AIAA JOURNAL

Vol. 41, No. 6, June 2003

Aerodynamics of a Double-Element Wing in Ground Effect

Xin Zhang¤ and Jonathan Zerihan†

University of Southampton, Southampton, England SO17 1BJ, United Kingdom

An investigation of a cambered, double-element, high-lift wing operating in ground effect was performed. The
effect of ground proximity and  ap setting was quantiŽ ed in terms of aerodynamic performance and off-surface
 owŽ eld characteristics. From that, it was found that the  ow is three-dimensional toward the wing tip with the
main element generating most of the downforce but retains quasi-two-dimensional features near the center of the
wing. However, at large heights the downforce increases asymptotically with a reduction in height. Then there
is either a plateau, in the case of a low  ap angle, or a reduction in downforce, in the case of a large  ap angle.
The downforce then increases again until it reaches a maximum and then reduces with decreasing height above
the ground. The maximum downforce is dictated by gains in downforce from lower surface suction increases and
losses in downforce caused by upper surface pressure and lower surface suction losses, with a reduction in height.
For the high  ap angle there is a sharp reduction just beyond the maximum, mainly because of the boundary layer
separating, and a resultant loss of circulation on the main element.
Downloaded by UNIVERSITY OF LEEDS on July 28, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.2057

Nomenclature of downforce, and hence the grip. It is important to maintain con-

b = wing span; 1100 mm sistent levels of front end grip, not only for performance reasons.
CD = drag coefŽ cient, D=q1 S It is not only important to have a car that handles well for perfor-
CL = lift coefŽ cient, L=q1 S mance reasons, but it is also a signiŽ cant safety issue. In addition
CP = pressure coefŽ cient, p=q1 to the aerodynamic performance of the front wing, another major
c = wing chord; 380 mm issue is the wake/vortices that it generates. The  ow to the under-
cf =  ap chord; 165.7 mm tray and diffuser in particular, but also the radiators and rear wing,
h = height above ground is severely affected by the front wing because they all operate in the
q1 = dynamic head, 12 ½1 U1 2 wake/vortices from the wing.
Re = Reynolds number, ½1 U1 c=¹ There has been a lack of data on inverted wings in ground effect
S = platform area that are supposed to simulate the  ow correctly. Ground effect is
U1 = freestream velocity normally felt within a height of one chord. Using a single-element
u; v; w = velocity components in x; y; z axis system wing, Zerihan and Zhang3 show that the use of a Ž xed ground would
u min = minimum u velocity component in wake result in a substantial reduction in the downforce, with a signiŽ -
u 0 u0 = turbulent stress cant drop below a critical height of around 0.3 chord (by as much
x; y; z = Cartesian coordinates, x positive downstream, as 25%). Above the critical height major  ow features would be
y positive up, z positive to starboard the same as in the freestream case. Below the critical height, new
® = incidence features emerge, for example, shear-layer instability and wall jet,
¹ = viscosity which will not be simulated correctly by the use of a Ž xed ground.
»; ³ = local coordinates, » tangential, ³ normal Among various studies, Knowles et al.4 were the Ž rst to study ex-
½1 = freestream density perimentally a single-element wing with the suction surface near
a moving ground. However, neither three-dimensional effects nor
Introduction off-surface  owŽ eld was studied. Recently, in a series of studies,

A WING operating in the proximity to the ground introduces Zerihan and Zhang3;5;6 conducted investigations of single-element
different  ow physics from that in freestream. Comparatively wings in ground effect, including three-dimensional effect and off-
little information is placed in public domain about an inverted wing surface  owŽ eld surveys, as well as numerical modeling.
in groundeffect,which hasits applicationin the automobileindustry. In practice, a typical geometry of the front wing is of a multi-
For example, the front wing of a racing car operates in ground effect, element conŽ guration, which would introduce additional features
at typical heights of 70–100 mm from the ground,1 and produces of importance. The study of multielement  ows is an area that has
about 25–30% of the total downforce of the car.2 The downforce challenged researchers for a long time. Smith7 described the Ž ve
works in conjunction with the mechanical grip to improve the ac- beneŽ cial effects of the gaps between the elements in multielement
celeration, braking, and cornering speed of the car. However, it is  ows: slat effect, circulation effect, dumping effect, off-the-surface
not only the overall level of downforce that is the important fac- pressure recovery, and fresh-boundary-layer effect. Ranzenbach
tor. The front wing changes height from the ground because of the et al.8 demonstrated the ground effect for a double-element airfoil
suspension movements on the car. This severely affects the level conŽ guration. Their work begins to address the topic, using two-
dimensional model tests in a Ž xed ground wind tunnel on a NACA
Received 4 April 2002; revision received 27 January 2003; accepted for 632 ¡215 Mod B section with a 30% slotted  ap for the double-
publication 27 January 2003. Copyright ° c 2003 by Xin Zhang and Jonathan element studies. Force reduction was observed. Jasinski and Selig9
Zerihan. Published by the American Institute of Aeronautics and Astronau- presented an experimental study of a three-dimensional multiele-
tics, Inc., with permission. Copies of this paper may be made for personal ment wing in ground effect, again using a Ž xed ground facility. Two
or internal use, on condition that the copier pay the $10.00 per-copy fee to trailing vortices were observed rolling up from the end plate, the
the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA size of which increased for the larger  ap de ection. In this study
01923; include the code 0001-1452/03 $10.00 in correspondence with the we aim to quantify the performance of a generic double-element
CCC.
¤ Professor of Aerodynamics, Aeronautics and Astronautics, School of wing in ground effect, employing model tests with correct ground
Engineering Sciences. Senior Member AIAA. conditions. Earlier results illustrate the large-scale unsteady and
† Research Student, Aeronautics and Astronautics, School of Engineering time-averaged  ow features of a high-lift single-element wing. It
Sciences; currently Aerodynamicist, BAR Operation Centre, Brackley, shows that the ground has indeed a profound effect on the aero-
England NN13 7BD, United Kingdom. dynamic performance. By associating  uid  ow measurements and
1007
 1008 ZHANG AND ZERIHAN

