Documente Academic
Documente Profesional
Documente Cultură
=
0.455
(
)
2.58
(2)
where C
f
is the skin-friction drag and R
l
is the Reynolds number.
2.2. Pressure Drag
Pressure drag is caused by the pressure differential throughout the
automobile. It is mainly caused by the shape and design of the
vehicle. As fluid (air) flows through a body, there will be a change in
pressure and velocity thus causing momentum changed [1].
= (3)
where D
p
is the pressure drag force, P is the pressure, is the
angle between relative velocity to the normal pressure force and A is
the frontal area [2].
Numerical Approach Using Computational Fluid
Dynamic (CFD)
Ansys Fluent 14 software is used to simulate the aerodynamics of the
automobile to obtain the drag coefficient numerically. A simplified
vehicle shape called Ahmed Body was used for verification and
validation of the software [3]. It is also used to determine the proper
meshing, turbulence model and CFD solver input settings for the
external flow simulation for the race car model. Table 1 shows three
different simulations on Ahmed Body using different turbulence
models; Realizable k- solves two transport equations to obtain
turbulent kinetic energy, k and dissipation rate, . Reynolds Stress
Model (RSM) solves 6 components of Reynolds stresses and
dissipation rate, and Large Eddy Simulation (LES) solves the large
eddies and model the smaller eddies [4]. LES Smagorinsky-Lilly
model was used for the simulation. The results dont differ much
from experimental value [3, 5, 6]. Hence, using less computational
time to obtain similar results is highly recommended.
The results from Table 1 shows that the numerical results is in
acceptable range hence the similar meshing and solver input
methods were used for the external flow race car simulations.
Table 1. Comparison of CFD results of Ahmed Body.
Realizable k- RSM LES
Drag coefficient (C
D
) 0.316 0.316 0.284
Experimental Drag Coefficient 0.299 0.299 0.299
C
D
accuracy (%) 5.3 5.3 5.2
Computational Time (hours) 0.5 3 8
3.1 Geometry and Meshing
Figure 2 shows a full-scale simplified drawing of Taylors University
Race Car named Imperica in Solidworks 2011. A computational
domain similar to a wind tunnel test section is created around the car.
A domain of 3 car lengths upstream and 5 car lengths downstream is
created to accommodate for the flow development at the front and
turbulence formation at the rear end. A hybrid meshing approach
tested using Ahmed Body is also used on Imperica. Prismatic layers
are created near the surface of the body and also the road. This type
of meshing is suggested by Marco Lanfrit using inflation of first
aspect ratio of 5, growth rate of 1.2 and a total of 5 layers [7]. A
rectangular box mesh with smaller element size is created near the
29
EURECA 2013 Calculation and Optimization of the Aerodynamic Drag of an Open-Wheel Race Car
body to capture the flow condition near the surface body and also the
wake region at the rear end [7]. Figure 3 shows the mesh condition
near the surface body of the car.
Fig. 2 Simplified full scale Fig. 3 Mesh condition near the
CAD drawing. surface body.
3.2 Turbulence Model
Realizable k- with non-equilibrium wall function is recommended in
this external flow simulation and it is also tested using Ahmed Body.
Non-equilibrium wall function is highly recommended for near wall
modeling as it can capture and predict near wall boundary flow
condition and also flow separation [7].
4.0 Results and Discussions
The first CFD simulation was conducted to obtain the total drag
coefficient of the current basic setup on the race car with the radiator
cooling channel relative to velocity. Figure 4 shows the relationship
of the drag coefficient relative to velocity obtained through CFD.
The figure shows that the drag coefficient doesnt vary much when
the velocity increases. The total drag coefficient obtained is
approximately 0.624 with the current race car setup. The results also
match with theory where the drag coefficient doesnt vary much
when velocity increased at subsonic speed [1].
Fig. 4. Variation of drag coefficient, CD, with velocity.
