Académique Documents
Professionnel Documents
Culture Documents
ABSTRACT
The new Porsche 911 Carrera went into production in Summer 2004. Both the drag
coefficient and the lift coefficient of the new 911 have been reduced significantly as
compared with the predecessor car.
One of the main reasons of the successful reduction of the drag coefficient by
CD=0.02 is the low cooling-air drag by which the 911 is distinguished.
The present paper describes the layout and development of the optimised coolingsystem concept of the new 911, which is essentially based on the cooling concept of
the highly successful predecessor model.
Following the brief description of the aerodynamic features and cooling requirements
of the new 911, the current know-how regarding the layout of low-drag cooling air
concepts is briefly addressed. Then, the layout philosophy, development and
optimisation of the cooling-air concept of the new 911 are described in detail. Finally,
a summary of the results obtained as well as the outcome of a benchmarking
campaign are presented.
1
Among the major engineering goals of the new Porsche Carrera 911 number
outstanding driving performances and excellent driving dynamics (Figures 1 & 2).
The significantly improved aerodynamic performances (Table 1) make an important
contribution to achieving these goals.
911 Carrera
MY 1997
MY 2002
MY 2005 911
911 S
Pe [hp]
300
320
325
355
CD
0.30
0.30
0.28
0.29
CLf
0.08
0.06
0.05
0.05
CLr
0.05
0.03
0.02
0.02
Af [m2]
1.95
1.95
2.00
2.00
CD x Af [m2]
0.59
0.59
0.56
0.57
0.30
0.25
0.20
0.15
0.10
0.05
911
Porsche 911
predecessor
MY 2005
MY 2002
0.00
0.38
0.36
0.34
0.32
0.30
0.28
0.26
Drag coefficient CD
Fig. 3: Drag & lift coefficients of the new 911 Carrera and its competitors (all data
determined in the full-scale Porsche wind tunnel under static conditions)
Figure 3 compares the aerodynamic coefficients of the new 911 Carrera and some of
its competitors. As can be seen, the aerodynamic evolution of the 911 series has
been successfully continued: With a drag coefficient of CD=0.28 for the basic version
the new 911 has reconquererd the top position in its market segment.
The improvement of the drag coefficient by CD=0.02 compared to its predecessor is
the result of thorough aerodynamic optimisations which were mainly focussed on the
outer shape, the almost 100% enlargement of the aerodynamic undershield and the
extremely low cooling-air losses compared with those of the cars competitors.
2
With a cooling drag of CDc=0.002-0.003 the first water-cooled 911 (MY 1997) has
almost zero cooling-system drag. This low drag level mainly results from the vertical
downward deflection of the outlet air ahead of the front wheels, which acts like a
virtual wheel spoiler and improves the flow conditions about the wheels [1]. The
resulting positive intereference effect almost completely compensates for the
overall-drag contribution resulting from the internal drag.
The predecessor 911 MY 2002 has a cooling drag of about CDc=0.006 which is
slightly higher than that of the first water-cooled 911 MY 1997. This results from the
roughly 15% higher cooling-air mass flow rate which is necessary to cope with the
higher cooling demand of the 320 hp engine.
The cooling-air system of the 911 MY 2002 is basically the same as that of the 911
MY 1997. The increase of mass flow rate was achieved solely by optimising the front
air-inlet area and inlet ducts. As the rest of the cooling-air ducts are the same for both
0.014
0.012
0.010
General
trendline ?
New 911?
0.008
0.006
911 MY 2002
0.004
911 MY 1997
0.002
0.000
80
100
120
140
160
[%]
Fig. 6: Extrapolation of the cooling drag for the new 911 based on the predecessors
At the beginning of the project, the trend illustrated in Figure 6 was used to roughly
estimate the cooling drag to be expected in the new 911. Based on the anticipate
engine-power increase by about 11%
Engine
Cooling air
911 Model
power
mass flow rate and the experience gained with the
MY 1997
Basis
Basis
predecessor
models
the
required
MY 2002
+7%
+16%
increase of the cooling-air mass flow rate
MY 2005 (911 S) +11%
+21%
was estimated at about 21% (Table 2).
Table 2: Estimation of the cooling air
demand of the new 911
Figure 6 shows that a further increase of the cooling-air mass flow by 21% relative to
the predecessor with unchanged cooling-air concept would have led to a cooling drag
of about 10 to 11 drag counts, corresponding to an increase by 4 to 5 drag counts.
