Active flow control for power enhancement of vertical axis wind turbines: Leading-edge slot suction

Vertical axis wind turbines (VAWTs) suffer from a poor power performance at low tip speed ratios, where their blade aerodynamics are dominated by unsteady separation and dynamic stall. Therefore, to enhance their aerodynamic performance, separation control is highly desired. The present study intends to suppress the flow separation on VAWTs using boundary layer suction through a slot located near the blade leading edge. High-fidelity computational fluid dynamics simulations extensively validated with experiments are employed. A characterization of the impact of the suction amplitude, 0.5% AS 10%, and the suction location, 8.5 XS/c 28.5, is performed. The dependency of the obtained power gain on operating conditions, i.e. tip speed ratio, 2.5 l 3.5, Reynolds number, 0.51 105 Rec 2.78 105, and turbulence intensity, 1% TI 25%, is studied. The results show that applying suction along the chordwise extent of the laminar separation bubble (LSB) can prevent its bursting, eliminate/postpone its formation, avoid the formation of the dynamic stall vortex and trailing-edge roll-up vortex, and delay the incipient trailing-edge separation. This will significantly increase the blade lift force, decrease the drag force, delay the stall angle and suppress the aerodynamic load fluctuations. For the reference turbine and for AS1⁄4 0.5% and XS/c1⁄4 8.5%, the power coefficient at l of 2.5, 3.0 and 3.5 is enhanced by 247%, 83% and 24%, respectively. The suction location is critical while a minimum amplitude, e.g. AS1⁄4 0.5%, suffices. The optimal suction location is insensitive to TI, weakly sensitive to l while comparatively more sensitive to Rec. © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
Vertical axis wind turbines (VAWTs) are promising candidates for floating offshore windfarms as well as for wind energy harvesting in urban areas, due to a number of specific advantages, especially their omni-directionality, scalability and low noise [1e5]. However, their aerodynamic performance, expressed in terms of power coefficient C P , is currently not comparable with horizontal axis wind turbines (HAWTs) [2,6]. Therefore, improvement of the aerodynamic performance of VAWTs can make them competitive with HAWTs and render them ideal options for wind energy harvesting in the aforementioned locations.
The comparatively low aerodynamic performance of VAWTs could be attributed to their complex unsteady aerodynamics, which has not yet been comprehensively elucidated. The blade aerodynamics of VAWTs are dominated by large excursions of the angle of attack and the relative velocity as well as rotational flow effects. Simultaneous occurrence of several flow phenomena, namely dynamic stall, blade-vortex and vortex-vortex interactions, flow rotational and curvature effects, greatly complicate the blade aerodynamics of VAWTs, where such complexity increases at lower tip speed ratios [7e14].
Variations of the angle of attack on VAWT blades at low tip speed ratio can significantly exceed the stall angle and, therefore, result in occurrence of deep dynamic stall on the blades, where the flow separation can considerably dominate the blade aerodynamics and adversely influence the turbine power performance [14e21].
Therefore, suppression of the unsteady separation and dynamic stall on VAWTs is highly desired [22e25] and can be an effective means toward enhancing their aerodynamic performance and bringing their power coefficient C P closer to that of HAWTs.
Flow control has emerged as a remedy to affect flow separation. In general, flow control is defined as local manipulation of flow features towards desired characteristics [26e28] and it has been employed with aims such as flow separation control, circulation and load control, transition control, drag reduction, heat transfer/ mixing enhancement and noise suppression [29e33]. Flow control actuators are categorized as passive (not requiring energy input, e.g. Gurney flap [24,34e37]) and active (requiring energy input, e.g. suction and blowing [38e41]) devices, where both have already been investigated/employed for numerous applications, such as aircrafts, helicopters, cars, wind turbines and turbomachinery [25,42e45].
Boundary layer suction is a classic flow control method where the near-wall fluid is sucked through porous surfaces or slots in order to reshape the boundary layer velocity profile. Boundary layer suction is normally employed to control separation and transition. Boundary layer suction was introduced and applied by Prandtl [46] to suppress flow separation on a circular cylinder. Since then, this method has been investigated, developed and employed for different applications including aircrafts, helicopters, cars and wind turbines [47e53].
Recently, the application of flow control devices on VAWTs has received attention, however with a focus on passive flow control mechanisms. The impact of leading-edge serrations on power performance of VAWTs has been numerically studied by Wang et al. [54,55], where a maximum improvement of 18.7% in turbine efficiency in terms of the power coefficient C P at the low tip speed ratio of l ¼ 2.0 has been achieved. Gurney flaps have also been numerically studied for VAWTs by Zhu et al. [56] and Bianchini et al. [57], where a maximum power gain of about 23% in C P has been achieved for l ¼ 3.3. Due to the inherently unsteady physics of VAWTs, passive measures are known to have limited power gain for VAWTs [58e60] and active flow control methods could be much more promising as their performance can be adapted to the local operating conditions of the turbine. Research on active flow control for VAWTs is very limited [61e64] and, to the best of our knowledge, the influence of boundary layer suction on the aerodynamic performance of VAWTs has not yet been investigated.
In this perspective, the current study aims to provide insights on the impact of boundary layer suction through a leading-edge slot on the suppression of unsteady separation and dynamic stall on blades of VAWTs in order to enhance their aerodynamic performance.
The objectives of the present work are as follows: -To understand the influence of boundary layer suction on the blade aerodynamics and the turbine power performance of VAWTs at low tip speed ratios; -To characterize the impact of suction characteristics, namely suction amplitude and location, on the blade aerodynamics and the turbine power performance of VAWTs at low tip speed ratios; -To analyze the dependency of the turbine power gain due to suction on operational parameters, namely tip speed ratio, turbulence intensity and Reynolds number; -To identify the optimal suction characteristics for various turbine operating conditions.
The aforementioned objectives are to be achieved using highfidelity CFD simulations. The CFD simulations are extensively validated with different sets of experiments in four different validation studies.
The outline of the paper is as follows. The computational

