The experiment is conducted in the low-speed low-turbulence tunnel (LTT) of Delft University of Technology

https://www.tudelft.nl/en/ae/organisation/departments/aerodynamics-wind-energy-flight-performance-and-propulsion/facilities/low-speed-wind-tunnels/low-turbulence-tunnel/ (last access: 8 March 2022).

The main element is a composite DU00-W2-401 airfoil model with a chord length of cmain=0.5 m. The model was mounted vertically in the test section on two aluminium attachment plates flush with rotating discs in the upper and lower tunnel wall that provide changes in angle of attack. The slat profile is a custom, cambered airfoil with a chord length equal to 0.3 cmain, as this was shown to lead to good performance in . The coordinates of the slat element are available in . The slat is 3D-printed with an internal structure consisting of stiffening ribs and a hexagonal steel rod, providing attachment to the aluminium plate and taking up the loads on the slat at high wind speeds (see Fig. , left). The slat surface is smoothed with sandpaper and finished with spray paint. In this study, the streamwise position of the slat trailing edge is fixed at sslat=0.151 cmain. The effect of two parameters are investigated, as shown in Fig. : (i) the gap width hslat between the main airfoil and the slat trailing edge and (ii) the slat angle βslat relative to the main airfoil. In order to vary these parameters, the steel rod has pre-manufactured set points in the mounting plate corresponding to pre-determined values of hslat and βslat. Figure  shows computer-aided design (CAD) views of the cross-section of the slat (left) and the end plate (right). The set of configurations investigated here, as well as their associated labels, is presented in Table . The values of the slat angle βslat are chosen based on previous works , which show that a slat angle of around 20∘ leads to the best performance. In order to assess the sensitivity of the results to βslat, three different values separated by 5∘ are considered here, namely βslat=16.4, 21.4, and 26.4∘. It is also expected that reducing the gap width hslat, while avoiding confluent boundary layers, increases the positive effect brought by the slat. However, the results obtained by with a 2 % gap width (i.e. hslat/cmain=0.02) led to smaller lift coefficients, smaller lift-over-drag ratios, and a similar stall angle than with a 4 % gap width. In this study, three values of the gap width varying between 2 % and 4 % are therefore investigated.

In order to mitigate interference effects due to the tunnel wall boundary layers, pairs of vortex generators (VGs) are installed on the main profile close to the walls, on both the pressure and suction sides of the main airfoil, as shown by Fig.  (showing the suction side only). The geometrical characteristics of these VGs, as defined in Fig. , are as follows: h/cmain=0.01, l/h=3.4, z/h=7, s/h=2, and β=16.4∘. As it is shown in Sect. , this helps reduce the three-dimensional flow effects on the measured pressure distributions. Since vortex generators are commonly used to improve the performance of wind turbine blades, this paper also presents results obtained when placing VGs on the main airfoil alone. In that case, the VGs are placed at x/cmain=0.35, which is a good compromise between increasing lift coefficient and reducing the associated drag penalty. The parameter values of these VGs are as follows: h/cmain=0.01, l/h=2.2, z/h=7.6, s/h=3.2, and β=15.8∘. The results obtained with the VGs alone are compared with the performance values obtained with a slat.

Finally, all the tests are performed with both a clean and a tripped main airfoil. For the tripped cases, a zig-zag turbulator is placed on the main airfoil, as shown on the right photograph in Fig. . The details about the setup are given in Sect. . Note that an analysis on different tripping methods has not been performed and could be useful for future studies.

