Downloaded from SAE International by Eindhoven University of Technology, Thursday, November 26, 2020

2009-32-0045 / 20097045

Simulation and experimental investigations of a direct-

injection combustion system for high speed – high
performance engines
Stephan SCHMIDT, Martin ABART, Oliver SCHÖGL, Dalibor JAJEVI,
Alexander TRATTNER, Roland KIRCHBERGER, Helmut EICHLSEDER
Graz University of Technology, Institute for Internal Combustion Engines and Thermodynamics

Copyright © 2009 SAE Japan and Copyright © 2009 SAE International

ABSTRACT x Short development cycles

x Increasingly strict emission regulations
This publication presents the development of a GDI
combustion system for high speed – high performance The following table gives an evaluation of the pros and
engines. The paper describes the development of the cons of GDI application in high performance engines
combustion process and its results, the development compared to external mixture preparation.
methodology, covering 1D and 3D CFD simulations,
Table 1: Evaluation of a GDI concept for recreational
and shows extensive experimental investigations. With
vehicles
3D CFD simulations the necessary deeper insights in
Criteria Evaluation
the mixture preparation process of a homogenous GDI
Performance improvement +
combustion system are gained.
The findings of the simulation are transferred to an Fuel consumption reduction +
injection and combustion system for a 4-cylinder test Scavenging losses reduction +
engine. The results of the experimental investigations Response +
of the 4-cylinder engine equipped with GDI technology
Drivability – Combustion stability at low +
show the potential of GDI applications for high
performance engines. speed and low load
Emission reduction +/-
INTRODUCTION Marketing (technological lead) +
Developments of modern combustion systems have to Costs -
be made against the background of the shortage of
In the course of a research project at the Institute for
energy resources and increasing restrictions of CO2-
Internal Combustion Engines and Thermodynamics at
emissions. Today, these demands are also related to
Graz University of Technology a GDI concept engine
high performance engines for recreational vehicles.
has been investigated. This paper presents the results
One approach for the reduction of CO2-emissions is
of a conceptual development of a GDI combustion
gasoline direct injection. This technology allows for a
system including methodology, simulation results and
reduction of fuel consumption and basically also for an
experimental outcomes.
increase of power [1].
While GDI is already used in series production
In the first section the paper shortly describes the used
passenger cars, the demand for high performance
methodology for the research and development of the
engines in the field of recreational vehicles accounts
combustion system. The differences to standard
for the development of a novel combustion system.
process structures and the set of building blocks for
The development faces the following boundary
this new integrated development process are
conditions and challenges:
described. The second section explains the
x The technological challenge of such a
combustion system concept, the boundary conditions
combustion system regarding the preservation
of the conducted investigations as well as the
of a satisfactory mixture preparation
respective results and findings with the following foci:
considering the big span between low speed
and low load and high speed and high torque x First, a description of the characteristics of the
x The layout conflict of high performance investigated test carrier including the essential
engines between emission cycle operating data of the basic engine is given and the
points and customer demands significant boundary conditions are explored.
x The specific demands of recreational vehicles x Then, the relevant physical processes for
regarding highest specific power output while internal mixture preparation are discussed via
maintaining low system complexity a description of the specific requirements of
SETC2009
 Downloaded from SAE International by Eindhoven University of Technology, Thursday, November 26, 2020

high speed engines for the development of a PRELIMINARY INVESTIGATION

GDI combustion system.
The preliminary investigations comprise a definition
x Finally, results of the conducted 1D and 3D
and an evaluation of the existing boundary conditions
CFD simulations and the corresponding
using CFD and CAD tools. This step ideally results in a
experimental findings are assessed by
thorough understanding of the respective system
focusing on the investigation and
demands providing the relevant input information for
understanding of the effects of different
the successive intrinsic development process. It
parameters on a high rpm combustion system.
delivers requirements on the engine concept itself as
The potential as well as limits of GDI
well as on the type and range of necessary
combustion systems for high performance
experimental and simulation investigations.
engines are evaluated.
Assumed that a close understanding of the influencing
parameters of a combustion system is available, the
DEVELOPMENT & RESEARCH listing of the requirements results in a very precise
METHODOLOGY “development need” – providing very specific
The development process of a combustion system is development targets already at a very early
very complex and full of interdependencies demanding development stage.
numerous tools and methods as well as a large
number of models. [ 2 ] To facilitate an efficient, SIMULATION AND EXPERIMENTAL
effective and consistent development process, a new INVESTIGATIONS
integrated development methodology using a modular
approach has been applied. For the development of the combustion system, 1D
The methodology is based on an intensive integration CFD and 3D CFD simulations are essential tools.
of simulation processes and their respective Based on the preliminary investigations, the
experimental methods [3]. It aims at an optimization of application of CFD simulation can be used for the
the consistency and efficiency of the development analysis of charge motion, mixture preparation, spray-
processes by replacing serial with parallel processes. wall interaction, different injection strategies, etc.
Typical development processes are characterized by
successive activities linked through optimization loops Experimental investigations have to be conducted for
as illustrated on the left side of Figure 1. This leads to the adjustment of the used simulation models. Due to
a time consuming R&D process where initial the complex physical phenomena during the mixture
assumptions can have big impact on the relevance of preparation process of high speed engines, the 3D
the research result. CFD simulation with turbulence modeling, its spray
breakup and wall interaction models has to be verified
by special experimental investigations (using flow
Serial process structure Integrated process structure bench, spray chambers, etc.). Additionally,
Boundary Conditions Boundary Conditions measurement results are necessary for the evaluation
of the simulation procedure and outcome, not only for
the final stages of a development project, but also for
Simulation Simulation Geometrical
Pre-investigations Investigations
the early and intermediate project phases.
COMBUSTION SYSTEM CONCEPT –
System Analysis/Description BOUNDARY CONDITIONS
Experimental
Investigations
Simulation Experimental BASIC ENGINE
Investigations
The described investigation of a combustion system
Development stage