Computational settings and parameters
2.1. Geometrical and operational characteristics Table 1 lists the geometrical characteristics of the reference turbine, the reference operating conditions and the reference characteristics of the leading-edge slot suction. Fig. 1 illustrates a schematic of the turbine with slot suction. Note that as the focus of the study is on the impact of boundary layer suction, for the sake of computational cost, the turbine is simplified to include only one blade, no shaft and no spokes. The impact of the number of blades and the turbine shaft has been investigated previously in detail in Refs. [60,65].
The suction slot is located on the blade inner side, which during the turbine first half-revolution corresponds to the blade suction side. The suction slot is positioned near the blade leading-edge aiming to suppress the dynamic stall phenomena, which dominate the blade aerodynamics of VAWTs at low tip speed ratios [13,14,19,60,66].

Computational settings
The employed computational settings are listed in Table 2 and are based on the guidelines for accurate CFD simulations of VAWTs [19,60,67e69]. The computational domain and grid are shown in Fig. 2.

Solution verification
Extensive solution verification analyses have been already performed to investigate the impact of computational settings for the CFD simulations of VAWTs [19,67e69,71]. Additional grid convergence analysis using uniformly-doubled grids and time-step sensitivity analysis performed for the VAWT with suction, see Fig. 3, show that the coarse grid with an azimuthal increment of 0.1 can be used for the rest of the study.

Vertical axis wind turbine
Three sets of validation studies have been performed for VAWTs with different geometrical and operational characteristics, where acceptable agreements between the CFD results and experimental data have been observed. The complete description of the validation studies is presented in Refs. [59,68,69] and for brevity is not repeated here.

Slot suction over a flat plate with turbulent boundary layer
Due to lack of high-resolution experimental data for VAWTs with slot suction, the experiments by van der Hoeven [72] are used for validation of CFD simulations of boundary layer suction. In the experiments, the surface pressure along a flat plate with slot suction in a turbulent boundary layer was measured. The incoming flow over the flat plate was ensured to be 2D. The momentum-thickness Reynolds number (Re q;0 Þ for the unperturbed boundary layer (i.e. suction off) at the location of suction was 10,100, corresponding to a local boundary layer momentum thickness of q 0 ¼ 5.6 mm. Description of the experiments and the computational settings are listed in Table 3. Fig. 4 illustrates a schematic of the suction slot along the flat plate and the computational grid. Note that the grid is selected based on a sensitivity analysis using a uniformly-doubled grid. The boundary conditions for the  simulations at the domain inlet, outlet, top and at the flat plate and suction slot are velocity inlet, zero static gauge pressure, symmetry, no-slip wall and velocity inlet (uniform inward). The domain inlet profiles are selected the same as the experimental data at x'/ W ¼ À30.0, given in Ref. [72]. Fig. 5 compares the CFD results with the experimental data for the surface pressure coefficient, CoP, along the flat plate with slot suction at 0 x'/W 1, where a very good agreement is observed.

