly at the lower values of the Reynolds number, below Re = 1 × 10 5 [10,11]. To reduce the secondary flow and the associated losses, various passive and active flow control applications in turbines are discussed in the literature [12]. Passive flow control initiatives, nonaxisymmetric endwall contouring [13], leading-edge fillets/bulbs [14], a boundary layer endwall fence [15], tubercles [16], undulated blade [17], etc., have all been pursued with varying degrees of success, however, they have the disadvantage of being permanent; thus, they produce an unwanted penalty loss at different operating conditions. As the active flow control initiatives, air suction [18], steady jet blowing [19], and pulsed jet blowing [20], have been demonstrated for secondary flow reduction. For example, Bloxham and Bons [18] found that the air suction of the endwall boundary layer yields up to a 28% reduction in total pressure loss, and Benton et al. [19] found that the steady vortex-generating jet blowing from the endwall yields up to a 40% reduction of the total pressure loss. The active flow control devices can be turned off when control is unnecessary, and avoid the negative effects of permanent passive flow control devices [21]. However, such air suction and injection devices impose substantial weight penalties.
A dielectric barrier discharge (DBD) plasma actuator is an attractive active flow control device, which is a simple, thin, light-weight actuator that converts electricity directly into flow acceleration with no moving parts. First, Roth et al. [22] measured the plasma wall jet induced by the DBD plasma actuator in quiescent air using a pitot probe. A recent review of the DBD plasma actuator was presented by Corke et al. [23,24] and Wang et al. [25]. Figure 2 shows the schematic illustration of the DBD plasma actuator. A common DBD plasma actuator configuration consists of an asymmetric pair of metallic electrodes, separated by a dielectric material. One electrode is exposed to the surrounding air, and the other electrode is completely encapsulated within the dielectric material. When a high voltage alternating current (AC) is supplied between the two electrodes, an electric discharge is initiated. The surrounding air is partially ionized, and charged particles are accelerated by the electric field formed around the electrodes, resulting in a two-dimensional one-way wall jet induced along the surface. Computational fluid dynamics (CFD) was performed on the models and applications of the DBD plasma actuators [26][27][28][29]. For the application of plasma actuator to the turbine flow, the separation control on the blade suction surface [30][31][32][33][34], and tip leakage vortex reduction [35][36][37][38][39][40][41][42] were conducted. However, there are few studies about the passage vortex reduction in the turbine cascade using the DBD plasma actuator. For the application of the DBD plasma actuator for the passage vortex reduction in a compressor cascade, De Giorgi et al. [43] investigated the effects of the actuator installed at the sidewall on the blade suction surface, and found successful reduction of total pressure loss owing to the passage vortex. In general, compressor blades are low-turning blades with weak secondary vortices, and hence, turbine blades are high-turning blades with strong secondary vortices. Therefore, the secondary flow fields, including the passage vortex, differ significantly for the compressor and turbine blades.
The aim of this study is to investigate the active flow control effects of the DBD plasma actuator on the inlet endwall for the reduction of the passage vortex generated in the linear turbine cascade. The velocity distributions at the outlet of the linear turbine cascade were analyzed using particle image velocimetry (PIV). The effects of input voltage of the plasma actuator and the freestream velocity (Reynolds number) are discussed in the paper. As mentioned, previous studies of DBD plasma actuator applications to turbine flows focused on the active flow control of the separation on the blade suction surface and the tip leakage flow through the blade tip clearance. To the best of our knowledge, the work presented in this paper is the first investigation of the passage vortex reduction of the high-turning turbine cascade by active flow control using the DBD plasma actuator. Figure 3 shows the test section of a linear turbine cascade. The overall view is shown in Figure  3a. The test section is located downstream of a low-speed, open circuit, blower-type wind tunnel. The top view of the test section and blade geometry is shown in Figure 3b. Table 1 shows the specifications of the linear turbine cascade. The turbine cascade consists of six blades, with a chord length of 58.65 mm, and blade height of 75 mm. The geometry of the linear turbine cascade is replicated in the turbine rotor hub shape of the annular turbine wind tunnel at National Institute of Advanced Industrial Science and Technology (AIST) [44,45]. The turbine cascade was made of high-transparent heatresistant resin (TSR-884B, CMET Inc., Yokohama, Japan) using a stereolithography machine (ATOMm-4000, CMET Inc.). The forming layer pitch of the machine was set at 100 μm. The turbine blades were sprayed with matt black paint in order to reduce both the reflection of laser radiation and the roughness of the blade surface. The surfaces of the blades were polished with ultra-fine grit abrasive paper which had an approximate average particle diameter of 10 μm. The centerline average roughness of the blade surface is estimated as Ra = 1.25 μm. The ratio of the centerline average roughness and blade chord length is Ra/C =2.1 × 10 −5 , which is considered a smooth surface [46]. Note that the surface roughness of turbine blade affects the total pressure losses [47]. The profile loss at a low Reynolds number is reduced by the surface roughness due to a reduction of separation on the blade suction surface, while the profile loss at a higher Reynolds number increases by the surface roughness due to a growth of the boundary layer transition [46,48]. However, the secondary loss is almost unaffected by the surface roughness [48]. A plasma actuator was installed in the acrylic endwall at 10 mm upstream from the leading edge of the linear turbine cascade. Details regarding the plasma actuator are described in Section 2.3.

