Technical information

ABB Turbocharging
Turbocharging medium speed diesel engines
with extreme Miller timing
2
Turbocharging medium speed diesel
engines with extreme Miller timing
E. Codan
a,1
and I. Vlaskos
a
Abstract
Miller timing is one of the few measures that can be applied in
an internal combustion engine to simultaneously reduce NO
x
emissions and fuel consumption a fact engine builders are
acknowledging by introducing it on almost all types of engine.
ABBs new-generation turbochargers allow operation at very
high pressure ratios and with very high turbocharger efficien-
cies. Engine builders can use the new potential these turbo-
chargers offer either to improve the Miller process or to further
increase engine output.
An overview of the theory of the Miller process is followed by
a look at how some important parameters of the gas
exchange and turbocharging system influence engine opera-
tion. Finally, the turbocharging system of a medium speed
diesel engine with extreme Miller timing up to 60 CA before
BDC is presented.
Key Words: Miller; emissions; two stage turbocharging
a
ABB Turbo Systems Ltd, Bruggerstrasse 71a, CH-5401 Baden, Switzerland
1
E-mail: ennio.codan@ch.abb.com, URL: www.abb.com/turbocharging
Translation of the paper:
Die Aufladung von mittelschnelllaufenden Dieselmotoren mit extrem frhen Miller-Steuerzeiten.
(9
th
Turbocharging Conference, Dresden, 23 24 September 2004)
Contents
1 Introduction 3
2 Basic principles of the Miller process 4
3 Turbocharging systems for engines with extreme Miller timing 10
4 Test engine 14
5 Summary and outlook 18
References/acknowledgements 19
3
The basic principle underlying the Miller process is that the
effective compression stroke can be made shorter than the
expansion stroke by suitable shifting of the inlet valve timing.
When both the engine output and boost pressure are kept
constant, this will reduce the cylinder filling and the pressure
and temperature in the cylinders will be lower.
The original purpose of the Miller process was to increase the
power density of engines without exceeding their mechanical
and thermal limits [8, 11]. In the 1990s, attention turned to
how it could be used to reduce the temperature in the cylin-
ders for a constant engine output, and to using this positive
effect to minimize NO
x
formation. In the case of gas engines,
an additional benefit is that the operating range can be
increased since there is less tendency for the engines to
knock.
The Miller process is one of the few options that engine builders
have for simultaneously reducing emissions and improving
engine efficiency. Since all engine builders strive to meet engine
emission limits without any loss of efficiency, practically every
modern engine is operated today with at least moderate
Miller timing.
The drawback is that ever-higher boost pressures are neces-
sary for a constant engine output, i. e. increasing demands are
made on the turbocharging system. This is clearly shown by
the boost pressure versus bmep diagram in Fig. 1. The lower
the engines charging efficiency (
l
), the higher the boost pres-
sure has to be for the engine to achieve its bmep.
The first part of this study presents some of the basic principles
of the Miller process and shows how the pressure ratio and
turbocharging efficiency affect engines operated with Miller
timing.
In a further section, results of experimental work undertaken
as part of the CLEAN project at Flensburg University of
Applied Sciences Institute of Marine Technology are presented.
Since the work carried out by ABB Turbo Systems within this
project focused on the theoretical optimization of the turbo -
charging system and on its experimental implementation on
the test engine, this section will:
discuss the possibility of supercharging the engine with
the Miller timings that were tested (
IC
= up to 60 CA
before BDC);
show the limits of single stage turbocharging; and
present and analyze, based on the results of the measure-
ments, controlled two stage turbocharging as a solution
that allowed the test engine to be operated with the
extreme Miller timing of
IC
= 60 CA before BDC.
All simulations were performed with the SiSy program
system [2], which makes use of the Woschni / Heider model [6]
to calculate the NO
x
formation.
1 Introduction
Fig. 1: Boost pressure dependence on bmep for large engines
Lines of constant
l
.
0
2
3
4
5
6
7
p
boost
/p
amb
[-]
bmep [bar] 5 10 15 20 25
2-stroke
4-stroke

