Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

SAE TECHNICAL
PAPER SERIES

2008-01-1610

Optimization of Fuel Injection Parameters

for Meeting Euro III Exhaust Emission
Norms on a Heavy Duty Diesel Engine
Using Taguchi Technique
C. L Krishnan, Y. S Vinod and C. V Sudhindra
BOSCH Limited

2008 SAE International Powertrains,

Fuels and Lubricants Congress
Shanghai, China
June 23-25, 2008
400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A. Tel: (724) 776-4841 Fax: (724) 776-0790 Web: www.sae.org

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

By mandate of the Engineering Meetings Board, this paper has been approved for SAE publication upon
completion of a peer review process by a minimum of three (3) industry experts under the supervision of
the session organizer.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise,
without the prior written permission of SAE.
For permission and licensing requests contact:
SAE Permissions
400 Commonwealth Drive
Warrendale, PA 15096-0001-USA
Email: permissions@sae.org
Tel:
724-772-4028
Fax:
724-776-3036

For multiple print copies contact:

SAE Customer Service
Tel:
877-606-7323 (inside USA and Canada)
Tel:
724-776-4970 (outside USA)
Fax:
724-776-0790
Email: CustomerService@sae.org
ISSN 0148-7191
Copyright 2008 SAE International
Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE.
The author is solely responsible for the content of the paper. A process is available by which discussions
will be printed with the paper if it is published in SAE Transactions.
Persons wishing to submit papers to be considered for presentation or publication by SAE should send the
manuscript or a 300 word abstract of a proposed manuscript to: Secretary, Engineering Meetings Board, SAE.
Printed in USA

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

2008-01-1610

Optimization of Fuel Injection Parameters for Meeting

Euro III Exhaust Emission Norms on a Heavy Duty
Diesel Engine Using Taguchi Technique
C. L Krishnan, Y. S Vinod and C. V Sudhindra
BOSCH Limited
Copyright 2008 SAE International

ABSTRACT
Diesel engine performance is being continuously
enhanced in the last decades in pursuit of high power
density, improved fuel economy and low emission
compliance. Developments in combustion technologies,
advanced electronic fuel injection, refined engine and
injection parameters have helped engineers tune ecofriendly diesel engines resulting in simultaneous NOxPM reduction. Understanding engine and injection
parameters and their effects on exhaust emissions is
vital for producing cleaner diesel engines and is
therefore a subject of intense research.
The scope of this work was to optimize the influence of
injection parameters on regulated emission constituents
of diesel exhaust to meet Euro III emission norms on a
heavy duty diesel DI engine. Carefully tailored design of
experiments and Taguchi technique were used to meet
the project objectives due to short development time.
The best injection parameters were isolated from
experimental results of L9 orthogonal array using
Analysis of Means (ANOM), Signal-to-noise ratio (S/N)
and Analysis of Variance (ANOVA). The results from
analysis are validated to meet Euro III emission targets
consistently with defined engineering margins on
emission components.
In this investigation, influence of parameters namely
cam velocity, nozzle through flow, nozzle tip protrusion
into the combustion chamber and number of spray
orifices were studied. Of the four injection parameters
evaluated, cam velocity affecting injection rate and
number of nozzle spray orifices affecting injection
pressures and atomization of spray played an influential
role in reduction of NOx and PM. In addition, nozzle tip
protrusion affecting the point at which nozzle spray jet
impinges into the combustion bowl only moderately
influenced the PM and HC emissions. Nozzle through
flow effect on emission constituents of diesel exhaust is
almost negligible for meeting Euro III exhaust emission
norms in the defined experimental region. Euro III
emissions targets were met consistently in totality with
more than 11.5% margins on NOx, 61.5% on CO, 76.5%

on HC and 23% on PM compared to federal Euro III

limits. Further, engine power and torque characteristics
were enhanced by 30% compared to base engine
performance together with 10.6% reduction in full load
SFC with intent of improving overall driveability and fuel
economy of the vehicle.

INTRODUCTION
The remarkable progress in compression ignition
engine, to its present form and diversity, has brought the
diesel engine to technological forefront. Its high
efficiency with lean combustion, resulting in minimum
fuel consumption, has made the diesel engine a subject
of intense research. This research, in a constant quest
to tap maximum power density with minimum fuel
consumption, has been a moving target for all
combustion engineers.
Last decades witnessed continuous attempts to enhance
diesel engine power and torque per unit displacement,
and simultaneously reduce specific fuel consumption
and emissions [1]. The focus of achieving this objective
is through the process of fine tuning fuel injection
equipment like injection pumps and nozzles [1-7]
together with improvements in combustion technologies
[3-5] [8-9]. This is accomplished by injecting as much
fuel as the engine can digest, at a given engine speed
and load, to achieve constant pressure combustion
without excessive increase in combustion pressures,
smoke and emissions.
The demand on diesel engines in late 1980s resulted in
birth of a new area of application very high speed DI
diesel engines [2]. High speed IDI diesel engines which
ruled passenger car market are now replaced by DI
technology due to its superior emission, fuel
consumption and driveability [1]. Concerns on noise in
DI diesel engines are addressed by injection rate
shaping, using of two spring injectors, multiple pilot and
post injections using common rail technology and better
understanding of engine acoustics [1][17-18]. High
speed DI diesel engine today has become undisputable

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

leader and its potential is enhanced using advanced

electronic fuel injection technologies [1][3-4].
Driving factors propelling countries worldwide to witness
rapid growth in diesel engine development include:
x
x
x
x

Demand for better engine performance and fuel

economy.
Growing awareness of harmful effects of engine
exhaust.
Refinement in federal laws controlling emission
regulations around the world.
Legislations controlling noise, vibration and
harshness characteristics.

The market demand for better power and fuel economy

together with legal demand for lower emission
compliance set mutually conflicting and challenging
targets for automotive engineers. Refer Figure 1 for
dynamic market trends since 1989 [19]. These goals, by
nature, impose tradeoffs within which the engine is
optimized. Typical tradeoffs include:
Performance
tradeoff:
Power-SFC,
Combustion
pressures-Durability, Start characteristicTemperature.
Emission tradeoff: NOx-PM, NOx-Soot, NOxHC.
Emission Performance tradeoff: NOx-Power, SmokeTorque, NOx-Injection timing.

operating environment of finalized fuel injection

parameters and guarantee high quality results reliably,
carefully tailored design of experiments and Taguchi
technique were used for experimentation and results
were analyzed using:
x
x
x

Analysis of means (ANOM).

