Race Car Aerodynamics Gregor Seljak April 8, 2008 1 Introduction First racing cars were primarily designed to achieve high top speeds and the main goal was to minimize the air drag. But at high speeds, cars developed lift forces, which a? ected their stability. In order to improve their stability and handling, engineers mounted inverted wings pro? les1 generating negative lift. First such cars were Opel’s rocket powered RAK1 and RAK2 in 1928. However, in Formula, wings were not used for another 30 years. Racing in this era 1930’s to 1960’s occured on tracks where the maximum speed could be attained over signi? ant distance, so development aimed on reducing drag and potencial of downforce had not been discovered until the late 1960’s. But since then, Formula 1 has led the way in innovative methods of generating downforce within ever more restrictive regulations. Figure 1: Opel’s rocket powered RAK2, with large side wings 2 Airfoils Airfoil can be de? nead as a shape of wing, as seen in cross-section. In order to describe an airfoil, we must de? ne the following terms(Figure 2) • The mean camber line is a line drawn midway between the upper and lower surfaces. • The leading and trailing edge are the most forward an rearward of the mean camber line. Compared to an aircraft 1 • The chord line is a line connecing leading an trailing edge. • The chord length is the distance from the leading to the trailing edge, measured along the chord line. • The camber is the maximum distance between mean camber line and chord line. • The thickness is the distance between the upper and lower surfaces. Figure 2: Airfoil nomenclature The amount of lift L produced by the airfoil, can be expressed in term of lift coe? cient CL 1 2 (1) L = ?? V? SCL 2 where V? denotes the freestrem velocity, ?? ?uid density and S the airfoil area. 2. 1 Flow over an airfoil

Properties of an airfoil can be measured in a wind tunnel, where constantchord wing spannes the entire test section, from one sidewall to the other. In this conditions, the ? ow sees a wing without wing tips. Such wing is called in? nite wing and streches to in? nity along the span. Because the airfoil section is identical along the wing, the properties of the airfoil and the in? nite wing are identical. Therefore the ? ow over an airfoil can be described as a 2D incompressible inviscid ? ow over an in? nite wing. Lift per unit span L? generated by an arbitrary airfoil(or any other body) moving at speed V? through the ? ud with density ?? and circulation ? is 2 given by Kutta-Joukowsky theorem L? = ?? V? ? . (2) Circulation around an airfoil, can be calculated with the concept of a vortex sheet, which was ? rst introduced by Prandtl an his colleagues. Consider an airfoil of arbitrary shape and thickness as shown in Figure 3. Circulation can be distributed over the whole airfoil area with surface density(vortex sheet strength) d? /ds = ? (s), where ? (s) must satisfy Kutta condition ? (trailing edge) = 0 (3) Entire circulation is then given by ?= ? (s)ds , (4) where the integral is taken around the complete surface of the airfoil.