1
4-stroke diesel engines,
IMO/ Miller, single stage
4-stroke low compression
engines, Miller, two stage
0.6
0.7
0.8
0.9
1.0
4-stroke
gas engines
Boot pressure/ ambient pressure
4
2 Basic principles of the Miller process
In an ideal cycle, late and early closing of the inlet valve are
equivalent. It can be seen from the p-v diagram (Fig. 2) that
both ideal processes begin with the effective compression at
point 1, i. e. their compression phase is shorter than the
expansion phase a characteristic of the Atkinson process.
The only difference between early and late closing is the
part between 1 and bottom dead center. In the first case,
i. e. early closing, sub-process 1-1-1 takes place in the
cylinder with closed inlet valves; in the second case, late
closing, (sub-process 1-1-1) with open inlet valves. Both
sub-processes exhibit zero area in the ideal case. Thus, all
the other parts are identical. The idea that the Miller process
reduces the charge temperature due to expansion of the
charge air in the cylinder is especially worth noting. This can
be clearly seen in the case of early closing, but would
seem to be missing in the case of late closing. In fact, it is
the virtual expansion from the charge pressure p
Rec
to
the initial pressure of the compression p
ac
that is relevant for
the reduction of the process temperatures. As long as the
same compression curve is achieved from the same charge
pressure and the changes of state continue to be isentropic,
then the changes of state necessary until the point 1'
(via 1-1-1 or via 1-1-1) will be irrelevant.
It should be noted in connection with the processes com-
pared in Fig. 2 that they show the same pressures but different
temperatures. As a result, the masses and outputs must be
different.
2.1 Efficiency of the Miller process
If only ideal processes are considered, then the overall effi-
ciency of the Miller process is worse than that of the process
with conventional valve timing. The reason is that the part
of the positive gas exchange loop lying between the com-
pression curve and the BDC (see Fig. 2) is cut off. This loss,
referred to in the following as Miller loss, can lead to a
reduction in engine efficiency of up to 0.5 %. In a real process
it can be expected that when the pressure behaves in the
same way in the high-pressure section, more output at higher
efficiency will be achieved since the gain from the lower heat
losses will more than compensate for the Miller loss. If the
output and all other process parameters are kept constant,
log p p
max
p
c
p
ac
p
ex
p
max
p
c
p
ex
T
ex
p
Rec
p
ac
= p
Rec
p
Rec
1 1
1
p
Rec
p
TI
p
TI
V
c
V
h
Displacement
Volume
log V
P
r
e
s
s
u
r
e
log T
T
c
T
c
T
ac
T
ac
Entropy log s
T
e
m
p
e
r
a
t
u
r
e

Conventional diesel cycle

Miller early cycle
Miller late cycle
Miller loss
T
ex
Fig. 2: Ideal processes.
5
the pressure level will be generally lower, thus improving the
efficiency of the high-pressure process. The lowering of the
peak pressure has, as a rule, the effect of freeing up some of
the engines mechanical potential for a further improvement in
efficiency. This improvement can take the following forms:
An increase in air / fuel ratio
V
An increase in compression ratio
An increase in the combustion pressure rise due to earlier
injection
In the investigated case the last of these proved most effec-
tive, as the engine already exhibited in the reference case a
high and no combustion pressure rise. However, depending
on the boundary conditions, the two other possibilities, either
alone or combined, could also be interesting.
To be able to evaluate the influence of the Miller process
on engine operation (i. e. independently of the inlet valve
cam profile), the term Miller Effect has been introduced.
This effect is described by (1
l
) and defined as follows:
Miller Effect [%] = (1
l
) 100
By simulating operation of the test engine described in sect. 4
with different Miller timings and assuming that the cylinder
output,
V
, and peak cylinder pressure remain constant, it
was possible to determine the influence of the charging
efficiency and thus of the Miller Effect on the overall engine
efficiency (Fig. 3). The efficiency has two components: a high-
pressure, i. e. closed cycle, component and a gas exchange
component:

engine
=
closed cycle
+
gas exchange
It is seen that the efficiency gain in the closed cycle increases
linearly with the Miller Effect. The efficiency of the gas
exchange component remains approximately constant up to
a Miller Effect of 15 20 %. A further increase in Miller Effect
causes a sharp increase in both the throttle losses and the
Miller loss. An optimum for the overall efficiency occurs at a
Miller Effect of about 30 %.
2.2 Early or late closing?
As already mentioned, in an ideal cycle late and early closing
of the inlet valves are equivalent. In real cycles, however, there
are differences:
With early closing the inlet valves must start to close very
early; therefore, the difference between the cylinder pressure
and the charge pressure is larger due to throttling; in the
case of late closing the intake and discharge of a portion of
the charge air also involves some throttling losses. Which of
these two effects is the more dominant will depend on the
valve geometry and cam profile.
With late closing it is also necessary to consider the heat
transfer: the charge air which is forced back has already
been heated up in the cylinder and this heat is stored in the
inlet channel until the inlet valve opens again in the next
cycle. This partly reduces the theoretical cooling of the
cylinder charge, compared with early closing.
0.02
0.01
0
0.01
0.02

