‘ ‘ H.

 

 

 

:3 4
4- .». .M.
‘3

2, .‘
“a“.uwmad.

.9 “hr
3.5.,3

Int.
:5 A.

:1?

.1 1...:
«yawn

é
may.
.

 

it 2. 5% 1353: 2% 9,5 a5 ..
.15...Fﬁo-a:9§ 1.. ...w..m..mwmwﬁ..mm.h‘ 1..

n ~O .4. hul‘ 9v

 

 

., zrhdlth I 9?: .1
, » ‘k\.\a::xx\w\$.
3.4\|.« 53...}. {IR
€11.39: 1‘ {3%.}
a 4! 3.1,: . El.
1.:

5:2..-
:tt1...

.1... a?
K tv
ix .3.

 

1...}...zﬂﬂ'm

.2. l 5.179324 x
& 13!...I‘gci‘...

 

1v. 1!: ‘
.23. k9

 

 

 

 

 

 

 

 

 

‘1‘? 3.70:
v
4

@»§%ﬁ .e._%%..

. 111...... .V :3!

  

'r-II‘ '-

 

i LIBRARY
0 Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

FLOW MEASUREMENTS AND IN-CYLINDER COMBUSTION
DIAGNOSIS IN AN INTERNAL COMBUSTION ENGINE
ASSEMBLY

presented by

Mayank Mittal

has been accepted towards fulﬁllment
of the requirements for the

Doctoral degree in Mechanical Engineering

WQWM

Major/Professor’s Signature
/2/ 3/0 7

Date

 

MSU is an Afﬁnnative Action/Equal Opportunity Employer

 

 

 

I
I PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

' MAY BE RECALLED with earlier due date If requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

SIDS KzlProj/AocaPresICIRCIDateDueindd

 

 

 

FLOW MEASUREMENTS AND IN —CYLINDER COMBUSTION
DIAGNOSIS IN AN INTERNAL COMBUSTION ENGINE
ASSEMBLY

By

Mayank Mittal

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

Mechanical Engineering

2009

ABSTRACT

FLOW MEASUREMENTS AND IN-CYLINDER COMBUSTION DIAGNOSIS IN AN
INTERNAL COMBUSTION ENGINE ASSEMBLY

By

Mayank Mittal

The flow ﬁelds inside the engine cylinder are extremely complex and exhibit
large cycle—to-cycle variations. It is the most important factor that controls the
combustion process. It governs the fuel-air mixing and burning rates inside the engine
cylinder. Therefore, it is desirable to improve the measurement techniques for velocity
measurements and scalar properties pertaining to the in-cylinder ﬂows. In this
dissertation, molecular tagging velocimetry (MTV) is used to obtain the multiple point
measurement of the instantaneous velocity ﬁeld inside the engine cylinder. MTV is a
molecular counterpart of particle-based techniques, and it eliminates the use of seed
particles.

In the ﬁrst part of this work, an experimental study is performed to investigate the
effects of charge motion control on in-cylinder ﬂovv (using MTV). It is found that the
charge motion control has a profound effect on cycle-to-cycle variations during the intake
and early compression; however, its inﬂuence reduces during the late compression. In-
cylinder engine ﬂow measurements are extended to obtain an instantaneous three-
component velocity ﬁeld using stereoscopic molecular tagging velocimetry (SMTV). The
image-processing technique, implemented to obtain the three-components of velocity,
involves two major steps: (i) calibration process and (ii) data acquisition and reduction.

Preliminary results show that the cycle-to—cycle variations are more prominent in the

 

velocity component perpendicular to the tumble plane, as opposed to the in-plane
components. Such new insights will help better understand the details of these ﬂows and
further improve CFD models for IC engines.

Experimental measurements provide useful information of the ﬂow ﬁelds inside
the engine cylinder. However, it is multi-dimensional numerical simulations that offer the
potential of signiﬁcant time and cost savings to design the engine with improved
performance. To date, numerical simulations of in-cylinder flows are performed with
assumed boundary conditions. Due to this, ﬂow measurements are performed inside the
intake manifold of an engine assembly that can provide real-time boundary conditions for
more accurate multi-dimensional numerical simulations. The geometry of the intake
manifold is simpliﬁed for this purpose. A hot-wire anemometer and piezoresistive type
absolute pressure transducers are used to measure the velocity and pressure, respectively.
In-cylinder ﬂow measurements are also performed (using SMTV) to validate the
modeling efforts.

In the second part of this work, in-cylinder combustion diagnosis is performed
inside an ethanol-gasoline, dual fueled, single-cylinder spark ignition engine. A dual fuel
injection system with both direct-injection (DI) and port-fuel-injection (PFI) is used. The
cycle-to-cycle variability is presented using the coefﬁcient of variation of indicated mean
effective pressure. Mass fraction burned and burn duration are determined from the

analysis of measured in-cylinder pressure data.

To my parents for their unconditional love and support,
to Kanu, the great companion of my life,

and to Mridu, my magniﬁcent daughter.

iv

ACKNOWLEDGEMENTS

First, I would like to thank my advisor, Dr. Harold Schock, for giving me the
opportunity to work at the Automotive Research Experiment Station (ARES). It has been
an enormous experience working under his supervision, which enabled me to think
independently and helped me to rise to the high standards of research. I thank him for
providing the ﬁnancial support during my graduate studies at Michigan State University.

I thank my thesis committee members, Dr. Manoochehr Koochesfahani, Dr.
Farhad Jaberi and Dr. Tim Hogan for their time to review my manuscript and make
valuable suggestions.

I am grateful to Dr. Reza Sadr, who advised me from the early stages of my
research on molecular tagging velocimetry. He also helped me to understand the concepts
of stereoscopic molecular tagging velocimetry and to develop various image processing
algorithms. I extend my thanks to Amir Naqwi for his valuable insights in calibration
process. This work would have not been possible without their valuable inputs. I am
thankful to Dr. Ahmed Naguib, who made the hot wire available for the velocity
measurements inside the intake manifold of the engine assembly. I would also like to
thank Dr. Guoming Zhu and Dr. David Hung for the discussions on combustion
diagnosis.

I thank my colleagues at ARES, for sharing their valuable time and helping me
with their valuable suggestions. I would like to specially thank Andrew F edewa and Tom
Stuecken for their help in experiments. I thank Jeff and Gary for understanding me and

providing the necessary help when required.

Finally, special thanks to my family. I would sincerely like to thank my parents
for the good things they taught me during my childhood, their unconditional love and
support throughout my life. Also, none of this would have been possible without the love,

patience and continual encouragement of my wife. She was always with me during the

stressful times.

vi

TABLE OF CONTENTS

LIST OF TABLES .................................................................................. ix
LIST OF FIGURES ................................................................................. x
LIST OF SYMBOLS .............................................................................. xv
LIST OF ABBREVIATIONS .................................................................... xvi
1. Introduction ....................................................................................... l
1.1. Background ................................................................................. I
1.2. Literature review ........................................................................... 3
1.2.1. Pre-combustion analysis .......................................................... 3

1.2.2. Combustion analysis ............................................................. 13

1.3. Contents of the Present Work ........................................................... 15

2. A study of cycle-to-cycle variations on in—cylinder ﬂow in an IC engine equipped
with CMCV using MTV ...................................................................... 18
2.1. Introduction ................................................................................ 18
2.2. Experimental setup and procedure ...................................................... 20
2.2.]. Experimental setup ............................................................... 20

2.2.2. Charge motion control valve .................................................... 26

2.2.3. Melecular tagging procedure ................................................... 26

2.3. Experimental Results ..................................................................... 31
2.3.1. Swirl measurement plane, 1500 rpm ........................................... 31

2.3.2. Swirl measurement plane, 2500 rpm ........................................... 42
2. 3 3. Tumble measurement plane, 1500 rpm ........................................ 46

2. 3. 4. Tumble measurement plane, 2500 rpm ........................................ 49

2. 4 Summary” ..... 52

3. In-cylinder ﬂow measurement using stereoscopic molecular tagging velocimetry. .55
3.1. Introduction 55
3.2. Experimental setup 57
3.3. Measurement Procedure 61

3.4. Experimental Results ..................................................................... 70
3.4.]. Flow ﬁeld at 90 CAD 70

