Você está na página 1de 12

Downloaded from SAE International by University of Michigan, Sunday, August 12, 2018

400 COMMONWEALTH DRIVE WARRENDALE, PA 15096

860218

Aerodynamic Model
for Wing..Generated Down
Force on Open..Wheel..Racing..Car
Configurations
Joseph Katz
NASA Ames Research Center
Moffett Field, CA

Reprinted from SP-656-


Aerodynamics: Recent Developments

International Congress and Exposition


Detroit, Michigan
February 24-28, 1986
Downloaded from SAE International by University of Michigan, Sunday, August 12, 2018

No part of this publication may be reproduced in any form,


in an electronic retrieval system or otherwise, without the
prior written permission of the publisher.

ISSN 0148-7191
Copyright 1986 Society of Automotive Engineers, Inc.

This paper is subject to revision. Statements and opinions Persons wishing to submit papers to be considered for
advanced in papers or discussion are the author's and are presentation or publication through SAE should send the
his responsibility, not SAE's; however, the paper has been manuscript or a 300 word abstract of a proposed manu-
edited by SAE for uniform styling and format. Discussion scriptto: Secretary, Engineering Activity Board, SAE.
will be printed with the paper if it is published in SAE
Transactions. For permission to publish this paper in full or 12 page booklet. Printed in U.S.A.
in part, contact the SAE Publication Division.
Downloaded from SAE International by University of Michigan, Sunday, August 12, 2018

860218

Aerodynamic Model for Wing=Generated Down


Force on Open=Wheel=Racing=Car Configurations
Joseph Katz
NASA Ames Research Center
Moffett Field, CA

erties of non separated regions in subsonic or supersonic


ABSTRACT flows (4,7,8). An additional advantage of the potential
flow problem emerges from the use of Green's theorem
A simplified panel model was constructed for de- (9). An integral equation is obtained which is solved
termining wing-generated· down force for open-wheel- only on the body's boundaries, rather than over a com-
racing-car configurations. The model required simula- plex grid spanning the whole fluid volume. Detailed
tion of the separated wakes emanating from the vehi- description of the mathematical principles and of the
cle's body and wheels. Separation line locations were numerical schemes used in these models are provided in
assumed to be known from experiments or observation References (7-12). These methods have been extended
and, for simplicity, were fixed throughout all computa- to include higher-order singularity elements (4,12), iter-
tions. Once the model for the vehicle body and wheels ation with viscous boundary layer solutions, wake-rollup
was established, inverted lifting surfaces were added to relaxations, and jet models (13) in order to cope with a
generate down force. The performance of these inverted variety of aircraft-related problems.
lifting devices was then numerically investigated and
For automotive purposes, where the estimation
compared with experimental data of a generic, Formula
of aerodynamic drag is one of the major tasks, fluid-
one, race car shape.
dynamic-computation methods (14-18) are still very lim-
ited. However, in the case of open-wheel- racing-car
INTRODUCTION configurations, the correct estimation of negative lift is
also important, since increased down force (due to aero-
dynamic devices) can considerably improve the perfor-
THE FLOW FIELD resulting from the high-speed mance of these vehicles (19-25). In spite of the complex
motion of ground vehicles is complex and time-dependent flow field about these racing car configurations, the pre-
primarily because of local flow separation, vortex flows. diction of their lifting properties by linearized methods
boundary layers, and their interaction with the ground. is a simpler task than an accurate drag computation (of
Analytical prediction methods for such flows, as well as both skin friction and separated flow elfects).
for flows about complete aircraft configurations are still
very limited (1)'. Consequently, aerodynamic perfor- In this present work, a potential-flow (panel) method
mance predictions for road vehicles are based mainly is used together with vortex wake models to simulate the
on experimental data. Because of the complexity of shear layers emanating from the separation lines, whose
the flow, any attempt to solve the detailed, nonlin- locations are assumed to be known from experiments.
ear, fluid dynamic equations (Navier Stokes) requires Once the separated flow field about the car body and
an elaborate and time-consuming computer code (com- wheels is adequately modeled, computation of the nega-
pared with linearized methods). On the other hand, tive lifting forces produced by the lifting surfaces is per-
during the past twenty years, considerable experience formed. The influence of several geometrical parameters
has been gained in solving the linear portion of these
equations (2-12). These solutions predict the fluid dy- * Numbers in parentheses designate references at
namics of inviscid flows which include the lifting prop- end of paper,