engine
0.7 0.8 0.9
l

closed cycle

engine

gas exchange
Fig. 3: Influence of Miller on engine efficiency.
6
These effects can be observed in the p-v- and T-v diagrams
(Fig. 4). The pressure curve has a direct influence on the gas-
exchange work, while the temperature curve influences the
closed cycle work due to the dependence of the heat losses
and the thermodynamic properties on the actual gas tempera-
ture. The possibility of further indirect influences on the com-
bustion also cannot be excluded, since the movement of the
charge in the cylinder can change in accordance with the
valve closing point.
The global effects can be derived from an analysis of the
results in Fig. 5. Since the engine has a camshaft with steep
flanks, the inlet closing time can be varied without changing
the maximum valve lift. In this case the advantage of engine
operation with Miller early can be clearly seen compared
with operation with Miller late. The second curve applies to
scaled inlet cams with reduced maximum valve lift, which
simultaneously results in a reduction of the inlet valve area.
The determining parameter:
C
VE
= c
K
A
K
A
VE max
(c
K
= mean piston speed, A
K
= piston area, A
VE
= effective
valve area) has increased by approx. 40 %. In high speed
engines it can occur that owing to the mechanical limitations
for valve acceleration, the maximum valve lift with Miller
early must be reduced. In such cases, better results can be
achieved with Miller late.
Previous considerations apply to the full load point. For engine
operation with constant speed it can be expected that all load
points exhibit a similar behavior. For operation with variable
engine speed, e. g. in propeller operation, the engine behavior
is better with Miller early at part load, since the charging effi-
ciency, in contrast to Miller late, is higher at reduced speed.
This effect can also be seen qualitatively in Fig. 5, since C
VE
is
proportional to the engine speed.
For operation with fixed valve timing, the part load behavior
is normally better with Miller early than with conventional
timings. Miller late is not suitable for operation with variable
engine speed.
2
3
4
5
6
p
cyl
[bar]
0.2 0.4 0.6 0.8 1.0 V
z
/V
h
Ref. = IC
560

IC
= 495 [deg. CA]

IC
= 560 [deg. CA]

IC
= 645 [deg. CA]
p
Rec,495
= p
Rec,645
p
Rec,Ref
= p
ac,Ref
EC
Ref
EC
645
EC
495
IO
645
IC
495
p
ac,495
= p
ac,645
IC
Ref
IO
495
IO
Ref
0
250
500
750
1000
T
cyl
[C]
0.2 0.4 0.6 0.8 1.0 V
z
/ V
h
EO
495
IC
Ref
IC
495
IC
645
EC
645
IO
Ref
IO
645
IO
495
EC
495

=

EC
Ref
T
Rec
EO
Ref
EO
645
Fig. 4: Cylinder pressure and temperature curves.
7
In operation with natural aspiration, i. e. with operating points
of very low load, in which the turbocharging system supplies
no pressure, a reduction of the air / fuel ratio compared with
conventional engines must generally be expected. With
extreme Miller timing and reverse scavenging the air / fuel ratio
can fall in such a way that the engine reacts with heavy
emission of smoke when starting and on the first application
of load. In this case, operation could be improved with Miller
late, since the residual gas is mixed with more fresh air and is
then partly exhausted from the cylinder.
It is known that the Miller process can supply very interesting
results with variable valve timing. Systems with full variability
have the highest potential, but are demanding. The Miller
process is simpler to apply to systems in which the entire inlet
valve opening phase can be displaced. These are not optimal,
however, since
IC
and valve overlapping are varied simulta-
neously.
These effects can be summarised as follows:
Miller early, full load: this provides the best possibility for controlling the
peak pressure, since the displacement of the cam in the early direction
simultaneously produces a reduction in charging efficiency and charge
pressure.
Miller early, part load: increase in air / fuel ratio in part load operation is
limited, since the displacement in the late direction increases the
charging efficiency, but reduces the charge pressure by reduction of the
scavenging.
Miller late, full load: the peak pressure is difficult to control, since
the Miller Effect is performed with reduced scavenging, which results in an
increase in the charge pressure.
Miller late, part load: the reduction in the Miller Effect provides a marked
increase in the air / fuel ratio here, since charging efficiency and charge
pressure increase simultaneously. This positive effect could be reduced,
however, by the pressure-reduction measures, which must necessarily be
introduced at full load.
8
4
0
4
500 525 550 575 600
IC
[deg. CA]
bsfc
[g/kWh]
1.0

l
[-]
0.5
1.5
3
4
5
6
7

R
[-]

V
*

[-]
A
VE, Ref.
A
VE, 70
(0.7 *A
VE, Ref.
)
Fig 5: A
VE
-
IC
variation.
8
2.3 Scavenging in engines with Miller timing
If an engine is to be designed for operation with extreme Miller
and fixed timing, the problems of valve overlap must also be
considered. The scavenging of an engine can be described by
three parameters: the delivery ratio
R
, the charging efficiency

l
and the scavenging factor:

1
=

l
An increase in
R
generally leads to a reduction in gas
exchange work, which has a negative effect on engine effi-
ciency. If the scavenging is reduced so far, however, that the
proportion of residual gas has increased considerably in the
cylinder, deterioration of the engine efficiency occurs, being
derived from the high pressure cycle. There is then an opti-
mum efficiency for every engine at the rated point, which cor-
responds to very moderate scavenging. Scavenged engines
are almost always designed with more scavenging, since this
improves the thermal loading and part load behaviour.
Variation of the inlet closing point
IC
from the conventional to
Miller timing with constant valve overlap results in equal
reductions of
R
and
l
, so that the scavenging factor tends to
rise. This can be explained by the following considerations:
With increasing Miller Effect and constant air / fuel ratio, the
air density before the cylinder is increased and the volume
to be filled is simultaneously reduced (
l
falls).
The scavenge volume with unchanged valve overlap rises
approximately proportional to the density before the
cylinder and the scavenging factor accordingly increases.
If the scavenging factor is to be kept constant, a reduction
in valve overlap is recommended, this resulting in a further
improvement in engine efficiency (see Fig. 6).
The reduction in valve overlap has a positive influence on
engine behaviour in the lowest load range, where, with
extreme Miller timing, the available fresh air is significantly
reduced.
The intersection of the lines of charge pressure and exhaust
gas pressure after the cylinders is displaced to areas of
lower load and the range in which the exhaust gas pressure
after the cylinder is higher than the charge pressure
becomes smaller.
The negative effects of the reversed scavenging are
reduced by the smaller valve areas during scavenging.
This is shown in exemplary fashion in Fig 7. Less valve overlap
significantly improves the air / fuel ratio in the part load range.
Smoke problems when starting and on application of load to
the engine can thereby be avoided.
Fig. 6:
VO
-
a
variation (Ref.:
VO
= 90 CA,
a
= 0.65).
16
12
8
4
0
4
bsfc
[g/kWh]
60 80
VO
[deg. CA]
1.1
1.0
1.2
1.3
1.4
1.5

a
= 0.65,
IC
= 560 deg. CA

a
= 0.65,
IC
= 495 deg. CA

a
= 0.70,
IC
= 495 deg. CA

a
= 0.75,
IC
= 495 deg. CA
Fig 8:
a
-
IC
variation.
9
2.4 Turbocharging efficiency
The results shown in Fig. 5 were calculated with a constant
turbocharging efficiency of
a
= 0.65. As a result the pressure
difference over the cylinder and therefore the theoretical value
of gas exchange work remains constant.
This corresponds to a turbocharger efficiency of 69 %, which
is an extremely good value for a small 4-stroke engine
turbocharger, even under normal pressure conditions. With
increasing Miller Effect the required pressure ratio rises and it
becomes increasingly difficult to maintain the turbocharger
efficiency at this level. Since, however, the turbocharging effi-
ciency has a significant influence on engine efficiency (see
Fig. 8) efforts will also be focused in future on achieving high
pressure ratios with high turbocharging efficiencies. The first
curve (
a
= 0.65) in Fig. 8 shows the optimum engine efficiency
with a Miller Effect of approx. 30 % and a compressor pres-
sure ratio of around 5. It is extremely difficult to achieve this
with the required turbocharger efficiency.
With higher turbocharging efficiencies it is possible to achieve
a further improvement in engine efficiency. The engine efficiency
optimum is displaced depending on the turbocharging effi-
ciency level to Miller Effect ranges of over 30 %. The current
standpoint, however, is that these results are only possible
with turbocharging systems in which higher efficiencies can
be achieved with intercooling.
Fig. 7:
v
at part load.
0.7
0.6
0.5
0.4
0.3
1.0
1.5
2.0
2.5
3.0
2 4 6 8 bmep [bar]
1.0
1.5
2.0
2.5
3.0
p
rec
[bar]
p
TI
[bar]
Valve Overlap (Ref. 20 deg. CA)
Valve Overlap, Ref.

l
4
0
bsfc
[g/kWh]
8
12
475 500 525 550 575 600
Ref.

IC
[deg. CA]
0.5
1.0
1.5
3
4
5
6
7

R
[-]

V
*[-]

l
[-]

a
= 0.65

a
= 0.70

a
= 0.75
10
3 Turbocharging systems for engines with extreme Miller timing
It was shown in the previous section that for implementation
of the Miller process in modern engines very high pressure
ratios and efficiencies are necessary. The possibilities for single
or two stage turbocharging with and without control are con-
sidered below.
3.1 Single stage turbocharging
The latest generations of ABB turbochargers for 4-stroke
applications permit pressure ratios greater than 5 in continuous
operation with aluminum compressors.
The product range covers two turbocharger families:
TPS: turbochargers with radial turbine for the power range
500 3200 kW [1]. The TPS. . -F versions are a further
development of the proven TPS. . -D/ E series. The range of
operation of TPS turbochargers and the cross-section of a
TPS. . -F turbocharger are shown in Figs. 9a and 9b.
TPL: turbochargers with axial turbine for the power range
from 1000 kW [10]. The TPL. . -C versions are a further
development of the proven TPL. . -A/ B series. The range of
operation of TPL turbochargers and the cross-section of a
TPL. . C turbocharger are shown in Figs. 10a and 10b.
4.0
4.4
4.8
5.2