3.4.2. Flow ﬁeld at 180 CAD ........................................................... 73

3.4.3. Flow ﬁeld at 270 CAD ........................................................... 75

3.5. Summary ................................................................................... 77

4. Intake manifold and in-cylinder ﬂow measurements inside an IC engine assembly ..79
4.1. Introduction ................................................................................ 79
4.2. Experimental setup ....................................................................... 81

vii

4.3. Experimental Results ..................................................................... 88

4.3.1. Flow measurements at 900 rpm engine speed ................................ 89

4.3.2. Flow measurements at 1200 rpm engine speed ............................. 100

4.4. Summary .................................................................................. 105
5. In—cylinder combustion diagnosis in an ethanol-gasoline, dual-fueled, spark ignition
engine .......................................................................................... 106
5.1. Introduction .............................................................................. 106
5.2. Experimental setup ..................................................................... 109
5.3. Measurement Procedure ............................................................... 111
5.3.1. Mass fraction burned analysis ................................................. 112
5.3.1.1. Model 1, Rassweiler and Withrow method ......................... 112

5.3.1.2. Model 2, Net Pressure method ....................................... 115

5.4. Experimental Results ................................................................... 116
5.4.1. PF I-E85 and DI-Gasoline system ............................................. 117

5.4.2. DI-E85 and PFI—Gasoline system ............................................. 121

5.4.3. PF I-Gasoline and DI-Gasoline system ....................................... 124

5.5. Improving the MFB calculation using net pressure method ....................... 126
5.6. Summary ................................................................................. 129

6. Final remarks ................................................................................. 132
7. References ..................................................................................... 135

viii

 

LIST OF TABLES
Table 2.1: Engine speciﬁcations ................................................................. 22
Table 2.2: Properties of biacetyl ................................................................. 28
Table 5.1: Properties of ethanol and gasoline ................................................ 107
Table 5.2: Test matrix ........................................................................... l 10

ix

LIST OF FIGURES

Figure 2.1: Optical engine assembly ............................................................ 21
Figure 2.2: Intake and exhaust valves lift ...................................................... 22
Figure 2.3: Tumble and swirl measurement planes inside the 1C engine cylinder ........ 23
Figure 2.4: Schematic of MTV optical setup for in—cylinder ﬂow measurement ......... 24
Figure 2.5: Front view of LPX2051 .............................................................. 25
Figure 2.6: Rear view of LPXZOSi ............................................................... 25
Figure 2.7: Charge motion control valve; (3) deactivated, and (b) activated .............. 26
Figure 2.8: Absorption spectrum for biacetyl .................................................. 28

Figure 2.9: Phosphorescence emission spectrum for biacetyl at four different excitation
wavelengths .......................................................................................... 29

Figure 2.10: Undelayed (leﬁ) and delayed (right) images for tumble measurement plane
at 138 crank angle degree ......................................................................... 30

Figure 2.11: Ensemble-averaged velocity vectors, contours of velocity ms for
deactivated (left) and activated (right) CMCV at (a) 121 (top), and (b) 300 (bottom)
CADs; swirl measurement plane at 1500 rpm .................................................. 33

Figure 2.12: Ensemble-averaged velocity vectors with activated CMCV at 121 CAD
considering (a) 1-100, (b) 101-200, and (b) 1-500 consecutive undelayed frames; swirl
measurement plane at 1500 rpm .................................................................. 34

Figure 2.13: Instantaneous velocity ﬁelds from consecutive engine cycles 99", 1001’ and
101C, and ensemble—averaged velocity ﬁeldd with activated CMCV at 121 CAD; swirl
measurement plane at 1500 rpm .................................................................. 36

Figure 2.14: Instantaneous velocity ﬁeld" and mean subtracted ﬂow ﬁeldb for 101 st cycle
at 121 CAD with activated CMCV; swirl measurement plane at 1500 1pm ............... 37

Figure 2.15: PDFs of normalized circulation at different CADs for swirl measurement
plane with CMCV deactivated at 1500 rpm engine speed .................................... 39

Figure 2.16: PDFs of normalized circulation at different CADs for swirl measurement
plane with CMCV activated at 1500 rpm engine speed ....................................... 40

Figure 2.17: Standard deviation of normalized circulation and turbulent kinetic energy at
different CADs for swirl measurement plane at 1500 rpm engine speed ................... 42

Figure 2.18: Ensemble—averaged velocity vectors, contours of velocity rms for
deactivated (left) and activated (right) CMCV at (a) 171 (mp), and (b) 221 (bottom)
CADs; swirl measurement plane at 2500 rpm engine speed ................................. 44

Figure 2.19: PDFs of normalized circulation at different CADs for swirl measurement
plane with CMCV deactivated at 2500 rpm engine speed .................................... 45

Figure 2.20: PDFs of normalized circulation at different CADs for swirl measurement
plane with CMCV activated at 2500 rpm engine speed ....................................... 45

Figure 2.21: Standard deviation of normalized circulation and turbulent kinetic energy at
different CADs for swirl measurement plane at 2500 rpm engine speed ................... 46

Figure 2.22: PDFs of normalized circulation at different CADs for tumble measurement
plane with CMCV deactivated at 1500 rpm engine speed .................................... 47

Figure 2.23: PDFs of normalized circulation at different CADs for tumble measurement
plane with CMCV activated at 1500 rpm engine speed ....................................... 48

Figure 2.24: Standard deviation of normalized circulation and turbulent kinetic energy at
different CADs for tumble measurement plane at 1500 rpm engine speed ................. 49

Figure 2.25: PDFs of normalized circulation at different CADs for tumble measurement
plane with CMCV deactivated at 2500 rpm engine speed .................................... 50

Figure 2.26: PDFs of normalized circulation at different CADs for tumble measurement
plane with CMCV activated at 2500 rpm engine speed ....................................... 51

Figure 2.27: Standard deviation of normalized circulation and turbulent kinetic energy at
different CADs for tumble measurement plane at 2500 rpm engine speed ................. 52

Figure 3.1: Basic conﬁguration of lens translation method for stereoscopic imaging ....57

Figure 3.2: Basic conﬁguration of angular displacement method for stereoscopic imaging

......................................................................................................... 57
Figure 3.3: Tumble measurement plane inside an IC engine cylinder ...................... 59
Figure 3.4: Schematic of SMTV optical setup ................................................. 60
Figure 3.5: Calibration Target .................................................................... 62

xi

Figure 3.6: Undelayed (left) and delayed (right) images at 90 CAD SMTV measurements

for right camera, time delay (At) = 10 us ...................................................... 65
Figure 3.7: Ensemble-averaged velocity (leﬁ) and out-of-plane mean vorticity (right) at
90 CAD .............................................................................................. 71
Figure 3.8: RMS of u, v and w velocity components at 90 CAD ............................ 73
Figure 3.9: Ensemble-averaged velocity (left) and out-of—plane mean vorticity (right) at
180 CAD ............................................................................................. 74
Figure 3.10: RMS of u, v and w velocity components at 180 CAD ......................... 75
Figure 3.11: Ensemble-averaged velocity and out~of~plane mean vorticity at 270 CAD
......................................................................................................... 76
Figure 3.12: RMS of u, v and w velocity components at 270 CAD ......................... 76
Figure 4.1: Experimental rig showing measurement planes and locations inside the intake
manifold for ﬂow measurements ................................................................. 82
Figure 4.2: Intake port geometry ................................................................. 82
Figure 4.3: Combustion analysis system (CAS) used for intake manifold flow
measurement ........................................................................................ 85
Figure 4.4: Experimental rig for in-cylinder ﬂow measurement using stereo MTV ......86
Figure 4.5: Tumble measurement plane inside the engine cylinder ......................... 86
Figure 4.6: Schematic of SMTV optical setup ................................................. 88
Figure 4.7: Averaged in-cylinder pressure trace at 900 rpm engine speed ................. 89

Figure 4.8: Averaged pressure trace at the circular section of the intake manifold at 900

rpm engine speed ................................................................................... 90
Figure 4.9: Averaged velocity proﬁle at different radial locations in the circular section
of the intake manifold at 900 rpm engine speed ................................................ 91
Figure 4.10: Intake valve lift and its ﬁrst derivative .......................................... 92

