Proceedings of PACAM XII 12th Pan-American Congress of Applied Mechanics January 02-06, 2012, Port of Spain, Trinidad WING DESIGN WITH A TWIST: OPTIMISED GEOMETRIC TWIST OF THE GRAND TOURING SPORTS CAR WING Pascual Marqués-Bruna, marquesp@edgehill.ac.uk Faculty of Arts & Sciences, Edge Hill University, St. Helen’s Road, Ormskirk, Lancashire, UK Abstract. Wing twist gives the Grand Touring (GT) sports car an eye-catching ‘aerodynamic shape’, improves downforce and can enhance aerodynamic efficiency if the twist is optimised. This study aimed to explore the influence of optimum total twist and optimum twist distribution upon wing aerodynamics, using numerical analysis. Straight baseline inverted wings with constituent NACA 651-412 and NASA LS(1)-0413 airfoils and linear taper were designed. Complete wings consisted of the baseline wings fitted with a Gurney flap and end plates. Optimum geometric twist was subsequently determined and applied. Aerodynamic parameters including downforce, induced drag and profile drag were computed using a modified version of the classical Prandtl’s Lifting-line Theory. Drag polars were constructed. The numerical analysis suggests that the twist-optimised wing duplicates the ideal aerodynamic performance of a wing of elliptic planform and yields high downforce and minimum possible induced drag. The findings support the implementation of the classical analytical lifting-line model and optimum wing twist at the conceptual stages of GT sports car design. Keywords: geometric twist, GT sports car, lifting-line theory, optimisation, wing design 1. INTRODUCTION Modern aeroplane wings are constructed with geometric twist in a washout configuration characterized by a progressive reduction in the angle of incidence β from wing root to wing tip (Anderson, 2007). Washout helps control wing stall, retain aileron function and reduce induced drag (Bertin, 2002). The wing lift coefficient −πΆπΏ (downforce in an inverted wing) and induced drag coefficient πΆDπ distributions vary along the span π. Similarly, the wing of the GT sports car requires optimised aerodynamics to attain high downforce, aerodynamic efficiency and improved safety while conforming to the Federation Internationale de l’Automobile’s (FIA) technical regulations for car design (FIA, 2010). However, there is a scarcity of research on the influence of optimised twist upon sports car wing aerodynamics. Straight wings of elliptic planform attain minimum πΆDπ and profile drag coefficient πΆD (Milne-Thomson, 1973), but are more expensive to manufacture than rectangular wings. The tapered wing is typically used as a compromise between the elliptic and the rectangular planform. However, Phillips et al. (2006) has explained that the tapered wing requires an optimum total wing twist Ωπππ‘ and optimum spanwise twist distribution π(π)πππ‘ to generate an elliptical −πΆπΏ distribution, minimum πΆDπ and replicate the high aerodynamic performance characteristic of a straight wing of elliptic planform (Phillips, 2004). Recent developments of Prandtl’s Lifting-line Theory (Phillips, 2004; Phillips et al., 2006; Phillips and Alley, 2007) allow computing Ωπππ‘ & π(π)πππ‘ for the design of a high performance wing. This study aimed to explore the influence of geometric Ωπππ‘ & π(π)πππ‘ upon −πΆπΏ , πΆDπ and πΆD in wings with linear taper. It was hypothesised that: H1 - application of Ωπππ‘ & π(π)πππ‘ yields a near-elliptical −πΆπΏ distribution that gives high downforce and minimum πΆDπ . The findings may be used for the implementation of optimised geometric twist at the conceptual stages of GT sports car wing design. 