c
[-]
V
298
[m
3
/s] 1.2 2.0 2.8 3.6 4.4
C
o
m
p
r
e
s
s
o
r

p
r
e
s
s
u
r
e

r
a
t
i
o

Volume flow
F31
F32
F33
T
P
S
4
8
-
T
P
S
5
2
-
T
P
S
5
7
-
T
P
S
6
1
-
TPS. . -E
TPS. . -D
Fig. 9a: Operating range TPS.. -F. Fig. 9b: Cross sectional view TPS. . -F.
11
Fig. 11 shows the development of the achievable pressure
ratio of ABB turbochargers over the course of the years.
The curve shows a levelling off up to the early nineties, since
the components were developed in mature products (VTR and
RR) with their corresponding limitations.
With the introduction of the new generations (TPS and TPL,
1996) it was possible to open up further development poten-
tial with new concepts. Shortly afterwards it was seen that
a compressor design with a simple trim concept was not
suitable for optimum coverage of the required flow rate range.
The introduction of different design areas according to the
mass flow rate, as well as the use of stabilizers [7] and the
latest tools for compressor development led to a further
improvement in pressure ratio.
This progress made it possible to turbocharge modern
engines with the highest mean effective pressure and with
a moderate Miller Effect. In engines not designed for the
highest mean effective pressures this results in an even
greater potential, which can be used for the introduction of
extreme Miller timing.
If, however, an engine is to be turbocharged for example at
mep = 28 bar with 30 % Miller Effect, a pressure ratio of at
least 6 is necessary. No suitable turbocharger is yet available
for this. Far higher pressure ratios are technically possible
with single stage turbocharging, but a technological leap
would be necessary with a change to better materials. This,
Volume flow
a
t

M
C
R
V [m
3
/s]
without compressor cooling option
T
P
L
6
7
T
P
L
7
1
T
P
L
7
6
T
P
L
7
9
5 15 10 20
3.0
3.5
4.0
4.5
5.0

c
[-] C32
C33
C34
C35
C36
C
o
m
p
r
e
s
s
o
r

p
r
e
s
s
u
r
e

r
a
t
i
o

1
2
3
4
2.3
3.6
3.9
4.1
4.5
5.0

c
[-]
1965 1975 1985 1995 [year]
P
r
e
s
s
u
r
e

r
a
t
i
o

Fig. 11: Achievable pressure ratio with single stage aluminum
compressor (50,000 h, base load) [9].
Fig. 10a: Operating range TPL. . -C (status: 2010). Fig. 10b: Cross sectional view TPL. . -C.
12
however, would significantly increase development and pro-
duction costs. Essential turbocharger requirements such as
high turbocharger efficiency, a wide compressor map and high
reliability and flexibility, also mean that there is a very high
technical risk involved. For these reasons it is also necessary
to consider two stage in addition to single stage turbocharging.
3.2 Two stage turbocharging
With two stage turbocharging, the air and gas side enthalpy
heads are divided between the two turbochargers. This divi-
sion alone enables higher turbocharger efficiencies to be
achieved in the individual stages owing to the lower loading of
the turbocharger components. The turbocharging efficiency
achievable can be even further increased, however, by addi-
tional measures:
Intercooling improves the process of air compression, since
the process of isothermal compression is more closely
approached. The theoretical gain (see Fig. 12) increases
with the pressure ratio.
At the turbine-end the sum of the individual isentropic
enthalpy heads of the two stages is always greater than the
enthalpy head of the overall stage, since the losses in the
high pressure stage at least increase the inlet temperature of
the low pressure stage. Optimal connection of the turbines,
however, would allow the outlet losses of the high pressure
stage, which comprise the greatest source of loss of a tur-
bine, to also be largely converted into pressure. The achiev-
able gain in efficiency of the two stage turbine can be 2 to
8 % points compared with the efficiency of the single stage.
Generally considered, turbocharging efficiencies up to around
80 % would be achievable with two stage turbocharging with
optimized efficiency.
The cost, weight and physical volume of a turbocharging sys-
tem of this kind, however, would be extremely high. And the
system complexity would increase. On top of this, the high full
load efficiency of the turbocharging system would make
sophisticated control intervention necessary for part load
operation of the engine to be practicable.
0
5
t
intercooling
= 25 C
t
intercooling
= 60 C
10
15

a
[%]
4 6 8 10
Pressure ratio
Fig. 12: Efficiency gain through intercooling.
3.3 Controlled two stage turbocharging
Solutions to the problems outlined above are offered by con-
trolled two stage turbocharging. A layout for optimum turbo -
charging efficiency leads to an even distribution of the pres-
sure ratios:

HP
=
LP
=
System
The result is that very large turbocharger stages are required.
If the pressure ratio is increased in the low pressure stage, the
result is as follows:
The low pressure compressor is smaller: it supplies the
same flow at a higher pressure.
The low pressure turbine is smaller: it must produce more
energy from the same exhaust gas flow.
The high pressure compressor remains the same or is
slightly smaller; the operating point moves parallel to the
surge limit. If low pressure ratios are necessary, smaller
compressor wheels with high specific swallowing capacity
can be used.
The high pressure turbine is larger: it must produce less
energy for the same inlet conditions.
13
Reference TPS xx