129
Downloaded from SAE International by University of Michigan, Sunday, August 12, 2018

on the aerodynamic performance of these inverted lift- sharp leading edges of these wings (30). Also, computed
ing devices is then investigated, utilizing a generic open- forces and wake shapes were in good agreement with ex-
wheel-racing-car shape which had been tested both in perimental results.
small and in full-scales.
An example of a computation with this modified
NUMERICAL METHOD vortex-rollup routine (which was exchanged with the
corresponding subroutines of VSAERO (5)) is presented
The numerical method used here is based on the in Fig. 1. Here the case of a two-dimensional flat plate
potential-flow, panel code VSAERO (5,13,18) (limited at an angle of attack of 60 degrees is shown, and de-
to 1000 surface panels), with a modifled wake-roll up tails of the time-dependent wake shedding and drag co-
routine. The principles of this vortex tracing method for efficient CD are presented. The computed drag force
simulating complex, unsteady wake deformations and flactuates about the value CD "" 1.2, which is close to
the resulting beneflts are discussed in Refs. 26 and 27. the value of time-averaged measurements. More details
Basically, in this vortex wake model, vortex elements are about this simplified example are presented in Ref. (28).
shed from specified lines (e.g., trailing edges of wings or Another two-dimensional computation, with the current
separation lines) and the displacement (t.x,t.y,t.z) of computer code and with simulation of flow separation,
each of these vortices ( )i during the time interval t.t in is presented in Fig. 2. The experimental data for the
the fluid is determined by the local velocity (u, v, w) : automobile shape (AUTO 2000), shown in the insert to
this flgure, was obtained with a two-dimensional model
in an open-jet wind tunnel and was reported by Buch-
heim and Rohe (31). Separation lines for the panel

( ~~) ,
t.z .
(:) i . t.t (Jor each vortex element)
-2.0
- - EXPERIMENT, REF. (311
PRESENT COMPUTATION

By applying this routine in a time-dependent man- -1.5 RANGE OF


ner, the two-dimensional shedding of separated wakes COMPUTED
behind stalled wings (28,29) was modeled (as in Fig. -1.0 PRESSURES
1). Application of this method for three-dimensional AT BASE
:;;
c. UPPER SURFACE ~ PANELS
flows about delta wings at high angles of attack showed ,c. -.5
c. • \ ..:
good modeling capability of the vortex roll up near the U

zle CD .5 -1.0 -
1.8
LOWER SURFACE -.5
1.6 •
1.4 "
m
0 3
0

4
1.2
1.0
. AUTO 2000 WAKE
.5
u
Ii:

.8
tM\~",~w
o L
•x
2 .6 o .25 .50 .75 1.0
x IL

Fig. 2. Comparison of computed and measured


pressures over a two-dimensional automotive shape

-4 L---'_--'_-'-_-'-_-'-_-'-_-'---l
o 2 4 6 8 10 12 14 16 model (which consisted of 77 peripheral panels) were
tV=/e assumed to be at the sharp corners of the car rear sec-
tion. The comparison between the experimental and
computed pressure coefficients C p shows good agree~
Fig. 1. Simulation of wake shape and drag force
ment over most of the surface. At the rear window area,
history behind a stalled flat plate
however, the kink in the experimental data indicates a

130
Downloaded from SAE International by University of Michigan, Sunday, August 12, 2018

more forward separation line than assumed in the panel wake-shedding panels on the body and wheels, created
model 1 and in this area the difference between the two by the difference in doublet strength of each adjacent
C p values is larger. This deviation can be corrected by panel. The hidden lines were not removed from this fig-
an additional iteration with a new assumption for the ure since the graphic subroutine for this operation was
location of the the separation line. In this case, how- not able to distinguish between body and wake panels.
ever, only the first iteration is presented to demonstrate Even with this sparse detailing computed wake loca-
1

the logic of the interaction between a designer and the tions are close to expe,rimental observations and com-
computation. Computed surface pressures 1 on the pan- puted forces (CD and C L '" 0.25), for the body only,
els at the back of the vehicle, were in the range of C" are close to values obtained \\'ith smooth automotive
= -0.50 to +0.25. This yields to an average base pres- shapes. This positive lift, for the body near the ground,
sure which is close to the measured experimental base was obtained with the computer code VSAERO (5) af-
pressures of about C p '" -0.3 shown in Fig. 2. ter the inclination of the base area (and thus the body's
circulation) was set, as shown in Figs. 3,4, and 5.
PANEL MODEL OF AN OPEN-WHEEL-RACING-CAR
CONFIGURATION