This was definitely not acceptable with regard to the ambitious aerodynamic goals
defined for the new 911.
Therefore a limit of CDc=0.005 was defined for the cooling drag of the new 911 and
stipulated in the Specifications. This meant that the cooling drag of the 911 had to be
reduced even below the drag level of the predecessor model.
The situation illustrated in Fig. 6 had made it clear that it would not be possible to
adopt the cooling-air concept of the predecessor unchanged and that considerable
modifications would have to be carried out.
The primary engineering target, however, was to guarantee a sufficiently high cooling
capacity regardless of the driving condition. For sports cars, circuit racing and highspeed driving are the most relevant operating modes with the highest loads on the
cooling system occurring during blower-assisted operation on racing tracks. During
aerodynamically relevant high-speed driving, the demand for cooling is considerably
lower. The different cooling-capacity and cooling-air demands under those operating
conditions had to be taken into due account. And in order to realize a most efficient
and demand-adapted cooling-air concept it was important not to overdimension the
cooling-air requirements either.
2.2
Regarding the 911, the question was how to increase the cooling-air mass flow while
reducing the cooling drag. Let us first have a look at the theoretical background:
The cooling drag and cooling-air flow rate
are dependent on a lot of parameters
which are shown in Figure 7 taking the
example of a closed cooling-air ducting.
Besides the air flow about the vehicle the
most important of these parameters for a
given inlet area Ai are the following: the
radiator core area Ac, internal losses (),
driving differential pressure (CpO) and the
cross-section of the cooling-air outlet AO.
V C
Ai
Vc
CpO
AO
Ac
Internal Drag
Interference
Drag
Fig. 7: Parameters relevant for cooling drag and cooling-air flow rate
When considering the cooling drag CDc it must be kept in mind that it is basically
composed of two elements: the drag due to internal flow resistance and the
interference drag:
CDc
C , ITD
+
Dc
Internal drag
C , IFD
Dc
(1)
Interference drag
Since the interference drag can be both positive and negative - with the latter being
welcome - there are two main approaches to optimise the cooling drag:
Low internal pressure losses and specific influencing of the interference drag can
only be obtained with a strictly ducted cooling-air system with air inlet and outlet
ducts as illustrated in Figure 7. Interferences of the cooling air flow with the air flows
about the car can occur anywhere - i.e. in the front-end region as shown here but
also in the wake of the car.
In the past, little consideration was given to the fact that the interference drag can
have a considerable influence - and be negative at that. According to the results of
pertinent investigations at Porsche, the interference drag can reach almost the same
order of magnitude as the internal drag. Negative interference effects occur in nearly
all Porsche sports cars and make a significant contribution to the low overall cooling
drag levels of the 911 model series (see Fig. 21).
It is very difficult, however, to specifically influence or predict such interferences.
While the interference contribution eludes even the most simple kind of formulation, it
is possible to formulate rather uncomplicated theoretical approaches at least for the
internal drag and the cooling-air volume flow rate, which clearly illustrate the
influence of the relevant parameters (see also Appendix A):
Internal drag:
CDc , ITD =
Vc
V
3
Vc
1
2
V Ac Af
1- C pO
1
+ 2
2
Ac
AO
(2)
(3)
As can be seen, the internal drag is influenced by the volume flow raised to the
power of three, the radiator area raised to the power of two and the pressure loss
coefficient. Thus formula (2) clearly underlines the priorities to be observed during
optimisation1:
According to equation (2) the internal drag increases by 1.21 = 1.772, i.e. by 77%, if the volume flow rate is raised by 21%
with otherwise unchanged parameters. Based on the 911 MY 2002 value of CDc=0.006 this means a cooling drag increase by
about 0.005 to CDc = 0.011. Thus the experimental trend shown in Figure 6 is clearly confirmed by theory.
So, the internal drag is influenced - with increasing priority - by internal losses, the
radiator core area and mainly by the cooling-air flow rate. The cooling-air volume flow
itself, on the other hand, depends on the internal losses, the radiator front surface
and - additionally - on the driving differential pressure (1-CpO ) and the outlet crosssection AO. Therefore, the relationships of equations (2) and (3) for the internal drag
and cooling-air volume flow rate cannot be treated separately.