Development stage

concept has been conducted on an existing naturally

Results
aspirated 4-cylinder high performance engine with
external mixture preparation and a cylinder
ok displacement of ~400 cm³. Table 2 gives a listing of
the basic engine parameters.
Results
ok
Figure 1: Research process structure
The application of an integrated process structure is
based on detailed preliminary investigations of the
underlying boundary conditions resulting in a profound
understanding of the examined system. The
development process, as shown on the right side of
Figure 1, consists of 3 major parts, with a close
integration of simulation and experimental
investigations. The single process sections are
described in the following:
SETC2009
 Downloaded from SAE International by Eindhoven University of Technology, Thursday, November 26, 2020

Table 2: Engine data

Cylinders 4 / inline
Specific power ~100 kW/liter
Maximum BMEP 13.5 bar
Maximum engine speed 11000 rpm
Bore/stroke ratio ~ 1.35
Compression ratio 13

BOUNDARY CONDITIONS

The given geometrical and technical boundary

conditions have been used as input data for Figure 3: Section through the cylinder middle plane
preliminary geometric and CFD investigations. with injector and intake port
Possible system concepts have been analyzed using
first geometric studies for a clarification of the available COMBUSTION SYSTEM CONCEPT –
possibilities i.e. for injector positions. 3-D CFD
simulations without real world boundary conditions for
REQUIREMENTS
the intake were performed in order to visualize several
spray targets for the solenoid multihole injector and the In order to design a suitable combustion concept, the
interaction between various piston bowl geometries information out of previous experiences and out of the
and the spray. first investigations has to be condensed to a list of
requirements and the definition of the relevant physical
The existing restrictions resulted in a decision for a processes (next chapter). As already stated, the
homogenous air-wall guided GDI concept with an technological challenge of a combustion system for
injector position beneath the intake port (Figure 2 and high performance engines mainly consist in the
Figure 3). Other positions of the injector were ruled out following demands:
due to geometrical constraints. Beside the
performance criteria, the decision for a homogenous 1.) The preservation of a satisfactory mixture
stoichiometric combustion concept was also based on preparation considering the big span between low
the demand for a less complex exhaust gas speed and low load and high speed and high
aftertreatment as well as a low system complexity. torque. Whereas the low charge motion level
hinders the mixture preparation at low load and
low speed, the short time span for fuel injection
and mixture homogenisation poses difficulties at
high engine speed. In Figure 4 the variation of the
turbulent kinetic energy and the available time for
mixture preparation shows the opposing relevant
trends. [4]

ETkin t

Figure 2: Sectional views of the concept engine in 3D Engine speed

CAD
Figure 4: Variation of turbulent kinetic energy and
available mixture preparation time over engine speed
2.) The layout conflict of high performance engines
between fuel efficiency and emission targets at
cycle relevant operating points (~2000rpm, <2bar
BMEP) and the costumer need for performance
optimization at full load at high engine speed
(>10000rpm, >12bar BMEP).
3.) The specific demands of power sport vehicles
regarding highest specific power output while
maintaining low system complexity basically favor
a homogenous GDI concept. Expensive
components such as Piezo-injectors or lean
exhaust gas aftertreatment are unjustifiable,
SETC2009
 Downloaded from SAE International by Eindhoven University of Technology, Thursday, November 26, 2020