However, there is no general solution for ? (s) for an airfoil of arbitrary shape and it must be found numericaly, but analytical solutions can be found with some aproximations. Figure 3: Simulation of an arbitrary airfoil by distributing a vortex sheet over the airfoil surface. 2. 2 Thin airfoil theory Here we discuss thin airfoil in freestream of velocity V? under small angle of attack ?. Camber and thickness are small in relation with chord length c. In such case, airfoil can be described with a single vortex sheet distributed over the camber line(Figure 4). Our goal is to calculate the variation of ? s), such that the chamber line becomes streamline and Kutta condition at trailing edge, ? (c) = 0, is satis? ed. 3 Figure 4: Thin airfoil approximation. Vortex sheet is distributed over the chamber line The velocity at any point in the ? ow is the sum of the uniform freestream velocity and velocity induced by the vortex sheet . In order the camber line to be a streamline, the component of velocity normal to the camber line must be zero at any point along the camber line. w ? (s) + V? ,n = 0 , (5) where w ? (s) is the component of velocity normal to the chamber line induced by the vortex sheet and V? n the component of the freestrem velocity normal to the camber line. Considering small angle of atack and de? ning ? (x) = dz/dx as the slope of the chamber line, V? ,n can be written as (Figure 5) V? ,n = V? ? ? dz dx (6) Because airfoil is very thin, we can make the approximation w ? (s) ? w (x) , (7) where w (x) denotes the component of velocity normal to the chord line and can be, using the Biot-Savart law, expressed as c w (x) = ? 0 ? (? )d? 2 ? (x ? ? ) (8) Substituting equations (6), (7) and (8) into (5) and considering Kutta condition, we obtain 1 2? c 0 ? (? )d? dz = V? ? ? x?? dx ? (c) = 0 undamental equations of thin airfoil theory. 4 (9) Figure 5: Determination of the component of freestrem velocity normal to the chamber line In order to satisfy this conditions , we ? rst transform our variables x and ? into c c x = (1 ? cos ? 0 ) (10) ? = (1 ? cos ? ) 2 2 and equation (9) becomes 1 2? ? 0 ? (? ) sin ? d? dz = V? ? ? cos ? ? cos ? 0 dx (11) with a solution that satis? es Kutta condition ? (? ) = 0 ? (? ) = 2V? A0 ? 1 + cos ? An sin(n? ) + sin ? n=1 (12) In order to ? nd coe? cients A0 and An , we substitute equation (12) into equation (11) and use the following trigonometric relations ? 0 sin(n? ) sin ? ? = ?? cos(n? 0 ) cos ? ? cos ? 0 (13) ? sin(n? 0 ) cos(n? )d? = cos ? ? cos ? 0 sin ? 0 (14) ? 0 and ? nnaly obtain ? dz An cos(n? 0 ) = (? ? A0 ) + dx n=1 5 (15) This equation is in form of a Fourier cosine series expansion for the function dz/dx. Comparing it to the general form for the Fourier cosine expansion we obtain 1 ? dz A0 = ? ? d? 0 (16) ? 0 dx 2 ? dz cos(n? 0 )d? 0 (17) An = ? 0 dx The total circulation due to entire vortex sheet from leading to the trailing edge is c cc ? (? ) sin ? d? (18) ? (? )d? = ?= 20 0 Substituting equation (12) for ? (? ) into equation (18) and carrying out the integration, we obtain ? = cV? ? A0 + A1 (19) 2 hence the lift per unit span, given by Kutta-Joukowski is 2 L? = ?? V? ? = c?? V? ? A0 + ? A1 2 (20) This equation leads to to the lift coe? cient in form cl = ? (2A0 + A1 ) = 2? ? + 1 ? ? 0 dz (cos(n? 0 ) ? 1)d? 0 dx (21) and lift slope dcL = 2? (22) d? Last two results are important. We can see, that lift coe? cient is function of the shape of the pro? le dz/dx and angle of attack ? , and that even symmetrical wing produces lift, when set under an angle of attack. Lift slope is constant, independently of the shape of the pro? le, while the zero lift angle lS ? ?L=0 = ? 1 ? 0 dz (cos(n? 0 ) ? 1)d? 0 dx (23) depends on the shape. The more highly chambered the airfoil, the larger is ? L=0 2. 3 Viscid ? ow By now, we have studied the inviscid incompressible ? ow. But in real case, ? ow is viscous. It is time to compare our theoretical results with real one. In Figure 6, we can see variation of lift coe? cient with the angle of attack. 6 At low angles of attack cl varies linearly with ? , as predicted by the theory. However, at certain angle of attack, cl reaches it’s maximum value cl,max and starts to decrease. This is due to viscous e? ect of the ? uid (air). First, the ? w moves smoothly over the airfoil and is attached over most of the surface, but at certain value of ? seperates from the top surface, creating a wake of turbulent ? ow behind the airfoil, which results in drop in lift and increase in drag. Figure 6: Variation of lift coe? cient with the angle of atack. To increase lift of the airfoil, we must increase cl,max . As we have seen, the cl,max of the airfoil primarily depends on it’s shape. Airfoil’s shape can be changed with use of multielement ? aps at the trailing edge and slats at leading edge. They increase chamber of the airfoil and therefore its cl,max .

The streamline pattern for the ? ow over such airfoil can be seen in Figure 7. 3 Finite wings Properies of airfoils are the same as the properties of a wing of in? nite span. However, all real wing are of ? nite span and the ? ow over ? nite wing is 3 dimensional. Because of higher pressure on the bottom surface of the wing, the ? ow tends to leak around the wing tips. This ? ow establishes a circulary motion that trails downstream of the wing. A trailing vortex is created at each wing tip. These wing-tip vortices induce a small downward component of air velocity, called downwash . It produces a local relative wind which is Figure 7: Flow over multielement airfoil. directed downward in the vicinity of the wing, which reduces the angle of attack that each section of the wing e? ectively sees ?ef f = ? ? ? i (24) and it creates a component of drag, de? ned as induced drag. 3. 1 Prandtl’s classical lifting-line theory The idea of lifting line theory, is to use two dimensional results, and correct them for the in? uence of the trailing vortex wake and its downwash. Let’s replace a ? nite wing of span b, with a bound vortex 2 extending from y = ? b/2 to y = b/2. But due to the Helmholtz’s theorem, a vortex ? ament can’t end in a ? uid. Therefore assume the vortex ? lament continues as two free vortices trailing downstream from the wing tips to in? nity(Figure 8). This vortex is due to it’s shape called horseshoe vortex. Downwash induced by such vortex, does not realistically simulate that of a ? nite wing, as it aproaches ?? at wing tips. Instead of representing the wing by a single horseshoe vortex, Prandtl superimposed an in? nite number of horseshoe vortices, each with an in? nitesimally small strength d? , and with all the bound vortices coincident along a single line, called the lifting line.