The paneled geometry of an open-wheel-racing- 1+-------4.05-------1


car configuration is shown in Fig. 3, and its general
dimensions are presented in Fig. 4. The total number

0.60
I~.---2.60---__+f1-_..j

Hi=iq:l1~
itl:Jo: 0.32

II
Fig. 3. Panel model of an open wheel race car II
(576 panels per one side)
QU
W
QM
of panels used to model one side of the model (sym-
REFERENCE CROSS SECTION AREA = 1.47 m 2
metry is assumed) was 576. This body shape has a
smooth planar underbody to comply with current For- Fig. 4. Geometrical details of the paneled model
mula 1 regulations and is similar to the generic shape
that was tested both in full and in one-tenth-scale (20).
The panel-code version used is limited to 1000 panels
and a maximum wake grid of 50 x 30 elements. This a. Wheels
latter limit for the wake geometry was the more critical,
and precluded a denser surface paneling. The separation
line for the vehicle's body was assumed to be at the rear One of the hardest problems in predicting the aero-
sharp edges; for the wheels it was along the backward- dynamic performance of open-wheel-racing-car configu-
facing panels; and for the wings it was along their trall- rations, either experimentally or numerically, relates to
ing edges. Fig. 5 shows the wake lines for the vehicle the geometric condition of the wheels relative to the
without the lifting surfaces. These lines represent the ground. An elevated, stationary wheel above ground or
cores of the "spanwise" vortices, emanating from the a rotating one on the ground will produce less lift than

131
Downloaded from SAE International by University of Michigan, Sunday, August 12, 2018

partially because the wakes of tbe model and its mirror


image (under the ground) crossed each other's ground
~ plane. Furthermore, since the total number of model
~:::~::- • and wake. panels was limited, only the simplified model
<' / ;/',c-~5~;.... shown (wIth sparse panelmg) could be attempted. Con·
·fJ.i1i)C!J¥f?!> ~// 7'~;';~~;~'
", ~/·.?~~~P::"·/I " sequently, the inviscid solution resulted
//~/~ ~// / D " / '
in very
/~ • lVI' high velocities (in the zero gap) between the lower pan-
a /~~-;:z: ~/- ~~F ;.~::~-:~~~:/,' eis and the ground plane, causing very low pressures un-
....-;
e2n~<.J> ~
:;g /~,(;t~ ~
.;;:4; .
. . ,/ 1.,;. ~f:S""'PI
.' .
..-c~ •• ,
·~17.
der the whee!. A solution that was made was to slightly
.
/ .... ).7P(~-;:'_·:" :x:p,.J} raIse the wheels above the ground plane and to sImu-
~;~/:' ~l'")~~ . ".'<d} late similar conditions (raised-wheel) in the wind tun-
,.//....~ <'i!?;'~~ ne!. Computed pressures at the wheel mIdplane are pre-
~;,,<; .~~/.
/..hf;lf.I. j'~p/ sented in Fig. 7 (the symbols represent computed pres-
:;i 0/
/ • sures at the panel's centroids, connected by estimated
lines). Because of the low number of circumferencial
panels the resolution is coarse, but the lower pressures
on the lower surface due to the gap beneath the wheel
Fig. 5. Longitudinal vortex wake lines behind can be clearly observed. Measured pressures over an
vehicle's body and wheels open wheel by Fackrell and Harvey (32) do not show
these low pressures at the lower surface, since in their
a stationary wheel contacting the ground. Therefore, in experimental setup the wheels were in contact with the
order to predict (numerically or by wind tunnel exper- ground. Their measured upper-surface pressures, how-
iments) actual road/track conditions, different correc- ever, are in better agreement with those of Fig. 7. Their
tions for each of these geometrical arrangements have measured base pressure of about C p = ~O.4 is within
to be developed. range with the results of this coarse model ( C p = -1.1
to +0.3 in Fig. 7). This computation can be refined
by adding more panels into this region, similarily to the
model of Fig. 2. As a result, computed forces for this

COMPUTED
-4 PANEL
PRESSURE
A - - UPPER SURFACE
0 - - LOWER SURFACE
-3 A
/ \
/ \
/ \
/
./ \
-2 / \
/ \
Cp
~. \
~
I
-1 t
Fig. 6. 99 panel model of a wheel and correspond-
ing wake lines