Wind Tunnel and Linear Turbine Cascade
(a) overall view (b) top view and blade geometry  The rotating speed of the wind tunnel blower was set from 113 Hz to 1125 Hz, in order to change the freestream velocity in the test section. As shown in Table 2

Particle Image Velocimetry (PIV) Measurements and Data Processing
PIV was employed to quantify the two-dimensional velocity field at the exit of the linear turbine cascade, using a 15 mJ/pulse, double-pulse Nd-YAG laser (NANO S30-15PIV, Litron Lasers Ltd., Rugby, England). Atomized dioctyl sebacate (DOS) oil with a mean particle diameter of 1 μm was injected upstream of the test section, via a pressurized oil chamber. Flow-image pairs were captured by a camera (PIV CAM 13-8, TSI, Inc., Shoreview, MN, USA) with 1280 × 1024 pixel resolution, and a frame rate of 3.75 Hz. A PIV software (Insight, TSI, Inc.) calculated the velocity vectors from the peak correlation of groups of particles between frames, using conventional cross-correlation algorithms on a 32 × 32 pixel grid. Time-averaged velocity distributions were analyzed using 300 instantaneous velocity pairs by PIV. Turbulence intensity was also calculated as follows: where ux' and uy' are the velocity random fluctuation components of x and y directions. The velocity and turbulence intensity were nondimensionalized by the freestream velocity at the turbine outlet, UFS,out., from 2.4 m/s to 25.2 m/s. Vorticity was calculated as follows: where Ux and Uy are the horizontal and vertical velocity components. Figure 4 shows the photographs of the top and bottom views, and cross-sectional schematic of the plasma actuator, along with its geometrical dimensions. The plasma actuator used in this study was created by a printed-circuit board (PCB) process. The exposed and encapsulated electrodes were designed by a CAD software and formed by etching a double-sided copper-clad laminate with a dielectric barrier layer made of silicone resin (CS-3975A, Risho Kogyo Co., Ltd., Osaka, Japan). The thicknesses of the copper exposed and encapsulated electrodes, and the silicone resin dielectric barrier are 0.018 mm and 0.44 mm, respectively. The spanwise width of electrodes is designed to be 150 mm. The streamwise lengths of the exposed and encapsulated electrodes are 5 mm and 15 mm, respectively. As shown in Figure 4, the electrodes are arranged asymmetrically in the streamwise direction and overlapped 0.5 mm in order to generate uniform DBD plasma near one edge of the side of the exposed electrode. The plasma actuator was excited with a sinusoidal waveform from a power supply (HAPS-10B40, Matsusada Precision Inc., Kusatsu, Japan). The amplitude of the input peak-to-peak voltage to the plasma actuator was varied from 6 kVp-p to 15 kVp-p, and the frequency of the input voltage was fixed at 10 kHz. The minimum input voltage, 6 kVp-p, was limited due to the lower voltage for plasma emission, and the maximum input was limited due to the power supply capability.    Figure 6 shows the vertical distributions of the velocity at the streamwise position of Z = 8.5 mm (red dashed lines in Figure 5). The peak and width of the velocity increases with the increased input voltage.     [49,50], and was normalized by the spanwise width of the plasma actuator (0.15 m). The power consumption of the plasma actuator with the spanwise width of 150 mm used in this study (red symbols) increased drastically, as the input voltage increased. The maximum power consumption at the maximum input voltage of 15 kVp-p is approximately 420 W/m. The measured power consumption data fit the typical curve of the 3.5 power of the input voltage, which was reported by Murphy et al. [51] and Hanson et al. [52]. The power consumption plots (blue symbols) of the standard plasma actuator with a gn spanwise width of 100 mm, named PAK-Ref03, is shown in the figure as a reference. The plasma actuator, PAK-Ref03, is provided by AIST in Japan as an activity of the Technical Section on Plasma Actuators, which was held in the Fluid Engineering Division (FED) of the Japan Society of Mechanical Engineers (JSME) [53]. The data of both the 150 mm and 100 mm spanwise width plasma actuators have a similar trend.          