considering vehicle prices around 10.000- 1) Break up and evaporation of the injected fuel
15.000€. mass: For a good mixture preparation, fast fuel
4.) Increasingly strict emission regulations specifying evaporation is necessary to guarantee a
exhaust emission levels similar to today’s homogenous stoichiometric mixture at ignition
passenger cars (EURO IV and V – sooner or later timing.
including OBD) intensify the above mentioned
2) Transport of the liquid and gaseous fuel
conflicts. In medium and long term, new emission
fractions in the combustion chamber: Although
concepts will be necessary whereby internal
the mixture preparation concept is intended to
mixture preparation can be a useful support.
be homogenous and quantity controlled, in
Furthermore, with EU V (~2025) a legal obligation
some operating points non-homogenous
for an onboard diagnosis system is probable -
zones and zones with residual gas can occur.
adding to the overall system complexity. Beside
In order to guarantee an ignitable fuel mixture
the emission topic, it can be expected that the
around the spark plug as well as high
pressure to reduce CO2 emissions will also
combustion stability, the injected fuel mass
increase for recreational vehicles.
has to be dispersed and guided through a
5.) The demand for short development cycles. To
directed air motion and/or a shaping of the
compete successfully on the market, short
piston top.
innovation cycles are increasingly necessary.
Therefore, an efficient and effective use of the 3) Combustion of the air-fuel mixture: For an
development tools is of great importance for time efficient and stable combustion over the whole
saving combustion development, especially under speed and load range (short ignition delay and
the aspect of increasing demands on the combustion durations) a high degree of
combustion system (see above). turbulent kinetic energy is necessary.
These physical processes can be described by
Derived from the mentioned specific requirements,
respective models as function of the underlying
combustion system-relevant questions have been
physical variables [2]. For the development of a
defined to help to evaluate the potential of GDI with
combustion system it is of great importance to
respect to the stated technological challenges. Beside
understand the system relevant models and
questions concerning the overall engine performance
parameters – on the one hand, to avoid a
(BMEP, BSFC, emissions, etc.), which will be
misinterpretation of the simulation results, and on the
separately treated (see chapter “Performance and
other hand, to allow an understanding for the
results”), the presented findings (see chapter “Findings
improvement of an existing combustion system.
from experimental and simulation investigations”)
Additionally, the correct choice of adequate
focus on the following questions:
experimental investigations for the calibration of the
simulation models is based on the factual knowledge
x Is it possible to improve the performance of about the processes.
the concept engine with internal mixture
preparation utilizing increased volumetric
TURBULENCE MODEL
efficiency and/or advantageous knocking
behavior?
A high degree of turbulent kinetic energy of the charge
x Is it possible to facilitate a satisfactory mixture motion is not only important for a good mixture
formation and combustion over the whole preparation, but also for an intense and
engine speed and load range and to improve thermodynamically efficient combustion leading to an
the combustion stability at low speed and low ideal premixed flame. The degree of turbulence of a
load compared to external mixture fluid flow can be described by the turbulent Reynolds-
preparation? Are modifications of the Number Ret using turbulent length and time scales
combustion chamber, using optimized piston
(with k, the kinetic energy, li the integral length, Q the
bowl shapes, sufficient beneficial regarding
mean kinematic viscosity and v the mean fluctuation
mixture preparation?
velocity) [2]:
x Is it possible to reduce the scavenging losses
by late direct injection compared to MPFI?
x Is it possible to improve the mixture formation 2k li vcli
Ret Equation 1: Turbulent Reynolds number
through double injection strategies? [5] Q Q
COMBUSTION SYSTEM CONCEPT – With a turbulent Reynolds-Number above 1, turbulent
RELEVANT PHYSICAL PROCESSES fluctuations, which can be modeled with several
approaches differing in complexity and detailedness,
occur. For 3D CFD simulations usually standard
Preliminary investigations and existing research
results explicitly show that the following three turbulence models like the k-H model are used. These
processes are of major importance for mixture fluctuations also cause a transport of momentum,
preparation and combustion (boundary condition: mass, and energy across the flow direction. Hence,
homogenous air-wall guided GDI concept). The they contribute decisively to the mixture formation and
understanding of these processes is of major combustion process. The turbulent kinetic energy is a
importance when using simulation tools. measure for its intensity. To achieve a high degree of

SETC2009
 Downloaded from SAE International by Eindhoven University of Technology, Thursday, November 26, 2020

turbulent kinetic energy, high inflow velocities of the injection for diesel combustion systems neglecting film
charge and squish-induced flow are supportive. High effects. Hence, it is clear that for the following reasons,
inflow velocities mainly depend on the piston speed the existing models have to be adapted for a GDI
and the design of the intake port. A tumble or swirl port combustion system:
allows to add a directed charge motion, further
- Significantly lower injection pressures
supporting mixture preparation and transportation.
compared to diesel DI engines.
- Significantly higher injection pressures
SPRAY BREAK UP MODELS
compared to MPFI gasoline engines.
The Weber-Number, which is dependent on the - Significantly higher wall temperatures
density U, the relative flow velocity vrel, the compared to the intake ports of MPFI gasoline
characteristic diameter dT and the surface tension V, engines.
characterizes the break up of a fuel spray – the higher Especially the influence of the surface temperature is
the number, the more intense the atomisation is: of main importance, due to temperature dependent
Uv rel ² d T physical surface phenomena, such as the Leidenfrost
We Equation 2: Weber number effect. [7], [8]
V
The relevant parameter for an intense break up is the The setting of the parameters of the spray-wall
relative velocity between the fuel spray and the air interaction model (calibration) can be conducted in a
motion. High velocities can be reached via a high spray chamber [ 9 ] using different simplified piston-
injection pressure as well as an intense counter flow of bowl geometries. Beside the influence of the bowl
the charge motion. For the simulation of the spray geometry, also the effects of the wall temperature and
break up, numerous models exist (e.g. Wave, the injection pressure have to be investigated. Figure 5
HuhGosman, etc.). Beside the choice of the correct shows a comparison of a spray chamber recorder data
model, a good adjustment of the model is decisive. (top picture) and the respective simulation results of
Therefore spray chamber measurements are a two different calibrated interaction models (middle and
necessity. bottom picture). The upper simulation picture shows a
rather good correlation with the experimental results,
SPRAY EVAPORATION MODELS whereas the lower one differs significantly.

The evaporation of the fuel mass depends on the heat

flux over the system boundary between the enclosing
charge and the liquid fuel. Based on the evaporation
model of Dukowicz [6] the mass flux can be written as:

dmd f
Q vs Equation 3: Evaporating mass flux
dt qs
The mass flux is dependent on the heat flux Q , the
local surface heat flux q s , and the vapour mass
flux fvs . High temperatures of the surrounding gas, a
large surface (small droplets) and high relative
velocities support a fast evaporation.