observation with force and pressure measurements, it is possible to image velocimetry (PIV) tests were performed at a constant speed
develop a greater understanding of  ow physics. of 30 m/s. The Reynolds numbers were in the range 0:735 £ 106
to 0:765 £ 106 based on the total chord. The tests were performed
Methods
transition free.
Test Facilities Two models were manufactured: a clean wing used for forces,
Tests were conducted in the University of Southampton  ow visualization,LDA, and PIV; and an identical model, with the
3.5 £ 2.5 m wind tunnel for laser Doppler anemometry (LDA) sur- addition of pressure taps. The surface pressures were measured by a
veys (Fig. 1) and 2.1 £ 1.7 m wind tunnel for other tests. Details can chordwisedistributionof pressuretaps, locatednear to the semispan.
be found in Ref. 10. Both tunnels are of a conventional closed-jet, It comprises 25 taps on the suction surface and 23 on the pressure
closed-circuit design. At 30 m/s the freestream turbulence is less surface. A similar conŽ guration was used for the 25 pressure taps
than 0.2% in both tunnels. The tunnels are equipped with a large, on the  ap. The chordwise group contains 13 taps on the suction
moving belt rig. surface and 12 taps on the pressure surface.
The forces and surface pressures acting on the wing have been
Wing Model
measured for the model at a wide range of ground heights, from
Earlier studies were performed on a generic single-elementwing, 1.97c to less than 0.05c above the ground. The height was deŽ ned by
with a cambered proŽ le.3 Current tests employ the single-element the distance from the ground to the lowest point on the wing with the
wing as the main element in addition to a large  ap. The main el- wing incidenceset to 0 deg. The effect of changingthe  ap de ection
ement has a modiŽ ed General Aviation-Whitcomb (GAW) airfoil angle was investigated at the different heights. The incidence of the
(Fig. 2). A detailed listing of the coordinates was given by Zeri- wing was varied using a rotation about the quarter-chord position.
han (see Table 3 in Ref. 10). The main changes include a forward The reference incidence of 1 deg at which all double-element tests
movement of the lowest point on the suction surface, a  attening were performed is the incidence correspondingto end plates parallel
of the pressure surface, and a modiŽ cation to the leading edge. The to the ground, with the wing in its datum  ap de ection, that is, a
Downloaded by UNIVERSITY OF LEEDS on July 28, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.2057

main purpose of the modiŽ cation is to minimize the wake behind true incidence of 14.1 deg.
the airfoil. A span of 1100 mm was used, which correspondsto less Off-surface results were also taken over a chordwise slice at the
than 75% of the width of the moving belt in the smaller wind tunnel, semispan of the wing with a LDA system to extract mean  ow and
to minimize effects at the edge of the belt. The wing has a constant turbulence data. LDA measurements were performed with a three-
chord of 223.4 mm. componentDantec systemwith a 5-W argon-ionlasergenerator.The
A  ap with a constantchord of 165.7 mm was used, the aft 35 mm system was operated in backscatter mode. The velocities measured
of which is the thin region, 1 mm thick. A detailed listing of the  ap in the beam axes were resolved into the tunnel coordinate system
coordinateswas given in Ref. 10. Two  ap angles were used (see the .x; y; z / using a matrix transformation.Seeding was introduced by
section on  ap location optimization). For the high  ap angle the three seeding generatorslocated downstream of the rolling road, be-
chord is 378.9 mm. For the lower  ap angle the chord is 381.5 mm. hind the model. The LDA signals were analyzed using three Dantec
In presenting the results, all of the length scales were normalized by Burst Spectrum Analyzers. On average, a total of 800 bursts (in-
380 mm. The combined chord correspondsto an aspect ratio of 2.89. stantaneous samples) were collected for each data point. The LDA
End plates were employed, which are rectangular with dimensions study was supported by PIV survey of the  ow between the  ap
400 £ 170 £ 4 mm (Fig. 2). and the ground, which provided insight into the state of the  ow
Tests immediately behind the main element. PIV was performed using a
All force, pressure,and  ow visualizationtests were performed at Dantec PowerFlow system. The laser for the PIV system was located
a constant dynamic pressure of 56.25 mm water. LDA and particle approximately1.6 m downstream of the center of the wing, after the
end of the rolling road. The region of the  owŽ eld including the
trailing-edge region, from the ground to above the  ap extending to
x=c D 1:2 at the wing semispan, was mapped. The total number of
samples recorded per run was 500. The analysis sequence used was
to cross correlate the data on 32 £ 32 pixels and perform a range
validation of the resulting vectors, generating a 157 £ 125 grid. De-
tails of the system can be found in Zerihan and Zhang.5 The range
of heights extended from 22 to 100 mm.