Figure 4.11: Averaged velocity at different radial locations 2 (0 mm), 3 (3 mm), 4 (21
mm), 5 (24 mm) and at the wall (26 mm) during the intake stroke at various crank angle
degrees; 900 rpm engine speed ................................................................... 92

xii

Figure 4.12: Averaged pressure trace at the circular and elliptical sections of the intake

manifold at 900 rpm engine speed ............................................................... 93
Figure 4.13: Calibration images of left (left) and right (right) cameras ..................... 95
Figure 4.14: Undelayed (left) and delayed (right) images at 120 crank angle degree for
left (upper) and right (lower) cameras at 900 rpm engine speed ............................. 96
Figure 4.15: Ensemble-averaged velocity ﬁled (left) and velocity nns (right) at 120 crank
angle degree and 900 rpm engine speed ......................................................... 97
Figure 4.16: Ensemble-averaged velocity ﬁled (left) and velocity rms (right) at 140 crank
angle degree and 900 rpm engine speed ......................................................... 98
Figure 4.17: Ensemble-averaged velocity ﬁled (left) and velocity rms (right) at 260 crank
angle degree and 900 rpm engine speed ......................................................... 99
Figure 4.18: Ensemble-averaged velocity ﬁled (left) and velocity rms (right) at 280 crank
angle degree and 900 rpm engine speed ....................................................... 100
Figure 4.19: Averaged in—cylinder pressure trace at 1200 rpm engine speed ............ 101
Figure 4.20: Averaged pressure trace at the circular and elliptical sections of the intake
manifold at 1200 rpm engine speed ............................................................ 101
Figure 4.21: Averaged velocity proﬁle at different radial locations in circular section of
the intake manifold at 1200 rpm engine speed ................................................ 103
Figure 4.22: Ensemble-averaged velocity ﬁled (left) and velocity rms (right) at 170 crank
angle degree and 1200 rpm engine speed ...................................................... 103
Figure 4.23: Ensemble-averaged velocity ﬁled (left) and velocity rms (right) at 210 crank
angle degree and 1200 rpm engine speed ...................................................... 104
Figure 5.1 : Experimental rig .................................................................... 1 l 1
Figure 5.2: Measured pressure data, Test C2 (PF I-gasoline 100 %) ...................... 114

Figure 5.3: Indicator diagram in logarithmic scales, Test C2 (PFI-gasoline 100 %) ....115

Figure 5.4: Mean IMEP (top) and COV.mcp (bottom) for PFI-E85 and DI-gasoline, and
DI-E85 and PF I-gasoline systems ............................................................... 117

Figure 5.5: Averaged Pressure (top) and MFB curves (bottom) for PFI-E85 and DI-
gasoline system at 3.3 bar IMEP and 1500 rpm .............................................. 119

xiii

Figure 5.6: Averaged Pressure (top) and MFB curves (bottom) for PFI-E85 and DI-
gasoline system at WOT and 1500 rpm ........................................................ 120

Figure 5.7: Averaged Pressure (top) and MFB curves (bottom) for PFI-E85 and DI-
gasoline system at WOT and 2500 rpm ........................................................ 121

Figure 5.8: Averaged Pressure (top) and MFB curves (bottom) for DI-E85 and PFI-
gasoline system at 3.3 bar IMEP and 1500 rpm .............................................. 123

Figure 5.9: Averaged Pressure (top) and MFB curves (bottom) for DI-E85 and PFI-
gasoline system at WOT and 1500 rpm ........................................................ 124

Figure 5.10: Averaged Pressure (top) and MFB curves (bottom) for PFI and DI-gasoline
system at 3.3 bar IMEP and 1500 rpm ......................................................... 125

Figure 5.11: Averaged Pressure (top) and MFB curves (bottom) for PF I and DI-gasoline
system at WOT and 1500 rpm .................................................................. 126

Figure 5.12: MFBs for PFI-gasoline and DI-gasoline system at 3.3 bar IMEP (top) and
WOT (bottom) .................................................................................... 129

xiv

LIST OF SYMBOLS

 

At Time delay
F Mapping function
¢ Fuel-to-air equivalence ratio
k Turbulent kinetic energy
,1 Air—to-fuel ratio
A Wavelength
n Polytropic index
”C Polytropic index of compression
"e Polytropic index of expansion
0) Vorticity
In—cylinder pressure
3,, Mean piston speed
0' Standard deviation
u, v, w x, y and 2 components of velocity
u', v', w' x, y and 2 components of ﬂuctuating velocity
(u) , (v) , (w) Ensemble averaged of u, v and w velocity components
V In-cylinder volume
V d Displaced in-cylinder volume
x, y, z Cartesian coordinates in physical space
X, Y Cartesian coordinates in image plane

XV

BDC

CAD

CAS

CCD

CFD

CI

CMCV

CO

COV

DI

E85

EOC

H WA

IC

IMEP

LDV

MFB

MON

MPP

MTV

 

 

LIST OF ABBREVIATIONS

Bottom dead center
Crank angle degree

Combustion analysis system
Charge coupled device
Computational ﬂuid dynamics
Compression-ignition

Charge motion control valve
Carbon monoxide

Coefﬁcient of variation
Direct-injection

Blend of 85% ethanol and 15% gasoline by volume
End of combustion

Hot-wire anemometry

Internal combustion

Indicated mean effective pressure
Laser doppler velocimetry

Mass fraction burned

Motor octane number
Measurement plane pointers

Molecular tagging velocimetry

xvi

NC
NOx
PDF
PF I
PIV
RMS
RON
RPM
SCV
SI
SMTV
SOC
SPIV
TDC
TKE
UV
VCT

WOT

 

 

Normalized circulation

Nitrogen oxides

Probability density function
Port-fuel-injection

Particle image velocimetry
Root-mean-square

Research octane number

Revolutions per minute

Swirl control valve

Spark-ignition

Stereoscopic molecular tagging velocimetry
Start of combustion

Stereoscopic particle image velocimetry
Top dead center

Turbulent kinetic energy

Ultraviolet

Variable camshaft timing

Wide open throttle

xvii

CHAPTER-1

INTRODUCTION

1.1 Background

Internal combustion (IC) engines, both spark—ignition (SI) and compression-
ignition (CI) types, are distinct from most other power producing and propulsion systems
in an important way. The working ﬂuid undergoes the processes essential for the

production of power—compression, fuel energy release by combustion, and expansion——

within a single chamber, known as the engine’s cylinder (Heywoodl, 1987). The majority

of internal combustion engines operate on the four-stroke cycle. Each engine cylinder
requires these four strokes of its piston—two revolutions of the crank shaft—to complete
the sequence of events that produces one power stroke. These four strokes are composed
of intake stroke, compression stroke, power stroke, and exhaust stroke:

1. The intake stroke starts with the piston at top dead center (TDC) and ends with
the piston at bottom dead center (BDC). During the intake stroke, a fresh mixture
is induced into the cylinder through the intake valve(s). The intake valve(s) opens
shortly before the stroke starts and closes after it ends to increase the mass
inducted.

2. During the compression stroke, both the intake and exhaust valves are closed and
the mixture inside the cylinder is compressed by the piston. Combustion process

is initiated towards the end of this stroke and the pressure inside the cylinder rises

rapidly.

 

 

3. The power stroke starts with the piston at TDC and ends with the piston at BDC.
During the power stroke, the high-temperature and high-pressure gases push the
piston down to do the mechanical work. About ﬁve times as much work is done

on the piston during the power stroke as the piston had to do during the
compression stroke (Heywoodz, 1988).

4. The exhaust stroke starts with the piston at BDC and ends with the piston at TDC.
The exhaust valve opens shortly before the power stoke ends (at BDC) and the
burned gases exit the cylinder due to the piston motion towards TDC. The exhaust
valve closes shortly after TDC and the cycle starts again.

To increase the work output of a four-stroke IC engine, with higher efﬁciency and

lower emissions, each stroke should be well understood and Optimized (Shen3, 2003).