Influence of boundary layer suction
In this section, a detailed analysis of the impact of boundary layer suction through a slot located near the blade leading edge on the blade aerodynamics and the turbine power performance is presented. The analysis is performed for the reference turbine at the reference operating conditions, see Table 1.
In this section, the focus of the analysis is on the turbine first half-revolution, i.e. 0 q 180 (see Fig. 2a), because during this period the blade inner side, where the suction slot is placed, corresponds to the blade suction side. Therefore, the boundary layer suction is expected to improve the blade aerodynamics and the turbine power performance. Fig. 6 shows the experienced angle of attack a and normalized relative velocity V rel,n during the turbine half-revolution. The method employed to calculate the experienced values of angle of attack and relative velocity from CFD results is presented in detail in Ref. [71]. Note that the presented values correspond to the reference turbine with no suction (suc. off). Nevertheless, the a and V rel,n values for the reference turbine with suction (suc. on) are  found to be overlapping.

Blade aerodynamics
The figure shows that for the three l values of 2.5, 3.0 and 3.5, a exceeds the static stall angle, a SS , of 14 , where a SS is based on Xfoil [73] calculations for NACA0018 at Re ¼ 10 5 . For l ¼ 2.5, the maximum angle of attack, a max , far exceeds a SS and reaches 23.1 at q ¼ 113 . For l ¼ 3.0, a max reaches 18.8 at q ¼ 109 . For l ¼ 3.5, a max hardly exceeds a SS by reaching 15.7 at q ¼ 105 .
Before elaborating on the influence of boundary layer suction, first, several observations for the reference turbine with suction off are discussed based on Figs. 7e10: - Fig. 8a e c: By decreasing l from 3.5 to 2.5, the LSB moves towards the leading-edge due to higher variations of a. Note that increasing a is known to shift the LSB upstream along the blade surface [59,74].
- Fig. 9a e c: By decreasing l from 3.5 to 2.5, the chordwise extent of the LSB decreases due to higher variations of a. -For all cases, for q > 60 , corresponding to a > a SS , the lift values increase while the drag remains lower than the turbine with suction off. This is due to significantly more limited separation along the blade for the 'suc. on' case.
-For l ¼ 2.5 and 3.0, the dynamic stall, which is recognized by a sudden reduction in the lift and an abrupt jump in the drag and consequent load fluctuations, is clearly avoided.
-For all cases, the slope of the C l e a curve during the upstroke remains almost untouched with respect to the 'suc. off' case. However, this slope is slightly higher than the 'suc. off' case during the downstroke. -For all cases, C l,max has significantly increased while C d,max has substantially decreased. The change is more noticeable at lower l where the dynamic stall is more dominant.   Table 4 presents the turbine power coefficient for the reference turbine with suction off and on and the overall power gain, in term of DC P , due to the boundary layer suction for the reference case, see Table 1. Fig. 14 shows the turbine instantaneous moment coefficient, C m , during the turbine half-revolution for different tip speed ratios.

Turbine power performance
It can be seen that by applying the suction, at l ¼ 2.5 where the turbine experiences a deep dynamic stall, the turbine C m is unchanged for q up to z53 . For q > 53 , the turbine C m is substantially improved due to the suction and consequently, the turbine C P is significantly enhanced by 246.6%. The peak in C m is also shifted from q ¼ 68