Errors and Uncertainties

The incidence of the wing was set to within §0:005 deg, and
the height above ground was set to within §0:2 mm. The constant
dynamic pressure was set to 56.25-mm water §0:05 mm. The uncer-
tainties in the force measurements were calculated using the addi-
Fig. 1 Model installation in wind tunnel. tion method and a 95% conŽ dence.11 C L and C D have uncertainties
of §0:003 and §0:0006, respectively. Uncertainties in the surface-
pressure results were calculated using the rss method as described
by Moffat11 : the worst case corresponding to a C p of §0:035. The
short-term repeatability was investigated; the highest uncertainties
were found to be at the suction peak and the transition bubble, the
worst corresponding to a C p of §0:075.
For the LDA survey the accuracy of the traverse is 0.01 mm,
but there is a gear backlash of 0.5 mm. Attempts were made to
reduce the effect of this by always approaching a boundary layer
or wake proŽ le from the same direction. Following an analysis by
Zhang,12 an estimate of the uncertaintyin the velocity measurement
gives u=U1 § 0:005 and v=U1 § 0:005. An estimate of the 95%
conŽ dence level has been performed following procedures given
by Benedict and Gould.13 In a typical test the worst uncertainty is
2 D
0.006 for u 0 u 0 =U 1 0:007, or less than 10% of the value. For the
Fig. 2 Schematic of a double-element wing in ground effect. u velocity the worst uncertainty is less than 0.1%.
 ZHANG AND ZERIHAN 1009

Results and Discussion

Flap Location Optimization
For the forces two  ap angles were used: the datum (high  ap
angle) and a de ection of ¡8:5 deg (low  ap angle). The point
about which the  ap was rotated was at a location of x=c D 0:567,
y=c D 0:076, that is, 2 mm downstream from the leading edge of the
 ap (see Tables 3 and 4 in Ref. 10 for the exact coordinates).
A series of tests were performed in order to Ž nd the  ap location
at which the maximum downforce was produced for a constant  ap
de ection.The gap and the overlapwere variedin steps of 2 mm. The
overlap was deŽ ned as the horizontal distance between the trailing a) h/c = 0.395, low  ap angle
edge of the main element and the leading edge of the  ap, with
a positive overlap for the  ap leading edge upstream of the main-
element trailing edge. The gap was deŽ ned as the vertical distance
between the trailingedge of the main element and the lowest point on
the  ap suction surface, with a positive gap for the  ap leading edge
abovethe main-elementtrailingedge. The  ap location optimization
was performed for the datum  ap de ection angle at an arbitrary
height of h=c D 0:263. Results of the optimization can be seen in
Fig. 3. The optimum location for the  ap can be seen to be an overlap
of 0.024c and a gap of 0.032c. These correspond to 9 and 12 mm,
Downloaded by UNIVERSITY OF LEEDS on July 28, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.2057

respectively. This was used as the location of the  ap for all further
tests at the different heights and  ap de ection.
b) h/c = 0.211, high  ap angle
Oil Flow Visualization Fig. 4 Oil  ow visualization on suction surface showing leading-edge
Oil  ow visualization was performed at various heights (see ex- lowermost.
amples in Fig. 4). Although it difŽ cult to see from the pictures, the
streaklines on both elements featured spanwise components, par- wing and x=c D 0:11¡0:16 elsewhere.On the  ap a reasonablylarge
ticularly near the tips. It seems, however, that over the center the transition bubble is again seen. It is broken up at three spanwise lo-
surface streaklines do not feature signiŽ cant spanwise components. cations where the brackets secure the  ap in place. The separation
The  ow over the central portion can be regarded as quasi-two- point was measured at x=c D 0:69¡0:72, with turbulent reattach-
dimensional. SigniŽ cant three-dimensional effect is observed near ment at x=c D 0:75. The leading-edge transition was not seen at
the tip, which is probably associated with the likely presence of the h=c D 0:211, and heights below this, for the low  ap angle. On the
edge vortices. In fact a recent single-element wing study suggests main element the detachment point in the bubble was measured at
that the breakdown of the edge vortices causes the change in the lift x=c D 0:12¡0:14, with reattachment at x=c D 0:17, that is, at ap-
slope between regions a and b (see later).14 It is conceivable that proximately the same location within the measurement uncertainty.
the pressure Ž eld could be affected near the center. (Further studies The bubble on the  ap was measured at the same position as that at
need to be performed on this.) Nevertheless, the main tenet of the h=c D 0:395.
current paper remains valid. For the high  ap angle there is a signiŽ cant difference (Fig. 4b).
At h=c D 0:395 (Fig. 4a) the separation bubble indicating tran- On the main element transition at the leading edge accounts for a
sition is clear over most of the span of the wing. The separation signiŽ cant portion of the span of the wing. In the central portion
point in the bubble was measured at x=c D 0:11¡0:13, with turbu- of the image, the leading-edge bubble can be seen, and on the right
lent reattachment at x=c D 0:16. However, near the center of the the bubble is farther back, at x=c D 0:11¡0:18. As at the low  ap
wing, a small region can be seen where the bubble is breaking up angle, this portion of the wing with leading-edge transition reduces
just to the left of the semispan and where the bubble is not present as the ground height is reduced. At h=c D 0:211 it is approximately
slightly to the right of the semispan. Close inspection of the patterns 36% of the span. This reduces from 57% at h=c D 0:395 to 46%
on the wing, difŽ cult to see in the Ž gure, reveals a small bubble at h=c D 0:263, 36% at h=c D 0:211, 20% at h=c D 0:158, and to
very close to the leading edge, x=c D 0:01¡0:02, where there is no virtually zero at h=c D 0:105. The transition location for the  ap is
main transition bubble. (This phenomenon is discussed further in now very close to the leading edge, at x=c ¼ 0:58. This was found
the following paragraph.) Hence, on the main element, transition to be the case for all heights tested for the high  ap angle.
is observed at two chordwise locations at this height for the low
 ap angle: x=c D 0:01¡0:02 for a small region at the center of the
Force Behaviors
The downforce and drag coefŽ cients as the ride height is varied
are given in Figs. 5a and 5b. It can be seen that the basic charac-
teristics of the downforce with height curve are broadly similar to
the single-element wing.3 The downforce increases as the height
reduces, and eventually a maximum downforce is obtained, after
which the downforce reduces sharply. For the low  ap angle the
maximum is reached at h=c D 0:066, and for the high  ap angle
h=c D 0:079. Below the maximum downforce height is the down-
force reduction region c.
However, for the low  ap angle at a height of h=c D 0:171 a dis-
continuity in slope can be seen as the trend of increasing downforce
with height reduction abruptly stoppings; just above this height the
gradient of the line is high, and just below this point the gradient
is low. As the height is reduced further, the slope of the line in-
creases again, in a similar manner to that at large heights, until it
gradually reduces, and the downforce eventually reaches the maxi-
Fig. 3 Effect of varying overlap and gap at h//c = 0.263; datum  ap mum, correspondingto C L D 2:588. To aid in describingthe aerody-
angle. namic characteristics,heights greater than and equal to h=c D 0:171
 1010 ZHANG AND ZERIHAN
Downloaded by UNIVERSITY OF LEEDS on July 28, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.2057