The question is, whether there are combinations of individual parameters which,
according to equations (2) and (3), result in a higher cooling-air volume flow rate
while reducing the internal drag at the same time. Based on the data of the MY 2002
model, corresponding parameter variations were performed which are given in
Appendix A. The results obtained confirm the afore-mentioned theory. Fig. 8 shows
some of the most efficient parameter combinations.
40
Volume flow rate
Internal drag
Change in %
20
0
-20
-40
-60
CpO= 0
-80
CpO= 0
CpO = 0
CpO = -0,25
CpO = -0,5
CpO = -0,5
CpO = -0,5
AO = -50%
AO = -50%
= -50%
AO = -50%
= -50%
AO = -50%
= -50%
AO = -50%
= -50%
Ac = +5%
AO = -50%
= -50%
Ac = +20%
Similar conclusions were derived in Ref. [2] based on the momentum approach.
As far as the new 911 was concerned, it was clear that an alternative air outlet in the
wheel house area would have to be provided for, allowing to unthrottle the air flow
downstream of the radiator and to control the air flow rate and internal drag via the
outlet area. However, earlier investigations with air outlet openings in the wheel
houses carried out within the scope of the predecessor development used to result in
strongly increased cooling drag and overall drag coefficients.
From an aerodynamic point of view, the question was which outlet concept would
allow similar positive interference effects to be reached as with the predecessor, i.e.
at least how to avoid negative interferences with a cooling-air outlet into the wheel
house.
To answer this question, three different outlet variants were tested, which are
illustrated in Figure 9 below.
Variant 1 (Basis)
Variant 2
Variant 3
Preliminary basic investigations with alternative cooling-air outlets were carried out in
a very early phase of the project using a 1:4-scale model with a radiator simulator.
The results showed similar lift and drag effects as in the original full-scale 911.
Subsequent tests with various front ends and air outlet variants furnished valuable
information about the cooling drags and interferences obtained with the styling and
outlet variants examined.
The first full-scale investigations with the three basic concepts as shown in Figure 9
were carried in the early concept phase using a predecessor model [5]. The results
showed that the mass flow target could be met, but none of the variants reached the
target for the cooling drag. However, with variant 3, it was possible - as expected - to
efficiently control the cooling-air mass flow by varying the cross-section of the air
outlet into the wheel house.
In the following, the outlet-air concepts were therefore examined in more detail.
These investigations were carried out in close cooperation with the thermodynamics
experts in order to obtain direct information about the cooling potential achievable
with the variants under examination. Wind tunnel measurements were performed in
order to determine and evaluate the cooling drag and air flow rates. With regard to
the thermodynamics, driving tests were carried out with the aim of evaluating the
alternative air outlet configurations. The test vehicle used was a predecessor model
fitted with a tuned-up engine and new sample radiators. The air outlet variants were
examined under both circuit-racing and top-speed conditions.
130
120
50%
100%
100%
100%
100%
100%
60%
50%
100%
12
10
Coolant temperature
Volume flow rate
110
Cooling drag
The first evaluation criterion checked was the cooling-water temperature. The
temperature changes and measured cooling-air flow rates were used as input data
for mathematical analysis based on a thermal model. Figure 10 sums up the most
important results obtained with the basic system and the two main variants.
Limit Tc 100
Target Tc
Safety
margin Tc
90
80
70
0
911 MY02
V1
V2
V3
V4
V5
Target
CDc
V6
Fig. 10: Wind-tunnel and driving-test results obtained with the air-outlet variants
All variants shown in Figure 10 were fitted with the tuned-up engine and new sample
radiators. Variant V1 which featured the cooling-air concept of the predecessor
clearly reflected the potential of the new radiators: Despite the tuned-up engine and
the lower cooling-air flow rate the coolant temperatures were practically identical with
those of the predecessor. As can be seen in Figure 10, the greatest potential with
regard to cooling-air flow rates and cooling capacities was offered by the variant V2
with the largest possible downward air outlet and the largest outlet into the wheel
house. On the other hand, this variant had the largest penalty for cooling drag.
However, from the results yielded by the other variants it was concluded, that the
cooling drag targets can be reached by reducing the cooling-air mass flow increase
relative to the predecessor to about 10% even though the coolant-temperature target
is still slightly exceeded.
