aerodynamics of a wing in ground effect in generic racing car wake flows

1

Aerodynamics of a wing in ground effect in genericracing car wake flowsM D Soso* and P A WilsonSchool of Engineering Sciences, University of Southampton, Southampton, UK

The manuscript was received on 17 March 2005 and was accepted after revision for publication on 7 September 2005.

DOI: 10.1243/095440705X69632

Abstract: In an effort to provide more detailed insight into the aerodynamic factors that mayinfluence the creation of overtaking opportunities in modern open-wheeled racing series, a setof wind tunnel experiments was initiated in the moving ground facilities at the University ofSouthampton. To generate data typical of one car following another, a single-element wing inground effect was tested downstream of a bluff body that incorporated a diffuser and rearwing. The tests included variations in the height and angle of attack of the wing, while datacollection was achieved via force and pressure measurements, flow visualization and flowfieldsurveys. The results were then compared with baseline data that were obtained without thepresence of the bluff body. It was found that, while behind the upstream body, the wingexperienced a decrease in its downforce values, with the amount of downforce lost dependingon its height above the ground. It was also shown that more downforce was lost from sectionscloser to the mid-span of the wing than was the case from sections closer to the tips of the wing.

Keywords: wing, ground effect, overtaking, diffuser, bluff body, aerodynamics, racing

1 INTRODUCTION carried out for a variety of drafting and passingpositions. Tunnel restrictions limited the maximumdistance between the models to two car lengths.The issue of the lack of frequent overtakingWithin this range, the force results that were pre-opportunities typical of open-wheeled racing seriessented showed that there was as much as a 37 persuch as Formula 1 has received a great deal ofcent reduction in the drag of the following car.publicity in recent times. It is common to hear racingSignificant changes in the distribution of the liftcommentary suggesting that the aerodynamics offorces at the front and rear axles of both models werea following car have been severely affected whenalso reported.travelling in the flow produced by a leading car of

Howell [2] preformed wind tunnel investigationssimilar specifications. In order further to investigateinto the resulting lift, drag, and pitching momentthis particular situation, and to provide informationforces occurring when two Can-Am racing car modelson strategies that could possibly be exploited towere in close proximity to each other. The maindevelop more robust aerodynamic packages, a seriesobjective was to highlight the fact that the followingof experiments were initiated to model the con-car could experience sufficient aerodynamic changesditions that may be experienced during a typicalto cause it to overturn. Data were provided torace, when one vehicle follows in the wake producedshow that the following car experienced reducedby another.drag and downforce values, along with increasedPrevious studies of this type of vehicle interactionpitching moment values, when in the wake of thewere carried out as early as the 1970s for the NASCARcar ahead.racing series in America. Romberg et al. [1] investi-

For the case of open-wheeled racing cars, Dominygated the aerodynamic forces on wind tunnel models[3] carried out a detailed investigation of the aero-of stock cars of the time. The experiments weredynamic effects occurring when quarter-scale modelsof a 1989 Formula 1 car followed in the wake of an

* Corresponding author: 2 Summerton Place, Chipping Norton, identical model. Measurements on the following carwere taken with it placed on a moving ground, whileOxon OX7 5AZ, UK. email: [email protected]

JAUTO56 © IMechE 2006 Proc. IMechE Vol. 220 Part D: J. Automobile Engineering

2 M D Soso and P A Wilson

the leading car was positioned less than a car length while the angle of stall was reduced, as the groundahead, on the floor of the test section. In the was approached. The computational comparisons,measurements of lift and drag that were presented, which were carried out with a two-dimensional air-it was shown that, when fully immersed in the wake foil in a panel method, were limited to pressureof the leading car, the downforce on the following distribution plots. The plots showed fairly goodcar was reduced by 36 per cent and the drag by 23 per agreement at low angles of attack and large ridecent. As the lateral offset between the two models heights.increased, both variables commenced to recover to Jasinski and Selig [13] performed an experimentaltheir freestream values, with the drag taking much ground effect study of wing and endplate com-longer to do so than the downforce. Data were also binations that were representative of Champ Carprovided to show that the centre of pressure changed and Formula 1 front wings at the time. A rolling roadfrom 68 per cent wheelbase to 90 per cent wheelbase was not used. They specifically looked at the effectwhen directly behind the leading car. This result of Reynolds number, flap deflection, flap planformimplied that more downforce was lost from the front shape, and endplate shape. The measurements thatof the car than from the rear. were taken included forces, pressures, and flowfield

Experimental tests aimed at investigating the down- data. They obtained results showing that the dragstream wake of vehicle models and vehicle-like bluff coefficient at constant C

Lwas relatively unaffected

bodies were carried out by a number of authors [4–8]. while increasing flap deflection, and that flap plan-The results that were obtained typically highlighted form and endplate shape had a large effect on thethe existence of a pair of counter-rotating vortices aerodynamics of the wing. Their flowfield data clearlyin the flow field. The vortices induced an upwash or a highlighted the presence of two vortices, the largerdownwash, depending on the manner in which they emanating from the bottom of the endplate and therotated, which in turn depended on the particular smaller from the top. The Reynolds number wasconfiguration. found to have the least effect on the aerodynamics.

