2.1 Attachment line contamination

The stability of attachment line contamination can be assessed based on the magnitude of the attachment line Reynolds number, $\bar R$ , based on the results of Poll and Paisley [Reference Poll and Paisley17]. The value of $\bar R$ is dependent on the initial pressure gradient, leading-edge sweep and flow condition. For $\bar R$ above a critical value of 245, any bursts or spots of turbulence are self-sustaining and will propagate along the attachment line. The gross contamination of the wing by the turbulence originating from the fuselage can be controlled by surface suction or devices such as a ‘Gaster bump’. The attachment line boundary layer is also susceptible to TS disturbances. These waves first appear when $\bar R$ reaches a value of approximately 580. If $\bar R$ exceeds 580 then the waves amplify as they travel along the attachment line and ultimately reach some threshold condition beyond which the waves break down to form localised turbulent spots. If the local values of $\bar R$ drop below 580, the waves will be damped and eventually die out. For the range of pressure distributions constructed for the parametric study and flow conditions relevant to civil transport, the value of $\bar R$ is between 245 and 580, as shown in Figs 3 and 4. The Reynolds number, Re, in this paper is based on wing chord. This indicates that the stability of the attachment line can be controlled by surface suction or passive devices such as a ‘Gaster bump’.

Figure 3.

Effect of initial pressure gradient on $\bar R.$

Figure 4.

Effect of sweep angle on $\bar R$ .

2.2 Crossflow instability

CF instabilities are due to the inflection point in the crossflow mean velocity profile. These instabilities are strongly dependent on wing leading-edge sweep and the initial flow acceleration corresponding to a favourable pressure gradient. To suppress CF instability, suction is applied over the initial steep pressure gradient region with aim of reducing the amplitude ratio below a given N-factor in the linear stability analysis method. The suction velocity, Vs, non-dimensionalised by freestream velocity is specified as an input to the laminar boundary layer code, BL2D [Reference Atkin15, Reference Cobbin16] for transition prediction in CoDS [Reference Atkin14] in the streamwise direction. The required suction velocity is sensitive to pressure gradient and wing sweep as shown in Figs 5 and 6. A constant value of Mn = 0.69 for the Mach number normal to wing leading-edge sweep has been used for the investigations. The results show that although the level of suction required varies with Reynolds number, it tends to become asymptotic and less dependent on the initial pressure gradient and wing sweep. This is beneficial for design in terms of determining a maximum required suction velocity. The maximum level of suction velocity required is about −0.007 for a wing leading-edge sweep of 27.8° for a range of initial pressure gradients.

Figure 5.

Effect of initial pressure gradient on suction velocity.

Figure 6.

Effect of sweep on suction velocity.

There could be a large difference in the value of N-factor for transition between that of flight and wind tunnel environment as outlined in Section 2. The range could be between 6 and 20. Apart from the differences in the flow environments that could affect the value of N-factor for transition, it may also depend on the transition prediction methods used. Surface roughness can also affect the N-factor for transition. All the results presented in this paper are for an N-factor of 10 for the purpose of transition control illustrations. The sensitivity of suction velocity required to suppress CF to N-factor has been investigated in Ref. [Reference Wong and Maina3]. It was found that assuming n = 8 rather than 10 implies a 10% difference in the suction velocity over most of Reynolds number range.

For a wing leading-edge sweep of 35.6° with a low initial pressure gradient of 40, which may be associated with a cruise Mach number of 0.85 for an inboard wing section with a larger leading-edge radius, the maximum suction velocity required is about −0.009 as shown in Fig. 6. The results have provided useful guidance in the maximum level of suction velocity required for controlling CF in HLFC wing design.

