The Development of the Audi 3.6-litre V8 Twin Turbo FSI Engine for Le Mans

The Development of the Audi 3.6-litre V8 Twin
Turbo FSI Engine for Le Mans
by Ulrich Baretzky, AUDI AG
Within a period of just 15 months,
Le Mans 24-Hour Race
the smallest details have to be
the race engine department of Audi
regarded as decisive factors for a
The Le Mans 24-hour race has
victory. But in this case, the race
Sport designed and developed a
considerably changed its charac-
distance of a whole F1 season of
ter in recent decades. The main
more than 5000 km is completed
homogeneous direct-injection engine
reasons were the modifications to
in just 24 hours.
based on the successful engine of
the rules and track profile as well
In order to create tight com-
as competition between the big
petition, the basic idea of the
1999 and 2000. Furthermore, all
manufacturers. The original idea
rules is to balance the chance of
– to prove the durability of cars
winning for all types of cars and
targets such as mixture homogene-
and pit crews even through major
engines. The mandatory type of
ity, power output and adaptation to
repair work during the race – has
fuel is provided by the organizer
been transformed into a 24-hour
and corresponds to "Super plus”
the car’s requirements were success-
sprint race with very fierce com-
gasoline, but with closer toler-
fully achieved within six months The
petition. As in Formula 1, even
ances. As a consequence, the
new engines went on to power the
two Audi R8 race cars to an impres-
Figure 2: Percentage of the throttle
sive victory in the Le Mans 24-hour
angles used in a Le Mans race lap
race in June 2001.
54
AutoTechnology 3/ 2002

---

C o v e r S t o r y
engine performance is limited by
Figure 3:
air-restrictor(s) upstream of the
Visible changes to the FSI engine
air intake system(s) and by addi-
compared to the basic engine
tional electronically controlled
boost limits for turbocharged
engines. The size of the restrictors
depends on the engine capacity
and technical complexity. For the
Audi 3.6-litre twin turbo engine,
the limits were defined by two air
restrictors each with a diameter
of 32.4 mm and by a maximum
boost of 1.67 bar.
All design and development
efforts were focused on the
requirements to be met by a suc-
The Basic Engine
•
One combined water-cooled
cessful Le Mans engine. Besides
oil radiator for the engine and
competitive performance and
The above-mentioned restric-
gearbox integrated into the V
high torque over all speed
tions have been in force since
ranges, reliability and low con-
2000 on engines with a capacity
The maximum performance
sumption also play a greater role
of 3.6 litres, two turbochargers
of this basic Multipoint
than in other races. A further
and 4 valves per cylinder
Injection (MPI) engine, which
important aspect is quick engine
installed in an open Le Mans pro-
was restricted as described
response and good driveability
totype car.
above, was 610 bhp at 6300 rpm.
under all weather and race track
The Audi engine was designed
The maximum torque of more
conditions.
as a fully stressed 90 degree V8
than 700 Nm was reached at
The basic engine, which was
DOHC unit with the following
5500 rpm. The usable speed
equipped with conventional
specifications:
range was between 3500 rpm
technology, had already reached
•
Bore: 85 mm
and 7800 rpm. The range with
the limits of its possibilities stip-
•
Stroke: 79.2 mm
90 litres of fuel (maximum per-
ulated by the rules. Only a com-
•
Compression ratio, e:
>11
mitted tank capacity) under race
conditions at Le Mans was 12 –
13 laps (one lap : 13.85 km).
Figure 4:
Geometrical array of
Engine Parameters
the fuel sprays of a
six-hole injector.
The choice of the injection
Only a completely
process was determined to a
new technical
large extent by the character of a
approach was able
race lap at Le Mans. Race analy-
to achieve the nec-
ses, Figure 2, showed that the
essary advantage in
track requires more than 70 % of
order to ensure
full load from an engine. A strat-
future victories.
ified mixture was not seen as an
effective mixture process, con-
sidering the absence of part load
sections. Therefore, a homoge-
neous mixture was seen as the
best solution for increasing both
performance and efficiency at
pletely new technical approach
•
Cylinder head in aluminium
the same time. Of the three pos-
was able to achieve the neces-
with a pent roof combustion
sible direct injection processes
sary advantage in order to
chamber
(spray-guided, wall-guided and
ensure future victories. Fuel
•
Cylinder block in aluminium
air-guided), the air-guided sys-
straight injection (FSI) was
with a closed deck and
tem offered the best possibilities
regarded as the technological
nikasil“-coated cylinders
and required the smallest modi-
key. In January 2000, design
without liners
fications to the mechanical sys-
work began on the new FSI
•
Bedplate in investment cast-
tem of the engine, Figure 3.
engine planned for the race in
ing including a sophisticated
The main target was to
2001. The first run on the test
scavenging system.
achieve the maximum perform-
bench was scheduled for
•
Eight-stage gearwheel-driven
ance possible with the available
November 2000 at the earliest,
oil scavenge and feed pumps
air. Therefore, the mixture of air
due to the lead time required.
•
Two independent water
and fuel was to be as homoge-
The development had to be com-
pumps in line with the oil
neous as possible with a ratio of
pleted by April at the latest.
pumps
between λ = 0.85 and λ = 1.0 .
AutoTechnology 3/ 2002
55