From the graph shown in Fig. 4, a velocity of 18.3 m/s is chosen
as an overall average speed for next simulation by considering the
race car is maintained at an average speed of 18.3 m/s throughout the
Melaka International Motorsport Circuit (MIMC). The effect of the
angle of radiator cooling channel is investigated through CFD at the
mentioned speed. The original setup of the cooling channel is angled
at 36 degree where it completely directs the flow into the radiator to
cool the engine. Multiple simulations were conducted to analyse the
effect of increasing the angle of the radiator cooling channel. The
effect of drag coefficient by varying the angle of the cooling channel
is plotted in Fig. 5.
Fig. 5 Drag coefficient as a function of angled radiator cooling channel.
The race car without any radiator cooling attached shows a drag
coefficient of 0.517. Malcolm Campbells Blue Bird achieved drag
coefficient of 0.45 for an open-wheel race car [1]. The result obtained
shows some similarity in drag coefficient. The current setup of the race
car which attached the radiator cooling channel at 36 degree produces a
drag coefficient of 0.625 which increased the drag coefficient by 20.8%
compared to the one without. Attaching the radiator cooling channel
produced negative pressure region at the rear of the car which increases
the drag coefficient. The cooling channel is then angled to reduce the
drag coefficient but on the other hand it reduces the air flowing into the
radiator. Increasing the angle of the cooling channel actually reduces
the drag coefficient towards the drag coefficient of the race car without
the cooling channel. The radiator cooling channel angled at 90 degree
shows similarity in drag coefficient with the one without the cooling
channel which is in the range of 0.51 to 0.52. There was less resistance
of air when the radiator cooling channel is angled at 90 degree similar
to the one without.
When the radiator cooling channel is angled at 72.5 degree, the
drag coefficient is 0.479 which is lower than the one without the
cooling channel attached. A reduction of 23.4% in drag coefficient
compared to the current setup is achieved. The drag coefficient after
this angle returns towards to the one without the cooling channel.
This phenomenon might be caused by the turbulence created from
the cooling channel. This particular angle introduced turbulence into
the negative pressure region at the rear end of the car which increases
the pressure coefficient, C
p
at the rear reducing the pressure
difference. 72.5 degree cooling channel setup has a C
p
of -1.5 at the
rear whereas the current setup has a C
p
of -2.0.
5.0 Conclusions
Based on this simulation, 72.5 degree cooling channel produced least
drag but the cooling of the engine need to be compromised. Hence,
an automated system with temperature sensor can be implemented to
optimize the drag produced and cooling of the engine. When the
temperature of the engine increases, the angle of the cooling channel
will be properly changed to direct more air into the radiator to cool
the engine and vice versa. Heat transfer effects from the radiator
could be taken into considerations for future work.
Acknowledgment
I would like to thank my colleagues for giving me valuable help and
support throughout this final year project.
References
1. Hoener, S. F. (1958). Fluid Dynamic Drag. New Jersey.
2. Zala, B. (2012). Comparative Assessment of Drag Force of Hatchback and
Sedan Car Model by Experimental Method. International Journal of
Advanced Engineering Research and Studies. Volume 1, Issue 3, pp. 181-
183.
3. Ahmed, S. R. (1984). Some Salient Features of the Time Average Ground
Vehicle Wake in Detroit. SAE International Congress and Exposition.
SAE paper 840300.
4. Versteeg, H. K., and Malalasekera, W. (1995). An introdution to
Computational Fluid Dynamics. Harlow: Person Limited.
5. Drage, P., Hormann. T., Meile. W., Gabriel. A., Brenn. G., and
Lindbichler. G. (2008). Efficient Use of Computational Fluid Dynamics
for the aerodynamic Developmet Process in the Automotive industry. 26
th
AIAA Applied Aerodynamics Conference. AIAA 2008- 6735.
6. Lienhart, H., and Becker S. (2003). Flow and Turbulence Structure in the
Wake of a Simplified Car Model in Michigan, USA. SAE 2003 World
Congress, SAE Paper 2003-01-0656.
7. Marco, L. (2005). Best practice guidelines for handling Automotive
External Aerodynamics with Fluent. Birkenweg: Fluent Deutschland
GmbH.
30