2.3 Tollmien-Schlichting instability

With the assumption that CF instability has been controlled and the onset of transition delayed to aft of the minimum pressure point, then TS instability would become the dominant mode for transition. Investigations carried out at ARA [Reference Wong and Maina3] have shown that the level of suction required for suppressing TS instability is much less than for CF. For most cases a suction velocity of about –0.0002 would be sufficient for a range of rooftop pressure gradients. It was also found that it is more effective to increase the suction panel length than the suction velocity in order to delay transition, as shown in Fig. 7. This implies the structural constraint of the suction panel forward of the front spar may limit the extent of laminar flow that could be achievable. This constraint may be overcome by the use of surface cooling instead of suction downstream of the front spar for suppressing TS instability. The cooling rates required were found to be low, generally about 10% below the ambient temperature is sufficient to delay transition significantly. A transition control methodology consisting of a successive number of suction panels forward of the front spar for suppressing the CF and TS instabilities, combined with cooling panels downstream of the front spar if further extent of laminar flow is required, as illustrated in Fig. 8, could be employed.

Figure 7.

Effect of suction velocity on TS transition.

Figure 8.

Effect of control methodology on disturbance growth.

There is current interest in the use of hydrogen fuel for engine propulsion for zero carbon emissions. The use of hydrogen as an alternative fuel remains a challenge and much research is required. The research could include investigations into the feasibility of using liquid hydrogen stored in low temperature, of the order of −250°C, to provide surface cooling to suppress the flow instabilities for delaying transition as shown in the transition control methodology in Fig. 8.

2.4 Single suction chamber and variable porosity technique

In the parametric studies described in Sections 2.2 and 2.3, the surface suction velocity, V_s, was specified directly in the transition prediction method, where the effect of the external pressure gradient on the surface suction velocity has not been taken into account. This effect is most pronounced in the leading-edge region due to the steep pressure gradient associated with the initial flow acceleration. The laminar boundary layer code, BL2D [Reference Atkin15, Reference Cobbin16], has been modified at ARA to allow the plenum chamber pressure and panel porosity to be specified as input data, rather than the suction velocity. The suction velocity is calculated by an empirical correlation based on the pressure difference across the porous skin [Reference Preist and Paluch18]. For a constant chamber pressure, the external pressure gradient shown in Fig. 2 can cause a large variation in the pressure difference (Δp) across the porous panel and hence the suction velocity distribution (V _s), as illustrated in Figs 9 and 10.

Figure 9.

Pressure difference variation.

Figure 10.

Suction velocity variation.

One approach to resolving this problem is the use of a number of separate suction chambers within a double sheet suction surface [Reference Horstmann, Schrauf, Sawyers and Sturn4], where the suction velocity in each chamber is controlled by orifices in the inner sheet and a constant speed suction pump directly coupled with the leading-edge box. To minimise the number of plenum chambers in order to reduce system complexity, ARA has developed an alternative approach to vary the suction panel porosity to balance the effect of pressure gradient [Reference Wong, Maina and Doig5]. A parameter that affects the panel porosity is the ratio of hole spacing to hole diameter (S _d), where a higher value would give a lower suction velocity. In order to provide a uniform suction velocity across the region with an initial steep pressure gradient, a higher initial value of S_d can be applied at the leading-edge with the value reduced linearly with distance, as shown in Fig. 11. It can be seen that the suction velocity distributions in Fig. 12 given by the porosity variation are now much more uniform across the initial pressure gradient region. The level of the suction velocity can be further controlled by the hole diameter and/or skin thickness.

Figure 11.

Porosity variation.

Figure 12.

Suction velocity variation.

5.1 Crossflow instability

The single suction chamber and variable porosity technique with a turbulent type of pressure distribution for HLFC design, has been applied to an Airbus A330 aircraft outer wing retrofit, under a EU-funded research programme called AFLONEXT (Active Flow, Load & Noise Control on next generation aircraft). The aim of the investigation of the single chamber and variable porosity concept was to establish the following potential advantages over the current multi-chamber concept:

1. 1. Reduced structural and system complexity.

2. 2. Reduced weight.

3. 3. Potential for incorporating different types of wing ice protection systems (WIPS), such as a conventional hot air WIPS, or a skin concept with electric wing ice protection systems (EWIPS).