Analysis of signal-to-noise ratio (S/N).
Analysis of Variance (ANOVA).

The simultaneous study of multiple injection parameters

using orthogonal arrays for studying the effects of
control factors on emission response was in contrast to
conventional combustion development utilizing onefactor-at-a-time experimentation.
The first objective was to enhance the engine
performance by exploiting engine design to produce
maximum torque and power per unit displacement
without significant increase in smoke at constant
combustion pressure. Table 1 shows details of existing
and proposed engine performance parameters.
Parameter
Engine
Power
Torque
Capacity
BxL
RC
Idle speed
Max speed
Swirl
Application

Euro II
Euro III
6 Cylinder, Inline, Water cooled with
Turbocharger and Intercooler
96kW@2400rpm 132kW@2400rpm
440Nm@1800rpm 600Nm@1400rpm
5.68 Liter
97 mm x 128 mm
19:1
17.5 : 1
650 r 50 rpm
2800 r 50 rpm
2.2 AVL
2.0 AVL
Heavy Commercial Vehicle (Bus,
truck)

Table 1: Existing Euro II and proposed Euro III

engine performance parameters

In order to meet the objectives of the project outlined

economically (in a short development time) and make
the engine performance robust to variations in the

Euro III
Target

Safety
margin

The present work is aimed at establishing the influence

of injection parameters on regulated emission
constituents of engine exhaust to meet Euro III emission
norms on a heavy duty diesel DI engine. This was
accompanied by enhancing the raw engine performance
of an existing Euro II compliant heavy duty diesel
engine.

Emission
Parameter

Euro III
Limits

AIM AND OBJECTIVE

Euro II
Limits

Figure 1: Market and legal demands of a heavy duty

diesel engine

The second objective was to meet emission regulations

for Euro III with defined safety margins. European
emission regulation 1999/96/EC [14] defines diesel
powered heavy duty vehicles of gross vehicle weight
greater than 3.5 tons (GVW>3.5 tons) must meet
emission regulation of NOx, CO, HC and PM within
prescribed limits shown in Table 2.

NOx (g/kWh)
CO (g/kWh)
HC (g/kWh)
PM (g/kWh)

7.00
4.00
1.10
0.15

5.00
2.10
0.66
0.10

4.600
1.785
0.561
0.085

8%
15%
15%
15%

Table 2: Euro II and Euro III emission regulations

and safety margins

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

To guarantee reliable results on subsequent engines

due to environmental variations, safety margins for each
emission component were defined based on past
experience on similar applications. The objective was to
achieve emissions lower than Euro III engineering
targets. This not only ensured meeting Euro III safely but
also defined the criterion for project's success.

NECESSITY OF ROBUST DESIGN

The first objective of engine performance enhancement
posed a rather easy task of working within engine
constraints (combustion pressure, smoke and torque
limits) to achieve enhanced performance by tuning
dynamic injection begin, injection duration and mass of
fuel quantity injected.
The second objective of meeting Euro III emission
regulations with safe margins on the other hand posed a
challenge, particularly for NOx and PM from existing
Euro II level maturity. Approximately 23% of NOx and
41% of PM emissions had to be reduced to meet Euro III
engineering targets. This called for a robust design
approach. Meeting CO and HC limits were not of serious
concern as excess oxygen present in lean burn diesel
combustion would eventually result in very low CO and
HC emission.
An important part of planning a robust design is to
identify the variables affecting the emission response
and deciding the direction in which to vary them.
Following an exhaustive literature survey that revealed
several variables influencing engine performance and
emissions, assigning each variable to a particular
category formed the next task.

IDENTIFICATION OF CONTROL FACTORS

Injection parameters whose effects on engine
performance were evaluated directly formed this class of
variables. As reported by Gerald G Hahn [10], these
variables significantly affect response (exhaust
emissions components) and create the basic need for
experimental investigation. Among fuel injection
variables surveyed in literature, those factors outlined in
Table 3 were optimized assuming 2 and 3 levels for
linear and quadratic loss function models respectively.

Experimental factors fall into two classes namely

qualitative and quantitative. All the experimental factors
used in this investigation were quantitative by nature. In
this experimental investigation, cam velocity, nozzle
through flow, nozzle tip protrusion in combustion
chamber and number of spray orifices of nozzles were
used to influence the injection parameters. These
injection parameters included injection rate, injection
pressures at pump and injectors, injection duration,
spray penetration, spray velocity, atomization etc. A
summary of these experimental variables affecting
different injection parameters and their effects on
emissions are presented in Table 4.
Control
Factors

Injection
parameter

Effects

Cam Velocity

x Injection rate.
x Injection pressure
at pump.

NOx, HC,PM
Torque, SFC

x Injection duration.

Soot, HC

x Spray jet targeting.

x Spray penetration.
x Injection pressure
at nozzle.
x Injection duration.
x Spray velocity.

NOx, HC, PM
Power, SFC

Nozzle
through flow
Nozzle tip
protrusion
No. of spray
orifices

SMD, Soot, HC,

PM

Table 4: Control factors affecting injection

parameters and effects on engine performance

BACKGROUND FACTORS
In contrast to control factors, several injection variables
and design features of injection systems, although not of
primary interest in this investigation affect emission
components and hence their effects could not be
ignored. It was of crucial importance that such
background variables be held constant or at specified
levels throughout the investigation without varying them
together with control factors as they would contaminate
the engine response with their influence. Madhav S
Phadke [11] emphasizes the fact that these factors could
be specified freely by engineers, and their settings or
levels can be selected and defined to minimize
sensitivity of products response (exhaust emissions) to
all noise factors or uncontrolled variables.