Impact of suction amplitude and location
In this section, the influence of suction amplitude and suction slot location is analyzed for the reference turbine at reference operating conditions given in Table 1. Table 5 presents the details of the studied cases. For the suction amplitude, a wide range of A S from 0.5% to 10% is studied. Regarding the suction location, four different locations near the leading edge of the turbine blade are analyzed. The center locations of the suction slots are X S =c ¼ 8.5%, 13.5%, 18.5% and 28.5%. The start and end locations of the suction slots are given in Table 5. As the impact of suction slot width is known to be minimal [75,76], therefore, in the present study the width of the suction slot is kept constant, i.e. W =c ¼ 2.5%.
First, the suction slot is fixed at X S =c ¼ 8.5% and the suction amplitude is varied. Table 6 presents the turbine power gain as a function of suction amplitude for A S ¼ 0.5e10%. Fig. 15 presents the spatiotemporal contour plots of skin friction coefficient, C f , along the blade suction side (inner side) during the turbine halfrevolution for A S ¼ 0.5% and 10%. Fig. 16 shows the turbine instantaneous moment coefficient during the turbine halfrevolution for A S ¼ 0.5% and 10%. Within the studied range, the impact of suction amplitude is found to be insignificant where the turbine power gain, the turbine instantaneous moment and the critical boundary layer events along the blade suction side are minimally influenced by suction amplitude. At l ¼ 2.5, by  increasing A S from 0.5% to 10% (i.e. a factor of 20 times higher), the power gain marginally increases from 246% to 259% (i.e. a factor of 1.05 times higher). The power gain with increasing suction amplitude is even less pronounced at higher l of 3.0 and 3.5. Similar observations are found (not shown here for brevity) when the suction slot is fixed at the other suction locations, i.e. X S =c ¼ 13.5%, 18.5% and 28.5%, and the suction amplitude is varied within A S ¼ 0.5e10%. Second, the suction amplification is fixed at A S ¼ 0.5% and the suction slot location is varied. Table 7 presents the turbine power gain as a function of suction location for X S =c ¼ 8.5%, 13.5%, 18.5% and 28.5%. Fig. 17 presents the spatiotemporal contour plots of skin friction coefficient, C f , along the blade suction side (inner side) during the turbine half-revolution for different suction locations. Fig. 18 shows the turbine instantaneous moment coefficient during the turbine half-revolution for different suction locations. Within the studied range, the impact of the suction location is found to be significant, where the turbine power gain, the instantaneous moment and the critical boundary layer events along the blade suction side are substantially influenced by suction location. At l ¼ 2.5, by shifting the suction slot downstream towards the trailing edge from X S =c ¼ 8.5%e13.5%, the power gain, in terms of DC P , reduces from 246% to 208%. The respective C m values also noticeably decrease for q > 80 . The skin friction plots reveal that by this shift, the suction location is moved slightly downstream of the location of the LSB for 'suc. off' case. Nevertheless, the dynamic stall is still avoided due to the suction. Further downstream shifts of the suction slot to 18.5% and 28.5% more substantially diminish the turbine power gain to 119% and 81%, respectively. Fig. 18a also shows that the C m values monotonically decrease due to the downstream shift. For the two most downstream suction locations, i.e. X S =c ¼ 18.5% and 28.5%, the suction slot is already too downstream of the LSB, which therefore cannot prevent its shedding and consequently the dynamic stall, recognized by the formation of the DSV, persists on the blade despite the suction.
The skin friction contours shown in Fig. 17 reveal that when the suction slot is located at 8.5%c, it is positioned right along the LSB for the turbine with suction off, therefore, applying suction can noticeably suppress the LSB and avoid the dynamic stall. However, when the suction slot is shifted further towards the trailing edge, its position moves to downstream of the LSB, thus applying suction will have much more limited influence on the critical flow features. These observations imply that in order to most effectively suppress the separation using boundary layer suction, the suction slot needs to be positioned along the LSB.
Overall, similar observations for the higher l of 3.0 and 3.5 are made although here DC P is much less sensitive to the suction location. The less sensitivity could be attributed to the larger chordwise extent of the LSB for the higher l, see Fig. 9b and c. In addition, for l ¼ 3.5, the DC P for suction location at 13.5% is marginally higher than 8.5%, unlike the lower l values. This is because for higher l, where the variations of a are more limited, the LSB is slightly shifted downstream for the turbine with suction off (see Section 4.1). Therefore, the optimum suction location is also a bit further downstream than the lower l. Similar observations are found (not shown here for brevity) when the suction amplitude is fixed at the other amplitudes, i.e. A S ¼ 1e10%, and the suction location is varied for X S =c ¼ 8.5%, 13.5%, 18.5% and 28.5%.