Published studies of racing car wings in ground Zerihan and Zhang [14–16] performed extensiveeffect were the result of experiments and com-

experimental investigations of a single element andputations in which the wing was placed in on-

then a double-element wing in ground effect incoming flow conditions that were undisturbed and

the moving ground facilities at the University ofof low turbulence intensity. Ranzenbach and Barlow

Southampton. The height above the ground of both[9–11] carried out a series of two-dimensional, com-

wings was varied, as was the angle of attack inparative computational and experimental studies

some tests. Among the important conclusions to beof a symmetric and a cambered airfoil in grounddrawn from the single-element wing investigationeffect. The work centred on the investigation of thewere that an increase in the angle of attack of theforce coefficients as the height of each airfoil waswing engendered greater maximum downforce, whilevaried. The authors also investigated the effect of areducing its sensitivity to changes in downforce,moving ground in the computational simulations.and that trailing edge separation increased withTheir results showed that, as the distance between thedecreasing ride height. Transition fixing was shown toairfoil and the ground was reduced, the downforceproduce less downforce on the wing and to produceexperienced by the wing increased to a maximum,force reduction at a greater ride height.and then decreased with further height reduction.

The current study focuses on assessing theThe drag was shown to rise monotonically for theaerodynamic performance of a wing in groundcase of the cambered airfoil. With regard to theeffect, when placed in different oncoming flow con-ground boundary, they showed that the airfoil experi-ditions. The different flow states were intended to beenced greater downforce coefficients when thatrepresentative of racing scenarios in which an open-particular surface moved with the freestream velocity,wheeled racing car was travelling in undisturbedas opposed to when it was stationary.freestream flow, and in the flow generated by a lead-Knowles et al. [12] carried out an experimental anding racing car. Focus was given to a wing in groundlimited computational study of a GA(W)-1 wing witheffect because it was the most forward part ofendplates, in ground effect. The experimental facilitythe vehicle that would experience changes in thewas equipped with a rolling road. They varied theoncoming flow, because the majority of the down-angle of attack of the wing at different heights aboveforce was typically lost from this component, andthe ground to generate lift curves, while recordingbecause the remainder of the car operated in theforce and pressure data. They obtained results show-

ing that the lift curve slope of the wing increased, wake that it generated.

JAUTO56 © IMechE 2006Proc. IMechE Vol. 220 Part D: J. Automobile Engineering

3Aerodynamics of a wing in ground effect

2 OUTLINE OF EXPERIMENT experience changes in the oncoming flow, becausethe majority of the downforce is typically lost fromthis component, and because the remainder of the2.1 Facilitiescar operates in the wake that it generates.

The experimental tests were carried out in theThe front wing was idealized as a single-element

large-scale moving ground wind tunnel facilities atLS(1)-0417 (GA(W)-1) wing, scaled to 40 per cent of

the University of Southampton. Flow visualizationthe dimensions of the front wing of a typical F1 car

images, along with force and pressure measure-for the year 2002. The resulting chord and span were

ments were gathered in the 2.1×1.5 m tunnel, while220 and 550 mm respectively. Endplates of 5 mm

laser doppler anemometry (LDA) measurementsthickness were attached to each end of the wing.

were taken in the 3.5×2.5 m R. J. Mitchell tunnel.The wing was specifically designed to pivot at the

Both tunnels are of a closed-circuit, single-returnquarter-chord point, while allowing the endplates to

design, with boundary layer suction mechanismsremain parallel to the ground at all times. It was

located just ahead of the moving ground belt.thought logical to establish baseline data with thisconfiguration before attempting to investigate multi-