Factor

Control
factor

Level

Magnitude

Unit

CONTROLLABLE BACKGROUND FACTORS

Cam Velocity

1.55,1.60

m/s

B
C
D

Nozzle
through flow
Nozzle tip
protrusion
No. of spray
orifices

440,460,480

cc/30se
c

1.0,1.5,2.0

mm

6,7

No.

Table 3: Control factors and their levels

In this investigation, environmental conditions in which

the engine was tested were controlled at specific levels
so that the influence of these factors on engine
emissions was minimized. To isolate the influence of
environment conditions like ambient temperatures,
pressures, humidity, etc. on exhaust emission
performance, ambient air that the engine aspirates was
conditioned and controlled to specified temperature,
pressures and humidity, as required by emission
regulations, using a series of complex instruments called
Sea Level Altitude Simulation System (SLASS). Other
background factors like fuel temperature, exhaust back

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

pressure, sulfur content in fuel, intercooler temperature,

oil temperature, etc. were maintained within the
specified limits simulating the engine operation on the
vehicle.
Dynamic injection timing : Held constant & optimized.
Diesel fuel type
: <500 ppm Low S
Fuel inlet temperature
: 403 C
Exhaust back pressure
: Max 605 mm Hg
Intercooler outlet temperature : 47 3 C
Water outlet temperature
: 85 5 C
Lubricant oil temperature
: Max 100 C
Relative humidity
: 50 1 %
Inlet air temperature
: 25 1 C
BACKGROUND FACTORS HELD CONSTANT
Variables which are not of primary interest of study, but
influence the engine emissions are sometimes held
constant. Holding the background variables constant
limited the complexity and size of the experiment. In this
investigation, following background variables affecting
engine emissions were held constant for all experiments:
x
x

Engine and engine parameters (Bore, stroke,

capacity, power, torque, compression ratio)
All nozzles used were Valve Covered Orifice (VCO),
conical spray orifice with hydro erosion type having
constant spray orifice location, constant spray cone
angle.
Discharge tubing diameter etc.

SYSTEM CONSTRAINTS AND BOUNDARIES

ENGINE CONSTRAINTS
Engines are normally constrained to operate within
certain boundary conditions of speed and load. The
maximum speed (2800 rpm) of the engine is limited by
flywheel design or end governing. The rated power is
maximum power engine safely develops (132kW) at
rated engine speed (2400 rpm). It is practically possible
to develop higher power at rated speed by injecting
higher mass of fuel. This results in higher fuel-air ratio.
Excess fuel, together with advanced injection begin
required at high speeds and loads result in high exhaust
gas temperatures and high combustion pressures.
These conditions overheat the exhaust manifold leading
to deteriorated engine performance and harmful
pollutants. The thermal limit constrains maximum power
(132kW) developed by the engine and guarantees
durability of engine components as maximum
combustion pressures (PMax=165bar) and exhaust
temperatures (TMax=650oC) are limited within safe
regions. Transmissions are designed generally to
transmit defined magnitude of maximum torque required
for vehicle motion. A maximum torque of 600Nm@1400
rpm produced by this engine was limited by the
transmission design of this vehicle.

UNCONTROLLABLE NOISE FACTORS

The performance of the engine varies due to a variety of
causes. Causes that are very expensive to control or
cannot be controlled, but influence the engine response,
significantly resulting in deviation of emissions from their
intended targets, formed the noise factors. These factors
fall into three categories: external, unit-to-unit variation
and deterioration of product.
To ensure that the uncontrolled noise factors do not
influence or bias the results, randomization was
introduced, as suggested by Gerald J Hahn [10].
Randomization involved in this work was sequence of
preparing experimental units, namely, the injection pump
and nozzles, assigning treatments like assigning the
experimental factors to orthogonal arrays, running order
of the emission tests, recording the measurements etc.
During execution of this project, parallel development
work of turbocharger and intercooler was planned. The
objective was to increase the airflow characteristics of
turbocharger for lower PM emission and obtain low
pressure drop across intercooler for lower specific
volume of air resulting in better fuel economy. Since use
of turbocharger and intercooler was subjected to
unplanned changes during experimentation, they were
treated as external noise.

Figure 2: Regime of operation in speed-load map

Limitations

Magnitudes

Thermal Limit

Max Combustion pressure = 165 bar

Max Exhaust gas temperature = 650 oC
Max Power = 132kW @ 2400 rpm

Transmission
limit
Smoke limit
torque
Speed limit
Safety limit

Max Torque = 600Nm @ 1400 rpm

Low end torque = 550 Nm @ 1000 rpm
(driveability limit)
Max rated speed = 2400 rpm
Max no load speed = 2800 rpm
Max. TOil & POil = 115C & 6 bar
Max. boost pressure: 2 bar

Table 5: Engine power, torque and safety limits

Availability of air at low engine speeds for combustion,
particularly in turbocharged engines, limit maximum

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

torque developed (550Nm@1000rpm). Producing higher

torque results in higher smoke due to limited air
available at low speeds. In contrast, a minimum torque is
ensured for good acceleration performance and
driveability of the vehicle. The engine speed and load
map shown in Figure 2 and Table 5 present regimes of
engine operation.
INJECTION SYSTEM CONSTRAINTS
High cam velocities used in the DI injection pumps
(1.55m/s and 1.6m/s) are desirable for achieving high
injection pressures (Max 1200 bar) and high injection
rates (28.5 and 30.0 mm3/stk/oCam) especially for a high
speed heavy duty DI application as reported by Max
Straubel and Klaus Krieger [2]. Higher cam velocities
over small cam angles translate as high cam
acceleration, which, together with high element
pressures (Max 800 bar), have a negative effect on the
durability and lead to increased drive torque (Max
115Nm) of the injection pump. In order to guarantee
reliable working life of injection system, constraints on
injection pressures at the pump and injector end are
imposed with limitation of drive torque.
Limitation

Magnitudes
Element pressure = 800 bar
Drive Torque = 115 Nm

Pump
limits
Injection
limits

Injector end pressure =1200 bar

Injector duration = 36 Crank
Injection begin (Max advance) = 14 BTDC
Injection begin (Max retard) = 5 ATDC

Table 6: Injection system constraints

Injection timing and injection duration play an important
role in engine performance. To ensure fuel temperature
is raised to auto-ignition point and total mass of fuel
injected participates in combustion process and
translates to useful power, fuel is injected slightly in
advance / before compression TDC. If ignition timing is
too advanced, combustion gases expand when the
piston is still approaching TDC. Expansion takes place
against compression work leading to loss of useful
power. This leads to high temperatures and combustion
pressures and increases NOx and knocking tendency.
While high speeds and loads favor advanced begin,
smooth engine operation with low combustion noise at
part loads, favors retarded timing. Excess retardation
results in unburnt or partially burnt fuel due to low
temperatures and results in higher HC, PM and smoke
with poor fuel economy. For safe engine operation,
preliminary combustion optimization was carried out to
evaluate ignition limits (injection timing and duration)
beyond which engine performance and emissions
deteriorate. These thresholds are presented in Table 6.