Fig. 5a Downforce with ground height.

Fig. 6 Chordwise surface pressures at wing center at h//c = 1.97.

a=b  ows, at h=c D 0:237, the general trend is for a sharper gradient
at low type a heights, and a lower gradient at greater type b heights.
This is somewhat more pronounced than for the low  ap de ection.

Chordwise Pressure Distribution

A comparisonof pressuredistributionat h=c D 1:97 is made of the
single-elementwing and double-elementwing (Fig. 6). The  ap has
the effect of introducinga Ž nite pressure over the trailing edge of the
main element. The pressures on the pressure surface are increased,
and the suction generatedon the suction surface also increases com-
pared to the single-elementwing. The increments are greater for the
high  ap angle than for the low  ap angle. Over the mid and aft
portions of the wing, the increases are broadly constant for both
 ap angles. However, to x=c ¼ 0:15 the shape of the distribution
changes more signiŽ cantly for the double-element wing compared
to the single-elementwing. On the pressure surface the acceleration
of the  ow from stagnation at the leading edge is smoother with
the  ap, in the region to x=c ¼ 0:1. For the single-element wing the
 ow reaches a velocity approaching freestream at x=c ¼ 0:02. On
the suction surface, the suction peak for the single-element wing at
Fig. 5b Drag with ground height. x=c D 0:08 remains at the same place. This will be known as the
suction peak. However, a suction spike near to the leading edge at
for the low angle  ap angle will be referred to as region a, and x=c D 0:01¡0:02 becomes apparent when the  ap is added, which
heights between h=c D 0:066 and h=c D 0:171 will be referred to as shall be referred to as the suction spike. The spike grows for the
region b. higher  ap angle. The bump that is present in pressure on the single-
For the high  ap angle the wing generates signiŽ cantly more element suction surface at x=c D 0:17 representing the transition
downforce than for the lower  ap angle. At large heights, greater bubble is not as apparent for the double-element wing.
than h=c D 0:237, similar characteristics can again be seen as The two  ap angles yield different types of distributions. For
the downforce increases asymptotically as the height is reduced. the low  ap angle the  ow is accelerated over the suction surface
However, a discontinuity in slope again exists in the curve, at a little from the higher than freestream velocity near to the main
h=c D 0:237, and the downforce actually reduces suddenly just be- element trailing edge. It remains at C P ¼ ¡1 until x=c D 0:72, and
low this height. As the height is reduced further, the portion of the then the recovery starts as the trailing edge is approached. For the
curve seems more linear. A maximum in the downforce occurs at high  ap angle the  ow is acceleratedrapidly from the leading edge
h=c D 0:079, correspondingto C L D 3:028. Below this height a sud- at x=c D 0:564 to a peak suction at x=c D 0:570. The  ow is then
den drop in the downforce is observed. In a similar manner to the retarded. The difference in suction between the high and low  ap
low  ap de ection, three  ow regions are deŽ ned. anglesreducesalong the chord of the  ap and is small for the Ž nal tap
The variation of drag with height is given in Fig. 5b. The curves at x=c D 0:880. On the pressure surface the  ow is accelerated from
for the two  ap de ections show a general trend of increasing drag stagnation at a greater rate for the low  ap angle. The pressures
as the height is reduced. At the low  ap de ection the curve is remain broadly constant for each conŽ guration but are greater in
relatively smooth, with the gradient of the line generally increasing magnitude for the high  ap angle than for the low angle.
as the ground height is reduced. However, just above h=c D 0:171, The chordwise C P distributions for the low  ap angle are pre-
at the lower boundary of the type a  ows at this  ap de ection, sented in Fig. 7. For the large heights (Fig. 7a), when the ride height
the gradient is greater than just below this height, that is, a similar is reduced, the suction increases on the lower surface of both the
characteristic to the downforce against height curve is observed. At main element and the  ap. The increase, however, appears greater
the high  ap de ection a signiŽ cantly greater drag is obtained at all on the main element when compared to the  ap. The suction near the
heights, than compared to the low  ap de ection. A broadly similar trailing edge of the main element increases with increasing ground
variation with height is observed. In the border between the type proximity, the magnitude of which is comparable to the increase in
 ZHANG AND ZERIHAN 1011

a) b)
Fig. 7 Chordwise surface pressures at wing center for low  ap angle.
Downloaded by UNIVERSITY OF LEEDS on July 28, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.2057

a) b)
Fig. 8 Chordwise surface pressures at wing center for high  ap angle.