Altogether, the results obtained were rather promising and confirmed that the
cooling-air outlet into the wheel house would allow a sufficiently high cooling capacity
to be realized. In addition, the measurements delivered a simple correlation between
the change of cooling-air mass flow and the change in cooling-water temperature.
This correlation was used to assess the subsequent wind tunnel tests.
For the aerodynamic tests performend after the concept phase, a full-scale throughflow concept vehicle whose outside panelling was continuously updated to the
respective styling status was used (Fig. 11).
The concept vehicle consisted of the basic structure of a stripped predecessor model
equipped with a GFRP outside panelling corresponding to the respective styling
status. The through-flow vehicle is a full vehicle equipped with radiators and air ducts
for engine cooling, engine-compartment cooling, brake cooling etc. Along with the
1:4-scale and 1:3-scale models it is Porsches main aerodynamic test device which
has yielded excellent results so far. The cooling-air mass flows are measured by
means of radiators fitted with cylindrical pressure sensors previously calibrated on a
flow test bench. The procedure [6] uses the static pressure difference across the
radiator as a basis (Fig. 12) and allows the volume flow rate to be measured with a
relative precision of about 3% [7].
p1
p2
The subsequent investigations were aimed at further optimising the cooling-air mass
flow and drag of the air-outlet variants favoured in the preliminary phase with the help
of the concept vehicle and to adapt them to the respective styling status. It had to be
made sure in particular, that the front-end air inlet openings designed by the styling
department were sufficiently large.
During the fine-optimising phase, more than 30 variants of the lower and wheelhouse air outlets with different dimensions and configurations were submitted to wind
tunnel tests.
The definite configuration chosen was the combined variant with downward air outlet
and the outlet into the wheel house as it offered the greatest scope for subsequent
optimisation. Another beneficial measure was to use pressure flaps in the fan shroud
which unthrottle the air flow downstream of the radiator and produce homogeneous
flow conditions about the radiator.
As far as thermodynamics were concerned, the air outlet variants were tested in one
of the first 911 prototypes. As of that stage, the primary evaluation criterion was the
absolute coolant temperature which was not allowed to exceed given limits.
Experience has shown that by the end of a project the coolant temperatures
frequently increase unexpectedly as a result of final modifications to the engine
management system. It was therefore decided at the start of the project to keep a
safe 5 degree distance from the thresholds.
Following the Styling-Freeze of the outer shape the fine tuning in the wind tunnel
could be tackled: The outlets leading the air downwards into the wheel house were
closed one after the other. The minimum air flow rate required to remain within the
given temperature limits was known from the correlation between the air flow rate in
the wind tunnel and the coolant temperature in the driving test. Thus, the individual
measures taken could be precisely evaluated. Figure 13 summarizes the influence of
these measures on the air-flow rate and the cooling drag.
140
130
120
100%
100%
70%
110
100%
50%
Predecessor + 5...10%
60%
100
90
80
0
10
12
Fig. 13: Fine tuning of the combined air-outlet variant in the wind tunnel
As illustrated in Figure 13, the cooling-air drag drops proportionally to the cooling-air
volume flow rate. For the initial status with maximum downward air outlet and
maximum outlets into the wheel house the cooling drag is CDc=0.009. Closing the
downward air outlets reduces the cooling drag by CDc=-0.002. By gradually closing
the remaining air outlets leading into the wheel house (from the vehicle center line to
the outside) leaving about 50% of the initital outlet area it is even possible to lower
the cooling drag to less than the specified limit of CDc=0.005 with the cooling-air flow
rate target derived from the correlation being only just reached.
The 50% outlet-variant with closed downward outlet which was developed in the
wind tunnel was subsequently submitted to corresponding driving test which fully
confirmed the afore-mentioned results at all operating points.
Fig. 16: Wheel-house outlet and undershield Fig. 17: Wheelhouse outlet (backside)
Figure 17 shows an enlarged view of the 50% air outlet into the wheelhouse. As
can be seen, the outlet air is horizontally deflected to the exterior at an angle of
approximately 45 by corresponding deflector fins. This configuration improves the lift
conditions at the front axle and has been used to comply with the targets in terms of
lift and lift balance.
To minimise the thermodynamic engineering risks before SOP, the achievable
maximum wheelhouse outlet opening was maintained. The aerodynamically
favourable 50%-outlet is realized by closing half of the openings (gaps) by means of
a thin plastic foil.