INTERACTION AND THEIR EFFECTS

When two or more variables are used in experimental
investigation, they may or may not interact. When effect

of response of one variable depends on the condition of

the other variable, the variables are said to interact. In
other words effect of first factor is not consistent under
all conditions of second factor.
Gerald J Hahn [10] emphasizes that the purpose of
designing the experiment is not only to study the effect
of control factors on response but also to obtain useful
information about the interaction among primary control
factors. This is accomplished by varying control factors
simultaneously in contrast to changing these factors one
at a time. If the predicted experimental conclusions are
not reproducible in confirmatory runs, the possible
causes according to [12] are:
x
x
x

Strong interactions between control factors.

Presence of strong noise factors.
Experimental errors.

Interactions in experimental designs can be categorized

into two groups.
x
x

Interaction between control factors and noise factors

relating to robustness of the system.
Interaction between control factors themselves
diminishing additivity of factorial effects.

Interaction between control factors causes inconsistent

effects when it exists in a system. The experimental
results therefore cannot be easily predicted, reproduced
or controlled. To overcome this, traditional methods of
counteracting interaction uses fractional or full factorial
experimentation. Robust design treats interaction
between control factors and noise factors by evaluation
of signal-to-noise ratio to achieve robustness.
Dr.Taguchi [12] emphasizes that engineers must take
advantage of interactions during development, however
moderate or insignificant to make systems robust. He
recommends use of any of the five countermeasures to
counteract interaction using robust design:
x
x
x
x

Proper determination of project scope.

Use of additive output response.
Proper selection of control factors and their levels.
Special orthogonal arrays ( L12 , L18 , L36 and L54 ).

Use of signal-to-noise ratios (S/N).

The intent in using these countermeasures is to

minimize control factor interactions and ensure that
control factor effects are stronger and dominant than
interaction effects resulting in reproducible experiments.
Among all the countermeasure techniques listed, signalto-noise (S/N) ratio is the most popularly used technique
as it addresses all aspects of noise, internal (control
factor interactions) and external (control factors with
external environment or noise). The S/N ratio is given
by:

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

Dr.Taguchis philosophy recommends maximizing S/N

ratio to maximize useful output energy and
simultaneously minimize all harmful output energies.
Traditional approaches emphasize minimizing only few
harmful output energies and therefore address only
partially the system responses. The S/N technique, on
the other hand, addresses both useful output and
harmful output and is therefore a complete approach.
S/N ratio countermeasure addresses interaction in any
form as harmful output. This technique was used to
overcome interaction.
Since exhaust emissions are harmful, ideally zero
emission is expected. The quality characteristic of
exhaust emission response follows smaller-the-better
type. The Signal-to-noise ratio (S/N or K ) follows the
relationship:

10 log 10

y 2i

i 1 n
i n

Control factors, cam velocity, number of nozzle spray

orifices at 2 levels and nozzle through flow, nozzle tip
protrusion into the combustion chamber at 3 levels were
planned for optimization:
The total degree of freedom= Df
Df = (Df)A+(Df)B+(Df)C+(Df)D.
The total degrees of freedom= Df = 1+2+2+1 = 6.
Minimum number of experiments =  Df + 1 = 6+1 = 7.
Minimum of 7 fractional factorial experiments was
necessary to evaluate effect of experimental factors and
their levels on response as defined in Table 3. The
nearest orthogonal array that can be used for
experimentation is L9.
A

Sl

Cam
velocity

Nozzle
throug
h flow

Nozzle tip
protrusion

No.
1
2
3
4
5
6
7
8
9

m/s
A1
A1
A1
A2
A2
A2
A3*
A3*
A3*

cc/30s
B1
B2
B3
B1
B2
B3
B1
B2
B3

mm
C1
C2
C3
C2
C3
C1
C3
C1
C2

EXPERIMENTAL INVESTIGATION - ESC TESTS

As defined in test procedure in emission directive
1999/96/EC [14] for Euro III, all precautions were taken
before start of emission test to ensure reliable and
repeatable measurements. Test conditions were
maintained before emission sampling of each ESC test.
Test apparatus, engine and its exhaust systems were
conditioned till stabilization of temperatures and
pressures were achieved followed by recording the
operating parameters and their influence on emissions.
OPTIMIZATION OF DYNAMIC TIMING

SELECTION OF ORTHOGONAL ARRAY

Factor

each of cam velocity and number of spray orifice were

considered for experimentation, dummy level technique
was used to modify the standard L9 orthogonal array to
accommodate 2 level factors. * represents a dummy
treatment where A3 = A1 and D3 = D1.

D
Number
of nozzle
spray
orifices
No
D1
D2
D3*
D3*
D1
D2
D2
D3*
D1

It was necessary to conduct 9 experiments with control

factors at levels defined by L9 orthogonal array. Each
experiment involved usage of specific combination of
control factors (cam velocity, nozzle through flow, nozzle
tip protrusion and number of spray orifices) at specified
levels and conducting ESC test traversing 13 distinct
speed-load points described by emission directive [17]. It
was therefore necessary to determine the best dynamic
injection begin for each emission mode (13 modes) for
each experiment suitable for ESC test. Only when the
best injection timing for NOx-PM and fuel economy is
used, would the ESC test be fruitful in determining best
injection parameters used for experiments.
Emission measurements were conducted based on
mass measurements of smoke and HC emission to
reduce soot emission; this formed an inexpensive way of
carrying out preliminary combustion work as suggested
by G Greeves and C H T Wang [15]. Based on analysis
done by R.H Hammerle et al, soot constitutes about
68% of particulate emission along with other
components of diesel exhaust like sulfates, oil and
unburnt fuel as shown in Figure 3.