Vmax
5.0
TPS xx:
VHP
=
Vmax
1
2 & TPS xx+1:
VLP
=
Vmax
1
2
two stage,
Vmax
> 10
TPS xx1:
VHP
=
Vmax
/ 4 & TPS xx:
VLP
= 4
two stage controlled,
Vmax
> 10
WG
Fig. 13: Turbocharger sizes for various turbocharging systems.
Obviously, for a compact solution the high pressure turbine is
critical. If, however, some of the exhaust-gas mass flow can
be made to bypass this turbine at the systems design point,
the turbocharging system will be compact and an ingenious
means of control is provided, effectively solving the part load
problems described.
Bypassing of the high pressure turbine is controlled by a
throttle device, which can be progressively closed in the part
load range. The boost pressure can therefore be increased
as required. In full-load operation, losses in efficiency must be
expected, but these will be much lower than with a conven-
tional waste gate, since the exhaust gas flow diverted is lost
over just a small part of the head.
The different sizes of the turbochargers for various turbo -
charging systems are shown in Fig. 13.
14
4 Test engine
The experimental work was performed as part of the CLEAN
partial project NO
x
-reduction in large diesel engines by appli-
cation of the Miller process [3] and the subsequent project
Particle and NO
x
reduction in large diesel engines by com-
bined application of charge pressure controlled high pressure
injection and turbocharging according to the Miller process.
ABB Turbo Systems, in close collaboration with the other
project partners ISF and FMC, carried out a thermodynamic
analysis of the influence of the Miller process and its effects
on engine operating values and NO
x
emissions, and also fitted
the engine with the appropriate turbochargers. In this way it
was also possible to confirm experimentally the engine opera-
tion and NO
x
reduction calculated from the thermodynamic
analysis.
The test plant was realized using the research engine ISF4524
of the Institute of Marine Technology at FH-Flensburg [4]. The
most important engine data are given in Table 1.
Table 1: Engine ISF4524
Bore 240 mm
Stroke 450 mm
Stroke/ bore ratio 1.875
Compression ratio 17.8 : 1
Cylinder capacity 240 kW 600 min
-1
Number of cylinders 3 (6)
Turbocharging system Constant pressure (single stage with charge air
cooler, two stage with intercooler and CAC)
Inlet closing 20 KW after, 45 and 60 before BDC
Injection system Two stage HDV system (feed pressure freely
adjustable in the range 30 500 bar) [5]
Fig 14: Controlled two stage turbocharging on the test engine and schematic arrangement.
Zyl 3
Zyl 2
Zyl 1
KB 1 VOL 1
VER 1
LLK 1
TUR 1
KB Constant boundary
VOL Volume
VER Compressor
TUR Turbine
BL Waste gate
Zyl Cylinder
LLK Charge air cooler
TUR 2
BL1
HP Stage
LP Stage
KB 2
KB 3
VER 2
LLK 2
KB 4
KB 6
15
The engine has six cylinders, but it was decided to run it on
three to keep costs down. ABB installed on this engine its
smallest turbochargers from the TPS series: A TPS48-D for
the reference measurement without Miller timing, a TPS48-E
for single-stage turbocharging, and a combination of TPS48-D
and TPS48-E for the two stage turbocharging (Fig. 14). Some
specially made components were necessary to optimize the
controlled two stage turbocharging. The control valve itself
(GloTech60) was provided by Woodward.
Since, in each of these configurations, and especially in the
last one, the turbochargers were operated well below their
capacity limits no conclusions can be drawn as to the real
potential in terms of turbocharging efficiency and compact-
ness. The results therefore serve only to confirm the different
configurations thermodynamic potential from the point of
view of engine operation. To optimize the turbocharging it
would be necessary to use smaller turbochargers or to run the
engine on six cylinders.
Test results
At the start of the project a variation in
IC
was calculated for
the engine for the full load point with various turbocharging
systems (see Fig. 15). This produced the target values and
forecasts shown in Table 2, which also shows the test results.
Fig. 15: Influence of
IC
variation on the operating parameters of the
research engine at full load.
6
4
2
8
p
rec
[bar]
13
14
15
NO
x
[g/ kWh]
475 500 525 550 575 600
IC
[deg. CA]
12
8
4
0
bsfc
[g/kWh]
450
475
500
525
T
HP-TI
[deg. C]
1-stage/ 3-Cyl.
2-stage/ 3-Cyl.
2-stage/ 6-Cyl.
Ref.
16
In general the forecasts were exceeded both for fuel con-
sumption and the NO
x
values. It was not possible to test the
planned version with inlet closing 30 CA bBDC owing to non-
availability of the parts. Instead a version with 45 CA bBDC
was tested experimentally for both single and two stage
turbocharging. The turbocharging systems were not optimized
for this purpose and therefore the engines charge pressures
were too low in case II.a and too high in case II.b (see Table 2).
The values specified apply to engine operation with a feed
pressure in the injection system of 300 bar. By varying this
feed pressure it is possible to move along a be-NO
x
trade-off
line of the engine. An increase in feed pressure in the injection
system leads to a reduction in fuel consumption and an
increase in NO
x
emissions, and vice-versa [5]. In the extreme
case: III. Miller /
IC
= 60 CA bBDC, a further significant reduc-
tion in NO
x
emissions, aimed at emissions optimization, can
be obtained without losing the entire efficiency advantage.
The configurations in Table 2 are generally only suitable for
generator operation. The engine was also operated, however,
according to the propeller law. The good experience so far
with the engine as well as close agreement between the
calculations and measurements (Fig. 16) indicate that the pro-
peller operation is reproduced well by the simulation.
Both the measurements and the calculations (see Fig. 17)
confirm that reliable propeller operation is not possible without
control intervention. It is anticipated that by re-specifying the
turbocharger for
Vmin
= 1.8 at part load and regulating the
charge pressure by means of a waste gate at full load, pro-
peller operation will also be possible with the extreme Miller
timing considered (Fig. 17).
Although the turbocharging efficiency is relatively low owing to
the considerably over-dimensioned turbocharger, the results
look interesting. Improvements can only be expected with a
version of the turbocharging group for the full engine.
Fig. 16: Comparison between simulation and measurement (propeller
law,
IC
= 60 CA bBDC, PVD = 300 bar).
4
8
1
2
3
4
5
6
Measurement
Simulation
0
100
200
20 40 60 80 100 [%]
Load
1.0
1.2
300
400
500
600
bsfc/
bsfc
Ref
T
LP-TO
T
LP-TI
[deg C]
T
HP-TI
SQC
[kg/kWh]
p
Rec
[bar]
p
HP-TI
[bar]
p
cyl., max
[bar]
17
Table 2 Forecast Measured
I. Standard/
IC
= 20 CA aBDC
Engine operation with TPS48-D/ single stage turbocharging
P
Rec
(100 %-load) = 3.4 bar
II. Miller /
IC
= 30 CA bBDC b
e
[g/ kWh] 3.8
Engine operation with TPS48-E/ single stage turbocharging NO
x
[g/ kWh] 0.66
Charge pressure at full load P
Rec
[bar] 4.5
II.a Miller /
IC
= 45 CA bBDC b
e
[g/ kWh] + 4.8
Engine operation with TPS48-E/ single stage turbocharging NO
x
[g/ kWh] 3.8
Charge pressure at full load P
Rec
[bar] 4.39
II.b Miller /
IC
= 45 CA bBDC
Engine operation with TPS48/ two stage turbocharging b
e
[g/ kWh] 11.6
TPS48-D (LP stage) + TPS48-E (HP stage) NO
x
[g/ kWh] 2.2
Charge pressure at full load P
Rec
[bar] 5.27
III. Miller /
IC
= 60 CA bBDC
Engine operation with TPS48/ two stage turbocharging b
e
[g/ kWh] 7.1 10.7
TPS48-D (LP stage) + TPS48-E (HP stage) NO
x
[g/ kWh] 1.41 4.4
Charge pressure at full load P
Rec
[bar] 6.0 5.68
Table 2: Fuel consumption and NO
x
emissions reduction with different inlet close timing compared with the basic engine version
(feed pressure = 300 bar).
18
Fig. 17: Controlled two stage turbocharging engine operating
parameters (propeller law,
IC
= 60 CA bBDC, feed pressure = 300 bar).
5 Summary and outlook
It was shown that the Miller process has very high potential
with regard to improvement of emissions and engine efficiency.
This potential can even be increased by introducing Miller
timings that produce far more Miller Effect than is usual today.
However, this requires very high boost pressures.
ABB Turbo Systems has developed two new turbocharger
generations that permit pressure ratios of over 5 in a version
with favorably priced aluminum compressors. At the same
time multi-stage solutions are being examined for even higher
pressure ratios.
It has been possible to confirm these concepts experimentally
in a research project lasting several years. At the same
time, the experience and knowledge gained generate new
ideas and increase confidence in the simulation tool, which
continues to yield information of ever-better quality.
Controlled two stage turbocharging has proved to be a highly
promising solution for the turbocharging of large diesel
engines, particularly for operation at variable speed and sub-
stantially increased charge pressures.
Modern turbochargers, however, are not designed for this
operation. An important task still to be undertaken is the
preparation of design solutions for optimized multi stage
turbocharging, therefore making it economically attractive.
There are also applications, as in the case of gas engines,
where the proposed control provides no benefit, since these
engines employ different control concepts. The definition
of suitable turbocharging systems for all high pressure appli-
cations is another demanding task requiring attention in the
future.
0.4
0.5
0.6
0.7
0.8
1
2
3
500
600
400
30 40 50 60 70 80 90 100 [%]
10
0
10
20
30
7
200
150
100
50
0
6
5
4
3
2
1
Without wastegate control
With wastegate control
Load
T
HP-TI
[deg. C]
p
c, max
[bar]
p
Rec
[bar]
p
HP-TI
[bar]
bsfc
[g/kWh]