The lift and drag coefficients on the airfoil and slat can be determined by using a combination of pressure measurements on the airfoil surfaces and using a wake rake. On the airfoil surfaces, the normal force coefficient Cn′ is obtained by integrating the pressure distribution from 94 and 13 pressure orifices in the main element and the slat, respectively; that is, 1Cn′=∫01Cp,l-Cp,udxc, where Cp,l and Cp,u are the pressure coefficients on the lower and upper surface, respectively, and c is the chord of the element being considered (either main element or slat). Note that primed coefficients denote uncorrected values. Also, the low number of pressure tabs on the slat leads to an integration error of about 11 %. This is quantified by comparing the integration of Cp distributions obtained from Xfoil using, on the one hand, all the slat coordinates and, on the other hand, only the data at the locations of the pressure tabs. The drag coefficient pre-stall is measured using a wake rake placed behind the airfoil. The wake rake has 67 total pressure tubes and 16 static pressure tubes. The uncorrected drag coefficient Cd′ measured with the wake rake is given by 2Cd′=2c∫wakeCp,t-Cp,s1-Cp,tdy, where Cp,t and Cp,s are the total and static pressure in the wake, respectively. When the wake becomes unstable or wider than the wake rake, the latter cannot be used to measure drag. Therefore, post-stall, the pressure measurements on the airfoil and slat surfaces are used instead. The angle of attack at which the switch is made between wake-rake drag and pressure drag is determined based on identifying flow separation from the measurements. These angles also correspond to Cl values that are slightly smaller than the maximum for a given case. The exact values are reported in Appendix . For small angles, the measured lift coefficient Cl′ is expressed as 3Cl′=Cn′cos⁡(α)-Cd′tan⁡(α), where α is the angle of attack. The measured lift, drag, and angle of attack are corrected following the methods described in and ; i.e. 4Cl=Cl′1-t1-t2+t3-t4, where 5t1=σβ2, 6t2=5.25σ2β4,7t3=2-M2β3Λσ1+1.1βctα2,8t4=2-M21+0.4M24β2chCd′, and σ=π2/48(c/h)2 is the wind tunnel blockage factor, β=1-M2 is the compressibility factor, M is the measured apparent upstream Mach number, c is the element chord length, h is the wind tunnel height, t is the airfoil thickness, and the body shape factor is set to Λ=0.9087. Independently of whether wake-rake or pressure drag is used, the resulting measured drag coefficient Cd′ is corrected as 9Cd=Cd′1-ΔCd-t3-t4, where the wake buoyancy correction is ΔCd=0 when using the wake rake, and otherwise 10ΔCd=Λσ1+0.4M2β31+1.1βctα2. It must be noted that the present corrections are only valid until about an angle of attack of 20∘. In situations where significant separation plays a role, these equations underestimate the effect of wake blockage and consequently may give too optimistic lift coefficients. As this is mainly a comparative study, no effort has been undertaken to account for the additional blockage due to separation. Also, all the results are shown up to an angle of attack of 20∘.

Data are recorded using an electronic data acquisition system and are online reduced to show corrected force and moment coefficients and pressure distributions. A thermal camera enables us to visualise the location of flow transition along with the pressure distribution. Wool tufts are also placed on the base airfoil to have a measure of flow separation and identify possible three-dimensional wall effects during testing.