Table 7: L9 orthogonal array with dummy technique

Figure 3: Composition of diesel particulates

L9 orthogonal array shown in Table 7 [13] advocates

systematic changes of many control factors
simultaneously. This ensures reliable and independent
study of main interaction effects. Since only 2 levels

A conservative estimate of 50% of contribution to PM

equal to 50% (0.1g/kWh) = 0.05 g/kWh from soot was
assumed. Dynamic injection begin was optimized in 13
ESC points for low NOx, soot emission and low SFC.

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

not be the best timing for fuel economy at that point.

Therefore, another set of timing was chosen based on
NOx-SFC trade off. With this timing and slight sacrifice
(marginal increase) in NOx emission, lower PM emission
and better fuel economy were expected. Similarly
injection timings were mapped at 13 ESC test points for
9 experiments. These unique injection timings were
used at each mode for each experiment (with defined
control factors as per L9 orthogonal array) with objective
of obtaining low emission in ESC emission cycle with
good fuel economy.
ESC EMISSION TEST RESULTS
ESC emission tests were conducted for 9 experiments
as defined by L9 orthogonal array and emission
response NOx, CO, HC and PM were measured. These
results are presented in Table 8.

A. Cam
velocity

B. Nozzle
through flow

C. Nozzle tip
protrusion

D. No. of nozzle
orifice

Emission Response

Experiment No

Control factors

m/s

cc

mm

No

NOx-PM Trade

NOx

Euro III Targets

Figure 4: Timing optimization: Experiments 1, 2 & 3

at 1500 rpm, 100% load.
Figure 4 presents a representative case of optimization
of dynamic injection begin at a typical ESC emission test
point. The engine was run at constant speed and load at
13 emission points as required by ESC cycle (in this
case 1500 rpm and 100% load). Dynamic injection begin
was varied within thermal, torque, smoke and other
limits constraining the system. Raw emissions like NOx,
CO, HC and smoke were measured. Since HC and CO
emissions were very low compared to Euro III limits,
NOx-Soot tradeoff was the main criterion for selecting
best injection timing. The timing which gives the best
NOx-Soot tradeoff at a particular speed and load may

CO

HC

PM

g/kWh

Engineering Euro III targets

4.60

1.79

0.56

0.085

1.55

440

1.5

4.29

0.89

0.18

0.08

1.55

460

2.0

4.12

0.85

0.14

0.12

1.55

480

2.5

6*

4.62

0.73

0.13

0.09

1.60

440

2.0

6*

4.58

0.78

0.13

0.08

1.60

460

2.5

4.58

0.93

0.14

0.08

1.60

480

1.5

4.32

0.91

0.18

0.11

1.55*

440

2.5

4.33

0.94

0.13

0.12

1.55*

460

1.5

6*

4.35

0.91

0.14

0.09

1.55*

480

2.0

4.37

0.96

0.15

0.09

Table 8: Euro III ESC Emission test cycle results of

L9 orthogonal array
It was observed from Table 8 that CO and HC emission
were relatively easier to meet due to presence of higher
concentrations of oxygen in intake air during diesel
combustion. CO and HC emission of 9 experiments
were significantly lower than engineering targets set for
Euro III. Comparing NOx and PM similarly, indicated that
they were within Euro III limits close to engineering
targets.
Investigations [3] [4] [5] reveal NOx and PM increase of
one emission component leads to reduction of the other.
This relationship NOx and PM in 9 experiments of L9
orthogonal array are presented pictorially in Figure 5.

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

By examining the columns A, B, C and D of orthogonal

array L9 in Table 8, it could be seen that all factor levels
are equally presented in 9 experiments. Thus mean
response is balanced overall for the experimental region.
To find main effect of a factor at a particular level on a
characteristic emission response, magnitude of deviation
between that factor level and overall mean on response
was determined. These results from Analysis of Means
(ANOM) are presented in Figure 6.
INFERENCES FROM ANALYSIS OF MEANS
Following inferences were drawn from main effect plots
presented in Figure 6.
x

CO and HC emissions for both injection timings are

not only lower than Euro III emission limits but far
lower than the engineering Euro III targets (15%
reduction in emission compared to standard Euro III
limits). Euro III engineering emission targets for CO
and HC can therefore be met with safe margins with
any combination of injection parameters used within
the experimental region.

While NOx-PM tradeoff timing is suitable for meeting

Euro III engineering NOx limits safely (>8%
margins), PM limits can be met, but with a safety
margin of <15%.

Increased cam velocities (1.6m/s), result in

increased injection rate. Consequently, higher
adiabatic flame temperature and high combustion
pressures result which favor NOx formation due to
dissociation of air into unstable nitrogen and oxygen
species that readily react to form NOx. Use of lower
number of nozzle spray orifices has a similar effect
on NOx emission as higher injection duration of
issuing fuel jet from nozzle orifice tends to burn the
fuel over larger crank angles, increasing the average
combustion pressures and temperature.

In contrast to NOx, PM emission reduces

significantly with high injection rate. Presence of
high pressures and temperature with high cam
velocities helps to burn injected fuel completely.
Unburned HC in dead volumes and traces of oil
particles present from blowby is oxidized fully
leaving no scope for formation of soot. Since the
tendency for formation of constituents of PM is
lower, PM emissions are lower.