Figure 13 shows the turbulence intensity distributions at the outlet of the linear turbine cascade at various input voltages, where UFS,out = 2.4 m/s. In Figure 13a, the no control condition, the maximum value of the turbulence intensity is 13% at the center of the passage vortex. Meanwhile, the minimum turbulence intensity in the outer area of the passage vortex is as low as 2%. The results indicate that the passage vortex is stable. In Figure 13b, in the condition of flow control at the input voltage of VAC = 6 kVp-p, the outer area of the passage vortex has a high turbulence intensity region, which means that the passage vortex becomes unstable due to the induced flow by the plasma actuator. In Figure  13c, at the input voltage of 7 kVp-p, the peak value of the high turbulence intensity in the passage vortex becomes 19%, and another high turbulence intensity region is observed at the mid-passage (Y = 30 mm) of the blade suction side after the trailing edge, where the peak turbulence intensity value is 18%. The high turbulence intensity region is generated by the instability of the boundary layer on the blade suction side due to the disappearance of the passage vortex. As the input voltage increases from 8 kVp-p to 10 kVp-p, in Figure 13d-f, the area of high turbulence intensity of the passage vortex decreases; however, another area of high turbulence intensity near the blade suction side is increased and strengthened (the peak value is 20%). At the higher input voltages, ranging from 11 kVp-p to 15 kVp-p, in Figure 13g-k, a new high turbulence intensity area is gradually spread in the downward flow near the blade pressure surface, with the maximum value of turbulence intensity of 13%.        Figure 16 shows the velocity distributions at the outlet of the linear turbine cascade at various input voltages, at an increased freestream velocity of UFS,out = 4.7 m/s. In the no control condition in Figure 16a, high velocity regions with strong secondary flow vectors, due to the passage vortices, exist near the endwall and blade suction side (SS) after the trailing edge. The maximum normalized velocity in the center passage is 0.38, which is smaller than that at UFS,out = 2.4 m/s (0.66). The reduction of the maximum normalized velocity is due to the reduction in strength of the passage vortex, or viscous effect, associated with the increased Reynolds number related to the increase in the freestream velocity. In the flow control conditions, at the input voltage of 6 kVp-p in Figure 16b, the high velocity regions and secondary flow vectors of passage vortices are slightly reduced by the plasma actuator. In Figure 16c, at VAC = 7 kVp-p, the high velocity regions in the right-side passage near the blade suction surface are divided into two peak areas. In Figure 16d,e, at VAC = 8 kVp-p and 9 kVp-p, the high velocity regions in the center passage move toward the endwall, and the high velocity regions in the right-side passage are divided into three areas. As the input voltage increases from 10 kVp-p to 15 kVp-p, in Figure 16f Figure 16. The value of the peak velocity in the no control condition is 0.38 (the black dashed line). By the flow control using the plasma actuator, the peak velocity is slightly increased from 0.36 at VAC = 6 kVp-p to 0.42 at VAC = 10 kVp-p, then the peak is gradually decreased to 0.30 at VAC = 15 kVp-p (21% reduction).  Figure 18 shows the turbulence intensity distributions at the outlet of the linear turbine cascade at various input voltages, at UFS,out = 4.7 m/s. In Figure 18a, the no control condition, the maximum value of the turbulence intensity in the passage vortex is 16%. The result indicates that the passage vortex is more unstable than at a lower freestream velocity, UFS,out = 2.4 m/s. As the input voltage increases from 6 kVp-p to 9 kVp-p, in Figure 18b-e, although the maximum value of the high turbulence intensity region in the center passage is constant (16%), the high turbulence intensity region is slightly reduced. At the input voltages of 10 kVp-p and 11 kVp-p, in Figure 18f,g, the size and magnitude of the high turbulence intensity region are gradually reduced. Meanwhile, a new high turbulence intensity region (the peak value of 15%) is observed near the mid-span (Y = 30 mm) of the blade suction side. At the higher input voltages, ranging from 12 kVp-p to 15 kVp-p, in Figure 18h-k, the high turbulence intensity region of the passage vortex is weakened and the new high turbulence intensity region along the blade suction surface is widely spread, with the maximum value of 14% for the turbulence intensity.        