Figure  shows the lift coefficient Cl as a function of the angle of attack α for different slat configurations, a clean main airfoil (i.e. no tripping), and at a Reynolds number of Re=1.5×106. The black dots represent the lift coefficient in the absence of a slat. In order to verify the experimental setup, the results of Xfoil simulations are also shown (continuous black line). There is a good agreement between Xfoil and experimental results in the linear part of the polar curve. The experimental results show that the maximum lift coefficient on the main airfoil is reached at around α=10∘ and remains close to unity as the angle of attack is further increased. At α≈13∘, the lift coefficient presents a slight increase which is not expected. This feature did consistently appear for multiple repetitions of the test, even after cleaning the main profile. Therefore, a possible reason for this small disparity could be related to three-dimensional wall effects. Additionally, running the test at a slightly higher Reynolds number (Re=2×106) made this feature disappear. For α>20∘, the main airfoil enters deep stall and the experimental results are disregarded. When a slat is added, the disparity around α≈13∘ also disappears. Tuft visualisations, such as the one presented in Fig. , were performed up to α=11∘ and demonstrate that the flow is rather two-dimensional as expected. The total maximum lift coefficient is significantly increased (factor up to 2.5) in comparison to the case without slat. For a given gap between the slat and the airfoil, the lift coefficient also increases as the slat angle βslat decreases. This is valid across the set of parameters investigated here. Note that with the smallest gap and angle investigated, i.e. hslat/cmain=0.02 and βslat=16.4∘ (configuration A), the flow stalls already at around α≈15∘, which is smaller than in the other cases. Changing the slat gap width, for a given slat angle, does not significantly change the lift coefficient at a given angle of attack. Similarly to Fig. , Fig.  shows the drag coefficient Cd as a function of the angle of attack. At all angles of attack, the drag coefficient with slat is larger than that without slat, which is expected as the slat introduces additional friction losses. The benefit of the slat on the overall aerodynamics is highlighted in Fig.  that shows the ratio of lift-to-drag coefficients as a function of α. Two types of symbols are used: dots represent the data for the ensemble airfoil + slat, whilst crosses are the lift-to-drag values integrated on the main airfoil only. It is clear that, for all the configurations considered here, the overall aerodynamic performance values of the ensemble airfoil + slat are improved in the presence of the slat, at least when α≥3∘. It is interesting to note that, for α≈8∘, the lift-to-drag ratio computed on the main airfoil only (coloured crosses) is smaller than the lift-to-drag ratio of the airfoil without slat (black dots). Thus, at small angles of attack, the increase in aerodynamic performance for the slat configurations is largely due to the slat itself. In particular, the presence of a slat decreases the lift force on the main airfoil, whilst it has a negligible effect on the drag force (not shown). This is in agreement with the description of . By contrast, at angles of attack larger than 8∘, the lift-to-drag ratio on the main airfoil alone (coloured crosses) is larger than that on the airfoil without slat (black dots). This means that the slat changes the flow field around the main airfoil in such a way that the aerodynamic performance of the main airfoil itself is increased. As expected, this increase is further amplified when the contribution of the lift force on the slat is added (coloured dots). Again, the slat has a negligible effect on the drag force experienced by the main airfoil, and the lift-to-drag ratios are consistent with those obtained by CFD simulations in similar conditions as shown by .

Figure  shows the pressure distributions of the clean airfoil (left) and the airfoil with slat in case A (right) at an angle of attack α=10∘. It is clear that the pressure distribution on the main airfoil suction surface is significantly impacted by the presence of the slat. Due to the circulation on the slat element, the low-pressure peak on the main airfoil is reduced, also reducing the adverse pressure gradient towards the trailing edge. This enables the flow to stay attached up to the trailing edge, while the model without slat shows separation right from the mid-chord location.

This is mimicked by the associated infrared images presented in Fig. , with flow coming from the right. The model without slat (left) shows a small laminar separation bubble followed by transition indicated by the thin dark band at about one-third of the chord. Separation may be hard to see in this picture. It is indicated by the lightest grey area. Both transition and separation locations may have been slightly influenced by the presence of the pressure orifices located in the lower part of the picture. Transition on the slat can be clearly seen in the right infrared picture, indicated by the dark band between the dashed lines. Its location coincides with the kink in the pressure distribution after the pressure peak on the slat leading edge. Due to the low Reynolds number of the slat, transition will almost certainly be realised through a small laminar separation bubble. It is followed by turbulent attached flow. Although the leading edge of the airfoil is not visible due to the presence of the slat, combining the pressure distribution and the infrared image shows that the main airfoil upper surface is laminar up to the mid-chord position, followed by an attached turbulent boundary layer.