The panel model used here to simulate the flow


field about the vehicle wheels is presented in Fig. 6.
0
-- ®V=

6"",,,,m, t
.!'1 ..
x
h/2R = 1120 %
The flow separations were modeled by enclosing the rear
section of the non-rotating wheel with wake panels as 1~ .25 .50 .75 1.00
shown by the figure. To partially account for wheel ro- x/2R
tation, the upper separation line was assumed to begin
slightly (5 deg.) ahead of the wheel's highest point. Fig. 7. Computed pressure distribution at the
Test cases that were run, using this wheel model placed wheel's midplane (with 11 circumferential panels only
on the ground plane, exhibited numerical instabilities, and [width/2R] = 0.69)

132
Downloaded from SAE International by University of Michigan, Sunday, August 12, 2018

case were CD"" 0.57, and C L "" -0.15. for aground clear-
ance of h/2R= I /20 (with wheel dimensions of R=0.32m
<> FULL·SCALE EXPERIMENT (20)
and width of 0.44m). This value of the drag coefficient
is within the range measured by Cogotti (33) and by
o SMALL·SCALE EXPERIMENT (20)

Fackrell and Harvey (32) in a wind tunnel with rotating COMPUTATION


wheels. The magnitude of the measured lift coefficient
for isolated wheels varies between C L = -0.17 to +0.76
(32,33), depending on test geometry (rotating or not) .6
and Reynolds number. A similar low lift coefficient was
also found during the small-scale tests of Ref. (20),
when the model wheels were raised above the ground .5
plane and negative lift forces equivalent to about C L =
-0.15 for a single wheel were detected (based on wheel
cross-section area). It is assumed, therefore, that for the .4
present study the above panel model reasonably sim-
ulates stationary and raised wheels above the ground
plane. Once the approximate flow field about the open .3
wheels is established, the effect of adding lifting surfaces
to the vehicle's body can be investigated. For compar-
ison with wind tunnel data, therefore, a correction to .2
the computed lift force has to be made to account for
the lower lift of the four wheels. The additional con-
tribution of the four wheels adds up to about .6.C L = .1
+0.10, and this amount is to be added to the results
obtained with the computation when predicting track
performance.
o -10 -20 -30
"'F' deg
b. Front Wing

Fig. 8. Vehicle's lift and drag increments due to


The front wing is probably the most effective aero- front wing angle-of-attack, and the shape of its airfoil
dynamic device for producing negative lift on racing cars section
(21). This is so because the air ahead of the vehicle is
undisturbed at this station. The effectiveness of the
front wing in producing down force is shown in Fig. 8 y
where the lift .6.C L and drag .6.CD increment due to -2.0
front-wing angle of attack aF are presented (all coeffi-
cients are based on the vehicle's reference area, given
in Fig. 4, except when noted otherwise). Experimen- -1.6
tal data to validate this computation was available only
at aF = -17 deg. (20) and, as can be seen in Fig. 8,
-1.2
it compares well with the predictions. The small dif-
ference between experiment and computation is due to -c~
camber effect, as a result of the slightly different air- -.8
foil sections used. In the full-scale test the wing section
had a slotted flap, whereas for the small-scale test the
slot was covered. For the computation, a NLF(I)-0414 -.4
airfoil section with flap setting of 10 deg. (as shown in
the insert in Fig. 8) was used since its zero-lift angle 0
of attack is close to the values obtained in the full- and .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
2y/b F
small-scale experiments.
The spanwise loading C/ (based on the wing's lo-
cal chord) of the front wing is shown in Fig. 9 for three
different angles of attack. Because of the aft sweep of Fig. 9. Spanwise load distribution of the front
the leading edge, the loading increases toward the outer wing (b F is front wing's span)

133
Downloaded from SAE International by University of Michigan, Sunday, August 12, 2018

sections of the wing. The measured chord wise pressure


-- h/e=~ } WITH
distribution about this airfoil seclion (Fig. 8) is very -- hie = 0.25 SIDE FINS
flat, and high two-dimensional lift coefficients are ob- 2.0
-- h/e=~ ) NOSIDE
tainable at typical racing speeds (section lift coefficient __ hie = 0.25 FINS
in excess of 1.7 was measured for angles of attack of 18
deg. at higher Reynolds numbers of about 8 x lOG).
The absolute spanwise lift coefficient values (based on
local chord), shown in Fig. 9 for "'F = -30 deg. are
slightly smaller than this two-dimensional value. There-
fore, based on these two figures, it is concluded that be-
1.5

- ----....-
----
....-
..........- ----
fore wing stall occurs L::.C L of about -0.4 (again, based -c£ 1.0
on vehicle's reference area) can be obtained with this
front wing.