WALL INTERACTION MODELS

Although the ideal conception of a modern internal

mixture preparation is that the injected fuel vaporizes
before getting on the wall, this condition can occur
during engine operation. Depending on several
parameters, such as droplet velocity, diameter, surface Figure 5: Calibration of spray-wall interaction model
temperature, surface roughness etc., the behavior of a Figure 6 depicts simulation results for two different wall
droplet - wall interaction may differ between simply interaction models showing the different spray
sticking to the wall, rebounding or splashing. Standard propagation on the piston surface. It is obvious, that
models for the description of these effects can be these uncertainties about the spray transport are
basically divided into wall jet and wall film models /6/. unacceptable for the layout of a GDI combustion
Originally, wall film models have been developed to concept, especially for the optimization of the piston-
describe the film formation in the intake port at bowl geometry. Therefore, an adjustment and
relatively low wall temperatures and low injection verification of the spray wall model using experimental
pressures. Wall jet models originate from the methods is required.
description of high pressure (up to +1600bar) direct
SETC2009
 Downloaded from SAE International by Eindhoven University of Technology, Thursday, November 26, 2020

ACHIEVABLE ENGINE PERFORMANCE AT WOT

The original engine with series configuration and port

fuel injection is used as basis for the comparison and
evaluation of the GDI combustion concept. Figure 7
depicts a comparison of the obtained WOT
measurement results from the engine test bench –
showing results of the GDI concept engine and the
reference measurement of the MPFI basis from 2000-
8000rpm. The figure shows results of the GDI concept
with a series production piston shape (without piston
bowl).
BMEP MPFI

specific power [kW/dm³]

17 120
BMEP GDI
16 spec. power MPFI 100
spec. power GDI
15 80
BSFC GDI
14 60
13 40
Figure 6: Spray propagation for different wall 12 20

BMEP [bar]
interaction models (operating point: 2000 rpm, 2bar 11 0
BMEP – at identical operating parameters) 10 110

Rel_BSFC [% of MPFI]
9 105
8 100
COMBUSTION SYSTEM CONCEPT -
7 95
DEVELOPMENT 6 90
5 85
The development of the combustion concept based on 4 80
the findings of the first investigations and the given 2000 3000 4000 5000 6000 7000 8000
Engine speed [rpm]
boundary conditions.
Figure 7: System comparison GDI - MPFI (BMEP,
With these data transient 3D-CFD simulations with
BSFC, Power)
moving piston and real world boundary conditions
were performed in order to investigate different
injection strategies, spray targets and injection BMEP/Engine Power at WOT
pressures as well as different combustion chamber
geometries including piston bowl variants. Evaluations Over a broad engine speed range the BMEP could be
of the simulation tools for these applications were raised with GDI while maintaining the maximum
conducted as described before. Additionally these engine power. Especially at low engine speed, the
optical experimental investigations of the interaction of results show a significant advantage concerning
piston bowl shapes and the injection spray were used internal mixture preparation. For engine speeds above
to intensify the understanding of the physical 8000rpm similar BMEP values compared to external
processes. mixture preparation can be achieved.
In parallel first experimental investigations were
performed with the basic engine concept with external BSFC at WOT
mixture preparation and these results were used for an
adjustment of the simulation and for an extraction of Especially at low engine speeds between 2000 and
boundary conditions. 4000rpm as well as higher engine speeds around
Out of many variants of piston bowl shapes, spray 7000 to 8000rpm a huge improvement regarding fuel
targets and injector positions, two piston bowl efficiency can be achieved. This efficiency gain is
concepts together with one injector position were mostly due to the higher power output whilst keeping
chosen for prototyping and subsequent experimental similar fuel mass flow (Figure 8). In addition, a leaner
investigations. The first investigations with internal mixture could be applied to several operating points
mixture preparation were performed with the original without knocking problems. On the contrary, high
piston shape in order to achieve an early adjustment specific HC emissions at low engine speed could be
between simulation and experiment. observed, preventing even higher fuel savings (Figure
In the following the results of the combustion concept 9).
development are presented.

COMBUSTION SYSTEM CONCEPT -

PERFORMANCE AND RESULTS

The following results have been gained with a basic

GDI setup with series piston shapes (no piston bowl).
The ignition timing and air aspect ratio have been
adapted for GDI. The injection timing was set towards
BMEP optimum.