Figure 21 shows the velocity distributions at the outlet of the linear turbine cascade at various input voltages with an increased freestream velocity ranging from UFS,out = 9.4 m/s to 25.2 m/s. The range of the corresponding Reynolds number is between Reout = 3.7 × 10 4 and Reout = 9.9 × 10 4 . The velocity distributions in the no control condition are shown in the upper side, Figure 21a,c,e,g. The velocity distributions in the flow control condition at the input voltage of VAC = 15 kVp-p are shown in the lower side, Figure 21b,d,f,h.      Figure 23b,d,f,h. At all freestream velocities, the high turbulence intensity region spreads on the suction side of the trailing edge. The peak value of the high turbulence intensity is mostly 12%. The high turbulence intensity region is slightly reduced and moves closer to the endwall in the flow control condition at VAC = 15 kVp-p.   Figure 24 shows the vorticity distributions at the outlet of the linear turbine cascade at various input voltages at an increased freestream velocity ranging from UFS,out = 9.4 m/s (Reout = 3.7 × 10 4 ) to 25.2 m/s (Reout = 9.9 × 10 4 ). The vorticity distributions in the no control condition are shown in Figure  24a,c,e,g. The vorticity distributions in the flow control condition at the input voltage of VAC = 15 kVpp are shown in Figure 24b,d,f,h.      The peak vorticity in the center passage at various freestream velocities in Figure 25 indicates that the plasma actuator is more effective for a lower outlet freestream velocity. The peak vorticity of the passage vortex at UFS,out = 2.4 m/s decreases by 74% due to the plasma actuator operation, while the reduction ratio gradually decreased as the outlet freestream velocity increased. A similar tendency, that is, an increased freestream velocity effect, was observed in the experimental and numerical studies regarding the active flow control of the separated flow on the blade suction surface of a low pressure turbine by a DBD plasma actuator [33,34]. In these studies, the flow separation completely disappeared at the lower freestream velocity of 1.5 m/s, while the flow separation was slightly reduced at the higher freestream velocity of 3.0 m/s. In general, the plasma actuator is less effective as the freestream velocity is increased. This occurs because the ratio of the wall jet velocity induced by the plasma actuator into the freestream velocity decreases when the freestream velocity (and Reynolds number) is increased [34]. Therefore, this result requires an improvement of the DBD plasma actuator system, including an increased input voltage excitation, as well as changes of the plasma actuator shape and setting location, for the active flow control of the passage vortex in higher freestream velocity conditions of actual industrial turbines.

Conclusions
Active flow control using a DBD plasma actuator was experimentally investigated as a potential technique for passage vortex reduction in a linear turbine cascade. The plasma actuator was installed on the endwall at 10 mm upstream from the leading edge of the turbine cascade. In order to drive the plasma actuator, a sinusoidal excitation with an input peak-to-peak voltage varying from VAC = 6 kVpp to 15 kVp-p under the fixed frequency of 10 kHz was applied. The plasma actuator induced tangential jets with the maximum velocity ranging from UPA,max = 0.6 m/s to 4.5 m/s. The freestream velocity at the outlet of the turbine cascade was set to range from UFS,out = 2.4 m/s to 25.2 m/s, which corresponded to the Reynolds number based on the blade chord length and outlet freestream velocity, ranging from Reout = 1.0 × 10 4 to 9.9 × 10 4 . PIV was used to measure the two-dimensional velocity, turbulence intensity, and vorticity fields at the outlet of the turbine cascade.