Use of lower number of nozzle spray orifices (6 no.)

also helps in lowering PM emission because
influences of interference between adjacent spray
jets are minimized as the jets are spread wide apart
with increased horizontal angles. Tendency for
formation of rich regions of fuel particles near the
periphery of the combustion chamber is also
minimized as an outcome. Furthermore, increased
injection pressures result when lower number of
spray orifices are used leading to lower PM
emissions. In addition to these factors, Euro III
engine design used wider combustion bowl for lower

Figure 5: Relationship between NOx-PM emissions

ANALYSIS OF MEANS (ANOM)

Effect of factor level is the deviation it causes from
overall mean. In each experiment four response
characteristics (emission components NOx, CO, HC,
PM) were observed with best injection timing for NOx,
PM and SFC. Contributions to main effects by the
control factors were analyzed in order to quantify their
magnitude of influence at each factor and its level on
response.

Figure 6: Main effect plots for different control

factors and their levels.

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

compression ratio with reduced swirl number

compared to Euro II engine. These factors, although
not evaluated, also partially contribute to lower PM
emission since they together reduce the angular
momentum of the spray droplets near the periphery
of the combustion chamber preventing interference
of spray jets.
x

Lower nozzle through flow increases injection

duration and injector end pressures and
consequently increases the tendency of constant
pressure combustion in diesel engine and helps
lower PM emission at the cost of increased NOx
emission. In contrast, higher nozzle through flow
shortens the injection duration, lowers flame
temperatures and combustion pressures and favors
NOx emissions. These effects can be clearly
interpreted from the overall trends of NOx and PM
emission in the main effect plots.
Neither too high nor too low values of nozzle tip
protrusion into combustion chamber are desirable.
While excessive protrusion of nozzle tip into
combustion bowl increases the tendency of
combustion chamber wall wetting, reduced spray
penetration and resulting in partially burned fuel due
to wall quenching effects leading to poor emission.
Inadequate nozzle protrusion results in spray jets
being targeted on piston crown above the
combustion bowl leading to poor emissions. 1.5 mm
nozzle tip protrusion into the combustion bowl was
found to be most optimum of different levels
experimented.

Table 9: Injection parameters and their predicted

emission responses
It can be inferred from Table 9 and Figure 7 that
minimum NOx configuration does not meet Euro III PM
limits and thereby the project objective. Since CO and
HC engineering limits can be met with any combination
of injection parameters, ranking of predicted emission
results was carried out based on NOx-PM tradeoff. The
injection parameter responsible for producing lowest PM
emission also significantly reduces NOx and therefore
ranked first.

SELECTION OF OPTIMUM FACTOR LEVELS

Use of following combinations of injection parameters
presented in Table 9 with NOx-PM tradeoff timing would
result in lowest NOx, CO, HC and PM respectively.
These conclusions are inferred from main effect plots in
Figure 6 by studying each control factor and its level
necessary for producing lowest emission for each
exhaust component.
PREDICTION OF EMISSION RESULTS
One of the primary reasons for using robust design was
to predict the exhaust emission response of the engine
with defined control factors and their levels through
analysis of data, thereby saving precious time and
resources and quickly converge at the results. Using the
combination of control factors and their levels necessary
for producing lowest NOx, CO, HC and PM respectively
the response characteristics of the system with NOx-PM
tradeoff timing was predicted. These results and
injection parameters required to produce them are
presented in Table 9.
The predicted emission results are ranked based on
severity, importance of emission component. Factors
like margins to meet Euro III engineering limits, criticality
of NOx-PM tradeoff, low criticality of CO and HC
emission were considered for ranking.

Figure 7: Predicted emissions Vs Euro III targets

ANALYSIS OF SIGNAL TO NOISE RATIO (S/N)

During the course of experimentation for meeting Euro
III, simultaneous development of turbocharger and
intercooler took place. A change in turbine diameter of
turbocharger was done in order to improve airflow
characteristics at high engine speeds and loads with
intent of improving PM emission. Likewise, the
intercooler development resulted in lower pressure drop
across the intercooler leading to lower specific volume
and low temperature at intercooler outlet. Consequently,
higher mass of compressed air per unit volume resulted
due to increase in mass density of air, altering the
emission characteristics, particularly NOx-PM.

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

Since these changes were unplanned, difficult,

expensive to control and beyond the scope of this
investigation, but at the same time drastically influenced
emissions, they were considered as noise. Other noise
factors during experimentation like deterioration of the
engine, gradual drift of measuring instruments, etc. were
considered less important and ignored, considering the
short time line for development and frequency of
preventive maintenance of measuring instruments.
Possibility of interacting control factors is also a form of
internal noise. Therefore Dr.Taguchi [11] proposes use
of signal-to noise ratio as countermeasure to overcome
both internal and external noise for recommending
optimum control factor and their levels by studying both
the main effects and signal-to-noise ratio.

noise ratio for NOx, response. In other words, use of

these factors and levels would result in lowest NOx
under influence of noise. During experimentation,
variation of turbocharger and intercooler were
considered as noise. Hence combination of these
injection parameters can be used in criticality of NOx
emissions.
x

As seen in analysis of means, use of control factor

combination A1, B2, C1 and D2 for low NOx
emission would result in predicted PM emission out
of Euro III limits and hence this combination was no
longer a viable option for meeting Euro III norms.

On the other hand, control factors A2=1.60m/s cam

velocity, B1=440cc/30s through flow, C1=1.5mm
nozzle tip protrusion and D1=6 orifice nozzle have
highest S/N ratio when PM emissions are
concerned. Hence this combination and levels of
control factors are more robust to influence of noise
than other factor levels and form an optimum
combination for low PM emissions. In addition to
producing low PM, predicted values of NOx, CO and
HC are lower than the Euro III engineering targets
(refer Figure 7). Hence, it was decided to use A2,
B1, C1 and D1 for confirmation run and validation of
analysis.

The objective function of signal-to-noise ratio was to

reduce all emission components, thus smaller-the-better
emission characteristic was used. Since the objective of
this analysis was to maximize the useful output energy
and minimize harmful output energy, maximize signal-tonoise ratio is desirable for each emission component.
Results of this analysis are presented in Figure 8.