2.2 Models and installationelement devices. A second, pressure-tapped modelwas also constructed. The pressure taps were locatedThe race scenario of one car following another was

replicated generically in the wind tunnel by the con- in the direction of the oncoming flow, at the spanwisestations of 2z/b=0.09, 0.49, and 0.89. There werestruction and use of simple experimental models that

were selected to represent the salient characteristics 44 taps at each station, 24 on the suction surface and20 on the pressure surface.of each particular vehicle. Starting with simple models

will allow for more complexity to be added in the The configuration adopted for the 2.1×1.5 m windtunnel is shown in Fig. 1. The wing was installedfuture, as a greater understanding of the inherent

interactions develops. The component of the follow- above the moving ground by attaching it to an over-head balance via movable struts. The struts hading car that was chosen to commence the study was

the front wing. This device was selected because it machined slots along their top ends that allowed adegree of freedom in the vertical direction in orderis the most forward part of the vehicle that would

Fig. 1 Experimental configuration adopted in the 2.1×1.5 m wind tunnel



to raise and lower the model. Machined blocks of 2.3 Methodology and uncertaintiesvarying thickness were then placed between the

The experimental tests involved placing the wing inbottom of the endplates and the top of the rolling

different oncoming flow conditions produced in theroad surface to set specific ride heights.

wind tunnel facilities. The baseline flow condition,The representation of the leading car developed

FC1 or clean air, was the uniform freestream thatin two stages. Firstly, a wing without endplates was

entered the test sections. The remaining flow con-used to idealize the upper elements of a typical rear

ditions were categorized as dirty air, since the bodieswing. This wing had a chord of 140 mm and a span

that were placed upstream of the wing disturbed theof 400 mm, as it was scaled to 40 per cent of the

oncoming flow.dimensions of the rear wing of a typical 2002 F1 car.

The dirty air flow condition was developed in aFollowing initial tests with this device, a generic

series of incremental steps. Each step was designeddiffuser was then incorporated into a bluff body

to add a physical component that would produce ashape so as to improve the model further. The rear

wake more closely resembling that of a generic racingwing was then attached to the diffuser bluff body via

car. As a result, the following subdivisions were used:endplates. The height and lateral position of the bluff

body were adjustable, as was the angle of the diffuser 1. FC2 – the wing used to simulate the rear wing oframp. Its width was 400 mm, as it was scaled to be a typical open-wheeled racing car was placed40 per cent of the actual width of a diffuser of a upstream of the test wing.typical 2002 F1 car. The entire length of the bluff 2. FC3 – the bluff body which incorporated a diffuserbody was 900 mm. The model was mounted to a and a rear wing was placed upstream of the testground board that was positioned just ahead of the wing.rolling road and suction box in the wind tunnel testsection. The distance from the base of the bluff body Owing to limits placed on the time available in theto the leading edge of the wing was 2160 mm wind tunnel facility, it was only possible to investi-(approximately 1.5 car lengths at full scale). A sketch gate FC1 and FC3 in an in-depth manner. FC2 wasof both models in the 2.1×1.5 m test section is investigated to a lesser extent.provided in Fig. 2. The physical procedure undertaken was to vary the

In the larger 3.5×2.5 m facility, where the LDA height and angle of attack of the test wing whensystem was permanently located, both the test wing placed in the different oncoming flow conditions.and the bluff body were mounted using the same The height of the endplate of the wing above thestrategy. The only difference was an increase in tunnel floor typically ranged from 2 to 169.6 mm.the distance between the two models owing to the The angle of attack ranged from −5 to 30°, in orderlonger moving ground belt system in that tunnel. The to ensure that the region of stall of the wing wasdistance from the base of the bluff body to the lead- covered. The investigations were carried out at aing edge of the wing was 3100 mm in this tunnel constant wind tunnel dynamic pressure of 25 mm

water, which corresponded to a freestream velocity of(approximately 2.2 car lengths at full scale).

Fig. 2 Partial schematic showing the relative locations of the diffuser bluff body and the testwing in the 2.1×1.5 m wind tunnel



20 m/s. Owing to variations in the ambient pressure to the tunnel test section in order to minimize theeffect of temperature changes on the transducer.and in the temperature of the wind tunnel, the

Reynolds number based on wing chord varied from For each ride height investigated, 22 values wererecorded at each tap. An average was then calculated300 000 to 309 000.

For some test conditions, the angle of the diffuser to provide the most representative value. Using thesame procedure as that used for the force coefficients,ramp was changed in order to generate downstream

flow typical of high-angle and low-angle diffusers. For the uncertainty in the pressure measurements wasestimated to be ±0.01.other test conditions, the height or lateral position

of the diffuser was varied, while keeping the ramp The LDA system at the University was previouslyquoted as producing uncertainties in u/U

2and v/U

2angle constant. In the baseline configuration, themodel was positioned 60 mm above the ground of ±0.005 [16].board and incorporated a 16.7° ramp. Techniquessuch as flow visualization, force measurements,pressure tapping, and LDA were then used to extractthe relevant data from the wing and its flowfield.