SETC2009
 Downloaded from SAE International by Eindhoven University of Technology, Thursday, November 26, 2020

equivalence ratio MPFI

equivalence ratio GDI 5% MBF timing MPFI
40

Equivalence ratio [-]

BMEP MPFI 5% MBF timing GDI
BMEP GDI 50% MBF timing MPFI
1.3 delivery ratio MPFI 1.08 35 50% MBF timing GDI
delivery ratio GDI 90% MBF timing MPFI
1.2 1.04 30 90% MBF timing GDI

Crank angle degress - ATDC

1.1 1.00

Volumetric efficiency [-]

25
1.0 0.96
20
0.92
15
0.88

14 0.84 10
BMEP [bar]

12 0.80 5
10 0.76
0
8 0.72
-5
6 0.68
2000 3000 4000 5000 6000 7000 8000 -10
Engine speed [rpm] 2000 3000 4000 5000 6000 7000 8000
Engine speed [rpm]
Figure 8: System comparison GDI – MPFI
(equivalence ratio, delivery ratio) Figure 10: System comparison GDI - MPFI (heat
release)

Emissions at WOT
COMBUSTION SYSTEM CONCEPT –
As expected, the specific NOx emissions of the MPFI FINDINGS FROM EXPERIMENTAL AND
show an increase due to a faster combustion and SIMULATION INVESTIGATIONS
therefore higher combustion temperatures. Due to the
rich mixture of the original MPFI, the external mixture In the following section, the findings of the combustion
preparation results in very low NOx emissions which system concept are discussed by means of the above
lead to much higher relative NOx emissions. This is stated questions.
true especially in speed ranges where the MPFI
engine is richer than the GDI. HC emissions showed IS IT POSSIBLE TO IMPROVE THE ENGINE
an increase which is mainly caused by scavenging PERFORMANCE WITH INTERNAL MIXTURE
losses due to the necessary early start of injection of PREPARATION?
this piston variant. For most of the operating points the
observed specific CO emissions are below the MPFI As seen before (Figure 7) the results show a
reference. significant advantage regarding the engine
performance at WOT with GDI. The improvement is
equivalence ratio MPFI
equivalence ratio GDI mainly based on two reasons:
NOx emission GDI
Equivalence ratio [-]

CO emission GDI
1.3 HC emission GDI 1000 - Advantages regarding ignition timing and
1.2 900 ignition delay: For most of the operating points
Rel. Emissions [% of MPFI]

1.1 800 the use of GDI allows an earlier 50% MBF

1.0 700 compared to external mixture preparation due
0.9 600 to advantages regarding knocking behavior
500 (Figure 10).
400
30 9.6
300
BMEP
50% MBF timing
MPFI Series Ign. Angle
200 27 9.4
100
50% MBF timing [CA ATDC]

24 9.2
0
2000 3000 4000 5000 6000 7000 8000
Engine speed [rpm] 21 9.0 BMEP [bar]

Figure 9: System comparison GDI – MPFI (emission) 18 8.8

15 8.6

12 MPFI 8.4

9 8.2
GDI
6 8.0
13 14 15 16 17 18 19 20 21 22 23
Ignition angle [CA BTDC]
Figure 11: Ignition timing with GDI (2000rpm, WOT)
As Figure 11 shows, the series ignition timing at
2000rpm WOT of the MPFI setup (~18° before TDC)
can be put more then 3° earlier with GDI due to a
better knocking behavior - resulting in an 8° earlier
center of combustion due to further advantages
regarding ignition delay and combustion duration.

SETC2009
 Downloaded from SAE International by Eindhoven University of Technology, Thursday, November 26, 2020

Thus the operating range (which is restricted at earlier

ignition angles by the knocking tendency) of the GDI
setup (green in Figure 11) can be extended in contrast
to the one of the MPFI (orange in Figure 11).
- Advantages due to increased volumetric
efficiency: With internal charge cooling an
increase of the intake air mass up to 4% is
measurable (Figure 8) similar to earlier
findings [10].
Also for WOT engine operating points above 8000rpm
the obtained test bench results show a satisfactory
engine performance. Figure 12 depicts a comparison
of BMEP measurement result of the above described
basic GDI setup and a BMEP curve for an adapted Figure 13: Charge velocity [m/s] - low rpm, low load
GDI setup (adjusted piston bowl and injector spray
pattern). The optimization focus for this GDI setup was
put on engine speed above 5000rpm. Thus, BMEP
values at low engine speeds are without relevance. At
high engine speeds the adjusted GDI setup shows
remarkable performance results with a further BMEP
increase at 8000rpm and a maximum BMEP at
8500rpm of around 13.4bar and a specific power
output of ~102kW/dm³ at 10000rpm.

14

13

12
BMEP [bar]

11
Figure 14: Charge velocity [m/s] - high rpm, WOT
10 base GDI

9
modified GDI Nevertheless, the simulation as well as experimental
results showed a good performance of the GDI
8
combustion system at low engine operating points,
7
founded in a significant contribution to the kinetic
6 energy level of the injection compared to the
2000 3000 4000 5000 6000 7000 8000 9000 10000
underlying charge motion. By using the series piston
Engine speed [rpm]
geometry, better combustion stability compared to
Figure 12: GDI engine performance at high rpm MPFI could already be reached (Figure 15). Even
more significant advantages could be observed by the
IS IT POSSIBLE TO FACILITATE A use of a piston bowl.
SATSIFACTORARY MIXTURE FORMATION AND
COMBUSTION OVER THE WHOLE ENGINE SPEED Figure 15 shows an improvement of the combustion
AND LOAD RANGE? stability (evaluated by the Coefficient of Variation –
COV) by a factor of 2 with GDI and the use of a piston
bowl compared to external mixture preparation.
Low engine speeds and low load Although the absolute values are high compared to
automotive standard, a reduction of the COV
The preliminary 3D-CFD simulation results already significantly below 10% is a good result for a high
revealed possible “problem-areas” regarding mixture performance motorcycle engine with a speed range
formation. Due to the layout of the intake ports for from 1500 to 12.000rpm and a possible BMEP range
maximum performance, especially at low engine from 1 to 13.
speed and low load, very little turbulence is generated
in the combustion chamber during the gas exchange
(Figure 13 and Figure 14). Whereas the charge motion
at high speed and WOT reaches velocities above
150m/s, maximum velocities at low speed and low
load are far below 50m/s, posing potential difficulties
for a sufficient mixture preparation.