---

C o v e r S t o r y
Figure 5:
The extremely short test time
of the basic cylinder head. The
the shape of the combustion
The CFD calcula-
required the development a
intake port includes an angle of
chamber, the position of the spark
tion was carried
sophisticated decision matrix in
almost 45 degrees.
plug and the protruding intake
out in order to
order to find the best combina-
valves, Figure 4.
gain information
tion of all the following parame-
In order to ensure sufficient
about the
ters:
The central axis of the
cooling of the injector itself and of
mixing process
•
the shape of the combustion
injector
the injector tip, the water flow
between the
chamber
through the cylinder block and the
injected fuel and
•
the number and position of
The central axis of the injector
cylinder head had to be modified.
the air forced
the sprays/injectors
had to be as close and as parallel
The modifications were based on
into the tumble
•
the spray angles
as possible to the centreline of
CFD calculations. The new water
motion.
•
the cooling of the injectors
the port. This made it possible to
jacket of the crankcase even
and injector tips
minimise the bend angles of the
allowed the adjustment of the
•
the fuel mass flow through
different sprays and, as a result,
main stream without any addition-
the injectors
the losses of the required mass
al machining. In the cylinder head,
•
the fuel mass flow and pres-
fuel flow. Furthermore, fuel was
water flow was directed more
sure generated by the fuel
to be prevented from making
towards the area of the injector
pumps in relation to the
contact with the cylinder walls.
between the two intake valve seats.
engine speed and load
The idea of using the exhaust side
The use of an already existing
•
the compression ratio
as a possible location for the
swirl injector was not possible, due
•
the inlet ports with different
injector had been dropped for
to the limited mass flow rate of 15
flow characteristics and dif-
safety reasons. Different spray
cm3/sec, as the performance of the
ferent tumble ratios.
angles were defined considering
engine required a mass flow high-
In addition, many assump-
tions had to be made due to a
lack of knowledge and experi-
ence of this technology in such
an engine.
The combustion chamber was
to remain as similar as possible to
the basic one. Therefore, all
investigations were based on the
existing 4-valve pent roof cham-
ber combined with an almost flat
piston crown. The determination
Figure 6: The optimum interaction between the spray geometry and the tumble is
of the injector position and orien-
the keyto achieving optimum homogenisation. This figure shows two different
tation was guided by the design
examples.
56
AutoTechnology 3/ 2002

---

C o v e r S t o r y
er than 25 cm3/sec. However, the variation of the injec-
tion phase at 6000 rpm and above required a shortening
of the injection time and, as a consequence, an increase
in that value even up to 50 cm3/sec. In order to achieve
such injection times in the range of a few milliseconds, a
much higher electrical performance than that offered by
the MPI low-pressure injectors had to be provided.
Therefore, external amplifiers were added to the original
Bosch MS 2.9 ECU.
Figure 7: Audi R8 with V8 FSI twin turbo engine
The fuel pressure generated by two three-piston
high-pressure pumps was limited to 110 bar and adjust-
ed by an electronically controlled valve at each fuel
rail. In order to avoid any possible bottlenecks, the first
test runs were performed with an electric drive for the
pumps. This step made it possible to determine the
boundary conditions for the fuel system independently
of the engine run. Furthermore, it was therefore possi-
ble to define the capacity of the pumps in accordance
with the number of revolutions. However, the two
pumps were later to be driven by the intake camshafts.
As a result, the fuel mass flow of each pump had to be
increased from 0.45 cm 3/rev to 0.65 cm 3/rev.
The optimum compression ratio could only be
determined in test runs. For this purpose, a number of
pistons with different piston crowns in combination
with corresponding cylinder head gaskets were pre-
pared. This made it possible to increase the compres-
sion ratio to an extremely high level without any
restrictions of the camshaft timing.
In order to define an inlet port with optimum flow
characteristics within a very short time, two parallel
approaches were pursued. The first involved the prepa-
ration of a number of cylinder heads with different port
designs in order to obtain results from running engines.
The second approach was to carry out CFD calculations
and simulations. This task was carried out by Cosworth
Technology, a subsidiary of Audi AG. The calculations
were based on port data from the flow bench, as well as
on injector data provided by Bosch and on the results
of a one-dimensional non-stationary calculation. An
example of one of these CFD runs is shown in Figure 5.
The CFD calculation was carried out in order to
gain information about the mixing process between the
injected fuel and the air forced into the tumble motion.
Figure 5 shows the velocity and distribution of the fuel
droplets in relation to the crankshaft angle in steps of
30 degree starting and ending at BTDC. The results are
AutoTechnology 3/ 2002
57