suction near the  ap leading edge. The fundamental shape of the Results for the high  ap angle (Fig. 8) show a similar effect of the
distributions does not change, and the main suction peak, and the ground on the pressure distributions. At the large heights (Fig. 8a)
spike very close to the leading edge, are still present as the ground is there is a large increase in suction on the main element lower surface
approached and remain at the same location within the spatial res- as the height is reduced. Over the  ap lower surface the increase
olution of the taps. The reduction in pressure from after the spike, in suction is not as signiŽ cant and is very small for h=c D 0:211.
to the main peak increases as the height is reduced. The distribution Again, the suction spike very near to the leading edge of the main
over the pressure surface varies little from h=c D 1:97 to 0.211. element provides the greatest overall suction at these heights. C P
Closer to the ground (Fig. 7b), the effect differs. The suction on distribution on the pressure surface experiences little variation as
the main element increases signiŽ cantly as the height is reduced, the height is reduced. At the lower heights (Fig. 8b), for all heights
especially in the chordwise region from x=c D 0:08 and down- apart from the lowest of h=c D 0:053, the lower surface pressures
stream. The increment starts to reduce from x=c ¼ 0:25 and is small on the main element reduce, most signiŽ cantly over the region from
at the trailing edge of the main element. For the lowest height, x=c D 0:1 to 0.25. The suction peak close to this chordwise position
h=c D 0:053, a reduction in suction is found from x=c D 0:25 com- moves aft, from x=c D 0:079 to 0.105 with the reduction in height.
pared to h=c D 0:079. The tap recording the suction peak moves Again, this suction peak contains the maximum suction over the
from x=c D 0:079 to 0.105 as the height is reduced. The character- entire wing for all heights lower than and including h=c D 0:132. At
istics of the spike near to the leading edge also change. Although greater heightsthan this, the maximum overallsuction is foundin the
not overly clear, at heights lower than and including h=c D 0:132 suction spike at the leading edge. For the  ow over the  ap suction
the maximum suction over the entire surface is found in the suc- surface at all heights apart from the lowest, the suction changes very
tion peak at x=c D 0:105, compared to the leading-edge spike at little as the height is reduced in this height range. There is a slight
heights greater than this. The peak suction at the lowest height cor- tendency for increasing suction in the peak upstream of x=c D 0:63
responds to C P D ¡8:7, much greater than the maximum for the and reducing suction downstream of this, as the height is reduced.
single-element wing. Over the lower surface of the  ap, the change For the lowest height of h=c D 0:053, the suction over the main
in suctionis small comparedto the main element.There is a tendency element lower surface reduces signiŽ cantly over the entire surface.
for the suction to increase near to the leading edge, with the lowest A constant pressure region is observed on the  ap from x=c D 0:7,
height case showing a slight reduction in suction from x=c D 0:65. also with reduced suctions on the  ap upstream of this. For both the
The distributionsover the pressure surface of both elements changes main element and the  ap, the pressure reduces very slightly over
very little, but the slight effect of reducing pressure with height can the upper surface with a reductionin height. An additionalreduction
be seen. is observed at the lowest height.
 1012 ZHANG AND ZERIHAN
Downloaded by UNIVERSITY OF LEEDS on July 28, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.2057

a) Low  ap angle b) High  ap angle

Fig. 9 The u/U1 contours.

a) Streamwise velocity b) Normal stress

Fig. 10 Wake surveys for low  ap angle at h/c = 0.211.

Off-Surface LDA Wake Survey show that the  ap wake would mix with the wake from the main
An LDA wake survey was taken at four streamwise locations of element, such that at x=c D 1:776 for h=c D 0:211; no sharp dis-
x=c D 1:066; 1:184; 1:381; and 1:776 behind the trailing edge of the continuity in the proŽ le exists, indicating that the wakes are fully
 ap. Boundary-layer surveys were taken along a line perpendicular merged at this location. At x=c D 1:066 the  ow velocity increases
to the suction surface, at the trailing edge of the  ap (see Fig. 2). from the wake to the ground. This vertical pressure gradient is as
Results were acquired at heights of h=c D 0:395; 0:211; and 0.105 would be expected from vertical traverses not perpendicular to the
for the low and the high  ap angles. curved surface of the  ap. Close to the trailing edge, the  ow is at
A complex turbulent wake  ow is produced by the wing (Figs. 9– a higher velocity than freestream in the region between the wing
11). Immediately downstream of the low angle  ap, two minima in and the ground. In the region from the wake to the ground, the
u were found (Figs. 9a and 10a); the more signiŽ cant in terms of adverse pressure gradient can be seen to reduce the velocities in
velocity deŽ cit and thickness appears to be caused by the main ele- the streamwise direction. The Ž nal point, at x=c D 1:776, shows
ment. As the wake developsdownstream, turbulentmixing increases a relatively constant velocity proŽ le in this region. The boundary
the size of the wake (Fig. 10b), as was found for the single-element layer very close to the ground can be seen and is more prominent
wing,5 and the maximum velocity deŽ cit reduces. The wake surveys than the single-elementwing. For h=c D 0:211 at the low  ap angle,
 ZHANG AND ZERIHAN 1013

a) Different heights at low  ap angle b) Effect of  ap angle at h//c = 0.211

Fig. 11 Wake surveys at x//c = 1.066.
Downloaded by UNIVERSITY OF LEEDS on July 28, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.2057

a) Tangential velocity b) Normal stress

Fig. 12 Boundary layer at  ap trailing edge.