ANALYSIS OF VARIANCE (ANOVA)

Different control factors (cam velocity, nozzle through
flow, nozzle tip protrusion, number of sprays orifices)
affect NOx, CO, HC and PM to different degrees. The
relative magnitude of factor effects can be judged from
Figure 8 that gives average signal-to-noise ratio of each
factor level.
In order to quantify and get a better feel of relative effect
of different factors, estimate the error variance and
prediction error, the results were analyzed further by
decomposition of variance called ANOVA. Results of this
analysis are presented in Table 10.
x

Referring to sum of squares column in Table 10,

number of nozzle spray orifices makes significant
contribution of 39.8% to the total sum of squares of
NOx emission. Nozzle tip protrusion into combustion
bowl and cam velocity contributes 27.5% and 20.0%
to total sum of squares for NOx emission. Similarly,
number of nozzle orifices and cam velocity
contribute 91.8% and 6.5% respectively for total sum
of squares for PM emission. Nozzle through flow
and nozzle tip protrusion together make only a
negligible contribution to total sum of squares for PM
emission.

Larger the contribution to the total sum of squares,

the larger is the ability of the factor to influence the
signal-to-noise ratio. This analysis reveals that
number of nozzle orifices influences NOx emission
significantly, followed by factors like nozzle tip
protrusion and cam velocity which influence the
signal-to-noise ratio by almost equal amounts. Both
number of nozzle orifices and cam velocity similarly

Figure 8: Signal-to-noise ratio of control factors and

levels for emission components
INFERENCES FROM S/N RATIO
The objective function is to maximize signal-to-noise
ratio, therefore the optimum factor is the one which
produces the highest value of signal to noise ratio in the
experimental region.
x

Factors A1=1.55m/s cam velocity, B2=460cc/30s

through flow, C1=1.5mm nozzle tip protrusion and
D2=7 orifice nozzle would give highest signal to

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

have significant influence on PM emissions and their

signal-to-noise ratios.

VALIDATION AND CONSISTENCY TESTS

The magnitude of factor effect relative to error

variance can be judged from calculated F-ratio. The
larger the F-ratio, the larger the factor effect is
compared to error variance. Also the larger the Fratio, the more important the factor is to that quality
characteristic. On these lines, the control factors are
ranked in descending order of their influence on
signal-to noise ratio for each exhaust emission
component.

Based on Analysis of Means (ANOM), Analysis of

signal-to-noise ratio (S/N) and Analysis of Variance
(ANOVA) optimum condition of control factors and their
levels required for producing emissions to meet the Euro
III engineering emission targets are :

Madhav S Phadke [11] recommends grading into

different categories. Control factors are graded
small when F- ratio < 1, moderate when F ratio >
2 and significant when F-ratio > 4. The grading of
factors and their ranks for each characteristic
emission component like NOx, CO, HC and PM is
shown in Table 10.
The standard F-ratio is determined using 95%
confidence interval F-Tables [16] based on degree
of freedom for each factor and overall error. In all
instances where calculated F-ratio is greater than
standard F-Table value, it is considered important.
This case occurs only in PM emission characteristic
with F-ratio of number of nozzle orifices being
greater than the value obtained from standard
tables. It can be inferred that the nozzle orifice
contributes to variation in emission significantly.

Cam velocity
Nozzle through flow
Nozzle tip protrusion
Number of spray orifices

= A2 = 1.6m/s
= B1 = 440cc/30s
= C1 = 1.5 mm
= D1 = 6 Numbers

Incidentally, this combination of injection parameter is

ranked first, based on maximum emission margins
between predicted emission results and engineering
targets with NOx-PM tradeoff timing. Overall signal-tonoise ratio also indicates that injection parameters and
levels chosen above are more robust than others. With
this basis emission results were validated and compared
against predicted emission results. ESC test was
conducted with A2, B1, C1, D1 configuration with same
background factors as done for all other experiments
during development, with NOx-PM tradeoff timing.
Results obtained are presented in Figure 9.
Further, ESC emission tests were repeated to ensure
that control factors and their levels (A2, B1, C1 and D1)
are the ones responsible for producing lower emissions
and contribution from errors to emission response is
minimized.

Figure 9: Validation and consistency test results.

INFERENCES -VALIDATION AND CONSISTENCY

Table 10: ANOVA for exhaust emission response

It can be inferred from Figure 9 that, emission

results of optimized configuration of injection
parameters A2, B1, C1 and D1 indeed produce
emission results within Euro III engineering targets.

Averaged emission components from consistency

tests indicate that the objective of meeting Euro III

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

emissions with more than 8% margins on NOx and

15% margins on CO, HC and PM have been
realized comfortably. In fact 11.5% margins on NOx,
61.5% margins on CO, 76.5% margins on HC and
23% margins on PM emissions have been achieved
compared to the Euro III limits.
x

Consistency tests indicate that injection parameters

and their levels used and emission measurements
are repeatable and reliable for meeting Euro III
exhaust emission norms with safe engineering
margins.

The predicted emission results through analysis,

and averaged emission results achieved by
experimentation, are in close proximity with each
other. This demonstrates that design of experiments
and Taguchi technique are powerful tools which,
when used properly, can meet the project objectives
effectively and efficiently.

The experimental error between predicted emission

components and actual measurements of emissions
are 1.65% for NOx, 7.5% for CO, 8.8% for HC and
6.8% for PM. This is a relatively small percentage
compared to contribution by the injection
parameters. This indicates that a major contribution
to the project has been through correct selection
and use of injection parameters.

This project has demonstrated yet another case of

successfully achieving the target objectives using
design of experiments and Taguchi techniques in
automotive domain, reinforcing the fact that these
techniques are an indispensable part of robust
design.

Figure 10 : Engine performance characteristic

comparison Euro II Vs Euro III

CONCLUSIONS

ENHANCEMENT OF ENGINE PERFORMANCE

Following conclusions are drawn from this investigation:

Using the finalized injection parameters, engine

performance characteristics were tuned. Factors A2 =
1.6m/s, B1 = 440cc/30s, C1 = 1.5 mm and D1 = 6 orifice
nozzles were used and injected fuel quantities and
dynamic begin of injection were tuned to achieve target
torque and power, low SFC, low smoke over entire
speed and load range of the engine. Figure 10 presents
comparison of Euro II performance and enhanced Euro
III performance characteristics.