3 RESULTS AND DISCUSSIONFor the flow visualization studies, a mixture of

invisible blue fluorescent pigment, paraffin, and oleic3.1 Diffuser flowfield

acid was used. It was then applied to the model withSince FC3 was representative of the conditions thata paint roller or paintbrush. When dried, the mixturemost closely resembled the flow generated by a lead-formed a white flaky coating which highlighted theing open-wheeled racing car, an outline of its mainflow of air around the object.features will be presented for the baseline case. FlowForces were measured on the overhead balancesvisualization was performed on the ramp and end-present in each wind tunnel. Typically, a series ofplate region of the diffuser bluff body. The resulting75 samples was taken at each ride height. An averagepattern, which is illustrated in Fig. 3, indicated thewas then calculated and output by the controllingpresence of vortex flow. The vortex flow was high-software. Uncertainties were calculated following thelighted by the swirling lines that trailed along themethods outlined by Coleman and Steele [17]. Theedge of the ramp, close to the endplate, as wasmaximum uncertainties in the downforce and dragpreviously described by Senior and Zhang [19]. Thecoefficients were estimated as being ±0.002 andflow was found to be symmetric about the bluff±0.0004 respectively.body centre-line and showed the characteristics ofPressures were measured by connecting the tubeshigh-angle diffusers as outlined by Rhurmann andthat emanated from the taps to a ZOC pressure trans-

ducer system [18]. The ZOC was located externally Zhang [20].

Fig. 3 Swirling lines on the diffuser ramp, indicating the presence of vortex flow



Smoke trails, which were released from a portable progressing through a plane perpendicular to the testsection centre-line.wand, showed that the flow in the vicinity of the rear

of the bluff body, close to the diffuser sideplates, was The wake deficit – highlighted by plots of u/U2

–is seen to have been greater at locations closer to thesucked in towards a plane that coincided with the

centre-line of the test section. The flow then pro- centre of the tunnel (2z/b=0) than at locationscloser to where the tips of the wing would have beengressed downstream, seemingly concentrated in the

middle of the test section as it did so. There was also positioned (2z/b=1). The plots for v/U2

highlightan upwash close to the centre of the tunnel. Awaya significant increase in the amount of audible noise

associated with this configuration, an indication of from the centre-line, the upwash gradually decreased,transitioning to a downwash by the location ofthe generation of turbulent flow [21].

Flowfield tests were carried out in the R. J. Mitchell 2z/b=0.80. At 2z/b=0 the maximum turbulenceintensity was found to be 11 per cent while atfacility, as it was equipped with a three-component

LDA system. The objective was to provide a more 2z/b=0.87 it was found to be 7.4 per cent.detailed map of the wake of the bluff body. Themeasurements, which were taken 3l downstream of

3.2 Effect of upstream bodiesthe bluff body, are presented in Figs 4(a) to 4(d) forclarity. It was only possible to extract two components The aerodynamic changes that the upstream bodies

induced on the downstream wing will be presentedof the velocity owing to a failure of some of thesystem hardware. The plots reveal a variation in the in the form of a comparison with the baseline data

that were obtained from the clean air case. Figure 5freestream and vertical velocity components while

Fig. 4 Velocity profiles 3l downstream of the diffuser bluff body in the 3.5×2.5 m wind tunnel



The variation in the drag coefficients with rideheight in the three flow conditions is presented inFig. 6. The plots indicate that the flow generated bythe upstream bodies caused an increase in the dragof the downstream wing. Above h

r/c=0.4, the highest

drag values occurred in FC3, while below this pointFC2 produced the highest. At ride heights belowh

r/c=0.153 in all flow conditions, fluctuations in

the values became evident. It is not entirely certainwhy this characteristic occurred. Flow visualizationimages did, however, highlight the presence ofrecirculating regions of flow at the junction of thewing and endplate. The regions were found tochange position and size with ride height changes.It is plausible to assume that this movement may

Fig. 5 Experimental downforce coefficients in ground have had some effect on the drag values.effect in the three flow conditions in which they An angle of attack variation was performed in FC1were measured and FC3 to highlight any changes that may have

occurred. The results are plotted in Fig. 7 for two rideheights, h

r/c=0.833 and h

r/c=0.153. It can be seenshows plots of the downforce coefficients while vary-

ing the height of the wing in all flow conditions. Its that, throughout the useful angle of attack range, thewing generated less downforce in FC3 than it did inangle of attack was held constant at 5°. On initial

observation, it is clear that there were successive FC1. Furthermore, previous authors reported that thelift (downforce) curve slope of a wing increased withdecreases in downforce as the oncoming flow

progressed from FC1 to FC2, and finally to FC3. decreasing ride height. It has now become evidentthat this trend also existed in dirty air conditions.It can be deduced that more downforce was lost

at greater ride heights than was the case at lower ride The results of Fig. 7 also highlighted the fact thatmore downforce was lost at greater ride heights thanheights. For example, at h

r/c=0.833, the downforce

value for FC2 was approximately 13 per cent less than at lower ride heights.The dirty air conditions also had the potential tothat of FC1, while the downforce value for FC3 was

approximately 33 per cent less than the same FC1 alter the stall characteristics of the wing. At hr/c=