Each stroke in an operating engine cycle is affected by several factors. During the intake
and compression strokes, the ﬂow ﬁeld inside the engine cylinder is highly turbulent and
affected by several factors including compression ratio; engine speed and load; intake
valve timing and lift; geometry of the intake manifold, intake valve and piston head; etc.
During the power stroke, the combustion process is primarily affected by the ﬂow ﬁeld
around the spark plug, air-to-fuel ratio, ignition timing, etc. The exhaust stroke is strongly
affected by the burned gas motion, exhaust valve timing and lift, geometry of the exhaust
manifold, etc.
In general, the ﬂow ﬁeld inside the engine cylinder is one of the most important
factors that control the combustion process. The ﬂame development inside the (SI)
engine cylinder and its subsequent propagation vary signiﬁcantly from cycle-to-cycle

because it depends on local mixture composition and motion. Note that these quantities

vary in successive cycles and therefore affect the performance of an engine. A
sufﬁciently turbulent ﬂow ﬁeld is required to ensure effective fuel-air mixing, rapid
ﬂame development and propagation during the combustion process. However, excessive
mixture motion and turbulence, beyond that required to achieve the desired burning rate,

is undesirable; it results in excessive heat losses to the combustion chamber walls
(Heywoodl, 1987). Consequently, the understanding of cycle-to-cycle variation is needed

to optimize them for improved engine performance. This requires continuous
improvement in measurement techniques for the velocity ﬁeld and scalar properties

pertaining to in-cylinder ﬂows and the methods for iii-cylinder combustion diagnosis.

1.2 Literature review

Cycle-to-cycle variations exist in every stroke of an operating cycle of a four-
stroke IC engine. Therefore, to optimize the cycle-to-cycle variations, each stroke of an
operating cycle should be studied. Based on the time sequence of a four~stroke cycle, the
analysis on the cycle-to-cycle variations in an IC engine can be classiﬁed into pre-
combustion analysis (intake and compression strokes), combustion analysis (late
compression and power strokes), and exhaust analysis (exhaust stroke). So far, most of

the efforts to optimize the cycle—to-cycle variations have been focused on the pre-

combustion and combustion analyses.

1.2.1 Pre—combustion analysis

Since the pre-combustion ﬂow has a strong inﬂuence on both spark ignition and

compression ignition engines, much insight into these flow ﬁelds has been gained using

various measurement techniques, i.e. hot-wire anemometry (HWA), laser doppler
velocimetry (LDV), particle image velocimetry (PIV), molecular tagging velocimetry
OVITV), etc. However, due to the variety of issues that are affected by in-cylinder ﬂows
and the difﬁcult measurement environment, in-cylinder velocity measurement techniques

and application details are often selected based on the kind of information required,

Farrell4 (2007). Therefore, the literature involving in-cylinder velocity measurements is

vast and is constantly being updated.

The earlier studies of in-cylinder ﬂow measurements are performed using hot-

wire anemometry, [5, 6, and 7]. Hassan and Dent5 (1971) used constant temperature

. . . . 6
HWA to measure the instantaneous gas velocrty in a motored 1C engine. Bahram et a1.

(1986) used HWA to study the effects of inlet conﬁguration, ﬂow rate, valve lift and
cylinder bore diameter on the velocity distribution around the intake valve. Their results
showed that the ﬂow entering the engine cylinder is fully turbulent. However, the

application of a hot-wire anemometer to measure the in—cylinder ﬂow has several

limitations, Witze7 (1980): (1) its response is strongly inﬂuenced by the properties of the

ﬂuid being measured. This is crucial for engine applications because of the highly
transient nature of the gas pressure and temperature during the compression, (2) it cannot
be used in a combusting environment, (3) it does not resolve the ﬂow direction, and (4) it

is intrusive, so that it can perturb the in-cylinder ﬂow which is highly transient and three-

dimensronal. Witze (1980) mentioned that accurate m-cyllnder hot-Wire measurements

are only possible for the intake and exhaust strokes.

Laser doppler velocimetry overcomes these difﬁculties and offers a signiﬁcant
advantage over HWA to measure the in—cylinder ﬂows. In the 19805, ‘905, and even

recently, laser doppler velocimetry has been used extensively for ﬂow measurements at

speciﬁc locations inside the engine cylinders [8-16]. Morse and Whitelaw8 (1981)

presented LDV measurements inside a single-cylinder, four-stroke reciprocating engine.
The authors masked the inlet valve over part of its periphery to promote swirl in the air
ﬂow during the intake stroke. The mean and root-mean—square (rrns) proﬁles of the swirl
velocity component were presented during the intake and compression strokes across the
radial location at distances of 10 and 25 mm below the level of the cylinder head. Their
results showed high levels of rms velocity ﬂuctuations during the intake stroke; however,

a continuous decrease in the levels of the rms ﬂuctuations was reported during the
compression stroke. Bicen et a1. 9 (1985) reported LDV measurements of air ﬂow through
the intake valve of a motored reciprocating engine. They concluded that the in-cylinder
ﬂow characteristics are strongly dependent on the piston interaction and ﬂow
unsteadiness. Hall and Bracco10 (1987) used LDV technique to evaluate the turbulence
intensities under both motored and ﬁring conditions in a ported homogeneous charge
spark ignition engine. The authors used zirconium oxide seeding particles in their work

with a nominal size of 1.5 pm. They found that TDC turbulence intensity was

approximately linear with engine speed; however, it was relatively insensitive to the

intake flow rate. Kent et al. 11 (1987) studied the effects of intake port design and valve

lift on swirl and rms velocity fluctuations near TDC of compression using LDV. They

concluded that depending on the port geometry, reduced valve lift may result in either

increased or decreased swirl. Their results showed that rms velocity ﬂuctuation was

relatively insensitive to the changes in valve lift. Fraser and Bracco12 (1989) used LDV

system in a motored, ported, single-cylinder 1C engine to provide both an ensemble and a
cycle-resolved analysis of integral length scales. Fluctuation integral length scale in their
work was based on the spatial correlation coefﬁcient of the ﬂuctuation velocity
(difference between the instantaneous velocity and the ensemble-averaged velocity at the
same crank angle) where as turbulence integral length scale was based on the turbulence
velocity (difference between the instantaneous velocity and the low-pass ﬁltered velocity
at the same crank angle and in the same cycle). Measurements were performed on the
mid-plane of the TDC clearance height from 43 crank angle degrees before TDC to 20
crank angle degrees after TDC. Three compression ratios (5.7, 7.6, and 11.4) were
considered, which correspond to the TDC clearance heights of 18.1, 12.8, and 8.2 mm,

respectively. They found that the trends versus crank angle of the ﬂuctuation length

scales are generally different from those of the turbulence length scales. Lee at a1. 13

(1993) studied the effects of four different cylinder head intake port conﬁgurations and
two piston geometries using a high speed ﬂow visualization technique. Two-component
LDV was used to compare the effects of piston geometry on the in-cylinder ﬂow. Their

results showed that cylinder head intake port conﬁguration plays a signiﬁcant role in the

generation of initial tumble motion in the early stage of the intake stroke. Yoo et a1. 14

(1995) presented simultaneous three-component LDV measurements in a four-valve

spark ignition engine. Their results conﬁrmed that the ﬂow ﬁeld generated in the engine

cylinder is complex, turbulent and three-dimensional. Hascher et a1. 15 (1997) conducted

three-component LDV measurement to calculate the turbulent kinetic energy (TKE)

inside a single-cylinder of a 3.5 L four-valve engine. The TKE is deﬁned by Equation
1.1.

.-%(<..2>.<.2>.<..2»

The authors found higher levels of TKE where intake ﬂows mix in high shear regions.
Their results showed that the TKE decay exponentially with time. LDV measurements

are very useful to generate detailed turbulence statistics due to their high temporal

resolution, see Miles et a1. 16 (2003).