The closed gaps are relatively easy to reopen. This allows maximum scope for
trouble shooting and offers a sufficiently great cooling-air reserve for further engine
tuning during subsequent model updating.
1:4 Model-Tests
25
20
SOP
15
10
0
Concept Development
Prototypes
Concept Validation
-5
0
20
40
60
80
100
120
140
160
180
200
220
240
Sequential Number
Fig. 18: Evolution of the cooling drag from the concept phase to SOP
12
3
10
Figure 18 illustrates the evolution of the cooling drag versus the engineering period
and up to SOP. At the beginning of the project, there is a large scatter to be seen
which results from the many concept variants examined. As the project progresses
the cooling drag took shape which was forecast by means of the through-flow
concept vehicle.
8
6
4
2
0
0
1
6
3
50
100
150
200
250
Figure 19 shows the cooling drags measured with a total of 30 production cars. The
average of these measurements is CDc=0.004-0.005 with a scatter of 0.001 to
0.007. This scatter results from the assembly tolerances and the measuring accuracy
in the wind tunnel. It must be kept in mind, however, that the variations of the 911
cooling drag are in the millesimal CD range. In Figure 20 the cooling drag for a
particular car has been plotted versus the wind velocity. To obtain this plot, a velocity
polare was carried out with the air inlets opened and closed, respectively.
4.2 Interference effects
When dealing with the cooling drag, distinction must be made between two different
interference effects: The most important one is the interference between the cooling
air flow and the air flow about the car which results in an additional - either positive of
negative -interference- related drag contribution. As already mentioned, interferences
can contribute considerably to the overall cooling drag level and be used to the
advantage of the system. The low cooling-air losses of the Porsche production cars
in particular are largely due to negative interference drags. Figure 21 shows the
individual drag contributions in the new 911 and its predecessors.
12
10
Internal drag
Interference drag
8
6
4
2
0
-2
-4
911 MY 97
911 MY 02
911 MY 05
-6
arisen, whether this method is really suited to determine the cooling drag correctly. In
the past already, it has been pointed out that the surprisingly low cooling-air losses of
quite a number of cars might be due to front-end interferences [4]. Such an
interference can be caused by a flow separation which occurs at the front end - e.g.
above the hood - if the air inlet openings are closed and which does not appear if the
the inlets are open. This means that the CD-value measured with the inlets closed is
too high and that the cooling drag determined on that basis is too low, i.e. it seems to
be low but in fact it is not.
To date, it has not been easy to prove that such effects occur in wind tunnel testing.
The existence of such front-end interferences was detected quite by chance some
years ago when examining a competitors car in Porsches wind tunnel. And it was
also found out that these interferences can be of considerable magnitude.
The car in question was tested using the standard method when all of a sudden a
negative cooling-air loss occurred, i.e. the drag coefficient increased when the inlets
were closed. The reason of this low cooling drag was supposed to be the abovementioned front end interference. The question was, how to measure and confirm
this phenomenon. To this end, a simple method was developed based on the
following assumptions:
If the front-end interference really exists it must be caused as explained above. In
that case, the drag level measured with the inlets closed includes and is unduly
increased by the drag portion resulting from flow separation. In order to determine
the actual cooling drag increase without this separation-dependent share, it must be
made sure that the separation occurs also when the inlets are open.
To this end, the front inlets of the car in question were fitted with circumferential,
sharp 50-mm-high separation edges3 as shown for the 911 in Figures 23 and 24.
Separation edge
Fig. 23 & 24: Measurement of cooling drag with separation edges (911 MY 05).
Due to their very typical look, these separation edges have been called fish-mouth by Porsches
aerodynamics experts. Accordingly, the aerodynamic test is called fish-mouth test.
16
14
without "fish-mouth"
with "fish-mouth"
12
10
8
6
4
2
0
-2
911 MY 02
911 MY 02 Turbo
911 MY 05
Competitor 400hp
Fig. 25: Influence of separation edges (fish-mouth) on the cooling drag of various
911 models and a competitor car.
Figure 25 illustrates the impacts of this measure on the competitor car in question
and on some 911 models - including the 911 MY 2005 - examined in the same way.
As can be seen, the effects measured with the competitor car are of a large
magnitude, whereas the influence on the cooling drag of the 911 models is
insignificant. With no separation edges, a negative cooling drag of CDc= -0.001 is
obtained. With separation edges, by contrast, there is a positive cooling drag of
CDc=0.014. This is precisely the magnitude one would expect from the theoretical
approach in equation (2). So, this result confirms that the car in question actually
shows a front-end interference.