2. METHOD 2.1. Airfoil and wing geometry Wing prototypes for the GT sports car were designed using National Advisory Committee for Aeronautics (NACA) 651-412 and National Aeronautics and Space Administration (NASA) LS(1)-0413 extensive laminar flow airfoils (Abbott and von Doenhoff, 1959). The design lift coefficient ππ0 is 0.4 for both airfoils and the zero-lift angle of attack πΌπΏ0 is -3° and -4° for the 651-412 and LS(1)-0413 airfoils, respectively (Abbott and von Doenhoff, 1959; Bertin, 2002). The leading edge of the 651-412 is of small-radius, which makes this airfoil aerodynamically efficient at low incidence (Milne-Thomson, 1973). The LS(1)-0413 is stall resistant and uses a concave pressure recovery that generates high downforce (Bertin, 2002). Straight baseline inverted wings with linear taper were constructed with the following geometry and aerodynamic efficiency: π = 1900 mm, chord π at the wing mid-span = 360 mm and at the wing tip = 324 mm, taper ratio π π = 0.90, aspect ratio π΄π = 5.56, planform area π = 0.65 m2, induced drag efficiency factor πΏ and lift efficiency factor π = 0.050, and Oswald efficiency factor π = 0.95. Straight complete wings included a Gurney flap on the pressure side of height βπΊ = 15 mm and end plates of height βππ = 150 mm to comply with technical regulations (FIA, 2010). Subsequently, the wings were twisted geometrically by applying Ωπππ‘ & π(π)πππ‘ twist mode (Fig. 1). The twist consisted of washout using π½ at the wing tips and π½ + Ωopt at the mid-span and was modelled numerically by Proceedings of PACAM XII 12th Pan-American Congress of Applied Mechanics January 02-06, 2012, Port of Spain, Trinidad elliptically modifying π½ along π (Anderson, 2007). Wing twist and aerodynamic computations were carried out using spanwise π station intervals of 0.01 m (Phillips and Alley, 2007). Thus, 8 wing prototypes were constructed: straight baseline 651-412, straight complete 651-412, twisted baseline 651-412, twisted complete 651-412, straight baseline LS(1)-0413, straight complete LS(1)-0413, twisted baseline LS(1)-0413 and twisted complete LS(1)-0413. Figure 1. A complete elliptically-twisted wing fitted with a Gurney flap and end plates installed on a GT sports car 2.2. Computation of wing aerodynamics 2.2.1. Straight baseline wings An elliptical distribution of the circulation Γ(π¦) was computed using a 4-stage numerical iterative method, based on Prandtl’s Lifting-line Theory (Anderson, 2007): 1 - An elliptical Γ(π¦) distribution designated Γπππ was assumed and the local induced angle of attack at each π station πΌπ (π¦π ) was calculated; 2 - The πΌπ (π¦π ) was used to obtain the effective angle of attack πΌeff at each π station πΌeff (π¦π ), which was subsequently used to find the lift coefficient ππ of the local airfoil πππ from the airfoil lift curve using AeroFoil 2.2 software; 3 - The πππ was used to calculate a new Γ(π¦), thus Γπππ€ , and agreement between Γπππ and Γπππ€ was attained by further input Γππππ’π‘ using Eq. (1) (acceptable accuracy was set at 0.01%); and 4 - The converged Γ(π¦) was used to obtain −πΆL and πΆDπ . Γππππ’π‘ = Γπππ + π·(Γπππ€ − Γπππ ) (1) where, the damping factor π· was 0.05. The Γ(π¦) at each π station was obtained from the function 2π¦ 2 Γ(π¦) = Γ0 √1 − ( ) π (2) where, Γ0 is the circulation at the origin (i.e., at the mid-span) obtained from Eq. (3) (adapted from Anderson, 2007) Γ0 = 0.5 ππ ππ π½∞ (3) where, the ππ was obtained using AeroFoil 2.2 software based on a Reynolds number π π of 1.08 × 106 and assuming International Standard Atmosphere (Bertin, 2002). The local wing chord ππ varied linearly along π and π½∞ represents the flow velocity vector. The πΌπ (π¦π ) was calculated using πΌπ (π¦π ) = π/2 (πΓ(π¦) ⁄ 1 ππ¦ )ππ¦ ∫ 4ππ½∞ −π/2 π¦π − π¦ (4) where, π¦ is the spanwise coordinate. Then, πΌeff (π¦π ) for the iterative solution was computed from πΌeff (π¦π ) = πΌ − πΌπ (π¦π ) (5) where, α is the angle of attack. The −πΆL and πΆDπ were obtained by integration using Γ(π¦), in the respective Eqs. (6) and (7), and the πΆπ· was computed using Eq. (8). Proceedings of PACAM XII −πΆπΏ = πΆπ·π = π/2 2 π½∞ π 2 π½∞ π πΆD = πππ + 12th Pan-American Congress of Applied Mechanics January 02-06, 2012, Port of Spain, Trinidad ∫ Γ(π¦)ππ¦ (6) −π/2 π/2 ∫ Γ(π¦)πΌπ (π¦)ππ¦ (7) −π/2 −πΆL2 πππ΄π (8) where, πππ is the airfoil drag coefficient ππ adjusted for induced flow. The πππ was obtained from graphical experimental data (Milne-Thomson, 1973; Anderson, 2007) using the adjusted airfoil lift coefficient πππ calculated from πππ = π0 (πΌeff − πΌL0 ) (9) where, π0 is the airfoil lift slope in radians given by π0 ≡ dππ /dπΌ (10) 2.2.2. Twist optimisation The normalised π(π)πππ‘ was given by the function π(π)πππ‘ = 1 − π ππ(π) 1 − (1 − π π )|πππ π| (11) where, π represents the spanwise angular coordinate. The π(π)πππ‘ was used to elliptically vary the local πππ , Γ(π¦) and πΌπ (π¦π ) consistent with the progressive non-linear twist along π (Milne-Thomson, 1973). The planform ππ and twist ππ coefficients for the lifting-line solution (Phillips, 2004) were obtained from the relations ∞ 2π΄π (1 + π π ) π ∑ ππ { + } sin(ππ) = 1 π[1 − (1 − π π )|cos(π)|] sin(π) (12) π=1 and π0 2(1 + π π )π΄π ππ = { π0 ππ [1 + ] 2(1 + π π )π΄π π1 − (1 − π1 ) πππ π = 1 (13) πππ π ≠ 1 where, π is the wing lift slope in radians computed using π= π0 π0 1+( ) (1 + π) ππ΄R (14) The Fourier coefficients π΄π were calculated using π΄π ≡ ππ (πΌ − πΌπΏ0 )πππ−π πππ − ππ Ω (15) where, the preset total wing twist Ω of 5° was given by Ω = (πΌ − πΌπΏ0 )πππ−π πππ − (πΌ − πΌπΏ0 )π€πππ π‘ππ (16) The −πΆL for the optimally-twisted wing was obtained using (Phillips, 2004) −πΆL = π΄π ππ΄1 (17) Proceedings of PACAM XII 12th Pan-American Congress of Applied Mechanics January 02-06, 2012, Port of Spain, Trinidad and Ωπππ‘ was given by Ωπππ‘ = ππ·πΏ ππ0 1 × 2ππ·Ω π (18) whereby, the lift-twist factor ππ·πΏ and twist factor ππ·Ω were obtained, respectively, from ππ·πΏ ≡ ππ0 πΏ (1 + π π )π (19) and ππ·Ω ≡ ( 2 ππ0 ) πΏ 2(1 + π π )π0 (20) The πΆπ·π for the optimally-twisted wing was calculated using Eq. (21) (Phillips, 2004) πΆπ·π = −πΆπΏ2 πΏ ππ0 Ω 2 + [−πΆπΏ − ] ππ΄π ππ΄π 2(1 + π π ) (21) 2.2.3. Complete wings The ππ and AR were adjusted to account for the increased downforce provided by the Gurney flap (Wang et al., 2008) and the flow control and increased π΄π effects achieved by the end plates (Phillips, 2004), respectively, using βππ = √ βπΊ π (22) and βππ π΄π πππ ππππ‘π = π΄π πππ‘π’ππ [1 + 1.9 ( )] π (23) where, βππ is the increment in ππ caused by the Gurney flap and π΄π πππ ππππ‘π is the effective AR attained using the end plates. A new −πΆL for the complete wings was obtained by modifying wing π to account for βππ and π΄π πππ ππππ‘π −πΆL = π(πΌ − πΌL0 ) (24) where, wing π was adjusted by using π΄π πππ ππππ‘π in Eq. (14). The πΆπ· and πΆπ·π associated with the installation of the Gurney flap and end plates were obtained using the new −πΆL and π΄π πππ ππππ‘π in Eqs. (8) and (21), respectively. 2.3. Modelling of wing aerodynamics Wing aerodynamics were simulated for -4° ≤ πΌ ≤ 12° at 1° intervals; based on wing-tip α in the case of twisted wings. Drag polars were plotted for the assessment of wing downforce and aerodynamic efficiency. 