0.833 in FC1 there was an abrupt stall at 23°. How-baseline value. At hr/c=0.401, the corresponding

losses in FC2 and FC3 were 7.6 and 25 per cent ever, in FC3, the stall became more gradual. This resultsuggested that the boundary layer characteristicsrespectively. At h

r/c=0.204, the corresponding losses

in FC2 and FC3 were 4.8 and 18 per cent respectively.The shape of the curves for FC1 and FC2 is con-

sistent with previously published data. There was thecharacteristic increase in downforce to a maximumvalue as the ride height was reduced to a certainpoint (h

r/c#0.09). Below this height, the downforce

then began to decrease, and continued to do so forfurther height reductions. The curve for FC3 followeda similar trend of increasing downforce between theride heights of h

r/c=0.83 and h

r/c#0.09. At ride

heights below this range, however, a new character-istic was revealed. The region of downforce decreasewas halted by an abrupt increase in downforce atvery low ride heights. The second region of down-force increase produced coefficients that were higherthan the maximum that was achieved during the firstregion of force increase. Tests that were carried outat a higher freestream velocity indicated that the Fig. 6 Experimental drag coefficients in ground effectsecond region of force increase commenced at a in the three flow conditions in which they were

measuredslightly greater ride height.



Observation of the pattern shows that there was alaminar separation bubble across the majority of thespan of the wing (a). At mid-span, it can be deducedthat the bubble trapped some of the flow visualizationfluid while it was drying (b), thereby causing anadditional obstruction to the flow, which then wenton to cause premature trailing edge separation (c).Premature separation was also aided by a blob ofunmixed solution (d). Close to each endplate, at thetrailing edge of the wing, a region of recirculatingbubble flow can be identified (e).

In comparison, Fig. 9(b) shows the flow overthe suction surface of the wing in FC3. It can bededuced that there is now a dramatic difference inthe surface flow when compared to FC1. At a con-

Fig. 7 Lift curves at hr/c=0.153 and 0.833 in flow

siderable distance either side of mid-span the laminarconditions FC1 and FC3separation bubble has disappeared (f) as the flow wasturbulent from the outset. The flow also appeared to

of the wing could potentially be altered in the dirty have been antisymmetric, as evidenced by the com-air flow. parable sizes of the regions labelled (g) and (h).

A series of measurements was taken while moving The asymmetry was also reflected in the slightlythe diffuser bluff body laterally away (indicated by different shape of the separation region at (i) and (j).2z/b) from its original position in front of the test The recirculating bubble regions (k, l), which nowwing. The resulting downforce curves at h

r/c=0.153 have distinctly different shapes when compared with

are plotted in Fig. 8. It can be seen that throughout each other, are still observed at either endplate, closethe prestall angle of attack range, the least down- to the trailing edge.force was generated when the bluff body was directly The flow on the pressure surface of the wing in FC1in front of the wing (2z/b=0). As the bluff body can be seen in Fig. 10(a). A large region of laminarwas moved laterally away, the downforce gradually flow is visible over the majority of the surface. Alsorecovered to the freestream values (as was similarly present was a laminar separation bubble (m), whichreported in reference [3]), and in the case of formed across the entire span.2z/b=1.5 it surpassed them. In comparison, Fig. 10(b) shows the flow over

Flow visualization was performed in FC1 and FC3 the same surface in FC3. There was again anotherin order to ascertain the physical effects of change striking difference when compared with FC1. Thein the oncoming flow conditions. Figure 9(a) shows laminar separation bubble was again eliminatedthe flow over the suction surface of the wing in FC1. from a considerable distance either side of the mid-

span of the wing (n), but was still clearly visibletowards the tips (o). In contrast to the overall flowon the suction surface, the flow on the pressuresurface appeared to be symmetric.