Dependency of optimal suction characteristics on operational parameters
In this section, the dependency of the turbine power gain due to the boundary layer suction through a slot near the leading edge on the operational parameters is investigated for the reference turbine. For every single operational parameter and each individual value, 20 different cases (consisting of 4 suction locations and 5 suction amplitudes) are investigated to find the individual local optimal suction characteristics at the given condition.
In general, the dependency of the optimal suction characteristics on operational parameters is important because for a turbine in real conditions, the operational parameters inevitably change. Therefore, if the suction system is to work optimally at a range of operational parameters, it might need to be designed with multiple locations and adaptable amplitude so that based on the real-time conditions the respective optimal suction is operated. Needless to say, such a system would need a set of sensors and a feedback control loop.
The suction characteristics are as given in Table 5. The studied parameters are (i) tip speed ratio, (ii) Reynolds number and (iii) turbulence intensity. Table 8 presents the details of the studied parameters.
i. Tip speed ratio: Three different tip speed ratios of 2.5, 3.0 and 3.5 are investigated. Note that although the results in Section 4 are presented for different l, however, the provided discussions did not aim to elucidate the impact of l on optimal suction characteristics, which is the focus of this section. The range covers low tip speed ratios, where the turbine blades experience a deep dynamic stall, and moderate tip speed ratios where the variations of the angle of attack during the revolution are more limited and light dynamic stall might occur. ii. Reynolds number: The chord-based Reynolds number varies from 0.51 Â 10 5 e2.78 Â 10 5 , where the range is corresponding to the operational regime for small-scale VAWTs [2].  Table 6 Turbine power gain as a function of suction amplitude. (X S =c ¼ 8.5%). iii. Turbulence intensity: Three values of approach-flow total TI are studied, i.e. 1%, 5% and 25%, to analyze the sensitivity of the findings to TI in a range from low TI levels relevant for wind-tunnel experiments to moderately high values corresponding to operational conditions in high turbulence areas. Table 9 presents the optimal suction location as a function of tip speed ratio for different suction amplitudes for the reference turbine at reference operating conditions, i.e. TI ¼ 5%, Re c z 10 5 . The table is prepared using 60 URANS simulations, i.e. the optimal values at each l are based on 20 URANS simulations. Fig. 19 shows the contour maps of the turbine power gain, in terms of DC P , in 'suction amplitude e suction location' space. The maps help to identify the optimal suction characteristics as a function of tip speed ratio. Within the studied range, the following observations are made: -In general, the power gain due to suction remarkably decreases with increasing l. For example, with a suction amplitude of 0.5%, the power gain at l of 2.5, 3.0 and 3.5 is 246%, 83% and 25%, respectively. The more prominent influence of suction at lower l is because at lower l the turbine aerodynamic performance is highly dominated by flow separation, thus, its suppression will have a strong impact on the overall power performance. Note that the changes in the absolute values of C P are: (i) at l ¼ 2.5, the C P value increases from 0.045 to 0.156; (ii) at l ¼ 3.0, the C P value increases from 0.137 to 0.251; and (iii) at l ¼ 3.5, the C P value increases from 0.254 to 0.316.

Impact of tip speed ratio
-The optimal suction location slightly shifts downstream as l increases. For l ¼ 2.5 and 3.0, the optimal suction location is at 8.5%c while this shifts to 13.5%c for l ¼ 3.5, where c is the blade chord length. Note that at l ¼ 3.5, the power gain due to suction at 13.5%c is marginally higher than 8.5%c, i.e. only 1%. Nevertheless, the shift is due to the modest downstream shift of the LSB at higher l, as discussed in Section 4.1.
-The optimal range of suction location, which can be inferred from the red regions in contour plots in Fig. 19, becomes wider as l increases. This is because of the increase in the chordwise extent of the LSB for higher l, as discussed in Section 4.1.
-The optimal suction location is independent of the suction amplitude.
In short, applying suction at X S =c ¼ 8.5% is found to be effective for the majority of studied tip speed ratios. At the highest l of 3.5, despite the slight downstream shift of the LSB, a suction slot with 8.5%c is still located along the LSB and provides results comparable to the optimal location of 13.5%c. Therefore, it can be concluded that for the reference turbine at the reference operating condition, i.e. TI ¼ 5% and Re c z 10 5 , a single leading-edge suction slot location can be selected to effectively suppress the flow separation at different tip speed ratios and to enhance the VAWT power performance. The sensitivity of the optimal suction location to tip speed ratio can be considered insignificant. In this regard, for typical constant-speed VAWTs, whose tip speed ratio inevitably changes with varying freestream velocity, the location of the suction slot can be configured near the blade leading edge to align the LSB location in order to effectively suppress the flow separation at low tip speed ratios.   Table 10 presents the optimal suction location as a function of turbulence intensity for different suction amplitudes for Re c z 10 5 . The table is prepared using 60 URANS simulations, i.e. the optimal values at each TI are based on 20 URANS simulations. Fig. 20 shows the contour maps of the turbine power gain, in terms of DC P , in 'suction amplitude e suction location' space. The maps help to identify the optimal suction characteristics as a function of TI. Within the studied range, the following observations are made: -In general, the power gain due to suction remarkably decreases with increasing TI at all l. For example, at l ¼ 3.0 with suction amplitude of 0.5%, the power gain at TI of 1%, 5% and 25% is 249%, 83% and 32%, respectively. The more prominent influence of suction at lower TI is because at lower TI, the boundary layer    on the blade is less resistant to flow separation, thus, the turbine aerodynamic performance is comparatively more dominated by separation and its suppression by suction will result in a stronger enhancement of the turbine power performance. -Within the studied cases (1% TI 25%), the optimal suction location is minimally affected by the TI. Note that by increasing TI, the LSB moves slightly upstream and its chordwise extent decreases. The upstream shift in the position of the LSB for higher TI might suggest that at high TI and low l, applying suction closer to the leading edge, X S =c < 8.5%, could be even more promising. Nevertheless, the suction location of 8.5%c is effective for the majority of conditions while providing comparable power gain to the optimal value for the other cases where 13.5%c is optimal. -The optimal range of suction location, which can be inferred from the red regions in contour plots in Fig. 20, becomes wider as TI increases. At l ¼ 2.5, for TI ¼ 1%, 5% and 25% the turbine with suction off operates in dynamic stall and applying suction at 8.5%c and 13.5%c can suppress that. Suction at 18.5%c is only effective to suppress the dynamic stall for TI ¼ 25% and suction at 28.5%c is ineffective in dynamic stall suppression for all TI. At l ¼ 3.0, for TI ¼ 1% and 5% the turbine with suction off operates in dynamic stall while due to the high turbulence level of the incoming flow, dynamic stall is already avoided at TI ¼ 25%.