a
19
References/ acknowledgments
[1] Born, H., M. Meier & C. Roduner, 2004, TPS. .-F:
A new series of small turbochargers for highest pressure
ratios, CIMAC Congress, Kyoto.
[2] Bulaty, T., E. Codan & M. Skopil, 1996, A flexible
system for the simulation of turbocharged diesel engines
and turbocharger systems, ASME Spring Technical
Conference, ICE-Vol. 26.3.
[3] Codan, E., H. Fiedler, H. Meier-Peter, 1999, Using
the Miller Process for marine diesel engines, MarPower
Conference, Newcastle, March 1999.
[4] Fiedler, H., 1992, Viertakt-Langhub-Forschungsmotor
Stand der Konstruktion und des Motoraufbaus,
14. Informationstagung des ISF, Flensburg/ Glcksburg.
[5] Fiedler, H., 2001, Shaping the combustion process by
utilisation of high pressure injection, CIMAC Congress,
Hamburg.
[6] Heider, G., G. Woschni & K. Zeilinger, 1998, 2-Zonen
Rechenmodell zur Vorausberechnung der NO-Emission
von Dieselmotoren, MTZ 59 11, p. 770.
[7] Hunziker, R., P. Jacoby & A. Meier, 2001, A new series of
small turbochargers for high flow rates and high pressure
ratios, CIMAC Congress, Hamburg.
[8] Miller, R., 1957, The Miller supercharging system for
diesel and gas engine operating characteristics, CIMAC
Congress, Zurich.
[9] Rohne, K.-H., 2004, Technologien fr grosse Hoch-
leistungsturbolader, 13. Aachener Kolloquium Fahrzeug-
und Motorentechnik, RWTH, Aachen (October 2004).
[10] Wunderwald, D. & K. Heinrich, 2004, Meeting
the requirements of modern diesel & gas engines:
The new TPL. . -C turbocharger generation,
CIMAC Congress, Kyoto.
[11] Zappa, G. & T. Franca, 1979, A 4-stroke high speed
diesel engine with two stages of supercharging and
variable compression ratio, CIMAC Congress, Vienna.
Acknowledgments
The authors wish to thank the German Federal Ministry of
Education and Research (BMBF) for its financial support for
the two research projects. They would also like to thank the
Institute of Marine Technology at the Flensburg University of
Applied Sciences (ISF) (Prof. Dr.-Ing. Hansheinrich Meier-Peter
and Dipl.-Ing. W. Eggert) and the FMC company (Dipl.-Ing. H.
Fiedler) for their productive collaboration, in addition to the
Woodward company for providing the control valve.
C
H
T
U
S
-
1
3
3
0
-
1
0
1
2
-
E
N