In this model, we have a continious distribution of circulation ? (y ) along the lifting line with the value ? 0 at the origin. The two trailing vortices in single horseshoe vortex model, have now 2 A vortex bound to a ? xed location in ? ow 8 Figure 8: Replacement of the ? nite wing with single horseshoe vortex. Figure 9: Superposition of an in? nite number of horseshoe vortices along the lifting line. became a continious vortex sheet trailing downstream of the lifting line,and the total downstream velocity w , induced at the coordinate y0 by the entire trailing vortex sheet can be expressed as w (y 0 ) = ? 4? b/2 ?b/2 (d? /dy )dy y0 ? y (25) The induced angle of attack at the arbitrary spanwise location y0 is given by ? i (y0 ) = arctan ?w (y0 ) ?w (y0 ) = , V? V? (26) where we considered V? ? w (y0) and arctan(? ) ? ? for small values of ?. Now we can obtain an expression for the induced angle of attack in term of the circulation distribution along the wing ?i (y0) = ? 1 4? V? 9 b/2 ?b/2 (d? /dy )dy y0 ? y (27) Combining results cl = 2? (y0) V? (28) and cl = 2? [? ef f (y0 ) ? ?L=0 ] (29) for coe? cient of lift per unit span from thin airfoil theory, we obtain ? ef f = ?(y0 ) + ? L=0 ?V? c(y0 ) (30)

Substituting equations (27) and (30) into (24), we ? nally obtain the fundamental equation of Prandtl’s lifting line theory. ? (y 0 ) = 1 ?(y0 ) + ? L=0 (y0 ) + ?V? c(y0 ) 4? V? b/2 ?b/2 (d? /dy )dy y0 ? y (31) Just as in thin airfoil theory, this integral equation can be solved by assuming a Fourier series representation for the distribution of vorticity N An sin n? ?(? ) = 2bV? (32) n=1 where we considered transormation y = (? b/2) cos ? , with 0 ? ? ? ? and coe? cients An must satisfy Equation (31). With such vorticity distribution, Equation (31) becomes ?(? 0 ) = N N 2b sin n? 0 nAn An sin n? 0 + ?

L=0 (? 0 ) + ?c(? 0 ) n=1 sin ? 0 n=1 (33) The total lift distribution is obtained by integrating equation for lift distribution over the span L= b/2 ?b/2 ?? V? ?(y )dy (34) Coe? cients of lift and induced drag3 , can be calculated via equations CL = and CD = 2 L = q? S V? S D 2 = q? S V? S 3 b/2 ?(y )dy (35) ?i (y )? (y )dy (36) ?b/2 b/2 ?b/2 Note the di? erence in nomenclature. For 2D bodies, coe? cients have been denoted with lowercase letters. In 3D case, we use capital letters 10 respecteviliy. Considering expressions (32) and (33), they can be written as CL = A1 ? AR (37) and 2 CL (1 + ? ) (38) ?AR here AR is aspect ratio of ? nite ? ng, de? ned as AR = b2 /S , and ? = N 2 2 (An /A ? 1) . Note that CL depends only on the leading coe? cient in Fourier series expansion and that ? ? 0. Therefore, the lowest induced drag will be produced by a wing where ? = 0, that is, n = 1. Such circulation distribution is given by ? (? ) = 2bV? A1 sin ? and is known as elliptical circulation distribution CD,i = 4 Ground e? ect The main di? erece between wing application in aviation and car racing is, that cars are in contact with the ground. Therefore, wing experiences some additional e? ects due to ground proximity.

Remember the wing tip vortices we mentioned at the beginning of the previous section. They do nothing but harm, as they increase drag and decrease lift at given angle of attack. When ?ying near to the ground, the ground partially blocks(Figure 10) the trailing vortices and decreases the amount of downwash generated by the wing. This reduction in downwash increases the e? ective angle of attack of the wing so that it creates more lift and less drag than it would otherwise. This e? ect is greater, the closer to the ground the wing operates. Figure 10: E? ect of the ground proximity on creation of the trailing vortices.