Impact of turbulence intensity
Applying suction at 8.5%c, 13.5%c and 18.5%c can suppress the dynamic stall at TI ¼ 1% and 5%. Suction at 28.5%c is only effective to suppress the dynamic stall for TI ¼ 5% and not at TI ¼ 1%.
At l ¼ 3.5, only for TI ¼ 1% the turbine with suction off  experiences dynamic stall and that is avoided by applying suction at 8.5%c, 13.5%c and 18.5%c and 28.5%c. -The optimal suction location is independent of the suction amplitude.
In short, the optimal suction location is weakly sensitive to TI. Therefore, an optimal suction location set for a turbine based on numerical/experimental analyses at lower TI, can also work optimally at higher TI in operational conditions. Table 11 presents the optimal suction location as a function of chord-based Reynolds number for different suction amplitudes for TI ¼ 5%. Note that, as detailed in Table 8, to modify Re c , the values of both the freestream velocity and the turbine rotational velocity are modified. The table is prepared using 60 URANS simulations, i.e. the optimal values at each Re c are based on 20 URANS simulations. Fig. 21 shows the contour maps of the turbine power gain, in terms of DC P , in 'suction amplitude e suction location' space. The maps help to identify the optimal suction characteristics as a function of Re. Within the studied range, the following observations are made:

Impact of Reynolds number
-In general, the power gain due to suction considerably decreases for higher Re c at all l. For example, at l ¼ 3.0 with suction amplitude of 0.5%, the power gain at Re c of z0.5 Â 10 5 , z1.0 Â 10 5 and z2.0 Â 10 5 is 193%, 83% and 36%, respectively. The more prominent impact of suction at lower Re c is because at lower Re c the boundary layer on the blade is more prone to flow separation, therefore, the turbine aerodynamic performance is comparatively more influenced by separation and its suppression using suction will more significantly enhance the turbine power performance. In this regard, the impact of increasing Re c is similar to that of increasing TI. -The optimal suction location is found to move upstream by increasing Re c from z0.5 Â 10 5 to z1.0 Â 10 5 . This is because by increasing Re c the chordwise extent of the LSB reduces while the LSB onset location remains almost unaffected. For the same reason, the optimal suction location could have potentially moved further upstream by increasing Re c to 2.0 Â 10 5 , however, X S =c ¼ 8.5% is the most upstream location investigated in the present study. -The optimal suction location is independent of the suction amplitude.
In total, the sensitivity of the optimal suction location to Re c necessitates that the design of the suction location is set with special attention to the target Re c .