SETC2009
 Downloaded from SAE International by Eindhoven University of Technology, Thursday, November 26, 2020

similar effects – showing a far higher induced turbulent

14.1 kinetic energy by the use of a piston bowl leading to a
GDI
faster evaporation of the injected fuel. The difference
Coefficient of Variation [%], GDI (with piston bowl)
at 2000rpm - 2bar BMEP
between the curves “turbulent kinetic energy” and the
MPFI
respective curves without the simulation of an injection
8.7 (no INJ) show the contribution of the fuel injection to
7.5 the turbulence level. As a consequence of the
advantages regarding combustion stability and
duration while maintaining the HC emission level of
MPFI, the application of a piston bowl leads to a
decrease of the specific fuel consumption of around
10% at 2000rpm and 2bar BMEP compared to
external mixture preparation.

Figure 15: System comparison GDI - MPFI (COV at

2000rpm, 2bar BMEP) 2000 rpm, start of injection 400 °CA
5.0E-06 25

The basic GDI setup shows significant higher HC

Turbulent kinetic energy

injected/evaporated [kg]
4.0E-06 20
emissions at 2000rpm and 2bar BMEP (Figure 16) due
to the necessity of an early start of injection for a

Fuel mass
3.0E-06 15

[m²/s²]
sufficient mixture preparation resulting in significant
scavenging losses. The application of an adjusted 2.0E-06 10
piston bowl enables a later injection timing avoiding
the extraordinary high scavenging losses while 1.0E-06 5

maintaining a high quality of mixture preparation and

0.0E+00 0
good combustion stability. 400 450 500 550 600 650 700
Crank angle degrees

GDI fuel mass injected fuel mass evaporated - flat piston

spec. exhaust emissions [% of MPFI],

fuel mass evaporated - piston bowl turbulent kinetic energy - flat piston
200% GDI (with piston bowl) turbulent kinetic energy - flat piston (no INJ) turbulent kinetic energy - piston bowl
MPFI turbulent kinetic energy - piston bowle (no INJ)
at 2000rpm - 2bar BMEP

150%
Figure 18: Simulation results - Evaporation of different
GDI concepts (2000rpm, 2bar BMEP)

100%
High engine speed

50% For the evaluation of the mixture preparation the soot

emission and combustion duration have been used as
main experimental criteria. As seen before, the
0%
evaluation of the performance values and the
HC CO NOx
combustion duration show acceptable results also at
Figure 16: System comparison GDI - MPFI (exhaust high engine speeds. Nevertheless, Figure 19 shows
emissions at 2000rpm, 2bar BMEP) unsatisfactory soot emission values for a broad engine
operating range. Since the quality of mixture
preparation does not only depend on the available
time and the prevailing turbulence of the charge
GDI motion (see above – section “Relevant physical
Crank angle degress - ATDC

80 GDI (with piston bowl) processes”) but also on secondary effects, the results
MPFI show a discontinuous characteristic with a noticeable
60 peak at 5000rpm.

40

20

0
5% MBF 50% MBF 90% MBF

Figure 17: System comparison GDI - MPFI (Heat

release at 2000rpm, 2bar BMEP)
Figure 17 depicts advantages of the GDI system with
the piston bowl regarding the combustion durations.
The respective simulation results (Figure 18) reveal
SETC2009
 Downloaded from SAE International by Eindhoven University of Technology, Thursday, November 26, 2020

35

Time between ignition and start

30
Time
Soot 25

of Injection [ms]
Soot emission

20

15

10

0
2000 3000 4000 5000 6000 7000 8000
Engine speed [rpm]
Figure 21: Interaction of injection spray and opening
Figure 19: Soot emissions at WOT intake valve
While sufficient time for the mixture preparation allows
for a homogenous charge at low engine speed, the IS IT POSSIBLE TO REDUCE THE SCAVENGING
intensified charge motion leads to a decrease of the LOSSES BY LATE DIRECT INJECTION?
detected soot emission at high engine speed. In
accordance with the results of the experimental A principal advantage of GDI is the theoretical
investigations, the simulation results also show possibility to avoid scavenging losses by the use of
disadvantages for the mixture preparation at 5000rpm. fuel injection after the exhaust valve closes. The
Figure 20 illustrates the characteristics of the simulation and experimental investigations show that
evaporated and injected fuel mass for different engine at many operating points the fuel injection has to start
speeds. At 5000rpm, the evaporation of the injected already before the exhaust valve closes (EVC at
fuel lasts over a longer time fraction of an engine ~400°CA) to provide sufficient time for the mixture
revolution than at lower or higher engine speed. preparation. Figure 22 shows the BMEP optimized
injection timing over the engine speed at WOT for two
different piston configurations (dashed line: piston with
bowl, continuous line: piston wo bowl). It shows the
Fuel mass injected/evaporated

demand for a start of injection (SOI) long before EVC

for nearly all operating points. The application of an
adjusted piston bowl enables the shifting of the
injection towards later SOI without completely avoiding
scavenging losses.
480
400 450 500 550 600 650 700 750 800
SOI flat piston
Crank angle degrees EOI flat piston
460
SOI piston with bowl
Fuel mass injected 2000rpm Fuel mass evaporated 2000rpm
EOI piston with bowl
Fuel mass injected 5000rpm Fuel mass evaporated 5000rpm 440
Crank angle degrees - ATDC