c. Rear Wing .5

M = 1.56
The geometric limitations on the rear wing, forced (XR = _20"
by current open-wheel-race-car regulations, restrict both
its width and its height above ground to 1 m. This re-
o .2 .4 .6 .8 1.0
quirement, together with the need to shift the down 2y/b R
force toward the rear I driving wheels, forces this wing
into a location where unfavorable interference with both
Fig. 11. Spanwise load distribution of the rear
wing (b R is rear wing's span)

wheels and fuselage might be expected. However, exper-


imental evidence (20), as well as current computations
(Fig. 10), show that a narrow and low vehicle body can
minimize these effects. The data presented in Fig. 10,
.7 similar to Fig. 8, is normalised by the vehicle's reference
area, and therefore the lift and drag data (L::.CL, L::.C D)
are presented as the rear-wing contribution to the ve-
.6 hicle's coefficients versus its angle of attack "'R . The
addition of side fins, as shown in Fig. 10 and 11, can
further improve its performance and also its drag/lift ra-
.5
tio. Part of the success in achieving these high lift coeffi-
cients, at the rear of the vehicll!', can be atributed to the
.4 slotted airfoil sections used. The paneled airfoil model
is based on the slotted section shown in Fig. 10, which
was found to be highly effective with small-aspect-ratio
.3 wing planforms (34). In this paneled model, for sim-
plicity, only the outer contours of the complete section
were modeled (as shown by the dotted line in the insert
.2
in Fig. 10). This has no major effect on the calculated
inviscid results for the nonstalled wing since the slotted
.1 flap design was devised to delay flow separations. The
"'F = 14°
spanwise loading (C I - local lift coefficient, based on lo-
AR = 1.56
cal chord, in Fig. 11) for the free-stream case (hie ->
00) is typical of rectangular planforms, and the addition
o -10 -20 -30
of the side fins increases the wing effective aspect ratio,
"'R' deg
resulting in a larger lift-curve-slope (Fig. 10). For the
Fig. 10. Vehicle's lift and drag increments due to actual configuration, however, the body's proximity to
rear wing angle-of-attack, and the shape of its slotted the wing (h/c=0.25) causes a loss of negative lift at the
flap airfoil section wing's centerline, as shown in Fig. 11. The maximum

134
Downloaded from SAE International by University of Michigan, Sunday, August 12, 2018

(absolute value) two-dimensionai-section lift coefficient ponents. The details of the computations show that,
that can be obtained with this airfoil is about 2.2 (with because of the rear-wing's proximitY1 the base pressures
the reference chord defined as in Fig. 10). It is assumed, of the body and of the rear wheels drop and therefore
therefore, that wing stall can be delayed to "'R = -35 their drag increases. This, of course, is a very simplified
deg. which, according to Fig. 10, amounts to a vehicle's model of the separated flow, which does not account
down force increment of about L:;.C L "" -1.00 due to the for the movement of the separation lines as a result of
rear wing. varying the wing's height. But the experimental results
The effect of rear-wing proximity to the vehicle's shown in Fig. 13 do show a larger drag increase between
body and wheels is investigated in Fig. 12. Al1 coef- configurations No. 2 and No. 3 that would have been
ficients are normalized by the vehicle's reference area expected by adding the drag contribution of the rear
and the subscripts ()rOT, ( )R stand for " vehicle" and wing only (based on Fig. 10).
The three configurations shown in Fig. 13 repre-
sent a baseline configuration with inherent positive lift
(No.1), and then two configurations with incremental
additions of front and rear lifting surfaces. Ful1-scale
1.2 data for configuration No. 2 were not available. The
lift coefficient Ci. is the value actual1y measured in the
ful1- and smal1-scale tests. For the prediction, however,
1.0 a correction of L:;.C L = 0.10 was applied (as indicated
in section a) because of the "raised-wheel-above-the-
ground model" that is used here. The prediction, in
.8
DESIGN POINT general, show slightly larger negative lift (with the cor-
rection L:;.C L =O.lO) and slightly high drag (for configu-
.6
--- -------------
.4