Although LDV data provide excellent temporal resolution, they are limited to a
single point or along a line. It is desirable to take simultaneous measurements at multiple
points, so properties such as vorticity and strain rate can be deduced easily. Therefore,
developments in the more mature single-point systems, either hot-wire anemometry or

laser doppler velocimetry, have been less rapid recently in 1C engine applications,

Farrell4 (2007). Particle image velocimetry has become the most widely used technique

for velocity ﬁeld measurements, as it can simultaneously measure two or three

components of the velocity vectors at multiple points in a plane [17-27]. Li et a1. 19

(2001) used PIV to investigate the tumble and swirl motions in the cylinder of a four-
valve SI engine. Tumble motion is deﬁned as the organized rotation of the charge in the
vertical plane of the cylinder. Similarly, swirl is the organized rotation about the cylinder
axis. These are used to increase the turbulence intensity and hence to speedup the

combustion process. A swirl ratio (SR) is normally used to deﬁne the swirl in an

operating engine (Heywoodz, 1988). It is deﬁned as the angular velocity of a solid-body
rotating ﬂow, (as, which has equal angular momentum to the actual ﬂow, divided by the

crankshaft angular rotational speed, we :

SR =-—§-
a) (1.2)

Li et a1. 19 (2001) expressed the swirl (or tumble) ratio in a planar PIV vector ﬁeld by

 

1w (1.3)

where I is the angular inertia;

N ..
’52 "M
(=1

-" . . . . .th . .
ri 15 the distance of the cell posmon of l velocrty vector to the vortex center, and m,- is

the mass of ﬂuid located in the cell. The authors suggested that the ﬂuid mass may be
assumed to be distributed uniformly with in the cylinder. Funk et a1. 20 (2002) also
studied the effects of low and high swirl in-cylinder ﬂows using particle image
velocimetry. Bevan and Ghandhi21 (2004) studied the effects of three intake port
geometries on in-cylinder ﬂows using two-component PIV measurements. They showed

that in-cylinder ﬂows generated by the three ports were highly complex and three-

dimensional. Signiﬁcant cycle-to-cycle variation was observed in the ﬂow ﬁeld. The

 

authors concluded that the orientation of the intake port also had a signiﬁcant effect on

the ﬂow ﬁeld. High frame rate measurements using particle image velocimetry in a

motored engine have been demonstrated by Ghandhi et a1. 22 (2005). Several studies have

been reported to measure the in-cylinder cyclic variations using particle image

velocimetry; see Towers and Towers23 (2004), Jarvis et a1. 24 (2006), Jarvis et a1. 25

(2006). Recently, Stansﬁeld et a1. 26 (2007) erforrned PIV in-cylinder ﬂow
P

measurements over a range of realistic engine speeds, i.e 750, 2000, and 3500 rpm.
Calendini et a1. 27 (2000) reported the in-cylinder engine flow measurements using

stereoscopic particle image velocimetry (SPIV).

Application of particle image velocimetry for in-cylinder measurements, however,
has some limitations. The presence of particle tracers can cause complications due to
their mechanical interference with the moving parts in the IC system. Moreover, the
particle tracers may not truly represent the highly turbulent ﬂow ﬁeld. A molecular
tagging approach overcomes these problems and offers an advantage over the particle-
based techniques, i.e. particle image velocimetry. The technique nonintrusively maps
fluid velocity simultaneously at multiple points over a plane, so properties like vorticity
and strain rate can be deduced easily. It works by premixing the ﬂowing medium with
molecules having a long-lived luminescence lifetime. Use of molecules as tracers, as
opposed to the seed particles, has the advantages of unbiased ﬂow tracking and clean
operation, i.e. elimination of the contamination of the engine cylinder by the tracer

particles. In addition, MTV measurement of in-plane velocity vectors is quite insensitive

to the out—of-plane velocity component. It is to be noticed that the performance and

accuracy of PIV measurement of in-plane velocity vectors may degrade in highly three-

dimensional ﬂows due to particle motion in/out of the plane of the laser sheet.

Koochesfahani et al. 28 (1996) demonstrated this aspect in an application of MTV to the

highly three-dimensional ﬂow of a forced wake. Details of various MTV techniques are

described previously in literature, for example by Gendrich and Koochesfahani29 (1996),

Hill and K1ewicki30 (1996), and Sadr and Klewicki3l (2003). Koochesfahani32 (1999)
discussed different molecular tagging methods and its applications to various ﬂow ﬁelds.
Bohl et a1. 33 (2001) reported the development of stereoscopic molecular tagging
velocimetry for measurement of the three-component velocity ﬁeld generated by a

propeller. Koochesfahani and Nocera34 (2001) discussed the details of various molecular

complexes suitable for gas and liquid-phase ﬂows in MTV measurements. In gas-phase
applications the authors discussed the use of biacetyl’s and acetone’s phosphorescence.
However, in liquid-phase ﬂows water soluble compounds such as caged ﬂuorescent
molecules and phosphorescent supramolecules are widely used.

Although the emphasis of this dissertation is on gas-phase ﬂows, it is worth
noting liquid-phase developments. Earlier MTV investigations in liquid-phase ﬂows have

relied on photochromic molecules, which required experimentalists to use organic

solvents such as kerosene as the ﬂowing medium. As discussed by Gendrich et a1. 35

(1997), in a photochromic process excitation by photons causes a change in the
absorption spectrum and therefore a change in the color of the solution (e.g. from clear to

dark blue). The color change can persist for several seconds to minutes, although the

10

 

tagging process occurs within nanoseconds. The photochromic process is reversible and
therefore the chemical is reusable. The use of photochromic chemicals requires two
photon sources: typically a pulsed UV source (e. g. X = 351 nm from an excimer laser) to
induce the color change and a white light source to interrogate the tagged regions. The
most signiﬁcant drawback in using photochromic chemicals is that the image is produced
by a change in absorbance, thereby requiring a measurement of the difference between
incident and transmitted light. Emitted light (against a black background) is more easily
and accurately detected than transmitted light; consequently, images based on

luminescence are better suited to MTV applications. The use of caged ﬂuorescein and
similar compounds was ﬁrst reported by Lempert et a1. 36 (1995). In this compound a

chemical group is attached to ﬂuorescein in order to render it non-ﬂuorescent. The caging
group is removed upon absorption of UV photons (7» = 350 nm), thereby creating regular
ﬂuorescein which ﬂuoresces with a very high quantum efﬁciency. Here the long-lifetime
tracer is the uncaged ﬂuorescein, which persists for a very long time and can be
interrogated at the time of interest through its luminescence upon re-irradiation. Two
sources of photons are therefore needed, one to break the cage and the other to excite
ﬂuorescence. It is to be noticed that the cage-breaking process is irreversible, so each

caged molecule can be tagged only once. The cageobreaking process is not rapid,

occurring with a time constant on the order of a few milliseconds. Gendrich et a1. 35
(1997) reported the development and application of more recent phosphorescent
supramolecules. Phosphorescent compound when used for molecular tagging, excitation

by photons produces a long-lived excited state which is interrogated through its

phosphorescence emission as the molecule radiatively returns to its ground state. The

11

long-lifetime tracer is the excited—state molecule itself. In this case only one source of
photons is needed; the tagging process occurs during the laser pulse; and the
excitation/emission process is reversible, which means the chemicals are reusable. The

difﬁculty is that the long-lived excited states suffer from O; and H20 quenching. Lum et

a1. 37 (2001) used phosphorescent supramolecules in the measurements of small-scale

ﬂows. Koochesfahani et al. 38 (2000) combined MTV with laser induced ﬂuorescence for

simultaneous measurement of the velocity and concentration ﬁelds using phosphorescent

supramolecules.
Stier and Koochesfahani39 (1998) used the molecular tagging velocimetry

technique to investigate the ﬂow ﬁeld inside a steady ﬂow rig model of an IC engine. The

geometry used by the authors is common to study the fundamental aspects of the intake

ﬂow. Goh4O (2001) and Koochesfahani et al. 41 (2004) showed the in-cylinder ﬂow

pattern during the late compression of a motored IC engine using two-component MTV.
The ﬂow ﬁeld inside an engine cylinder is highly transient in nature and exhibits

large cycle-to-cycle variations. The high cycle-to-cycle variations affect the fuel-air

mixing and therefore it can inﬂuence the engine performance. Consequently, the

understanding of cycle-to-cycle variation is needed to optimize the engine design.

Reuss18 (2000) introduced the probability density function to characterize the cyclic
. . . . . 23 42
vanatrons of 1n-cy11nder ﬂows. Towers and Towers (2004), Schock et al. (2003), and

lsmailov et al. 43 (2006) investigated the cycle-to-cycle variations inside the engine

cylinder. It was found that probability density functions of the normalized circulation

12

 

calculated from instantaneous planar velocities are most suitable to characterize in-
cylinder cycle-to-cycle variations.

Experimental measurements of in-cylinder ﬂows (either MTV or PIV) provide
useful information that can be used to validate the ongoing efforts in the development of
multi-dimensional numerical simulations of in-cylinder ﬂows. Note that experimental
measurements are limited over a plane and do not provide detailed ﬂow information over
the complete volume of the engine cylinder. It is the multi-dimensional numerical
simulations that provide this information, and offer the potential of signiﬁcant time and
cost savings to design the engine with improved performance. With the advent of

powerful computers, several studies of in-cylinder ﬂows have been reported using multi-

dimensional numerical simulations, i.e. Luo et al. 44 (2003), Shojaeefard and Noorpoord'5

(2008). However, boundary conditions are assumed in the modeling efforts. Experimental
information that can provide the real-time boundary conditions is necessary to perform

more accurate multi-dimensional numerical simulations of complex in~cy1inder ﬂows.