Fuel mass injected 8000rpm Fuel mass evaporated 8000rpm

Fuel mass injected 11000rpm Fuel mass evaporated 11000rpm 420
EVC
Figure 20: Simulation results of the fuel evaporation for 400

2000-11000rpm at WOT 380

Additionally, an increase of the engine speed also 360

causes secondary disadvantageous effects due to the 340

IVO
relatively long injection period. Since the injection 320
period covers a wide crank angle span, fuel mass also 300
has to be injected at the time of the maximum opening
280
of the intake valves resulting in a valve wetting, as the 2000 3000 4000 5000 6000 7000 8000 9000 10000
simulation results reveal (Figure 21). Moreover, piston Engine speed [rpm]

wetting resulting in pool fires could be detected by Figure 22: Injection timing at WOT for 2 piston
means of optical investigations. Piston wetting is also configurations
dependent on the injection timing, as well as on the For the operating range at lower engine speeds and
respective piston position. As commonly known, part load, the use of a piston bowl allows significantly
effects of wall-wetting inside the combustion chamber later SOI - also later than EVC. A SOI variation at
significantly contribute to soot formation. [11] 2000rpm and constant fuel mass (appr. 2.5bar BMEP)
showed a complete avoidance of scavenging loses
and a minimum of HC emissions at an injection start at
400° after TDC (Figure 23). The remaining HC
emissions are unburnt fuel fractions. The amount of
unburnt HC increases further with SOI later than 400°
ATDC due to insufficient mixture preparation.

SETC2009
 Downloaded from SAE International by Eindhoven University of Technology, Thursday, November 26, 2020

12
25
BMEP

Turbulence Kinetic Energy [m²/s²]

spec. HC emission
10

Fuel mass injected/evaporated

BSFC

at 2000rpm and 2bar BMEP

spec. HC emission
20
8 4.0

6 3.6 15

BMEP [bar]
4 3.2
10
2.8

2.4 5

525 2.0 0
BSFC

400 450 500 550 600 650 700

450
Crank angle degrees
375 fuel mass injected fuel mass evaporated
fuel mass injected - split injection fuel mass evaporated - split injection
turb. kinetic energy turb. kinetic energy - split injection
300
340 350 360 370 380 390 400 410 420 430 440 450 460 470 480
Start of injection [CA] Figure 24: Simulation results split injection - at
Figure 23: SOI variation with constant fuel mass at 2000rpm and 2bar BMEP
2000rpm piston with bowl
500

INFLUENCE OF INJECTION TIMING AND 480

Combustion chamber temperature - split
STRATEGY 460
injection
440 Combustion chamber temperature

Temperature [K]
420
The analysis of the evaporation rate of the injected fuel
400
shows a significant dependency on the injection time.
380
The main influential parameter beside the intensity of
360
the charge motion is the temperature inside the
340
combustion chamber. Both factors have a strong
320
correlation to the crank angle and the piston motion,
300
respectively. For a fast evaporation a high turbulence 400 450 500 550 600 650 700
level as well as a high temperature is favorable. Crank angle degrees - ATDC
Shortly after the top dead center both parameters, Figure 25: Simulation results split injection - at
temperature and kinetic energy, have their maximum 2000rpm and 2bar BMEP
during the intake stroke. Therefore, the earliest
possible fuel injection is principally advantageous for a CONCLUSION AND OUTLOOK
good mixture preparation. For reasons of the spray-
piston interaction and exhaust gas after treatment, In the course of the above described research
different injection timings could be more favorable. activities a GDI combustion system concept for a high
speed high performance engine has been developed
and investigated. The findings focus on the evaluation
of the combustion concept at WOT and part load at
Previous investigations [5] showed the potential of stationary operating points not accounting for possible
double injection strategies for GDI combustion response advantages of a GDI concept compared to
systems at lower loads. Therefore, beside the timing of MPFI.
a single injection, also split injection strategies have
been investigated by means of 3D CFD simulation. The derived results show a significant improvement of
The aim of the investigations is to evaluate the the BMEP characteristic as well as reduced fuel
possibility to induce an additional charge motion consumption at WOT over a wide engine speed range
through a late injection of a part of the fuel mass. The due to an increase of the volumetric efficiency and
results (Figure 24) show effects corresponding to the advantages regarding knocking behavior. The findings
above mentioned principles. Due to the reduced prove the principal applicability of a GDI combustion
temperature and lower kinetic energy level during the system at high engine speeds implicating a sufficient
second part of the injection, the application of a split mixture preparation up to 11000rpm. For low load and
injection results in a worse mixture formation. For late low speed operating points a significant improvement
injection timings around bottom dead center not only of the combustion stability has been achieved with the
the turbulence kinetic energy level is about 4 times use of GDI and the adjustment of the piston geometry.
lower but also the temperature is about 150° lower
(Figure 25), resulting in a significantly slower fuel The main design features of the GDI combustion
evaporation. concept are an optimization of the combustion
chamber including the piston bowl. This is
accompanied by an appropriate choice of the injector
position and angle and an optimization of the spray
target in order to improve the charge motion and the
spray / charge as well as the spray wall interaction.