A comparison of the surface pressure distributionsin FC1 and FC3 can be seen in Fig. 11. Measurementswere taken at the ride heights of h

r/c=0.833, 0.401,

0.153, and 0.077, but only those at hr/c=0.153 are

shown. The data indicate that changing the flowfrom FC1 to FC3 resulted in a decrement in thepressure distribution at each station investigated.The decrement appeared to be greater at stationscloser to the centre of the wing than at stationsfurther away. Also evident was the fact that themajority of the loss occurred from the suctionsurface of the wing, especially at 2z/b=0.09 andFig.8 Downforce curves at h

r/c=0.153 while moving

0.49. These results also held for the other ride heightsthe bluff body laterally away from its originalposition in front of the wing investigated.



Fig. 9 Comparison of the suction surface flow in (a) FC1 and (b) FC3 for hr/c=0.153. Leading

edge uppermost in image

The pressure distribution data also show the exist- not support its existence. That is, the boundary layerwas now turbulent. It can be inferred that the greatestence and disappearance of the separation bubbles.

A typical location of the bubble can be pinpointed loss in pressures on the suction surface occurredbecause of the elimination of an extensive region ofby a plateau-like region followed by a steep drop,

both of which produce an area that does not appear laminar flow when progressing from FC1 to FC3. Incontrast, between x/c#0.6 and x/c#0.95 the lossto fit with the natural curvature of the plot. For

example, the region bounded by x/c#0.45 and in pressure was not as great, suggesting that theturbulent boundary layer was not that sensitive tox/c#0.6 at 2z/b=0.09. When the flow was changed

to FC3, the region just described vanished. A similar the change in the characteristics of the oncomingflow.scenario existed on the pressure surface between

x/c#0.65 and x/c#0.75. The two-dimensional sectional downforce coeffi-cients were estimated for each station, at eachA more detailed analysis of the same plot can

provide the reader with further insight into the ride height at which measurements were taken.The trapezium rule was used to integrate the distri-aerodynamic changes experienced by the wing. The

presence of a separation bubble on the suction butions between x/c=0 and x/c=0.9. Owing tomanufacturing constraints, no pressure tap wassurface in FC1 has just been highlighted. This surface

flow feature implied that the boundary layer prior to located beyond x/c=0.9 on the wing pressure surfaceand beyond x/c=0.95 on the suction surface. Theits existence was laminar, while the boundary layer

after the bubble was turbulent. A comparison with integration was therefore not carried out for the last10 per cent of each section. Table 1 summarizes thethe corresponding curve for FC3 will show that, in

general, the greatest difference in values between the changes as the oncoming flow evolved from FC1 toFC3. It is clear that, at each ride height, most lift wastwo plots occurred between x/c#0.01 and x/c#0.6.

As the bubble had disappeared in this flow condition, lost at 2z/b=0.09. This was followed by the sectionat 2z/b=0.49, then 2z/b=0.89.it was evident that the flow prior to that point did



Table 1 Sectional downforce coefficient values force than the inboard sections. As the ride heightfor each of the ride heights investigated was further reduced, this trend gradually reversed.

At hr/c=0.153 and 0.077, the inboard sections gen-2z/b=0.09 2z/b=0.49 2z/b=0.89

erated more downforce than the outboard sections.h

r/c=0.833 From the data presented thus far, a more detailed

Cl: FC1 0.779 0.750 0.648

understanding has emerged regarding the changesCl: FC3 0.361 0.472 0.572

%DCl

53.7 37.1 11.7 experienced by the downstream wing. To supplementh

r/c=0.401 these data further, the sketches in Figs. 12 and 13 have

Cl: FC1 0.956 0.916 0.766 been produced. The first image shows a plan view of

Cl: FC3 0.626 0.680 0.679

the experimental configuration in the 2.1×1.5 m%DCl

34.5 25.8 11.4

wind tunnel. The arrows on the image show the mainhr/c=0.153

Cl: FC1 1.54 1.45 1.14 path taken by the flow immediately surrounding the

Cl: FC3 1.17 1.17 1.05 bluff body, as revealed by the smoke wand flow%DC

l24.0 19.3 7.89

visualization tests. The entrainment of the outer fluidh

r/c=0.077

by the diffuser vortices led to a significant increaseCl: FC1 1.44 1.30 0.613

Cl: FC3 1.15 1.10 0.580 in turbulent flow along the centre-line of the tunnel.