In order to assess the sensitivity of the results to surface roughness, the boundary layer on the main airfoil is tripped using a zig-zag turbulator tape placed at the 10 % chord location on both the pressure and suction sides. The tape height is calculated according to using a critical Reynolds number of 400. This leads to a tape of thickness 0.2 mm. Furthermore, the tape has a width of 6 mm and an angle of 30∘. Figure  shows the lift coefficient Cl as a function of the angle of attack α for different slat configurations. The black dots again represent the lift coefficient in the absence of slat. As expected, but still highly undesirable, tripping the boundary layer significantly reduces the lift coefficient at low angles of attack. The negative effect of the zig-zag tape can be alleviated by using vortex generators, as shown in Fig.  for the main airfoil without slat. Note that the small abnormality in the lift coefficient at α≈13∘ for the clean case is not observed in the tripped case (with and without VGs). When the slat is mounted on the main airfoil, the zig-zag tape also leads to a loss of Cl at small angles of attack, although this effect is less pronounced than in the absence of slat (Fig. ). This observation is also made on the lift contribution coming from the main airfoil (not shown). At negative angles of attack, the lift coefficient on the ensemble airfoil + slat is almost identical to that of the airfoil without slat. Consequently, the presence of the slat decreases the contribution of the lift force coming from the main airfoil. This is not the case for positive angles of attack. In particular, for α>5∘, the slat has no influence on the lift force experienced by the main airfoil only, as that contribution is identical to the lift force on the main airfoil without slat (not shown). This explains why the lift coefficient on the ensemble airfoil + slat is then larger than that of the airfoil without slat (Fig. ). Additionally, similar trends than with the clean airfoil are obtained, namely an overall increase in Cl in the presence of a slat, with larger Cl as βslat decreases. Also, there is a small decrease of Cl as hslat increases, although this is very small.

The drag coefficients are presented in Fig. . As expected, drag is larger in the tripped case compared to the clean results. However, the presence of the slat slightly decreases Cd at all positive angles of attack before stall. As for the results on the clean airfoil, the contribution of the slat to the total drag force is very small. The positive effect that the slat has on the overall drag reduction is therefore mostly due to the fact that the slat changes the flow field around the main airfoil. Figure  shows the lift-to-drag ratio for all the tripped configurations. As already mentioned before, the overall aerodynamic performance of the tripped system is smaller than in the clean case. The addition of a slat has a positive influence on the overall aerodynamic performance. However, as opposed to the clean cases, the ratio Cl/Cd computed on the airfoil alone (in the presence of a slat) is rather similar to that of the airfoil without slat (not shown here for the sake of graph clarity).

Figure  shows infrared images of the flow on the airfoil at α=10∘, without (left) and with (right) slat. In the right figure, the red dashed line indicates the slat leading edge, whilst the blue dotted line shows the slat trailing edge. It can be seen that, in both cases, the flow is rather two-dimensional as expected. Also, the presence of the slat alleviates flow separation on the main airfoil. This is also apparent on the pressure coefficient shown in Fig.  for the same conditions.

Figures  and  show the polars for the lift and drag coefficients, respectively, when the main airfoil is equipped with VGs instead of a slat (red dots). For comparison, results obtained with the slat under configuration A are also shown in these plots (blue dots). Again, both clean (left) and tripped (right) conditions are shown. In both conditions, it is clear that the lift coefficient is considerably increased due to the presence of the VGs, which suppress separation. The increase is however smaller than when a slat is used. Furthermore, at small positive angles of attack, the lift-to-drag ratio is reduced in the presence of the VGs in the clean case (Fig. , left), which is not the case with a slat. The aerodynamic performance values obtained with VGs are larger than with the slat only in the tripped conditions and for 6∘<α<14∘, where the small gain in Cl brought by the slat is not enough to compensate for the small associated increase in Cd. Outside of this range of angles of attack, the slat seems to be aerodynamically more advantageous than VGs, at least for the conditions investigated here. Of course, the slat brings additional challenges for the structural design of the blade, which are not taken into account in this study.