3. RESULTS 3.1. Airfoil and wing aerodynamics and normalised π(π)πππ‘ Installation of the Gurney flap changes the πΌπΏ0 from -3° to -5.11° and from -4° to -5.86° in the 651-412 and LS(1)0413 airfoils, respectively, and causes a Δππ of 0.205 and an increase in wing πΏ and π to a value of 0.055. Mounting the end plates increases the effective AR from 5.56 for the baseline wing to 6.39 for the complete wing. The Ωπππ‘ for the 651-412 wing is 4.6° and for the LS(1)-0413 is 4°. The normalised spanwise π(π)πππ‘ is near-elliptical, thus the rate of geometric twist increases rapidly in the regions near the wingtips, which causes a near-elliptical −πΆL distribution. Proceedings of PACAM XII 12th Pan-American Congress of Applied Mechanics January 02-06, 2012, Port of Spain, Trinidad 3.2. Wing πΆDπ and drag polars The use of the LS(1)-0413 airfoil, application of wing twist and installation of the Gurney flap and end plates are all factors that augment πΆDπ (Fig. 2). Complete wings and twisted wings generate greater −πΆπΏ than baseline wings and straight wings, at the expense of additional πΆD and lower aerodynamic efficiency (Fig. 3). In Fig. 3, data at α = 5° has been marked using arrows to help compare data for different wing prototypes. Wings built with a constituent 651-412 airfoil attain higher aerodynamic efficiency but wings constructed with an LS(1)-0413 airfoil yield greater −πΆπΏ in all 8 wing prototypes. Straight baseline 0.6 0.6 Twisted baseline Twisted complete 0.4 CDi CDi Straight complete 0.2 0.4 0.2 651-412 0.0 LS(1)-0413 0.0 -4 -2 0 2 4 6 8 10 12 -4 -2 0 2 α (°) 4 6 8 10 12 α (°) Figure 2. The πΆDπ for the prototype wings 2.5 2.5 Straight wings 2.0 2.0 α = 5° 1.5 α = 5° 1.5 Baseline 651-412 1.0 -CL -CL Twisted wings Baseline 651-412 1.0 Complete 651-412 Complete 651-412 Baseline LS(1)-0413 0.5 Baseline LS(1)-0413 0.5 Complete LS(1)-0413 Complete LS(1)-0413 0.0 0.0 0.0 -0.5 0.1 0.2 CD 0.0 0.3 -0.5 0.1 0.2 0.3 CD Figure 3. Drag polars for the straight and twisted wings (-4° ≤ πΌ ≤ 12°) 4. DISCUSSION Use of the LS(1)-0413 airfoil for wing construction results in higher downforce primarily due to the larger leading edge radius and 1% greater thickness than in the 651-412 airfoil (Abbott and von Doenhoff, 1959), however such geometric features augment πΆDπ and πΆD (Figs. 2 & 3). The twisted complete LS(1)-0413 yields the largest −πΆπΏ and πΆDπ of all wings (Figs. 2 & 3) due to the large β and α in the mid-span region, the presence of the Gurney flap and the use of the LS(1)-0413 airfoil. The larger mid-span β and α of the twisted wings causes an increase in πΆDπ (Fig. 2). Nonetheless, the increase in πΆDπ can be considered small and solely attributed to mid-span β and α. This suggests that a twistoptimised wing with a linearly-tapered planform for the GT car nearly duplicates the high performance of an elliptic wing and satisfies the requirement for minimum possible πΆDπ , consonant with theory (Milne-Thomson, 1973; Phillips and Alley, 2007); thus, H1 was accepted. This is because, with the exception of an elliptic planform, a straight wing is optimised for a design lift coefficient of zero. However, Ωπππ‘ & π(π)πππ‘ considerably reduces πΆDπ when the wing operates at any other −πΆπΏ (Phillips, 2004). The Gurney flap forces the wake downwards and is primarily responsible for increased −πΆπΏ with further increments in downforce achieved by the installation of the end plates (Wang et al, 2008). In fact, the use of end plates reduces Oswald efficiency, however the augmented effective π΄π associated with the end Proceedings of PACAM XII 12th Pan-American Congress of Applied Mechanics January 02-06, 2012, Port of Spain, Trinidad plates curtails the additional πΆDπ and πΆD caused by the Gurney flap and helps maintain efficiency (Milne-Thomson, 1973). In a high-downforce racing circuit with slow turns, the additional downforce achieved by the use of the twisted complete LS(1)-0413 will compensate for the performance losses caused by the extra induced drag (Wang et al., 2008)). Partial