Details of the investigation are described in Ref. [Reference Wong and Walters7]. This paper will only outline some aspects of the findings. Various degrees of pressure gradient for the upper surface and their effect on transition control and aerodynamic performance have been investigated [Reference Green and Wong8]. The 2D section geometry that has been used as a basis for the study has been derived from a ‘normal-to-leading-edge’ cut of the outboard end of the HLFC panel. The design constraints were to maintain the maximum thickness/chord ratio and aft part of the geometric profile same as the baseline section. A range of design lift coefficient, C _L, from 0.48 to 0.63 for the section in the streamwise direction, for a cruise Mach number of M _∞ = 0.82 has been specified by Airbus for the HLFC study. For the results presented here, a sectional C _L of 0.58 in the streamwise direction within the range of the design target has been considered. The corresponding C_L and Mach number normal to the wing leading-edge sweep is 0.76 and 0.708, respectively, as shown in Fig. 21. The pressure distributions for three of the sections for a same design C _L that have been derived based on the baseline geometry in the design investigation are shown in Fig. 21. The wing section, Design A, has a favourable type of pressure distribution, and Designs B and C both have a turbulent type of pressure distribution with an adverse pressure gradient and a different suction peak level in the leading-edge region. A lower leading-edge suction peak would have a higher suction velocity for a same suction chamber pressure than a higher suction peak. It can be seen from Fig. 22, for a given design C _L, the lift to drag ratio, L/D, for Designs B and C with a turbulent type of pressure distribution is higher than Design A with a favourable gradient type of pressure distribution, in line with the findings in Section 4. The turbulent designs also have a weaker shock and lower drag rise than the design with a favourable type of pressure gradient, as shown in Figs 21 and 23, respectively. The results presented in Figs 21–23 are with HLFC, but the scales on the axes could not be shown due to commercial confidentiality.

Figure 21.

2D pressure distributions, M = 0.75, M_2D = 0.708, C _L = 0.76, Re = 19.81 × 10⁶.

Figure 22.

L/D performance.

Figure 23.

Drag performance.

For the designs with a turbulent type of pressure distribution, it is possible to use a single suction chamber with variable porosity to delay transition as far downstream as the shock location. For the pressure distributions shown in Fig. 21, the steep pressure gradient occurs close to the leading edge, within the initial 2% chord, and its variation with distance over this region is nearly linear. Based on the transition control methodology established in Ref. [Reference Wong, Maina and Doig5] outlined in Section 2.4, a higher value of S _d can be applied at the leading edge and then reduced linearly with distance. The variation of porosity with distance could be defined in various ways, and it was shown that two types of porosity variation as shown in Fig. 24 could be used to provide an appropriate suction velocity distribution for controlling transition. One of these cases is a linear reduction in S _d by 50% over the initial 2% chord followed with a constant porosity to 20% chord, defined as ‘two-part’ variable porosity distribution, and the other is simply a linear variation from the leading edge to 20% chord.

Figure 24.

Porosity distributions, d = 50 µm.

Figure 25.

Design C suction velocity distribution, chamber pressure = 13250 Pa, d = 50 μm, t = 3 mm.

Figure 26.

Effect of skin thickness on suction velocity distribution, Design C, chamber pressure = 13250 Pa, d = 50 μm.

For surface suction, the chamber pressure needs to be lower than the external surface pressure distribution, otherwise blowing would occur, which could cause transition. A lower chamber pressure would give a higher suction velocity and mass flow but requires a higher suction pump power. For the single chamber and variable porosity technique, the suction pump power requirement could be minimised by setting the chamber pressure at a level just low enough to avoid blowing. This chamber pressure is sufficient to provide an appropriate suction velocity across the porous panel for controlling the flow instabilities to delay transition. This is illustrated in Fig. 25, with a higher suction velocity in the initial steep pressure gradient region for controlling CF instability and a reduced level downstream of this region for controlling TS instability. The level of suction velocity distribution can be further controlled by a combination of the porosity variation, suction hole diameter and skin thickness, in order to achieve an appropriate level for delaying transition. For example, Fig. 26 shows that the level of suction can be increased by reducing the skin thickness. It should be noted that for the constant porosity distribution, it has not been possible to achieve an appropriate level of suction velocity distribution; it is either too high in the initial steep pressure gradient region (between 0 to 2% chord), which may have the problem of over-suction, or too low downstream of this region to delay transition. It was found that transition is mainly due to CF in the leading-edge region. A multi-chamber system could be used, but it would be more complex and difficult to manufacture and have a greater weight penalty than a single-chamber variable porisity system as shown in Ref. [Reference Wong and Walters7]. The single chamber and variable porosity concept can be easily adapted for different porosity distributions with various hole diameters and skin thicknesses. Future research is required to study the trade-off between the structural stiffness, weight and laser drilling in terms of hole spacing, hole diameter and skin thickness for the porous skin.