P
D
F

2
0
1
0

A
B
B

T
u
r
b
o

S
y
s
t
e
m
s

L
t
d
,

B
a
d
e
n
/
S
w
i
t
z
e
r
l
a
n
d
ABB Turbocharging Service network
ABB Turbo Systems Ltd
Bruggerstrasse 71 a
CH-5401 Baden/ Switzerland
Phone: +41 58 585 7777
Fax: +41 58 585 5144
E-mail: turbocharging@ch.abb.com
www.abb.com/turbocharging
Aberdeen Adana Alexandria Algeciras Antwerp Baden Bangkok Barcelona Bergen Bremerhaven Brisbane Buenos Aires Busan Cairo Cape Town
Casablanca Cebu Chennai Chicago Chittagong Chongqing Colombo Copenhagen Dakar Dalian Dar es Salaam Davao Delhi Dortmund Douala
Dubai Dunkerque Durban Faroe Islands Fukuoka Gdansk Genova Gothenburg Guangzhou Guatemala City Haifa Hamburg Helsinki Ho Chi Minh City
Hong Kong Houston Incheon Istanbul Izmir Jakarta Jeddah Jubail Kaohsiung Karachi Kobe Lahore Las Palmas Le Havre Lima Limassol
Los Angeles Lunenburg Madrid Malta Manaus Manila Mannheim Marseille Melbourne Miami Montreal Mumbai Naples New Orleans New York
Onomichi Oporto Oslo Panama Perth Piraeus Qingdao Quito Reykjavik Rijeka Rio de Janeiro Rotterdam Saint Nazaire Santo Domingo Santos
Seattle Shanghai Singapore Southampton St. Petersburg Suez Sunderland Surabaya Talcahuano Tallinn Telford Tianjin Tokyo Vadodara Vancouver
Varna Venice Veracruz Vizag Vladivostok