wing stall is an important parameter in wing design that may contribute to safety in motor racing by preventing a sudden loss of downforce when the car decelerates upon entering a slow curve (Bertin, 2002). Geometric twist may be effective in achieving a partial stall of the wing since the mid-span region, which is set at a higher πΌ, will stall first (Anderson, 2007). Greater safety, as well as higher downforce, can be attained by using the LS(1)-0413 airfoil, which is stall resistant (Bertin, 2002). Also, the small π π helps preserve π length and prevent a low Re and the stall in the wing tip region (Abbott and von Doenhoff, 1959). Thus, the use of a LS(1)-0413 airfoil, geometric twist and a small π π help control, and perhaps prevent, wing stall. The data in Fig. 3 help prioritise either downforce or aerodynamic efficiency in wing design to adjust the GT car to the characteristics of the racing circuit (Wang et al., 2008). Greater downforce is attained using the twisted complete LS(1)-0413 wing but higher efficiency is achieved using the straight complete 65 1-412 wing. The higher aerodynamic efficiency attained when using the 651-412 airfoil over the full range of πΌ (Fig. 3) is due to the small-radius leading edge and 1% thinner profile of the constituent airfoil (Abbott and von Doenhoff, 1959). Phillips et al. (2006) demonstrated the validity of the modified lifting–line numerical solution for the prediction of the πΆDπ associated with optimised wing twist. The computational fluid dynamics validation performed by Phillips et al. (2006) suggests that the lifting-line method slightly underestimates the decrease in πΆDπ caused by optimum twist. The complexity of flow over complete optimally-twisted wings may be examined in a wind tunnel to further validate the numerical methods. 5. CONCLUSIONS An optimally-twisted wing is complex to fabricate, as the twist varies elliptically along π. However, the findings suggest that a twist-optimised wing with a linearly-tapered planform for the GT car replicates the high performance of an elliptic wing and fulfils the requirements for high downforce and minimum possible πΆDπ . A twisted complete LS(1)-0413 wing yields high −πΆπΏ , at the expense of greater πΆDπ and πΆD , and is suitable for a high-downforce racing circuit. The geometric twist and small π π help achieve a partial wing stall that contributes to safety in motor racing. The complete straight 651-412 wing achieves superior aerodynamic efficiency and may be used in fast circuits. 6. ACKNOWLEDGEMENTS The author acknowledges the financial support provided by Marques Aviation Ltd for this project. 7. REFERENCES Abbott, I.H. and von Doenhoff, A.E., 1959. “Theory of Wing Sections”, Dover Publications, New York. Anderson, J.D., 2007. “Fundamentals of Aerodynamics”, McGraw-Hill, London. Bertin, J.J., 2002. “Aerodynamics for Engineers”, Prentice Hall, New Jersey. FIA, 2010. Technical Regulations for Grand Touring Cars (Group GT1 and GT2), Annexe J, Appendix J, Article 257, Paris, pp. 1–15. Milne-Thomson, L.M., 1973. “Theoretical Aerodynamics”, Dover Publications, New York. Phillips, W.F., 2004. “Lifting-Line Analysis for Twisted Wings and Washout-optimized Wings”, J. Aircr., Vol. 41, No. 1, pp. 128-136. Phillips, W.F. and Alley, N.R., 2007. “Predicting Maximum Lift Coefficient for Twisted Wings Using Lifting-Line Theory”, J. Aircr., Vol. 44, No. 3, pp. 898-910. Phillips, W.F., Fugal, S.R. and Spall, R.E., 2006. “Minimizing Induced Drag with Wing Twist, Computational-fluiddynamics Validation”, J. Aircr., Vol. 43, No. 2, pp. 437-444. Wang, J.J., Li, Y.C. and Choi, K.S., 2008. “Gurney Flap: Lift Enhancement, Mechanisms and Applications”, Prog. Aerospace Sci., Vol. 44, No. 1, pp. 22-47.