Of course, the separation-edge or fish-mouth test described above is no absolutely
perfect method as the separation edges might also have an effect on the pressure
conditions and air flow about the cars. However, it indicates whether there is a
distinct front-end interference or not. In addition, it allows the true cooling-air drag to
be at least approximately measured.
5
Figure 26 compares the 911 production status with the predecessor models. In
comparison with the direct predecessor MY 2002 the cooling air mass flow rate could
be increased by about 10 to 15%, while the cooling drag was reduced to CDc = 0.004
for the 911 basic model and to CDc=0.005 for the 911 S. This increase of the cooling
air mass flow rate was achieved inspite of the significant 60% reduction of the
cooling-air outlet area when compared with the predecessor MY 2002, see Figure 27.
10
9
140
130
130
8
120
115
100
100
80
5
4
60
3
40
2
20
160
0
911 MY 97
911 MY 02
911 MY 05
911S MY 05
Fig. 26: Cooling drags and cooling air mass flows of the new 911 and its
predecessors
Cooler
911 MY 2002
911 MY 2005
Inlet
0
Outlet
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Fig. 27: Cooling-air duct areas of the new 911 and its predecessor
Figure 28 compares the cooling-air efficiencies of various models of the 911 family.
The efficiency criterion used is the cooling drag related to the velocity ratio =VC/V .
With the cooling-air velocity VC resulting from the cooling-air volume flow and radiator
area, the latter is implicitly taken into account. Of two cars, having identical cooling
drags and cooling-air volume flows, the car which does with the smaller radiator core
area is the most efficient one.
0.14
= Vc / V
0.12
0.10
0.08
0.06
0.04
0.02
0.00
911 MY 97 911 MY 02
Fig. 28:
911 Turbo
MY 02
911 MY 05
911S MY
05
Competitor
400hp
From Figure 28 it can be seen that the efficiency criterion allows the various coolingconcepts to be more easily compared:
As can be seen, the new 911 yields approximately the same cooling air efficiency as
the first water-cooled 911 MY 1997, whose cooling drag level of about 0.002 is still
setting the standard today against which all the other Porsche cars and their
competitors have to match themselves.
50
Ai /Ac = 40 %
30
20
10
0
Boxster
MY 1996
Boxster
MY 2001
911 Carrera
MY 1996
911 Turbo
MY 2000
911 Carrera
MY 2002
911 Carrera
MY 2005
Competitor
400hp
Fig. 29: Relative inlet cross-sections of various Porsche cars and a competitor
Another criterion - though mostly a stylistically relevant one - is the size of the inlet
openings in the front end. Big inlets which are not necessarily negative from an
aerodynamic point of view are often rejected for styling reasons. Other automotive
manufacturers apply the 40% rule - i.e. the target size of the air inlet areas is 40% of
the radiator core area.
By contrast, clearly smaller inlet areas can be realized by providing for efficient
cooling-air concepts, see Figure 29.
0.05
0.04
0.03
0.02
0.01
0.00
100
911 MY 05
200
300
911S MY 05
400
500
CONCLUSIONS
One of the major engineering challenges in the aerodynamic development of the new
911 was to realize the required increase of the cooling-air volume flow rate while
preventing the cooling drag from growing. To solve this problem, the air ducting
concept of the predecessor with vertically downward cooling-air outlets ahead of the
front wheels had to be given up. Under strict observance of the given theoretical
layout principles, an alternative low-drag air-ducting concept with air outlets opening
into the wheel houses was developed.
The aerodynamic part of the production-car development was performed in the wind
tunnel with the help of conventional methods. The test car used was Porsches
service-proven full-scale through-flow concept vehicle. The entire development
process drew benefit from the close interactions between the aerodynamic concept
layout in the wind tunnel and the thermodynamic driving tests.