the velocity deŽ cit appears similar as the  ow moves downstream, more signiŽ cant as the height is reduced, as the layer becomes
with a minimum velocity of u=U1 ¼ 0:95. The layer does, however, thicker, and possibly becomes more extreme in terms of the velocity
appear to have grown in thickness. deŽ cit.
2
The turbulent stress u 0 u 0 =U 1 distribution (Fig. 10b) shows two Both the velocity deŽ cit and the wake thicknessare greater for the
distinct peaks at x=c D 1:066, representing the wakes from the high  ap angle, not only because of the main element, but also from
 ap and the main element. Here, the contribution from the  ap is the  ap (Figs. 9a and 11b). From the wake surveys it can be seen
more signiŽ cant than that of the main element. At the next point, that, for the high  ap angle, the wakes are further from merging than
x=c D 1:184, the perturbationsfrom the main element have reduced for the low  ap angle. The  ow between the wing and the ground
only slightly in the portion of the wake from the main element but is accelerated to a greater extent for the high  ap angle for the
signiŽ cantlyin the portionfrom the  ap. At x=c D 1:381 the  ap con- results near to the wing. For h=c D 0:105 the contours show that the
tribution falls again and is more like a plateau, until at x=c D 1:776 velocity deŽ cit very close to the ground is greater for the high  ap
the results show advanced merging, with only a small bump. High angle, both in terms of the velocity and the thickness. At the greater
levels of the perturbationvelocity,which decrease signiŽ cantly with heights it is clear that the layer thickensas it moves downstream, but
distance downstream, can be seen in the boundary layer close to the it is difŽ cult to compare the velocities directly. The general effect
ground. of changing the height on the  owŽ eld and the development of the
As the height of the wing is reduced (Fig. 11a), the wake from the wake downstream is similar to those found for the low  ap angle.
wing increases as was found for the single-element wing. However, Boundary-layerproŽ les (Fig. 12a) conŽ rm that at the trailingedge
the portion of the wake from the  ap does not change signiŽ cantly, of the  ap the wake from the main element is indeed separated from
in terms of the velocity and the thickness. The portion from the the boundary layer of the  ap and that the merging of the layers, if
main element is the cause of the increase; the wake thickens, and any has happened,is small at this streamwise location.As the height
the velocities reduce as the height is reduced. In a similar manner is varied for the low  ap angle, the boundary layer directly from the
to the single-elementwing, it is the portion from the suction surface  ap changes little, and the results are within the positional accuracy
(of the main element) that changes, and the pressure surface con- of the equipment used. The velocity at the con uence between the
tribution does not vary signiŽ cantly. Similar values for the velocity layers is u » =U 1 ¼ 1:06 for the three heights for the low  ap angle.
are found in the region of accelerated  uid between the wake and The minimum velocity caused by the wake from the main element
the ground. The retarded  ow very close to the ground becomes reduces as the height reduces, and the location of this moves farther
 1014 ZHANG AND ZERIHAN

a) b)
Downloaded by UNIVERSITY OF LEEDS on July 28, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.2057

c) d)
Fig. 13 Off-surface PIV survey: u/U1 contours; a) h/c = 0.058, high  ap angle, b) h/c = 0.132, high  ap angle, c) h/c = 0.058, low  ap angle, and
d) h/c = 0.132, low  ap angle.

away from the surface, as the wake thickness increases. The effect glass end plate. Figure 13 gives mean streamwise velocity contours
of the overall boundary-layer thickness increasing as the ground is at two typical heights: one in the force reduction region and another
approached is caused by the contribution from the main element. in the enhancement region.
For the high  ap angle at h=c D 0:395, the boundary layer directly The PIV surveys generally conŽ rm the observations made with
from the  ap is thicker.In addition,the merging of the wake from the the LDA surveys with additional new information. The wake from
main element with the  ap boundary layer is less developed than the main element is seen to dominate, and its thickness increases as
for the results at the low  ap angle. The main element boundary the ground heightis reduced,as observedin the LDA measurements.
layer is both thicker and more signiŽ cant in terms of the velocities, The  ow accelerates between the ground and the main element
compared to the low  ap angle. to reach a speed higher than the freestream before the maximum
2
The turbulent stress u 0 u 0 =U1 is shown in Fig. 12b for the suction point (not shown in the PIV images); hence the high velocity
boundary-layer proŽ les. For the low  ap angle concentrations in region between the groundand the wing. After the maximum suction
2
u 0 u 0 =U1 are found in the region close to the  ap surface. The high- the wing effectively forms a diffuser section with the ground, and
est values are found at values of ³ less than 0:002c from the surface, the  ow decelerates downstream. For the majority of test cases, the
and this decreases to minima at ³ ¼ 0:015c, a height near the merg-  ow stays attached at the trailing edge of the main element, unlike
ing of the main-element wake and the  ap boundary layer. In the the single-element case.3 The Kutta condition at the trailing edge
2
region of the main-elementwake, u 0 u 0 =U1 increasesto a maximum. is satisŽ ed, and the wake off the trailing edge of the main element
For all three heights this maximum is at a location farther away than follows an upward path (Fig. 13b). Nevertheless,the wakes from the
the center of the main-element wake, and the magnitude increases main element and the  ap do not merge immediately after the  ap,
as the height is reduced. Some evidence of a second peak of pertur- and the main-element wake does not touch the  ap for all cases.
bations exists at a location closer than the center of the boundary For the high  ap angle, at the low ground heights of h=c D 0:058
layer, but this is not well deŽ ned. The curve then drops as the edge of and 0.066, a different  ow pattern emerges. The  ow on the suc-
the boundary layer is approached. For the high  ap angle the peak tion surface of the main element is separated at the trailing edge,
within the  ap boundary layer is of a greater magnitude than the although the extent of the separation is small. (See Fig. 8b and note
low  ap angle at h=c D 0:395. The perturbations then drop sharply that the separation is not shown in the PIV images because of re-
to practically zero, conŽ rming that this is out of, or very close to, the  ection.) Nevertheless, with the appearance of the separation, the
edge of the main-element wake. The curve increases to a Ž rst small Kutta condition is not satisŽ ed at the trailing edge. As a result, the
peak, on the inner side of the center of the wake, drops slightly, and lower bound of the wake from the main element now experiences
increases again to a second peak, farther out from the center of the a downward trend. In fact, a wall-jet-like  ow now exists between
wake, then falling off to the edge of the boundary layer. the ground and the wing. Inspection of the mean velocity distribu-
tion immediately points to the existence of a minimum point in the
Off-Surface PIV Survey streamwise velocity in the wake from the main element. The po-
Although the LDA survey provided both qualitative and quanti- sition of the minimum, though, differs for the low ground heights
tative data of the wake development,it does not extend to the region and high ground heights. At h=c D 0:058 and 0.066 the minimum
underneath the  ap and above the ground. Features in the region is located immediately after the main element and beneath the  ap,
would provide insight into the force reduction at the low ground whereas for the higher ground heights for the high  ap angle and
height. PIV surveys were therefore performed in the region between for the low  ap angle cases the minimum is located after the  ap
the  ap and the ground. The survey was made feasible by the use of (see Table 1).
 ZHANG AND ZERIHAN 1015