Another way to create downforce is to create low pressure area underneath the car, so that the higher pressure above the car will apply a downward force. The area between car’s underbody and the ground, can be thougth as an example of Venturi nozzle. The Venturi e? ect may be derived from 11 a combination of Bernoulli’s principle and the equation of continuity. The ? uid velocity increases through the constriction to satisfy the equation of continuity, while it’s pressure decreases due to conservation of energy. The gain in kinetic energy is supplied by a drop in pressure.

The main advantage of ground e? ect is, that it produces almost no drag. 5 Applications in car racing Now summarize what we have learned so far. The coe? cient of lift increases with increasing angle of attack. At some angle, ? ow seperates from the wing, which causes drop of lift coe? cient. With use of multidimensional ? aps, we increase chamber of the airfoil and thus maximum coe? cent of lift. In 3 dimensional case, vortices appear at wing tips. They reduce wing’s e? ciency and increase drag. The lowest drag can be achieved with elliptically shaped wing. Dimensions of the wing are also important.

Wing with greater surface, produces more lift and wing with higher aspect ratio induces less air resistance. In the next sections, we will see, how engineers used this principles at developing the main aerodynamical parts of racing cars. 5. 1 Rear wing First rear wing appeared in 1966, when Jim Hall equiped his Chaparral 2E with a rear wing. From then on, use of wings grew quickly. First wings were mounted high over the rear end of the car to operate in indisturbed ? ow. They were also mounted on pivots, so the driver was able to change the angle of attack during the ride. High mounted wings often broke o? uring the race and were therefore prohibited by FIA. In Formula 1, wings were ? rst introduced in 1968 at the Belgium grand prix, when Ferrari used full inverted rear wings, and Brabham did likewise, just one day after the Ferrari’s wings ? rst appeared. Modern rear wings produce approximately 30-35 % of the total downforce of the car. A typical con? guration(Figure) consists of two sets of airfoils connected to each other by the wing endplates. The most downforce is provided by the upper airfoil. To achieve the greatest possible lift coe? cient, it consists of multiple high aspect ratio elements, which prevent ? w separation. Angle of attack depends on circuit con? guration. On tracks with many turns, more downforce is needed, therefore the wing is set at higher angle of attack. Conversely, on tracks with long straights, wing has small angle attack, thus reducing air drag and allowing higher top speeds. Lower airfoil section ac12 Figure 11: Chapparal 2E (left) and Ferrari 312 (right). tually reduces the downforce produced by total rear wing, but it creates a low-pressure region just below the wing to help the di? user4 to create more downforce below the car. Ususally it consists of two elements.

Another important part of rear wing are endplates . They provide a convenient way of mounting wings, but also have aerodynamic function. They reduce the 3D e? ect of the wing by preventing air leakage around the wing tips and thus formation of trailing vortices. An additional goal of the rear endplates is to help reduce the in? uence of up? ow from the rear wheels. The U-shaped cutout from the endplate further alleviates the development of trailing vortices. 5. 2 Front wing The front wing on the car produces about 1/3 of the car’s downforce and it has experienced more modi? ations than rear wing. It is the ? rst part of the car to meet the air mass, therefore, besides creating downforce, it’s main task is to e? ciently guide the air towards the body and rear of the car, as the turbulent ? ow impacts the e? ciency of the rear wing. Front wings appeared in Formula 1 just two weaks after the ? rst rear wings, on Lotus 49B. First front wings were quite simple with single rectangular airfoil with ? at vertical endplates to reduce wing tip vortices. First improvement appeared in 1971, with so-called Gurney ? ap. This is a ? at trailing edge ? p perpendicular to the chord and projects about 2% of the chord. It can improve the performance of a simple airfoil to nearly the same level as a complex design. The same year, the concept of elliptical wing was applied. March equiped it’s 711 with elliptical front wing. Two years later Ferrari avoided wing-body interaction with wing mounted quite far ahead 4 See section 5. 3 13 Figure 12: Modern rear wing consists of upper(2) an lower(3) airfoil section mounted on endplates (1) with U-shaped cutout (4). from the body. Multi element wings were introduced in 1984 by McLaren.

The angle of attack of the second element was allowed to be modi? ed so that the load applied on the front wing could be changed to balance the car according to the driver’s wishes. In 1990 Tyrell raised the nose of it’s 019 to increase the ? ow under the nose cone and improve ? ow conditions under the car. This concept avoids wing-body interaction and allows the front wing to operate in undisturbed ? ow. It also enlarges e? ective area of the wing. After Imola 1994, the FIA regulations do not allow any chassis parts under a minimum ground height. This clearance is di? erent between the centre and the side of the car.