---

C o v e r s t o r y
Figure 8:
1000 km Without Any
Detail image of
Problems
the Bosch high-
pressure injection
pump on the
In February 2001, the first test in
cylinder head.
a car was performed for more
than 1000 km without any major
problems. This meant that it was
already possible to start adapting
the part load maps. The develop-
ment programme on the test
bench continued parallel to the
road tests, with the focus now on
increasing the power and torque.
λ was raised to 0.93 and above
and the ignition angles were opti-
mised under qualifying condi-
tions. For the first test bench
runs, λ had been fixed at around
0.90 to avoid any possible knock-
The optimum
based on model calculations, tak-
was increased to a value above
ing damage. Lean maps with λ
interaction
ing into account such aspects as
12. Due to the increased efficien-
values above 1.0 were created for
between the
the wall effects between the
cy, the exhaust temperature fell
race periods with lower power
spray geometry
droplets and the piston surface, as
by 50 °C. Therefore, the insula-
requirements (pace car, yellow
and the tumble
well as the vaporization speed
tion of the exhaust system and
periods) and were successfully
is the key.
and the required heat transfer. The
the turbine housing had to be
tested. The durability of the new
influence of the tumble is clearly
improved in order to maintain a
engine was checked under race
visible at 150 degrees. Evaporated
sufficient energy supply to the
conditions in the car and on the
droplets are no longer depicted.
turbine at lower engine speeds,
test bench in several test runs
The optimum interaction
Figure 7.
over periods of more than 30
between the spray geometry and
The next development step
hours.
the tumble is the key to achieving
was the adaptation of the engine
optimum homogenisation. Two
to the car. The main requirements
Results
different examples are shown in
were autonomous starting, an
Figure 6 . In the image on the left,
instantaneous reaction to dynam-
In the Le Mans Pretest in May
a very lean mixture (blue region)
ic processes and the ability to shift
2001, all results previously
around the spark plug inhibits
gears up and down at full throttle.
achieved on the test benches in
safe ignition and good combus-
Fuel pressure during the runs var-
Ingolstadt and Neckarsulm were
tion. The image on the right, on
ied between 40 and 100 bar. In
proved under real Le Mans condi-
the other hand, shows good
contrast, when the engine was
tions:
homogenisation (green region)
started, only a limited fuel pres-
•
an increase in performance by
almost everywhere in the com-
sure of 8 bar was available, and
up to 9 % between 3000 rpm
bustion chamber. This was
was provided by an electrical in-
and 8000 rpm
achieved by using a six-hole
tank pump. Nevertheless, it was
•
a reduction in fuel consump-
injector in combination with an
possible to start the engine with-
tion by up to 8 – 10 % result-
inlet port with a tumble value of
out any delay under cold condi-
ing in at least one additional
almost 4. These calculations were
tions and, even more important
lap at Le Mans between two
subsequently verified in test runs.
for the race, under hot conditions.
pit stops for refuelling
The mapping of the engine on
• excellent driveability
The Test Programme
the test bench convincingly
proved the expected benefits of
The Le Mans 24-hour race in
In November 2000, the first FSI
the FSI technology. No additional
June 2001 finished with a con-
engine was fired up on the test
fuel was required during accelera-
vincing victory for the FSI tech-
bench at the Audi Neckarsulm
tion to avoid knocking. The real
nology. Under extremely difficult
plant. To begin with, a hybrid
time reaction of power require-
rainy conditions, Audi once again
injection system was used. A
ments during gear changes
impressively confirmed its slogan
conventional MPI system was
reduced the gear shifting time.
of "Vorsprung durch Technik".
installed parallel to the FSI. This
Furthermore, the FSI technology
allowed the development steps to
also made it possible to control
be compared with the actual type
highly dynamic transition func-
of engine which had powered the
tions. The possibility of a cycle-
We would like to thank Dr. Ullrich,
Audi R 8‘s to a one-two-three
to-cycle adjustment of the injec-
H. Diel , W. Kotauschek and E. Weil
victory in June 2000.
tion resulted in an additional
for their critical reading and helpful
After the optimisation of the
considerable reduction in fuel
discussions.
fuel pump, the compression ratio
consumption.
58
AutoTechnology 3/ 2002

---