1.2.2 Combustion analysis

The primary purpose of combustion in a piston engine is to generate the pressure
for shifting the expansion process away from the compression process and produce the
work cycle. In the analysis of combustion in piston engines, the principal components are
considered as hydrocarbon ﬁiel and air. They are ﬁrst combined into a molecular
aggregate, and then transformed by an exothermic reaction to form the products. The
process of this exothermic reaction is conventionally referred to as “heat release”. In an

internal combustion engine, this transformation is observed as a measurable cylinder

13

pressure rise manifesting the essential outcome of combustion. Thus, cylinder pressure is

frequently utilized as a direct parameter to characterize the combustion process in IC

engines; see Shen et al. 46 (2002), Shen et al. 47 (2003), and Oppenheim48 (2004).
Burgdorf and Denbratt49 (1997) compared various knock detection methods based on in-

cylinder pressure data. Mittal et al. 50 (2009) used recorded in-cylinder pressure data to

study the effects of pre-injection on combustion characteristics of a single-cylinder diesel
engine. One important measure of cycle-to-cycle variations, derived from in-cylinder

pressure data, is the coefﬁcient of variation (COV) in indicated mean effective pressure

(IMEP), covimep; Heywood2 (1988). It deﬁnes the cyclic variability in terms of

indicated work per cycle.

In—cylinder pressure is important not only for itself as a quantitative characteristic
parameter, but also because some combustion-related parameters, such as mass fraction
burned (MFB), can be determined from the recorded pressure data. Mass fraction burned
shows how in-cylinder combustion progresses as a function of crank angle. It gives a
direct indication of the quality of combustion and quantiﬁes the cycle-to-cycle variations
during the combustion process. Several methods have been suggested to evaluate the

mass fraction burned in gasoline engines based upon the measured pressure data. The

most popular method stems from the classical paper of Rassweiler and Withrow51

(1938). The authors reported that the percent of pressure rise due to combustion is

approximately equal to the percent of charge burned (by weight) at the corresponding

0 . u I 5
instants in the combustion period. Stone and Green-Annytage 2 (1987) compared the two

14

methods; (i) two-zone combustion and (ii) Rassweiler and Withrow methods, to evaluate
the mass fraction burned as a function of time in a spark ignition engine. Their results of

the complex two-zone combustion model showed good agreement with a simpler model

presented by Rassweiler and Withrow. Brunt and EmtageS3 (1997) compared the

performance of ﬁve alternative MFB models using simulated and experimental data.

They preferred the Rassweiler and Withrow model to produce the best results to

determine the mass fraction burned in their comparative tests. Shayler et al. 54 (1990)

investigated the best form to implement the Rassweiler and Withrow method and
discussed the effect of uncertainty assumptions. They found that the end of combustion
determination, the selection of polytropic index, and the effect of signal noise are
important parameters for calculating the mass fraction burned using the Rassweiler and

Withrow method.

1.3 Contents of the Present Work

In this dissertation, pre—combustion in—cylinder flow analysis is performed using
the molecular tagging velocimetry technique in gas—phase ﬂows. Working ﬂuid in the
experiments is nitrogen due to the fact that presence of oxygen quenches the MTV
chemicals. Note that MTV is a molecular counterpart of particle-based techniques, and it
eliminates the use of seed particles. Biacetyl is used as the tracer molecule. An
experimental study is performed to investigate the cycle-to-cycle variations on in-
cylinder ﬂow inside an optical internal combustion engine assembly. Both tumble and
swirl measurement planes are considered at different engine speeds. Effects of charge

motion control valve (CMCV) are presented and compared with the similar cases when

15

 

 

CMCV was deactivated. Probability density functions of the normalized circulation are
calculated from the instantaneous planar velocity to quantify the cycle-to-cycle variations
of in-cylinder ﬂow. In—cylinder engine ﬂow measurements are then extended to obtain an
instantaneous three—component velocity ﬁeld using stereoscopic molecular tagging
velocimetry (SMTV). A novel image processing technique is implemented to obtain the
velocity data. The technique has the advantage that it eliminates the geometric details
required to obtain the three components of the velocity ﬁeld. The procedure involves two
major steps: (1) calibration process and (ii) data acquisition and reduction. More details of
the technique are presented in Chapter 3 of this dissertation. Ensemble-averaged velocity
and the ms of three-component velocity ﬁeld are presented inside the engine cylinder.
The three-component planar velocity ﬁeld conﬁrmed that during the intake stroke in-
cylinder ﬂow is highly turbulent, complex and three-dimensional. This work can be
utilized as a tool for engine developers where only the logistics of operation need to be
described.

Experimental measurements provide useful information of the ﬂow ﬁelds inside
the engine cylinder; however, it is multi-dimensional numerical simulations that offer the
potential of signiﬁcant time and cost savings to design the engine with improved
performance. To date, multi-dimensional numerical simulations of in-cylinder ﬂows are
performed with assumed boundary conditions. Real-time boundary conditions are
necessary to perform more accurate multi-dimensional numerical simulations of complex
in-cylinder ﬂows. Therefore, velocity and pressure measurements are performed inside
the intake manifold of an engine assembly that can provide these necessary boundary

conditions. The geometry of the intake manifold is simpliﬁed for this purpose. A hot-wire

l6

anemometer and piezoresistive type absolute pressure transducers are used to measure the
velocity and pressure, respectively. In-cylinder ﬂow measurements are also performed to
validate the modeling efforts. Stereoscopic molecular tagging velocimetry is used for this
purpose.

In-cylinder combustion diagnosis is performed in a dual fueled, single-cylinder,
spark-ignition engine. Cycle-to-cycle variability is presented using the coefﬁcient of
variation in indicated mean effective pressure. Mass fraction burned and burn duration
are determined from the analysis of measured in-cylinder pressure data. The Rassweiler
and Withrow method (Model 1), with a new linear model for the polytropic index, is used
to obtain the MFB curves. Start of combustion and end of combustion are determined
using a least—square ﬁt algorithm. In addition, results of mass fraction bumed using a net
pressure method (Model 2) are presented and compared with the results of Model 1, i.e.
Rassweiler and Withrow method. Model 2 is of interest as it offers an advantage that the

data processing time is short enough to allow for online processing.

17

CHAPTER-2

A STUDY OF CYCLE-TO-CYCLE VARIATIONS ON IN-CYLINDER FLOW IN
AN IC ENGINE EQUIPPED WITH CMCV USING MTV

2.1 Introduction

As discussed in Chapter 1, the ﬂow ﬁelds inside the engine cylinder are extremely
complex and exhibit large cycle-to-cycle variations. It is to be noticed that the high cycle-
to-cycle variations affect the fuel—air mixing and ﬂame propagation during the
combustion process, and hence inﬂuence the engine performance. Therefore, an
understanding of cycle-to-cycle variations is needed to optimize the engine design. Flow
inside an engine cylinder is affected by several factors including engine geometry, speed,
load, valve timing and lift. In addition to these factors, the intake charge motion control
valve is an important factor that affects the ﬂow inside an engine cylinder. It is expected

that the CMCV imparts an angular momentum to the charge entering the engine cylinder.

Clarke and Stein55 (1999) combined the variable valve timing with charge motion control

valve. Variable valve timing was obtained using dual equal variable camshaft timing

(VCT) strategy. More details about the dual equal VCT can be found in Stein et al. 56

(1995). The combination of dual equal VCT with a CMCV allows an engine to be
operated either at or near stoichiometry or at lean conditions, which allows the use of a
NOx trap for the purpose of further reducing the air pollution. Authors further discussed
that the synergy between the CMCV and the dual equal VCT allows the fuel consumption

to be less than the fuel consumption during lean operation at standard valve timing. This

18

 

is due to the fact that CMCV increases the in-cylinder charge motion and hence improves

the combustion and the ability to handle the charge dilution which occurs from increased

. . . . . . . 57
levels of internal exhaust gas recrrculatron resulting from valve timing retard. L1 et al.