Although a major step towards pre-mass production

development of a GDI system for high performance
SETC2009
 Downloaded from SAE International by Eindhoven University of Technology, Thursday, November 26, 2020

engines has been done, the investigations confirm MPFI: Multi Point Fuel Injection
some presumed difficulties for the layout of a
NOx: Nitrogen Oxides
combustion system for the wide operating range of
such engines. The results reveal that not all SOI: Start of Injection
advantages of a GDI application can be exploited for
several reasons. One of the major drawbacks is the (A)TDC: (After) Top Dead Center
necessity of an early timing of the fuel injection over a WOT: Wide Open Throttle
broad operating range to guarantee a sufficient
mixture preparation. Hence, on the one hand, it is not REFERENCES
possible to completely avoid scavenging losses, and,
on the other hand, problems such as valve and piston
wetting have to be faced. This situation is worsened by 1 Eichlseder H., Klüting M., Piock W., „Grundlagen
the typical large valve overlapping of high performance und Technologien des Ottomotors“, Springer, Wien,
engines. The late timing of EVC results in a 2008, ISBN 978-3-211-25774-6
development trade-off between an acceptable quality 2 Pischinger R., Klell M., Sams T., „Thermodynamik
of the mixture preparation and a reduction of the der Verbrennungskraftmaschine – Der
exhaust, especially HC, emission. These findings lead Fahrzeugantrieb“, Springer Verlag, 2002
to the necessity of variabilities in the air path such as 3 Schmidt S., Schoegl O., Rothbauer R.J., Eichlseder
variable valve timing or variabilities regarding charge H., Kirchberger R., „An Integrated 3D CFD
motion. These could assist to overcome the stated Simulation Methodology fort he Optimization of the
difficulties. Mixture Preparation of 2-Stroke DI Engines“, SAE
Paper 2007-32-0029 (2007)
Another major challenge for the development of such a 4 Schmidt S., Margelik R., Kirchberger R., Eichlseder
combustion system are the relatively high soot H., „Hochdruck-Direkteinspritzbrennverfahren am
emissions especially at medium engine speed - Beispiel eines Hochleistungs- 2-Takt Motors“,
operating points with still little charge motion but Hochleistungs- und Rennmotoren, Haus der
already relatively short available mixture preparation Technik, Essen, 2004
time. Beside an intensification of the charge motion 5 Klawatsch D. et al, “Improvement of the Operation
through the noted variabilities, an increase of the Characteristics of Gasoline Direct Injection Engines
injection pressure with improved spray atomization by the Application of Double Injection”, 8th
would be a promising step. Symposium “The Working Process of the Internal
Combustion Engine”, Graz, 2001
CONTACT 6 Fire Manual: FIRE v2008 - ICE Physics &
Chemistry Users Guide
Stephan Schmidt: Email: schmidt@vkmc.tugraz.at 7 Elsässer G., „Experimentelle Untersuchung und
Phone: 0043 (0)316 873 7591 numerische Modellierung der freien Kraftstoffstrahl-
ausbreitung und Wandinteraktion unter
Martin Abart Email: abart@vkmc.tugraz.at
motorischen Randbedingungen“,
Phone: 0043 (0)316 873 4585
Forschungsberichte aus dem Institut für
Oliver Schoegl Email: schoegl@vkmc.tugraz.at Kolbenmaschinen, Universität Karlsruhe, Band 5,
Phone: 0043 (0)316 873 4597 2001
8 Mills A. A.,Fry J.D., “Rate of evaporation of
Roland Kirchberger Email: kirchberger@vkmc.tugraz.at
hydrocarbons from a hot surface: Nukiyama and
Phone: 0043 (0)316 873 8090
Leidenfrost temperatures”, European Journal of
Helmut Eichlseder Email: eichlseder@vkma.tugraz.at Physics Vol. 3, 1982
Phone: 0043 (0)316 873 7201 9 Fimmel W., Jauk T., Fuchs C., Wimmer A., „Optical
Graz University of Technology, Institute for Internal
analysis and simulation of diesel sprays in a high
Combustion Engines and Thermodynamics, Inffeldgasse pressure and high temperature spray box”,
21A, 8010 Graz, Austria Proceedings of ICES2006-1376, Aachen, 2006
10 Spicher U. et al, „Ottomotor mit Direkteinspritzung“,
DEFINITIONS, ACRONYMS, ABBREVIATIONS Vieweg Verlag, Wiesbaden, 2007, ISBN 978-3-
8348-0202-6
BMEP: Break Mean Effective Pressure 11 Stevens E., Steeper R., „Piston Wetting in an
Optical DISI Engine: Fuel Films, Pool Fires, and
BSFC: Break Specific Fuel Consumption Soot Generation“, SAE Paper 2001-01-1203 (2001)
CFD: Computational Fluid Dynamics
CO: Carbonmonoxide
COV: Coefficient of Variation
EVC: Exhaust Valve Close
GDI: Gasoline Direct Injection
HC: Hydrocarbons
SETC2009