.2
"F = _14'
"R = _20'
1.0
o .4 .8 1.2 1.6 2.0 .8
hie
.6
.4
Fig. 12. Effect of rear~wjng proximity to vehicle's .2
body (Design point=actual clearance for "'R=-200)
Cl. a
-.2
"rear-wing-only" contributions, respectively. When the -.4
~ 1/10-SCALE TEST
rear-wing height h is reduced, the absolute value of its -.6
negative lift is slightly decreased as a result of the loss -.8 [&]1/1 TEST (ROTATING WHEELS)
of the wing's section lift coefficient at the centerline re- -1.0
gion, as shown in Fig. 11. The vehicle's total negative [&] COMPUTATION
lift has changed by a similar amount, as indicated by the 1.2 O::F = - 11"
figure. Experiments by Scibor-Rylski (24) who investi- 1.0 O::R = - 24°
gated the effect of rear-wing proximity to the vehicle's .8
body with a one-tenth-scale model did show a larger .6
CD
loss in the down force for similar h values. This might
.4
be explained by the thick (partial1y separated) bound-
ary layer originating at the windshield of their model, .2
producing a wake which further reduced the rear wing's a CONFIGURATION 1 2 3
midspan lift. The drag of the vehicle, however, increases
with reduced height h in spite of the marginal change Fig. 13. Comparison of computed lift and drag
in the rear-wing drag. This indicates an unfavourable forces with experimental values obtained in small-scale
interference of the fear wing with the vehicle's corn- (stationary wheels) and in fuJI-scale tests (rotating wheels)

135
Downloaded from SAE International by University of Michigan, Sunday, August 12, 2018

ration No.3)' when compared with the full-scale tun-


nel tests. Because of the lower Reynolds number of the
small-scale tests (compared to full-scale)' the boundary
layer on the body and on the lifting surfaces is thicker.
The flow, therefore, will separate at smaller (absolute
"') incidences, resulting in higher drag and less lift (for
the larger down force conditions (20)). Cosequently, the
comparison of the computations with the small-scale
R
tests is also reasonable for configurations No. 1 and
No.2. For configuration No.3, the rear-wing flow
separates at "'R =-24 (in the small-scale tests), and [%1
noticably higher drag and less down force were mea-
sured. In conclusion, the lift increments due to the front
and rear wings are well predicted by the computation
(when examining the increments between the configu-
rations and excluding stalled wings). Of course, none of
these prediction techniques (including the wind tunnel
experiments) can accurately simulate actual road con- F
ditions. But, as indicated here, all of these methods
can be very useful in evaluating the effectiveness of the
various negative-lift devices.
For vehicle stability considerations it is desirable
to control the down force distribution between the front o -10 -20 -30
and rear wheels. Figure 14 shows the computed aero- aR' deg
dynamic load distribution, in percent of the total down
force, between the front ( F ) and the rear ( R) axles, for
this particular model. This calculation of the front/rear Fig. 14. Influence of front and rear wing inci-
lift distribution was not a direct output of the com- dences on the aerodynamic down force distribution (in
puter code and was calculated from the individual force percents) between vehicle's front (F) and rear (R) axles
and moment contributions of each component (wings,
wheels, and body). The only experimental data for this
configuration are obtained from Ref. 20 for "'F = - hody have been modeled in this way, the performance
17 deg. and D:R = -24 deg. and are presented by the of nonspparated lifting surfaces can be sat isfactorily cal-
symbols in the figure (which were positioned by the pa- culated. The method is much more suitable for cal-
rameters "'R and [%]). The corresponding calculated culation of the negative lift, produced by the ""ings,
axle loads for the same "'F and "'R are shown by the than for studying the resulting drag force. Therefore,
tips of the small arrows, indicating the difference (in its application for investigating pasgenger~car drag re~
terms of Ll.[% ] or Ll."'F) between the calculation and duction needs extensive further refinemenL both in the
the experiment. Since the dimensions and wing IDea· potential-flow solution and by adding a boundary-layer
tion for many open-wheel-racing-car configurations are solution. ..
close to the geometry of this model, the slopes of these Computed results show that a typical rare-car body,
curves are representative of a variety of vehicles of this with wheels and with flat underbody surface. has a pos-
class. Consequently, by using the slopes of these curves, itive lift of about C L = 0.25. A front wing, with this
a spectrum of front/rear load distributions can be al- present geometry, is capable of producing a negative lift
tered for different vehicles and conditions. increment up to Ll.C L = -0.40, while a rear wing with the
slotted flap section can produce a tlC L of -1.00. This
CONCLUDING REMARKS implies that down force on the order of CL '" -1.15 can
be obtained when both wings are added to the race-car,
and without the ground effect of the body.
The results of this study show that, for the specific
The use of such computational fluid dynamic meth-
purpose intended, a simplified panel model can be con-
ods for preliminary configuration investigation can re-
structed for the complex flow field about open-wheel-
duce design time and the cost of full-scale wind tunnel
racing-car configurations. The basic modeling has to
investigations. By applying this technique, large por-
rely on some prior knowledge of separation-line loca-
tions of the proposed geometric changes can be sorted
tions so that the shear layers can be simulated. Once
out, in advance, so that only a limited number of pa-
the bluff-body flow behind the wheels and the vehicle's
rameters need be evaluated in the final wind tunnel test.