(2000) investigated the effects of swirl control valve (SCV) on in-cylinder ﬂow
characteristics using laser doppler anemometry. They found that SCV closed condition

produced both swirl and tumble motion. in the cylinder and the mean velocity was nearly

doubled in comparison to the SCV open condition. Schock et al. 42 (2003) studied the

effects of tumble and swirl port blockers on cyclic variations of ﬂow in a piston cylinder

assembly. They found that swirl port blocker strengthens both tumble and swirl motions;

however, tumble port blocker only strengthens the tumble motion. Kim et al. 58 (2005)

investigated the effects of injection timing and intake port ﬂow control on fuel wetting
inside the engine cylinder. They found that a tumble mixture-motion plate inside the
intake port signiﬁcantly reduced cylinder liner and piston top fuel wetting. This is

because the use of tumble mixture-motion plate provided more turbulence, which
effectively enhanced the mixing during the intake process. Lee and Heywood59 (2006)

studied the effects of CMCV on combustion characteristics and hydrocarbon emissions.
The authors concluded that CMCV improved mixture preparation due to increased swirl
and tumble intensities, which enhanced fuel transport, distribution and evaporation.
CMCV in closed condition allowed reduced fuel injection and retarded spark timing
strategies that reduced hydrocarbon emissions signiﬁcantly during cold start due to

greater fuel evaporation and faster burning rate.

19

 

 

 

Overall, previous investigations show that a charge motion control valve is an
important factor that controls the combustion process and hence inﬂuences the engine
performance. However, studies of charge motion control valve on in-cylinder flows are
limited. Therefore, an experimental study is performed to investigate the effects of charge
motion control on ﬂow measurement inside an internal combustion engine assembly.
Molecular Tagging Velocimetry is used to obtain the multiple point measurement of the
instantaneous velocity ﬁeld. A two-component velocity ﬁeld is obtained at various crank
angle degrees for swirl and tumble measurement planes inside the optical engine
assembly at 1500 and 2500 rpm engine speeds. Effects of charge motion control are
studied considering different cases of: (i) Charge motion control valve deactivated and
(ii) CMCV activated. Both the measurement planes are used in each case to study the
cycle-to-cycle variability inside the engine cylinder. Probability density functions of the
normalized circulation and the turbulent kinetic energy of ﬂow are calculated from the
instantaneous planar velocity to quantify the cycle-to—cycle variations of in—cylinder
ﬂows. Different geometries of CMCV produce different effects on the in-cylinder ﬂow
ﬁeld. It is found that the CMCV used in this work has a profound effect on fuel-air
mixing; however, its inﬂuence is not as signiﬁcant during the late compression.
Therefore, it can be assumed that CMCV has less contribution to enhance the ﬂame

speed during the combustion process.

2.2 Experimental setup and procedure

2.2.1 Experimental setup

20

 

The motored optical engine assembly used for the experiments is a three-valve,
two intakes and one exhaust, 0.675 1 single-cylinder engine; see Figure 2.1.
Speciﬁcations are listed in Table 2.1 with a bore diameter of 90 mm and a stroke length
of 105.6 mm. Figure 2.2 shows the intake and exhaust valves lifts. As shown in the
ﬁgure, intake valves open shortly before the top dead center at the start of intake stroke
and close after the intake stroke ends. The exhaust valve opens before the power stoke

ends (at BDC) and closes shortly after TDC of intake.

 

Figure 2-1. Optical engine assembly

21

 

 

 

 

 

 

 

 

Speciﬁcations Measure

Bore (mm) 90
Stroke (mm) 105.6
Compression ratio 10
Stroke/Bore ratio 1.17
Connecting rod length (mm) 169

 

 

 

Table 2-1. Engine speciﬁcations

 

 

 

 

 

 

Intake Compression Power Exhaust
TDC <———-—-———o BDC a-——-—-—-v TDC +—————-—-> BDC <————————>TDC
12 L .
l i , ‘
J 5.: 1‘}.
g 8 A i :1" i,
E 1 .11" is.
g 6 " l t"
- . l a;
4 . -—lntake 11ft (mm) ,1 i . 1‘.
-9- Exhaust Valve Lift (mm) ”3 i}
J ‘ W i U U
2 , :2"
o ‘ , a 5L 5 '
0 90 1 80 270 360 450 540 630 720
CA (degrees)

Figure 2-2. Intake and exhaust valves lift

Figure 2.3 shows the locations of the swirl and tumble measurement planes inside

the engine cylinder. As shown in the ﬁgure, the swirl plane is located 24.5 mm below the

head deck level. Measurements with 1500 rpm engine speed are performed at 121 crank

22

 

angle degree during the intake stroke and at 192, 257, and 300 crank angle degrees during
the compression stroke. Similarly, measurements are performed at 138 and 171 crank
angle degrees (CADs) during the intake stroke and at 221 and 253 CADs during the
compression stroke at 2500 rpm engine speed. Zero crank angle degree represents the
start of the intake stroke, i.e. piston at top dead center of the intake. 180 crank angle
degree represents the BDC of the intake stroke. Similarly, 360 CAD represents the TDC

of the compression.

 

 

 

 

Head deck level
0..» -- — .-
Piston top at TDC
-20 Swirl plane
location
-40
E
E '60 Tumble plane
a” at 121 CAD
-80

 

 

-100 Pistontop‘atBDC

 

 

O 20 40 6O 80
x(mm)

Intake

ve
valves Exhaust val

Tumble plane
location

 

Figure 2-3. Tumble and swirl measurement planes inside the IC engine cylinder

The optical setup used for the molecular tagging velocimetry measurements msrde

the engine cylinder is shown in Figure 2.4. It consists of a pulsed ultraviolet (UV) laser,

23

 

 

whose light beam is used to create a grid pattern in the cylinder. The UV laser beam used
in this work is produced by a Lambda Physik excimer laser device, LPX205i, ﬁlled with
XeCl. This laser device provides laser pulses at a wavelength, 7», of 308 nm and repetition
rates up to 60 Hz. Figures 2.5 and 2.6 shows the front and rear views of the LPX205i.
The laser beam passes through numerous optical elements (including mirrors, a beam
splitter and cylindrical lenses) that generate two light sheets. Beam blockers are used to
transform the light sheets into sets of laser beams that are introduced into the cylinder

through the optical cylinder and the optical piston head.

CCD
Camera
Beam

Mirror blocker N

 

,. Cylindrical Laser -'”'”'":-—

a: ‘6"5”=5°mm beams
C7 ———— Cylindrical lens,

f: 150 mm
1
Beam
splitter
\_ Mirror
‘\
Cylindrical

lens, f: 50 mm
Incoming laser
beam

Figure 2-4. Schematic of MTV optical setup for in-cylinder ﬂow measurement

24

 

,/Mirror
JBeam shutter
(in closed position)

 

Laser ON warning light
,Air intake
Figure 2-5. Front view of LPX205i
HV input Exhaust port Vacuum pump connection

Data connections

Main switch

  
 

Cooling water inlet

‘1 Cooling water outlet

Key switch

Gas connections

Figure 2-6. Rear view of LPX205i

25

2.2.2 Charge motion control valve
It is believed that charge motion control valves have a profound inﬂuence on the

ﬂow ﬁeld, even to late compression which affects the fuel-air mixing and hence the
combustion process, Mittal and Schock60 (2009). The nature of CMCV inﬂuence

depends on its geometry which likely produces different ﬂows with different geometries.
Figure 2.7 shows the charge motion control valve device used in this experiment,
installed between the intake manifold and the intake port. Experiments are performed
with both the conditions: CMCV deactivated and CMCV activated. It is expected that
charge motion is enhanced, both in tumble and swirl measurement planes, with the
activated CMCV. This work quantiﬁes the magnitude of this enhanced motion at

different crank angle degrees.