Discussion
Including a suction system on the blades of a wind turbine might impose several practical concerns regarding the structural strength of the blade surface near the suction location, the space needed for installation of the pumping system and the connecting tubes, the location to exhaust the pumped air and the input power for the pumping system.
In order to estimate the required input power for the suction system, a potential pumping system is considered (see Table 12). Note that the pressure losses in the tubing are not considered. For the reference operating condition, see Table 1, an estimation of the new power gain due to the use of the suction system is presented in Table 13. The turbine net power gain due to the suction at l ¼ 2.5, 3.0 and l ¼ 3.5 is 219.6%, 74.3% and 19.6%, respectively. Due to the low volume flow rate needed for the vacuum pump, the turbine net power gain by the suction is still highly significant and, thus, well justifies its employment.
In the present study, a constant suction amplitude is applied during the whole turbine revolution assuming that the pumping system provides a fixed flow rate regardless of the static pressure along the suction slot. In practice, if the flow rate of the pumping system is a function of the surface pressure along the suction slot, then the suction amplitude might vary during the turbine revolution. However, as shown in Section 4.3, the impact of suction amplitude on the turbine power gain due to suction is negligible.
Note that the sucked air can be exhausted via the blade/shaft tips on the top and bottom, which will negligibly influence the turbine structural and aerodynamic performance.

Conclusions
Applying suction through a slot near the blade leading edge, along the chordwise extent of the laminar separation bubble (LSB) and upstream of its bursting location, can effectively (i) prevent the LSB bursting; (ii) eliminate the LSB formation or postpone it to downstream of the suction slot; (iii) avoid the formation of the dynamic stall vortex (DSV) and trailing-edge roll-up vortex (TEV) and suppress the dynamic stall phenomena; and (iv) delay the incipient trailing-edge separation.
As a result, the blade lift force and the C l,max significantly increase, the drag force and the C d,max noticeably decrease, the stall angle is postponed, the aerodynamic load fluctuations are suppressed, and the slope of the C l e a curve during the downstroke increases. Consequently, the instantaneous moment remarkably grows and the turbine power coefficient is very much enhanced.
Applying suction with an amplitude of A S ¼ 0.5% for the reference turbine, near the blade leading edge at X S /c ¼ 8.5%, operating at the reference conditions, Re c z 10 5 and TI ¼ 5%, results in a power gain of DC P ¼ 247%, 83% and 24% at different tip speed ratios of l ¼ 2.5, 3.0 and 3.5, respectively. The average power gain over all l is 118%.
The study of the impact of suction characteristics, namely suction amplitude and location, on the blade aerodynamics and the turbine power performance shows that: -The impact of suction amplitude on the turbine power gain due to the suction, the turbine instantaneous moment coefficient and the skin friction coefficient along the blade suction side is insignificant within the range studies, where the minimum value of suction amplitude, i.e. 0.5%, is enough to suppress the flow separation. -The location of the suction slot has a significant influence on the obtained turbine power gain using suction. Such a location needs to be along the chordwise extent of the LSB, optimally towards the downstream end of the LSB. If the slot is positioned further upstream, it still can suppress the separation but the turbine power gain is not optimal. On the other hand, if the slot is positioned too downstream, not overlapping the LSB, the dynamic stall will not be suppressed and the suction will not work effectively.

Table 12
Features of a potential pumping system for a single turbine blade with suction settings given in Table 1.
-Parameter -Value -Unit Width, W 0.0015 m Suction velocity, V W 0.0465 m/s Volume flow rate per blade with unit span 4.185 l/min Power consumption of an off-the-shelf vacuum pump (flow rate 5 l/min, vacuum deeper than 80 kPa, voltage 12 V, e.g. Ref. [77]) 4e6 W Regarding the dependency of optimal suction characteristics on the operational parameters (namely tip speed ratio, chord-based Reynolds number and turbulence intensity), the LSB moves slightly upstream and shrinks in chordwise extent for lower l, higher TI and higher Re c . The analysis shows that the optimal suction location is insignificantly influenced by TI and minimally influenced by l while more sensitive to Re c . Therefore, the suction location needs to be set with special attention to the target Reregime. When the operational range of l, TI and Re c are known, the optimal suction location is best to be set based on the most separation-dominant regime, i.e. lower l, TI and Re c , as this is the regime where suction will most effectively enhance the turbine power performance. Of course, in the ideal case, multi-slot suction system needs to be developed so that based on the operating conditions, the optimal suction location is actuated.