%DCl

20.1 15.4 4.89The concentrated turbulent flow in this region wasresponsible for the distinctive patterns that appearedin the central portions of the wing, as previouslyIn FC3, the shape of the load distribution of theshown in Figs 8(b) and 9(b).wing changed depending on its ride height. At

Plane a–a is shown in the second image. In thishr/c=0.833 and 0.401 it can be deduced that the

outboard sections of the wing generated more down- view, the hypothetical positions of the diffuser vortices

Fig. 10 Comparison of the pressure surface flow in FC1 and FC3 for hr/c=0.153. Leading edge

uppermost in image



The vortices are seen to rotate in such a manneras to induce an upwash along the wing centre-line,while inducing some degree of downwash close tothe tips. The upward velocity along the centre-linewould have been noticeably greater than the down-ward velocity closer to the wing tips owing to theadditive effect of the velocity components from bothvortices. This flowfield structure is supported by theLDA measurements that were presented in Fig. 4.

The structure of the flowfield in which the wingwas placed can be used to help explain the reductionin downforce that it experienced when placed down-stream of the diffuser bluff body. In the dirty air, thedownforce reduction followed from the generationof lower C

Pvalues, and hence lower total forces.

The lower CP

values were induced by a reductionin the freestream velocity, an upwash of the flow,and increased turbulence. The reduced freestreamvelocity essentially lowered the Reynolds number atwhich the wing was operating, while the upwash hadthe effect of reducing its effective angle of attack. Bothscenarios would have contributed to lower pressures.The increased turbulence in the oncoming flow pro-moted early laminar-to-turbulent boundary layertransition on the wing. Consequently, the boundarylayer was much thicker from the outset, thereforehaving a greater decambering effect on the profile.That is, the wing effectively lost camber, which would

Fig. 11 Pressure distributions in FC1 and FC3 at thehave also resulted in lower pressures.spanwise stations of 2z/b=0.09 (a), 0.49 (b),

Also evident from the results was the fact that moreand 0.89 (c), for the ride height of hr/c=0.153

downforce was lost at greater ride heights than wasthe case at lower ride heights. This observation canbe linked to the change in upwash with increasinghave been sketched. The figure is used to visualizedistance above the ground, when in the flow pro-the flow structure that was reasoned to have beenduced by the diffuser. As the wing ride height wasoccurring at that spatial location, in the vicinity of theincreased, it operated in a region of greater upwash,wing. The circular dashed lines are used to representwhich continued to reduce the effective angle ofthe vortices generated by the upstream diffuser,attack at which it operated. At greater ride heights,while the arrows signify the vertical component of

velocity in the plane. there was also continued slowing of the oncoming

Fig. 12 Plan view of the experimental configuration in FC3



Fig. 13 View through plane a–a of Fig. 12

flow, in addition to increased turbulence levels. The wing in ground effect experienced a decrease inits downforce values and an increase in its dragend result for the wing was therefore lower overall

downforce values. values.2. When varying the height of the wing in the dirtyComparisons at the spanwise stations at which

pressure distributions were recorded indicated that air flow, more downforce was lost at greater rideheights than was the case at lower ride heights.more lift was lost from the mid-span of the wing than

was the case from the ends of the wing. The mid- 3. When in the dirty air flow, more lift was lost fromsections closer to the mid-span of the wing thanspan was thought to have experienced a greater loss

in downforce than the tips because the majority of was the case for sections closer to the tip of thewing.the disturbed flow appeared to be concentrated in

that region, and because that region experienced an 4. When in the dirty air flow, the shape of the span-wise load distribution altered, depending on rideupwash, while the tips experienced flow with a neutral

to slight downward component of velocity, in addition height. At greater ride heights, the load distri-bution was lower at mid-span than it was towardsto a slightly higher freestream velocity.

The increase in the drag of the wing can be the wing tips. At lower ride heights, the trendreversed.explained by the resultant interaction of a number

of factors. Firstly, flow visualization in the disturbed 5. The downstream wing was affected by the upwashflowfield of the upstream diffuser bluff body. Theconditions highlighted the fact that the laminar

separation bubble was eliminated from the middle presence of an upwash would have resulted in anincrease in the induced drag experienced by theportion of both the upper and lower surfaces of the

wing. This result in itself should have accounted for wing.6. The disturbed flow emanating from the upstreama decrease in the drag, but this decrease may have

been outweighed by drag increments from increased body had the ability significantly to alter the sur-face flow patterns on the downstream wing. Theinduced drag [22] and the increased extent of the

turbulent boundary layer. Increased induced drag altered characteristics included earlier laminar-to-turbulent boundary layer transition, and theresulted from the fact that the wing operated in the

upwash of the upstream diffuser bluff body. The elimination of laminar separation bubbles.increased extent of the turbulent boundary layerwould have caused increased skin friction drag.