5.2 Structural integration

The single chamber and variable porosity suction system has been integrated on the computer-aided three-dimensional interactive application (CATIA) into the leading-edge wing structure of the Airbus A330 aircraft outer wing retrofit, with the WIPS and high lift system [Reference Wong and Walters7]. The structural integration was based on the modification of an existing multi-chamber geometry, with most of the structural features maintained. The resultant structural system is less complex than the multi-chamber geometry, and with an appreciable weight saving as shown in Table 1.

Table 1.

Comparison of weight estimates

The single chamber consists of an outer porous skin and a solid inner skin with suction holes. The multi-chamber internal structures have been totally removed to create a single chamber, with the depth of the chamber between the outer and inner skin remaining the same. The multi-chamber system cannot be shown here for comparison due to company proprietary considerations with the company that has developed it. The outer skin with constant porosity of the multi-chamber geometry has been replaced with a skin with variable porosity distribution in the chordwise direction. There are three rows of internal ribs in the suction chamber to support the outer skin under load deflection as shown in Fig. 27. These ribs are evenly spaced across the span, with cuts to enable air to pass through in the spanwise direction. These ribs could be placed across the span, if necessary, instead of in the chordwise direction as currently shown. The inner skin has four rows of suction holes to minimise the spanwise variation in suction as shown in Fig. 27. The suction air is passed through these suction holes into the collector ducts underneath the inner skin as shown in Fig. 28. The collector ducts are connected to two trunk pipes via two collector pipes, one for the suction air supply and the other for the hot air supply for the WIPS.

Figure 27.

Inner skin with suction holes and internal ribs.

Figure 28.

Collector ducts and collector pipes underneath the inner skin.

The weight estimates for the part of the suction skin structural system for the single chamber and variable porosity concept illustrated in Fig. 28 are shown in Table 1. The weight for an equivalent structural system for the multi-chamber concept is also included for comparison. The weight estimation carried out for the HLFC structure as shown in Fig. 28 is purely for assessing the relative weight between the two concepts. It can be seen in Table 1 that the weight for single chamber concept is about 44% lighter than the multi-chamber concept with a wing span of 2.31 m. For the whole wing structure of an aircraft, the single chamber and variable porosity concept would have a significant weight saving compared with the multi-chamber concept.

7.1 Surface imperfections

Surface imperfections in the wing leading-edge region could result from the structural integration of various components such as the high-lift devices and anti-icing system, where small steps or grooves may occur between these components. The effect of steps and grooves on transition has been investigated in the ARA TWT [Reference Lawson and Ciarella10] under the Advanced Laminar Flow Enabling Technologies (ATI/IUK ALFET) programme led by Airbus, which aimed to develop the understanding and tools required to design and build a natural laminar flow wing. A ring wing model shown in Fig. 35 was designed and manufactured with an NLF aerofoil section and provision to install a variety of surface imperfections. The model was manufactured with a high-quality surface finish and fitted with instrumentation to measure laminar flow extent, including an insulating panel for use with IR thermography and hot films as shown in Fig. 36.

Figure 36.

Side view of the installed measurement system.

Figure 37.

Infra-red example data.

Figure 38.

Effect of groove parameter on transition onset location.

Figure 39.

HLFC wind tunnel model.