136
Downloaded from SAE International by University of Michigan, Sunday, August 12, 2018

REFERENCES 13. Maskew, B., Strash D., Nathman, J., and


Dvorak, F. A., "Investigation to Advance Prediction
Techniques orthe Low-Speed Aerodynamics of V/STOL
1. Chapman, D. R., "Computational Aerodynam- Aircraft," NASA CR-166479, Feb. 1983.
ics Development and Outlook," AIAA J., Vol. 17, No.
12, 1979, pp. 1293-1313. 14. Ahmed, S. R., and Hucho, W. H., "The Cal-
culation of the Flow Field Past a Van with the Aid of
2. Kraus, W.) "Panel Methods in Aerodynamics," Panel Method," SAE Paper No. 77-0390, 1977.
Numerical Methods in Fluid Dynamics, H. J. Wirz and
15. Losito, V., De Nicola, C., Albertoni, S. and
J. J. Smolderen, Eds., McGraw-Hili Book Co., 1978, pp.
Berta, C., "Numerical Solutions of Potential and Vis-
237-297.
cous Flows Around Road Vehicles," Int. J. of Vehicle
3. Rubbert, P. E., and Saaris, G. R., " A Gen- Design, Technological Advances in Vehicle Design Se-
eral Three-Dimensional Potential-Flow Method Applied ries, SP3, Impact of Aerodynamics on Vehicle Design,
to V/STOL Aerodynamics," SAE Paper No. 68-0304, pp. 429-440, 1983.
1968. 16. Chometon, F., " Calculating Three-Dimensional
4. Carmichael, R. L., and Ericson, L. L., "PAN Separated Flow Around Road Vehicles," Int. J. of Ve-
AIR - A Higher order Panel Method for Predicting Sub- hicle Design, Technological Advances in Vehicle Design
sonic or Supersonic Linear Potential Flows About Arbi- Series, SP3, Impact of Aerodynamics on Vehicle Design,
trary Configurations," AIAA Paper No. 81-1255, June pp. 374-386, 1983.
1981. 17. Markatos, N. G., "The Theoretical Prediction
5. Maskew, B., " Program VSAERO, A Computer of External Aerodynamics of Road Vehicles," Int. J.
Program for Calculating the Nonlinear Aerodynamic of Vehicle Design, Technological Advances in Vehicle
Characteristics of Arbitrary Configurations," NASA CR- Design Series, SP3, Impact of Aerodynamics on Vehicle
166476, Nov. 1982. Design, pp. 387-400, 1983.
6. Margason, R. J., Kjelgaard, S. 0., Sellers, W. 18. Summa, J.M., and Maskew, B., "Predicting
L. 3rd., Morris, C. E. K. Jr., Walkey, K. B., and Shields, Automobile Aerodynamic Characteristics Using an It-
E. W., "Subsonic Panel Methods - A Comparison of erative Viscous/Potential Flow Technique," SAE Paper
Several Production Codes," AIAA Paper 85-0280, Jan. No. 83-0303, 1983.
1985. 19. Dominy, J. A., and Dominy, R. G., "Aerody-
7. Woodward, F. A., "Analysis and design of namic Influences on the Performance of the Grand Prix
Wing-Body Combinations at Subsonic and Supersonic Racing Car," Proceedings of the Institution of Mechan-
Speeds" , J. of Aircraft, Vol. 5, No.6, 1968, pp. 528- ical Engineers Vol. 198 D, 1984, pp. 1-7.
534. 20. Katz, J., "Investigation of Negative Lifting
8. Hess, J. L., "Calculation of Potential Flow Surfaces Attached to an Open-Wheel Racing Car Con-
About Arbitrary Three- Dimensional Lifting Bodies," figuration," SAE Paper. No. 85-0283, Feb. 1985.
Final Technical Report MDC J5679-01, McDonnell Dou- 21. Katz, J., "Calculation of the Aerodynamic
glas, Long Beach, CA, Oct. 1972. Forces on Automotive Lifting Surfaces," J. of Fluids
9. Hess, J. L., and Smith, A. M. 0., " Calculation Eng., Vol. 107, No.4, 1985.
of Potential Flow About Arbitrary Bodies," Progress in 22. Pancini, G. F., and Di Giusto, N., "Exper-
Aeronautical Sciences, Vol. 8, Pergamon Press, 1967, imental Methods for Wind Tunnel Testing of Racing
pp. 1- 138. Cars with Ground Effect," Int. J. of Vehicle Design,
10. Woodward, F. A., "An Improved Method for Technological Advances in Vehicle Design Series SP3,
the Aerodynamic Analysis of Wing-Body-Tail Configu- Impact of Aerodynamics on Vehicle Design, pp. 480-
rations in Subsonic and Supersonic Flow," NASA CR- 492, 1983.
2228, May 1973. 23. Sawyer, R. A., and Lewis, M. S., "Wind Tun-
11. Morino, L., " A General Theory of Unsteady nel Tests on a 1/4 Scale Formula 2 Racing Car," Pro-
Compressible Potential Aerodynamics," NASA CR-2464, ceedings of the 4th Colloquium on Industrial Aerody-
1974. namics, Aachen, June 1980, pp. 235-245.