ACKNOWLEDGEMENT

The authors would like to recognize the support of4 CONCLUSIONSthe School of Engineering Sciences at the Universityof Southampton, Highfield, Southampton, UK, inThe data and associated analyses that have been pre-providing a scholarship to author Michael Soso.sented were aimed at cataloguing the aerodynamic

changes that may be experienced by a single-elementwing when it operated in ground effect, downstream

REFERENCESof a leading racing car. The following conclusions canbe drawn from the research:

1 Romberg, G. F., Chianese Jr, F., and Lajoie, R. G.1. In the flow produced by bodies used to simulate Aerodynamics of race cars in drafting and passing

situations. SAE paper 710213, 1971.a leading racing car, a downstream single-element



2 Howell, J. Catastrophic lift forces on racing cars. 16 Zhang, X. and Zerihan, J. Aerodynamics of a doubleelement wing in ground effect. Am. Inst. Aeronaut.J. Wind Engng and Ind. Aerodynamics, 1981, 9,Astronaut. J., June 2003, 41(6).145–154.

17 Coleman, H. W. and Steele, W. G. Experimentation3 Dominy, R. G. The influence of slipstreaming on theand uncertainty analysis for engineers, 1999 (Johnperformance of a Grand Prix racing car. Proc.Wiley).IMechE, Part D: J. Automobile Engineering, 1990,

18 ZOC22B and ZOC23B instruction and service manual204, 35–40.(Scanivalve Corporation).4 Ahmed, S. R. Wake structure of typical automobile

19 Senoir, A. E. and Zhang, X. The force and pressureshapes. Trans. ASME, J. Fluids Engng, March 1981,of a diffuser-equipped bluff body in ground effect.103, 162–169.Trans. ASME, J. Fluids Engng., March 2001, 123,5 Ahmed, S. R. An experimental study of the wake105–111.structures of typical automobile shapes. J. Wind

20 Ruhrmann, A. and Zhang, X. Influence of diffuserEngng and Ind. Aerodynamics, 1981, 9, 49–62.angle on a bluff body in ground effect. Trans. ASME,6 Hamidy, E. The structure of wakes of 3D bluffJ. Fluids Engng., March 2003, 125, 332–338.bodies in proximity to the ground, PhD Thesis,

21 Gad-el-Hak, M. Flow control – passive, activeDepartment of Aeronautics, Imperial College ofand reactive flow management, 2000 (CambridgeScience, Technology and Medicine, London, 1991.University Press).7 Bearman, P. W. Near wake flows behind two- and

22 McCormick Jr, B. W. Aerodynamics of V/STOL flight,three-dimensional bluff bodies. J. Wind Engng and1967 (Dover Publications).Ind. Aerodynamics, 1997, 69, 33–54.

8 Lienhard, H., Stoots, C., and Becker, S. Flow andturbulence structures in the wake of a simplified car

APPENDIXmodel (Ahmed model), Lehrstuhl fur Stromungs-mechanik (LSTM), Universitat Erlangen Nurnberg,Erlangen, Germany. Notation

9 Ranzenbach, R. and Barlow, J. B. Two-dimensionalb wing span (m)airfoil in ground effect, an experimental and com-c wing chord (m)putational study. SAE paper 942509, 1994.

10 Ranzenbach, R. and Barlow, J. B. Cambered airfoil CD

coefficient for the horizontal componentin ground effect – wind tunnel and road conditions. of the resultant force on the wingAIAA-95-1909-CP, 1995. C

ltwo-dimensional sectional lift coefficient

11 Ranzenbach, R. and Barlow, J. B. Cambered air- CL

coefficient for the vertical component offoil in ground effect – an experimental and com-

the resultant force on the wingputational study. SAE paper 960909, 1996.C

Ppressure coefficient12 Knowles, K., Donoghue, D. T., and Finnis, M. V. A

FC1 clean air flowstudy of wings in ground effect. LoughboroughFC2 flow generated by the upstream wingUniversity Conference on Vehicle Aerodynamics,

1994, pp. 22.1–22.13. FC3 flow generated by the upstream diffuser13 Jasinski, W. J. and Selig, M. S. Experimental study bluff body

of open-wheel race-car front wings. SAE paper hr

ride height (m)983042, 1998. l length of the diffuser bluff body (m)

14 Zerihan, J. D. C. An investigation into the aero-U2

length of the diffuser bluff bodydynamics of wings in ground effect, PhD thesis,u, v, w Cartesian velocity componentsUniversity of Southampton, April 2001.x, y, z Cartesian coordinates15 Zerihan, J. and Zhang, X. Aerodynamics of a single

element wing in ground effect. J. Aircraft, November–December 2000, 37(6). a angle of attack (deg)


aerodynamics of a wing in ground effect in generic racing car wake flows

Documents