The optimised cooling-air concept of the new 911 covers all of the highly differing
demands of circuit racing and high-speed driving. Inspite of the roughly 15% increase
of the cooling-air flow through the radiators, it has even been possible to reduce the
cooling drag level of the new 911 to below that of the predecessor model. The low
cooling-air losses of the 911 are primarily due to the integrally designed cooling-air
ducting from the inlet to the outlet openings, the large radiator areas, low internal
pressure losses and mainly to the control of the cooling-air mass flow rate and
cooling drag through the clearly reduced outlet cross-sections. Similar to all watercooled variants of the 911 family, negative interference drags, too, are contributing to
the low cooling drag of the new 911. With a share of about 1.4% in the overall drag,
the cooling drag of the new 911 is one of the lowest among its competitors.
TERMINOLOGY
A
Ac
Af
Ai
AO
CD
CDc
CDc
CDc,IL
CDc,IF
CpO
CL
CLf
CLr
D
DC
GFRP
MY
p
pt
P
Pc
Pe
Tc
V c
Vc
VO
V
area
radiatior core area
projected frontal area
duct inlet area
duct outlet area
vehicle drag coefficient, CD=D / (V2/2 Af)
drag coefficient contribution due to the cooling system
measured cooling drag, CDc = CD,inlets open - CD,inlets closed
cooling drag due to internal pressure losses
interference drag
static pressure coefficient at outlet
total lift coefficient
lift coefficient front axle
lift coefficient rear axle
aerodynamic drag force
aerodynamic drag due to the cooling system
glass-fibre reinforced plastic
model year
static pressure
total pressure
power
power to overcome the cooling losses
engine power
coolant temperature
cooling air volume flow rate
radiator normalizing or core speed
outlet air speed
free stream air speed (synonymous with car speed on road)
exit-flow inclination angle to vertical
air density
pressure loss coefficient, =pt / (Vc2/2 )
velocity ratio, =Vc / V
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
APPENDIX
The basic relationships between the cooling drag and volume flow rate are
represented by the following simple equations:
Cooling drag:
CDc
CDc , ITD
(A1)
CDc , IFD
3
V A
c c
V Af
CDc , ITD =
Vc = Vc Ac
Vc
=
V
(A2)
(A3)
1- C pO
A
+ c
AO
(A4)
When considering the internal drag only, the question arises whether there is a
particularly suited combination of parameters, which reduces the internal drag inspite
of the volume flow increase. In order to get a proper idea of the basic relationships,
the parameters were varied. Using the data of the 911 MY 2002 each parameter was
varied by +/- 50%, Figure A5:
70
60
Parameter +50%
CpO, AO, Ac ,
50
40
Parameter -50%
CpO, AO, Ac ,
30
20
10
0
-10
-20
-30
-40
Cooling drag
-50
-60
-70
CpO
AO
Ac
Ac
AO
CpO
Fig. A5: Influence of the parameters in equations (2/3) on the cooling drag and
cooling air flow rate
As can be seen, the greatest possible internal drag reduction with lowest possible
negative impacts on the volumetric flow is achieved by reducing the air outlet areas.
On the other hand, the cooling-air volume flow can be increased with almost no rise
in drag by lowering either the pressure loss coefficient or enlarging the radiator area.
Accordingly, it should be possible to reach the desired target by cleverly combining
these parameters as shown in Fig. 8.
P = D V = CD
V3 Af
(B1)
V3 Af =
V3 ( CD - CDc , ITD ) Af
(B2)
Pc = Dc Vc =
Overall power requirement:
Vc3 Ac
P = P ' + Pc
(B3)
(B4)
By inserting equations (B1)-(B3) in equation (B4) we obtain the following equation for
the internal drag coefficient:
3
CDc , ITD
V A
= c c
V Af
(B5)
Equation (B5) was already derived by H. Schmitt in 1940 [9]. It can be considered as
equivalent to the momentum approach. Momentum changes or the the influence of
the outlet angle of the cooling air exit flow (see Fig. 7) which is usually taken into
consideration in the momentum approach are implicitly accounted for in the pressure
loss coefficient .
However, there is an interesting difference: According to equation (B5) the drag due
to internal flow resistance cannot be negative as the total pressure loss coefficient
is not negative if there is no energy input from outside. Thus according to equation
(A1), the overall cooling drag of a car can only be negative if there is a negative
interference drag which must be higher than the internal drag.
The main advantage of equation (B5) over the ususal momentum approach is in the
simplicity of the represenation and, above all, in the fact that the parameters used are
easy to measure and allow the internal drag to be determined in a simple manner.
Using equation (A1) the overall cooling drag can then be split up into the internal and
interference drags as shown in Figure 21 for the new 911 and its predecessors.