Table 1 Measured minimum in streamwise velocity in  ap circulation, reducing the main-element circulation. The PIV
in the main-element wake measurements suggest that the Ž rst is the case. The existence of
High  ap angle Low  ap angle the trailing-edge separation leads to a downward movement of the
lower bound of the wake from the main element and the presence
h=c u=U1 x=c y=c u=U1 x=c y=c of a velocity minimum in the wake immediately after the main ele-
0.058 0.04 0.80 0.06 0.30 1.03 0.12 ment and beneath the  ap. This creates an effective diffuser section
0.066 0.16 0.82 0.08 0.37 1.04 0.13 between the wake and the  ap and places a higher pressure recovery
0.071 0.04 1.06 0.17 0.40 1.06 0.13 demand on the  ap suction surface  ow, leading to separation and
0.079 0.08 1.07 0.18 0.44 1.08 0.14 downforce loss.
0.084 0.14 1.08 0.19 0.46 1.09 0.14
0.092 0.20 1.09 0.20 0.47 1.07 0.14
0.097 0.25 1.08 0.19 0.48 1.10 0.15
Summary
0.105 0.31 1.11 0.21 0.49 1.09 0.15 The aerodynamic behavior of a cambered, double-element,high-
0.118 0.37 1.12 0.21 0.51 1.11 0.16 lift wing has been studied using model tests. Techniques em-
0.132 0.41 1.10 0.21 0.53 1.12 0.16 ployed include force balance, surface oil  ow visualization,surface-
0.158 0.44 1.12 0.22 0.55 1.11 0.16 pressuretaps, off-surfaceLDA surveys, and PIV measurements.The
0.184 0.47 1.14 0.23 0.64 1.12 0.16
effects of ground proximity and  ap angle de ection are quantiŽ ed.
0.211 0.48 1.13 0.23 0.64 1.12 0.17
It was found that the main element produces most of the down-
force and dominates the turbulent wake development. The ground
proximity does not alter the turbulent wake from the  ap signiŽ -
Further Discussion cantly, in terms of the velocity deŽ cit and the thickness.An increase
Adding the  ap to the main element induces a greater circulation in the wake thicknessand a reductionin the velocitydeŽ cit are found
Downloaded by UNIVERSITY OF LEEDS on July 28, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.2057