The model was tested over a range of Mach numbers from M _∞ = 0.574 to M _∞ = 0.785 for grooves of different depths, defined by a groove parameter. An example of an IR image showing transition onset locations is illustrated in Fig. 37. The location for transition onset and completion were obtained from the intensity ratio distribution from the IR data as shown in Fig. 37 for a slice at location 5.2.1. A large quantity of transition data for various groove parameters and Mach numbers were obtained. An example test result showing the effect of the groove parameter on transition onset location is presented in Fig. 38. The data are widely scattered, and it is difficult to deduce a definitive effect of various groove parameters on transition. It is acknowledged that uncertainty limits in the data should be indicated but not available. The line in Fig. 38 is an average curve fit through the data to aid visually for any trends in the scattered data points. However, it was found that the groove has the effect of forcing transition forward more strongly at lower than at higher Mach numbers [Reference Lawson and Ciarella10]. The transition onset location and definition of the grove parameter cannot be disclosed in this paper due to commercial confidentiality. The example of results presented here are to show that ARA has acquired the wind tunnel testing capability for measuring the effect of small features on transition in NLF, and the knowledge gained could be extended to HLFC to investigate the effect of suction on the disturbances due to various degrees of surface imperfections in the future.

Figure 40.

Two-dimensional pressure distribution, M _2D = 0.70, M _∞ = 0.85, Λ = 35°, α = 1.8°.

7.2 HLFC and variable porosity

A HLFC wind tunnel model with a variable porosity surface has been tested in ARA TWT [Reference Lawson, Ciarella and Wong11]. The HLFC wind tunnel model was specifically designed in a modular form, with the mid-span leading-edge region interchangeable for various porous panels, as shown in Fig. 39. The aim of the wind tunnel experiments was to demonstrate, for the types of pressure distribution relevant to civil transport aircraft design, the effectiveness of a range of transition control techniques. Additional test objectives included generating a data set for method validation and to evaluate the suction properties of porous plates. The pressure distribution of the aerofoil section normal to leading-edge sweep of 35° for the design condition at M _∞ = 0.85 is shown in Fig. 40. It was not possible to compare the design pressure distribution with measurement as the wind tunnel was not able to run at M = 0.85 during the test due to mechanical problems. Three different panels were tested: a solid panel, a constant porosity panel and variable porosity panel. The porous panels had 45 $\unicode{x03BC}$ m diameter holes, providing porosity from the leading edge to 20% chord on the upper surface. The constant porosity panel had a hole spacing to hole diameter ratio (S _d) of 12, while the variable porosity panel had a linear variation of S_d from 12 at the leading edge to 8 at 5% chord returning back to 12 at 20% chord, as shown in Fig. 41. The hole diameter and porosity were determined based on the concept established in earlier research described in Section 2, to provide an appropriate level of surface suction to control CF and TS to delay transition sufficiently downstream in wind tunnel conditions. Wind tunnel tests using the HLFC model are described in Ref. [Reference Lawson, Ciarella and Wong11], and the analysis of the results using local linear stability [Reference Atkin14] and linear parabolised stability equations analysis [Reference Mughal23, Reference Mughal24] are given in Ref. [Reference Ciarella, Lawson and Wong21]. This work was undertaken as part of the Aerodynamic Research Testing Capability and Data (ARCADE) programme, a research project funded by the UK Department of Business, Energy & Industrial Strategy (BEIS) and Innovate UK (IUK).

Figure 41.

Hole spacing to diameter ratio distribution for variable and constant porosity panels.

Figure 42.

Transition N-factor correlated with I-R at M _∞ = 0.7.

In linear stability analysis for transition prediction, there are various approaches in determining the N-factor for transition. For example, in the analysis of the wind tunnel test data carried out in Ref. [Reference Mughal24], separate values of N-factor for transition onset due to CF and TS instability modes have been derived, whereas a variation in the N-factor between CF and TS instability modes for transition has been used by various researchers. In the analysis of ARA TWT test data [Reference Allen, Lawson, Maina and Alderman9] as outlined in Section 6.1, a single value of N-factor of about 7.5 for transition has been determined.

Figure 43.

Wave angle at transition location at M _∞ = 0.7.

Figure 44.

