Beruflich Dokumente
Kultur Dokumente
INTRODUCTION In the present paper, results are shown from a simple study in
which the aerodynamic characteristics of a typical racecar in
Racecar aerodynamics has evolved into an important technology extreme pitch attitude are examined. Specific attention is paid
that often makes the difference between success and failure on to the nature of the flow over the car outside of the ground
the racetrack. Today all serious racecars are tested in wind effect, especially at high angles of attacks. Attention is also
tunnels with the well-funded teams spending significant time drawn to vortical structures that develop on the lower sides of
and money in continuous development. However, due to the the car when placed at such attitudes.
highly competitive nature of the sport, individual teams closely
guard their results. Consequently, little is information is A 4,200-panel model of the typical configuration is used for
available in the open literature on such efforts. estimating the surface pressure distribution and streamline
pattern over the car. These streamline patterns are examined
As speeds have increased gradually over the years, reliance on and compared with those obtained via flow visualization in a
ground effect in maintaining traction has become the water tunnel.
cornerstone of many designs. It is not unusual to see a
configuration in which the negative lift generated by the body at
racing speeds approaches or even surpasses the vehicle weight1. METHOD OF ANALYSIS
In an effort to maximize the ground effect, little attention is paid
to off-design behavior of the configuration, especially in those Computational Method
cases where the car separates from the track. Consequently, loss
of ground effect during a race can have catastrophic The panel code used for the present analysis was originally
consequences. developed at NASA Ames Research Center under the name of
PMARC7 (Panel Method of Ames Research Center). The
In recent years, a few accidents have pointed out the necessity original code, developed in FORTRAN, lacked the flexibility of
for a closer examination of these configurations outside of their dynamic memory allocation. Therefore, the code was converted
operational envelope. During the 1999 Le Mans 24-hour race, a into C language and labeled CMARC. The new code that can
be executed on a personal computer has been supported since by
AeroLogic6 as the analysis module of the PSW.
The viscous flow modeling employed by the code, being two- The work described here was performed in the water tunnel at
dimensional, is expected to break down in regions of severe the National Institute for Aviation Research (NIAR), located on
cross-flow. Such regions abound on a racecar where aspect the campus of Wichita State University (WSU). This facility is
ratios are small, the surface is composed of compound curves, a closed-loop horizontal tunnel containing approximately 3,500
and there are many fittings and sharp corners that can trip gallons of water. The clear test section, visible from five
boundary layers. Therefore, the results of the boundary layer directions, is 2-ft deep, 3-ft high, and 6-ft long. A 2.5-ft segment
analysis performed by the code can be used only qualitatively at with constant cross sectional area connects the test section to the
best. Furthermore, since this study focused on high angles of diffuser. This section can also be used as an extension of the
attack, large regions of separated boundary layers were present test section. Water speed can be varied from 0.05 ft/s to 1.0 ft/s
in the flow. Consequently, no reliable force data could be with very low levels of turbulence. However, the flow speeds of
obtained with this method. 0.2 to 0.6 ft/s result in the least amount of turbulence.
Computational Model The majority of the runs used for the current research were
made at tunnel speeds of 0.5 ft/s, corresponding to a Reynolds
A typical Le Mans category racecar was modeled for the number of approximately 50,000 per foot. The lowest level of
purpose of the present study. Figure 1 shows the configuration free stream turbulence was visually observed in the
and its grid distribution. The geometry was represented with neighborhood of this speed, which also corresponded to the
approximately 4,200 quadrilateral panels. lowest tunnel vibration levels measured with an accelerometer
in an earlier experiment. Flow visualization was performed by
The numerical model included as much detail as possible. dye injection on the surface of the model. Figure 2 shows the
Nevertheless, certain elements of the geometry had to be schematic of this facility.
approximated. These included the wheels that were modeled as
extensions of the lower surface, without any gap between the The model was mounted on a C-strut that protruded into the
tire and the car body. No attempt was made to model details flow from the top surface. The strut allowed controlling the
such as the recess at the base of the windshield, rear-view model orientation in pitch and in yaw. Pitch attitudes from –45
mirrors, and gaps in the body surface. Air intake and exhaust degrees to +60 degrees could be attained. However, only
ports could be modeled with panels of user-defined porosity. positive pitch angles were used in the present study. Yaw angle
However, such modeling was not used for the purpose of could be varied over ± 45 .
accurate comparison with the experimental results.
Water Tunnel Model
The code could account for the ground effect using the method
of images, without generating additional panels. Typical run The water tunnel model was composed of a 1/24-scale model of
times were of the order of 5 minutes on a personal computer the car. The model was fitted with 7 dye ports on the top and 6
with 128MB RAM and a 733 MHz Pentium III processor. ports on the bottom for flow visualization. The dye ports were
flush mounted and all were located symmetrically to allow
detecting any asymmetry due to unintended yaw. The model
2
Test Section
Expectedly, no vortex filament was modeled by the inviscid These results, however exploratory, clearly demonstrate the
flow solution, but the presence of this type of vortical flow undesirable characteristics of a well-designed racecar in off-
structure was experimentally verified, as indicated in Figure 8. design attitudes. It is evident that a configuration optimized for
In this case, because of the absence of dye ports at proper maximum aerodynamic efficiency, which includes strong
locations on the model, flow visualization was accomplished ground effect, tends to act as a wing of small aspect ratio once
with the aid of the dye wand. separated from the ground plane. For the safety, of the occupant
and of the other participants, it may be adviseable to consider
The cases shown in Figures 7 and 8 were generated without a adding active devices to destroy organized flow structures in
ground plane. It would be logical any ground effect would such events.
magnify these phenomena. Furthermore, although not observed
experimentally on this model, it is possible for some of these
vortex filaments to find their ways up over the top of the car at CONCLUSIONS
high angles of attack, as indicated by the numerical model. This
behavior would lead to persistent lift generation well past An exploratory investigation of the aerodynamic characteristics
attitudes that should result in aerodynamic stall. of a typical Le Mans class racecar was presented, in an effort to
better understand its behavior in off-design attitudes. Primary
The above situation can be aggravated further if the emphasis was placed on extreme pitch orientations outside of
configuration is equipped with a splitter plate. The model used the ground effect. Results from a combination of computational
for most of the experimental investigation reported above lacked and experimental modeling were shown. Computational
a splitter plate. However, on many racecars this device is used modeling was performed with a low-order panel method, while
to gain better ground effect. flow visualization was accomplished with dye injection in a
water tunnel using a 1/24-scale model.
Generally, the splitter plate is a relatively thin sheet mounted
under the front of the car, extending some distance ahead of the Streamline patterns obtained from the two methods agreed very
nose. If the car encounters a high incidence angle, well at small angles of attack, despite the large differences in
Reynolds numbers. Experimental results indicated the presence
of attached boundary layers over a large part of the car at very
high angles of attack. Computational results suggested this
behavior to be due to the favorable pressure gradient over the
upper surface at high incidence angles.
REFERENCES