Teams used this to curve front wing in the centre of the span and regain some of the lost ground e? ect. In 1998, regulations decreased the width of Formula 1 car, so the front wings overlapped the front wheels. This created unnecessary turbulence in front of the wheels and reducing aerodynamic e? ciency of the wing. With reducing wing’s span this could be avoided, but it would also decrease wing’s aspect ratio. Insted this, teams use wing tips to direct the air around the wheels. 14 Figure 13: Con? guration of modern front wing. Two element airfoil (1 & 2) is mounted under the nose of the car (5).

Endplates (4) direct air around the wheels and curved area (4) under the nose increases wing’s e? ciency. 5. 3 Ground e? ect The second revolution in Formula 1 aerodynamics occurred about a decade after the ? rst, with the introuction of the Lotus T78 in 1977. Lotus T78 and it’s further development, Lotus T79, were ? rst cars to use ground e? ect. The underbody took shape of inverted wing pro? le(Figure). The decreasing crosssectional area accelerated the air? ow and created low pressure underneath the car. The gap between the bottom of the sidepods and the ground was sealed with so-called sidepods. They helped to maintain 2D ? w characteristics that provide increased downforce and reduced drag, compared to a typical 3D wing. Skirts enabled very high cornering speeds and were prohibited by the rules, due to safety reasons and from 1983 onwards, the tehnical regulations in Formula 1 require the underbody panel between the wheels to be completely level. The ? ow wolume between the vehicle and the ground is strongly dependent on the car’s attitude relative to the ground. This correlation is illusrtated in Figure. Very small ground clearence results in positive lift, since there is almost no air? ow between the underbody and the ground. With in- 15

Figure 14: Some historical milestones in front wing develpment. Lotus 49B, March 711, Ferrari 312 T2 and Tyrrell 019. Figure 15: Lotus T79 and sketch of it’s underbody creasing ground clearence the air? ow produces low pressures causing overall lift to be lowered to negative values and then to rise again as ground clearence continues to increase. This is due to the fact that the ? ow velocity under the car decreases as ground clearence increases. More downforce can be generated using a di? usor between the wheels at the rear of the car. The air enters the di? user in a low-pressure, high-velocity state after accelerating under the 16 ar. By gradually increasing the cross-sectional area of the di? user, the air gradually slows down and returns to its original free-stream speed and pressure. The di? user’s aim is to decelerate the air without it separating from the tunnel walls, which would cause a stall, reducing the downforce and inducing a large drag force. By installing an inverted wing close to the di? user exit 5 it is possible to create a low-pressure area, which essentially sucks the air from the di? user. The di? user and wing combination permits a higher air mass ? ow rate through the di? user, thus resulting in higher downforce.

Sharp edges on the vertical tunnel walls generate vortices from entrained air and help con? ne the air through the di? user and reduce the chance it will separate. Figure 16: Correlation between lift coe? cient and ground clearence(left) and di? user on Ferrari F430(right) Again Chaparral, showed completely new way to create downforce. The Chaparral 2J in 1969 used two rear fans to suck in air from under the car, thus creating low pressure under the car. Big advantage of this concept is, that downforce can be generated independently of the speed. Fans were also used in Formula 1. Brabham BT46 used a rear mounted fan driven o? he gearbox. It won it’s debut race in 1978, but was promptly banned by the governing body. Barge boards were ? rst seen in 1993 and their purpose is to smooth the air? ow around the car and into the radiator intakes. They are most commonly mounted between the front wheels and the sidepods (See Figure) . Their main purpose is to direct relatively clean air into the sidepods. Clean air is from the low section of the front wing where air? ow is fairly una? ected 5 See rear wing section 17 Figure 17: Two cars which used fans to create downforce. The Chaparral 2J "sucker car" (left) and Brabham BT46 "fan car" y the wing and far away from tires, which may throw stones and debris in to the radiator. Bargeboards also produce vortices, to seal the area between the sidepots and the surface. They work as a substitude for skirts. Figure 18: Bargeboards on McLaren MP4/8 6 Conclusion References [1] J. D. Anderson; Fundamentals of Aerodynamics 18 [2] Applied Aerodynamics: A Digital Textbook, http://www. desktopaero. com/appliedaero/preface/welcome. html [3] W-H Hucho: Aerodynamics of Road Vehicles [4] Peter Wright: Formula 1 Technology [5] Milliken,Milliken: Race Car Vehicle Dynamics [6] F. Mortel: Cran? eld Team F1: The Front Wing 19