around the main element, as can be seen by the greater suction on in the portion from the suction surface of the main element, as the
the lower surface and the increased pressures on the upper surface. wing is moved to the ground. Both the velocity deŽ cit and the wake
The fact that there is a Ž nite suction at the trailing edge of the main thickness are greater for the high  ap angle, not only because of
element implies that the pressure recovery from the suction peak on the main element, but also from the  ap. In the center of the wing,
the main element to the trailing edge is not as severe. Large regions the  ow can be regarded as quasi-two-dimensional. Three regions
of separated ow were observed for the single-elementwing, but for are identiŽ ed on the downforce with height curve. A force reduc-
the double-element wing separation is not as widespread. The two tion region c, similar to that of a single element wing, is presented
effects, the circulationeffect and the dumping effect,were described for the two  ap settings. Above the force reduction region c, there
by Smith7 as contributing factors to the beneŽ t of a multielement are two distinct regions. At large heights, region a, the downforce
conŽ guration. increases asymptotically with a reduction in height. Then there is
For the double-element wing a force reduction region c is iden- either a small plateau, in the case of the low  ap angle, or a reduc-
tiŽ ed, similar to the single-element wing.3 Above the region c two tion in downforce, in the case of the large  ap angle. The downforce
regions of force behavior are identiŽ ed: regions a and b. Above then increases again, region b, until it reaches a maximum, and then
region a heights, at and above h=c D 0:171 for the low  ap angle reduces. In the case of the low  ap angle, the maximum downforce
and at and above h=c D 0:237 for the high  ap angle feature an is dictated by gains in downforce from lower surface suction in-
increasing downforce with an increasing proximity to the ground. creases and losses in downforce caused by upper surface-pressure
The slope increases as the height reduces. Below this, at region b losses and lower surface suction losses, with a reduction in height.
heights, the curve also increases with reducing ground height until For the high  ap angle there is a sharp reduction just beyond the
the maximum downforce is obtained. For the low  ap angle there is maximum because of the boundary layer separating and a resultant
an asymptotic increase and then a smooth reduction in the gradient loss of circulation on the main element.
to the maximum downforce, followed by a reduction in downforce.
For the high  ap angle the curve is more linear, and there is a sharp Acknowledgments
reductionbelow the maximum downforce.At the boundarybetween
J. Zerihan is supported by an Engineering and Physical Sciences
the type a=b  ows, there is a discontinuityin slope in the downforce
Research Council studentship. The authors thank BAR for support
curve. For the low  ap angle this is manifested as a plateau region
and W. Toet for discussions during the course of the study and
at the large type b heights. For the high  ap angle there is a sudden
J. McManus and S. Mahon for performing the PIV survey.
reduction in downforce, compared to the lowest type a heights.
At the center of the wing, the contribution from the  ap at the
low  ap angle increases by about 30% as the height is reduced from References
1
h=c D 0:592, until a maximum at h=c D 0:105, and then reduces a lit- Agathangelou, B., and Gascoyne, M., “Aerodynamic Considerations of
tle below this. Similar results are foundfor the high  ap angle,where a Formula 1 Racing Car,” Society of Automotive Engineers, Publ. 980399,
the height at the maximum sectional downforce is h=c D 0:158. The Feb. 1998.
2 Jeffrey, D., and Alperin, M., “Aspects of the Aerodynamics of Year 2000
pressure distributions suggest that the small reductions in down-
Formula One Racing Cars,” 3rd MIRA International Vehicle Aerodynamics
force are caused by reductions in pressure on the pressure surface
Conference, MIRA, 2000.
and small reductions in suction in the region x=c D 0:65 ¡ 0:8 on 3
Zerihan, J., and Zhang, X., “Aerodynamics of a Single-Element Wing-
the  ap suction surface. The general increase in downforce as the in-Ground Effect,” Journal of Aircraft, Vol. 37, No. 6, 2000, pp. 1058–1064.
ground is approached for the  ap is signiŽ cantly smaller than that 4
Knowles, K., Donahue, D. T., and Finnis, M. V., “A Study of Wings in
for main element. As the height of the wing is changed, the  ap Ground Effect,” Loughborough University Conference on Vehicle Aerody-
is farther from the ground and therefore less sensitive to changes namics, Royal Aeronautical Society, London, 1994, pp. 22.1–22.13.
in ground height than the main element. There is a more signiŽ - 5 Zerihan, J., and Zhang, X., “Unsteady Turbulent Wake Behind a Single-

cant reduction in downforce at the lowest height for the high  ap Element Wing in Ground Effect,” 10th International Symposium on Ap-
angle, which is caused by the boundary layer separating over the plications of Laser Techniques to Fluid Mechanics, Lisbon, July 2000,
 ap. The lower circulation imparted by the  ap on the main element Paper 8-1.
6 Zerihan, J., and Zhang, X., “A Single Element Wing in Ground Effect:
can be seen represented by the lower suction on the main element
Comparisons of Experiments and Computations,” AIAA Paper 2001-0423,
suction surface for the lowest height. According to LDA measure- Jan. 2001.
ments, two hypotheses are presented. First, the boundary layer can 7 Smith, A. M. O., “High-Lift Aerodynamics,” Journal of Aircraft,
separate from the main-elementsuction surface for the lowest height Vol. 12, No. 6, 1975, pp. 501–530.
with the high  ap angle, preventing the wake from  owing over the 8 Ranzenbach, R., Barlow, J. B., and Diaz, R. H., “Multi-Element Airfoil in
 ap, leading to the  ap boundary layer separating and a lower  ap Ground Effect—An Experimental and Computational Study,” AIAA Paper
loading. However, the  ap  ow can separate itself, causing a loss 97-2238, June 1997.
 1016 ZHANG AND ZERIHAN

9 Jasinski, W. J., and Selig, M. S., “Experimental Study of Open-Wheel pp. 344–354.
Race-Car Front Wings,” Society of Automotive Engineers, Publ. 983042, 13 Benedict, L. H., and Gould, R. D., “Towards Better Uncertainty Es-

Nov. 1998. timates for Turbulence Statistics,” Experiments in Fluids, Vol. 22, No. 2,
10 Zerihan, J., “An Investigation into the Aerodynamics of Wings in 1996, pp. 129–136.
14
Ground Effect,” Ph.D. Dissertation, Aeronautics and Astronautics, School of Zhang, X., Zerihan, J., Ruhrmann, A., and Deviese, M., “Tip Vor-
Engineering Sciences, Univ. of Southampton, Southampton, England, U.K., tices Generated by a Single-Element Wing in Ground Effect,” 11th Interna-
April 2001. tional Symposium on Applications of Laser Techniques to Fluid Mechanics,
11 Moffat, R. J., “Contributions to the Theory of Single-Sample Un- Lisbon, July 2002, Paper 4-1.
certainty Analysis,” Journal of Fluids Engineering, Vol. 104, June 1982,
pp. 250–260.
12 Zhang, X., “An Inclined Rectangular Jet in a Turbulent Bound- A. Plotkin
ary Layer-Vortex Flow,” Experiments in Fluids, Vol. 28, 2000, Associate Editor
Downloaded by UNIVERSITY OF LEEDS on July 28, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.2057