Variation in NOx emission is caused by different

control factors, significant factors being number of
nozzle orifices, nozzle tip protrusion and cam
velocity which influence NOx by 39.8%, 27.5% and
20.0% respectively.

Number of nozzle orifices influences PM emission

significantly. It contributes almost 92% to PM
variation. Cam velocity influences variation of PM
emission by 6.5%. Changes in injection rate as a
consequence of change in number of nozzle orifices
or cam velocity play a significant role in increasing
the injection pressures and consequently reducing
PM emission. Nozzle tip protrusion and nozzle
through flow have negligible effect on PM emissions.

Injection parameters like cam velocity, nozzle

through flow, nozzle tip protrusion and number of
spray orifices can be used to meet Euro III with safe
margins. In addition, Euro III engineering targets are
achieved with electronic diesel control systems,
permitting flexible timing, high injection pressures
with suitable modifications on engine parameters
like compression ratio, swirl and combustion bowl as
shown in Figure 11.

It is noticed from enhanced Euro III engine performance

that the torque and power over the full load range have
is increased. Low end power was increased by 30% for
better driveability and acceleration performance, in
addition power at maximum torque speed (volumetric
efficiency zone) and at rated speed were increased by
33% and 38.5% for better load carrying capacity and
facilitate higher vehicle speed. Further overall reduction
in full load fuel consumption was achieved by reducing
SFC by a maximum 10.6%.

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

[3]
Soorajith R, Vinu H R, Electronic Diesel Control
A Strategy for Euro 3 Optimization, SAE paper 200428-0025
[4]
R M Petkar, C A Kardile, P V Deshpande,
Influence of Increased Diesel Fuel Spray Velocities and
Improved Spray Penetration in DI Engines SAE paper
2004-01-0031
[5]
Daniel W Dickey, Thomas W Ryan, Andrew C
Matheaus, NOx Control In Heavy Duty Diesel Engines
What Is The Limit ? , In-cylinder diesel particulates and
NOx Control, SAE paper 980174.

Figure 11: Euro II to III trends in engine parameters

and injection systems
x

In contrast to set objectives of 8% emission margins

on NOx and 15% each on CO, HC and PM, Euro III
targets are achieved consistently and successfully,
as shown in Figures 9 and 11, with 11.5% margins
on NOx, 61.5% on CO, 76.5% on HC and 23% on
PM emissions compared to Euro III limits.
Saturation tradeoffs between NOx-PM-SFC have to
be simultaneously studied so that engine does not
suffer from excess emission or poor fuel
consumption.
Carefully planned and tailored orthogonal arrays,
DOE and Taguchi technique, with clearly set
objectives, significantly helps engineers to
investigate problems and quantify effects of
experimental factors in a short time frame
economically.

ACKNOWLEDGEMENTS
The author would like to thank Motor Industries
Company Ltd (BOSCH Group) Management for having
given the opportunity to publish the findings of this
research in the form of a technical paper.
The author also expresses gratitude to his superiors,
peers and colleagues who were instrumental in
providing their rich experiences, suggestions and
guidance which resulted in shaping the technical paper
to current form.

REFERENCES
[1]
Klaus Krieger, Diesel Fuel Injection Technology
An Essential Contribution Towards An Environment
Friendly Powerful Diesel Engine 2000-01-1429
[2]
Max Straubel and Klaus Krieger Distributor
Type Injection Pump for High-Speed Diesel Engines with
Direct Injection SAE paper 872222

[6]
S Kampmann, B Dittus, P Mattes, M Kirner, The
Influence Of Hydro Grinding On VCO Nozzles On The
Mixture Preparation In DI Diesel Engines, Fuel Spray
Technology and Applications, SAE Paper 960867
[7]
C Heimgartner and A Leipertz, Investigation Of
Primary Spray Breakup Close To The Nozzle Of A
Common Rail, High Pressure Diesel Injection System,
Spring Fuels and Lubricants meeting and Exposition
SAE paper 2000-01-1799.
[8]
C.Arcoumanis, ST Cho, M Gavaises and HS Yi,
Spray and Combustion Development in a Four-Valve
Optical DI Diesel Engine, Imperial College of Science,
technology and medicine, 2000-01-1183
[9]
Lu Lin, Duan Shulin, Xiao Jin, Wu Jinxiang &
Gao Xiaohong, Effects Of Combustion Chamber
Geometry On In-Cylinder Air Motion & Performance In
DI Diesel Engine In-cylinder diesel PM & NOx control,
SAE paper 2000-01-0510.
[10]
Gerald J Hahn, Some Things Engineers
Should Know About Experimental Design, Journal of
Quality technology, Vol 9, No 1, January 1977.
[11]
Madhav .S. Phadke, Quality Engineering Using
Robust Design, AT&T Bell Laboratories, Prentice Hall,
Englewood Cliffs, New Jersey, 1989.
[12]
American Supplier Institute Interaction and
Countermeasures, 1998.
[13]
N Logothetis, Excerpts from Managing for total
quality-From Deming to Taguchi
[14]
European Union directive 1999/96/EC of the
European parliament and of the council, 13th Dec 1999.
[15]
G Greeves and C H T Wang, Origins of Diesel
Engine Particulate Emissions, SAE Paper 810260,
1982.
[16]
F-Distribution
Tables
from
Douglas
C
Montgomery, Design and Analysis of Experiments
John Wiley and Sons, 5th Edition, 2001
[17]
N Guerrassi and P Dupraz, A Common Rail
System For High Speed Direct Injection Diesel Engine,
SAE paper 980803.
[18]
Noboru Uchida, Kiohiro Shimokawa, Yugo Kudo
and Masatoshi Shimoda, Combustion Optimization By

Downloaded from SAE International by Kirloskar Oil Engines Ltd, Saturday, June 13, 2015

Means Of Common Rail Injection System For Heavy

Duty Diesel Engines, 982679
[19]
Klaus Krieger, Excerpts From Innovative
Injection Systems for Clean Diesel Engines, Robert
Bosch GmbH

CONTACT
Krishnan C.L
(Product Development)
BOSCH Limited
PO Box 3000, Hosur Road, Adugodi,
Bangalore - 560030
E Mail Krishnan.CL@in.bosch.com