12. Johnson, F. T.," A General Panel Method for 24. Scibor-Rylski, A., " Experimental Investiga-
the Analysis and Design of Arbitrary Configurations in tion of the Negative Aerodynamic Lift Wings Used on
Incompressible Flows," NASA CR-3079, May 1980. Racing Cars," Advances in Road Vehicle Aerodynamics,
BHRA Fluid Engineering, pp. 147-154, 1973.

137
Downloaded from SAE International by University of Michigan, Sunday, August 12, 2018

25. Wright, P. G., "The Influence of Aerodynam- NOMENCLATURE


ics on the Design of Formula One Racing Cars," Int. J.
of Vehicle Design, Vol. 3, No.4, 1982, pp. 383-397. A reference area (1.47 m')
26. Katz, J., "Method for Calculating Wing Load- AR aspect ratio (area/b')
ing During Maneuvering Flight along a Three-Dimensional bF front wing span
Curved Path," J. Aircraft, Vol. 16 No. 11, 1979, pp. bR rear wing span
739-741. c wing chord
27. Katz, J., and Weihs, D., " Hydrodynamic Propul- CD drag coefficient (drag/0.5pV;'A)
sion by Large Amplitude Oscillation of an Airfoil with C, section lift coefficient (section lift/0.5pV;'c)
Chordwise Flexibility," J. Fluid Mech., Vol. 88, part 3, CL lift coefficient (lift/0.5pV~A)
1978, pp. 485-497. Cp pressure coefficient (p-p=/0.5pV;')
h vertical clearance
28. Katz, J., "A Discrete Vortex Method for the
Non-Steady Separated Flow Over an Airfoil," J. Fluid L vehicle length
P pressure
Mech., Vol. 102, 1981, pp. 315-328.
R wheel radius
29. Katz, J., "Large Scale Vortex-Lattice Model t time
for the Locally Separated Flow Over Wings," AlAA J., V 00 free-stream velocity
Vol. 20, No. 12 1982, pp. 1640-1646. XS,Z body coordinates
30. Katz, J., " Lateral Aerodynamics of Delta " angle of attack
Wings With Leading Edge Separation," AIAA J., Vol.
22, No.3, 1984, pp. 323-328. Subscripts
31. Buchheim, R., and Rohe, H., "Kann die Stro- F
front
mungsberechnung in Zukunft zur Besseren Aerodynamichen R
rear
Entwicklung von Pkw Beitragen," Verein Deutscher In-
genieure, Berichte, pp. 261-288, 1985.
32. Fackrell, J. E., and Harvey, J. K., "The Flow
Field and Pressure Distribution of an Isolated Road
Wheel," Advances in Road Vehicle Aerodynamics, BHRA
Fluid Engineering, 1973, pp. 155-165.
33. Cogotti, A., " Aerodynamic Characteristics of
Car Wheels," Int. J. of Vehicle Design, Technological
Advances in Vehicle Design Series, SP3, Impact of Aero-
dynamics on Vehicle Design, pp. 173-196, 1983.
34. Nichols, J. H. Jr, and Taylor, D. W., "De-
velopment of High Lift Devices for Application to Ad-
vanced Navy Aircraft," Proceeding of the V/STOL Air-
craft Aerodynamics, Vol. 2, May 1979, Monterey, CA,
pp. 777-849.

138

Você também pode gostar