Measured pressure distributions in the freestream direction at M _∞ = 0.7, α = −1.24°, α = 1.82°.

In the analysis of ARA TWT data for the HLFC model with a solid panel surface [Reference Ciarella, Lawson and Wong21], it was found that the N-factor for transition varies with angle of incidence as can be seen in Fig. 42, which is contrary to a single N-factor value that was previously assumed [Reference Allen, Lawson, Maina and Alderman9]. The wave angle at transition location for these cases is shown in Fig. 43. Based on the wave angles, transition is due to CF at negative angle of incidence and TS at higher angle of incidence. The results are consistent with the pressure distribution on the upper surface, with a favourable gradient at negative angles of incidence for CF and an adverse gradient at positive angles of incidence for TS, as shown in the pressure distributions presented in Fig. 44 for a case at M _∞ = 0.7. It should be noted that the adverse pressure gradient featuring in the initial 20% chord region shown in Fig. 40 and the α = 1.82° case in Fig. 44 was specifically designed to give TS transition at positive angles of attack for the wind tunnel tests in ARA TWT. It was apparent that the N-factor is particularly lower for conditions corresponding to cases with the presence of both strong CF and TS waves, identified as oblique waves in Fig. 42 for cases near zero angle of incidence. The results suggested that the low N-factor may be due to the limitation inherent in the linear methods for cases where there is a possibility of interaction between CF and TS waves. A preliminary investigation using linear and non-linear parabolised stability equations (PSE) methods [Reference Mughal23, Reference Mughal24] has also been carried out [Reference Ciarella, Lawson and Wong21]. The non-linear method requires an initial inflow state to seed the simulation, and the initial condition can affect the growth of the instability waves. Although the results with an assumed initial inflow state have provided support for the most likely instability waves that are responsible for the onset of transition as identified in the linear method, it requires detailed information of the tunnel environmental fluctuation measurements in order to provide a more accurate simulation. Detailed measurement of the tunnel environment and fluctuations within the boundary layer are required in the future to provide data for transition prediction investigations using methods such as non-linear PSE.

Figure 45.

Chamber pressure difference across panel, M _∞ = 0.7, α = 1.82°, chamber pressure = 59825 Pa.

For porous panel cases, it was shown that the variable porosity panel has been able to provide a higher suction velocity than the constant porosity panel, as the design has intended. See Fig. 41 for the porosity distribution, where the variable porosity panel has a lower S_d than the constant porosity panel. This is illustrated in the results presented in Fig. 45 for M _∞ = 0.7 and α = 1.82° case with a chamber pressure of 59825 Pa, where the total suction mass flow is higher for variable porosity than constant porosity. It can be seen that for this chamber pressure, blowing is occurring between 2 and 6% chord indicated by the top curve in Fig. 45, where the pressure difference across the panel is negative. The pressure difference shown in Fig. 45 is given by the chamber pressure minus the pressure on the external surface of the porous panel, hence blowing occurs when the pressure difference is negative. The pressure on the external surface of the porous panel is given by the pressure distribution of the wing section measured by pressure taps, and the chamber pressure measured by pressure transducers. For this case, it is the pressure distribution at α = 1.82° shown in Fig. 44. In HLFC, the chamber pressure needs to be set low enough to avoid blowing, as blowing can cause transition. The suction mass flow rate for the variable porosity panel is higher than the constant porosity panel as shown in Fig. 45, due to the fact that the overall hole space to hole diameter ratio, S_d, for the variable porosity panel is lower than the constant porosity panel as shown in Fig. 41. For the constant porosity panel, blowing has caused transition within the porous panel region based on infra-red (IR) images recorded in the wind tunnel test. However, for the variable porosity panel case, although blowing was present as shown in Fig. 45, the amount of suction achievable downstream of blowing was able to stabilise the flow delaying the onset of transition at about 50% chord according to the IR images recorded in the wind tunnel test [Reference Ciarella, Lawson and Wong21]. This finding is unexpected and should be further investigated in the future, as this would have important implications on the possibility of using surface suction to stabilise the instability due to the effect of surface imperfections on transition.