   

(b)

Figure 2-7. Charge motion control valve; (a) deactivated, and (b) activated

2.2.3 Molecular tagging procedure
Molecular tagging velocimetry relies on molecules that can be turned into long

lifetime tracers upon excitation by photons of an appropriate wavelength, Koochesfahani

26

 

 

and Nocera61 (2007). Biacetyl (CH3COCOCH3) is used in this work as a tracer

molecule. It is worth noting a brief description of biacetyl properties relevant to this
dissertation. Table 2.2 shows the physical properties of biacetyl [62]. Figure 2.8 shows

the absorption spectrum of biacetyl, which has a peak at about It = 420 nm; Stier and

Koochesfahani63 (1999). It is to be noticed that the absorption increases below it = 330

nm and has a second peak at about it = 270 nm. This shows that the laser device used in
this study with It = 308 nm is a suitable source of excitation. The working ﬂuid used in
this experiment is nitrogen, since the presence of oxygen quenches the phosphorescence
of biacetyl. The tracer molecules are premixed into nitrogen gas by an evaporation
process in a preparation tank before entering the IC engine. The excitation of biacetyl
leads to both ﬂuorescence and phosphorescence emissions; the latter being used for

velocimetry. The phosphorescence lifetime of biacetyl is reported to be as high as 1.5 ms;

Sidebottom et al. 64 (1972). Stier and Koochesfahani63 (1999) measured a lifetime of

about 0.1 ms in their work using an excimer laser (A = 308 nm). The observed lifetime

may be lower unless all oxygen is completely purged from the ﬂow system. Stier and

Koochesfahani63 (1999) compared the phosphorescence emission spectra of biacetyl at

different excitation wavelengths; see Figure 2.9. They found that the wavelength with
more absorption leads to more phosphorescence emission. The authors reported that It =

308 nm is also a suitable source of excitation for phosphorescence emission.

27

 

 

 

 

 

 

 

 

 

 

 

Parameter Measure

Molecular formula C 4H 602

Appearance Liquid with a butter-like odor
Melting point -2 to -4 C

Boiling point 88 C

Vapor density 3 (air = 1)

Vapor pressure 52 mm Hg at 20 C

Synonyms diacetyl, 2,3-butanedione

 

 

 

Table 2-2. Properties of biacetyl

 

'o

v rw—T‘i v ‘7‘ Tj’vﬁ’ﬁj ‘7‘ VT—rTT‘Yﬁ—TVYf

Absorption (a.u.)
9 .9 .0
A Ch 00

.3
N

 

 

 

 

 

250 300 350 400 450 500 550
Wavelength (nm)

Figure 2-8. Absorption spectrum for biacetyl (Stier and Koochesfahani, 1999)

28

 

 

 

 

 

 

 

3:“ , . i .......... : . . I 14201111) ‘
f .' 3 v—v3SSmn ‘
’7 > ' r' g °'"‘°308nm ‘
f; > 5 3 : H27Onm
v 2.........-...._.?... ' .. a .. . :. .-..-..... -......,,,e
t: . '
.2
.2
E .
m l“
O ‘ ‘ - ‘ e A ‘7 - - x
450 500 550 600 650 700

 

Wavelength (nm)

Figure 2-9. Phosphorescence emission spectrum for biacetyl at four different excitation
wavelengths (Stier and Koochesfahani, 1999)

The MTV images at two successive time levels are recorded using a Xybion ISG
intensiﬁed CCD (charge coupled device) camera. Each image is an 8 bit image with the
size of 640x480 pixels. The reference (undelayed) image is recorded as soon as the grid
pattern is written and the delayed image after a certain delay, 12-35 us. Figure 2.10
shows a sample of undelayed (left) and delayed (right) images for the tumble
measurement plane at 138 crank angle degree. Similar images are obtained for the swirl
measurement plane at various crank angle degrees. The images are processed using a data
reduction program to ﬁnd the displacement at each grid point. At each measured crank
angle degree, 200 undelayed frames are recorded to obtain the averaged undelayed
image. This is to reduce the noise in undelayed image. Note that the displacement at each
grid point is zero when an undelayed image is spatially correlated with the averaged

undelayed image. 500 delayed frames are considered at each crank angle measurement to

29

 

obtain the instantaneous velocity ﬁelds in this work. A spatial image correlation

technique is used to obtain the displacement vectors within one pixel for the tagged
region, Gendrich and Koochesfahani29 (1996). More details can be found in Section 3.3.

Sub-pixel accuracy is obtained using a fourth order polynomial which requires a
minimum of 15 data points from the correlation ﬁeld. The location corresponding to the

maximum correlation coefficient is found using a decoupled method as presented by
Zheng and Klewicki65 (2000). Artiﬁcial images, with signal to noise ratio of 5, are used

to obtain the accuracy of the data reduction program with a known spatial distribution of
displacements. These images are processed using the data reduction program to obtain
the displacements. The obtained values are then used to estimate the accuracy of the
measurement technique. The uncertainty in determining the displacements is less than
0.05 pixels. Similar results are obtained using a ﬁfth order polynomial; however, a third

order polynomial is insufﬁcient to ﬁt the correlation coefﬁcient data.

 

....II...‘
...—.....I

Figure 2-10, Undelayed (left) and delayed (right) images for tumble measurement plane
at 138 crank angle degree

30

 

2.3 Experimental Results

Measurement results of complex in-cylinder ﬂow are presented at various crank
angle degrees during the intake and compression strokes. Both swirl and tumble
measurement planes are considered. Inﬂuence of charge motion control valve is
presented and its results are compared with similar cases when engine was operated with
deactivate CMCV. Experiments are performed at 1500 and 2500 rpm engine speeds. 500
delayed frames are considered to obtain the instantaneous velocity ﬁeld at each crank

angle degree measurement, unless speciﬁed. Then velocity statistics are calculated.

2.3.1 Swirl measurement plane, 1500 rpm

For the swirl plane at 1500 rpm engine speed, measurements are performed
during the intake stroke at 121 crank angle degree and during the compression stroke at
257 and 300 crank angle degrees. Figure 2.11 shows the ensemble-averaged velocity
vectors and the contours of rms velocity when engine was operated with deactivated (left)
and activated (right) CMCV at 121 (upper graphs) and 300 (lower graphs) CADs,

respectively. The vectors in the plots show the magnitude of ensemble-averaged velocity

vectors (u), Equation 2.1, and (v). Note the notation, (), used here for ensemble-

averaged.
] n
(HF; 'Z1ui (2.1)
I z

where n is the total number of delayed frames.

31

 

As shown in the ﬁgure, the velocity magnitudes are higher at 121 CAD during the intake
stroke and reduce during the compression stroke. Similarly, the rrns values decay during
the compression stroke. It is to be noticed that the rms cyclic variations are considerably
enhanced with the activated CMCV at 121 CAD, and they are as high as 20 m/s. The
higher rrns values at 121 CAD show that the ﬂow at this crank angle is highly turbulent
in nature with both the intake valves open. The cyclic variations decay during the

compression stroke. These results are consistent with the earlier literature; see Jarvis et al.

24 (2006). They found high rrns values during the intake stroke at 149 CAD. These

values reduce as the compression stroke is traversed. There are some counterclockwise
rotations that can be observed toward the upper right half of the plane and some
clockwise rotations at the lower left half of the plane (at 121 CAD) with activated

CMCV.

Figure 2.12 shows the inﬂuence of considering different number of consecutive

cycles to evaluate the ensemble-averaged velocity vectors, (u) and (v). The three plots, a,

b and c, in the ﬁgure are obtained considering 1-100, 101—200 and 1-500 consecutive
frames, respectively, with activated CMCV at 121 crank angle degree. The ﬁgure shows
that the mean ﬁeld is quite stable when calculated with 100 consecutive frames. In
general, more frames are required to calculate the higher order turbulence statistics, for
example, the probability density function of the normalized circulation, turbulent kinetic
energy, etc. Therefore, 500 delayed frames are considered in our further investigation to
obtain the instantaneous velocity ﬁeld at each crank angle degree and then the velocity

statistics are calculated.

32

'm A *7. wa‘”

 

Cylinder

 

 

 

 

 

 

 

  

 

 

20
0r
7:
-20 '
2Q "If . rms contour ,3 8 20\m{5 J rms contour g’,
4 80
20 40x (mm) 60 80 20 0x (mm) 60
(a) 121 CAD
CMCV deactivated CMCV activated
\ / \
20 _ t’ms _ l1" :
, (m/s) ‘
10
.0- .-
E t
N
-20. .
4 _ .z.
J, \5 m/S rrns contour /
20 40 60 80 20 40 6O 80
x (mm x (mm)
(b) 300 CAD

Figure 2-11. Ensemble-averaged velocity vectors, 0011101115 0f “ﬂoaty ms for
deactivated (left) and activated (right) CMCV at (a) 121 (101’): and (b) 300 (bottom)
CADs; swirl measurement plane at 1500 rpm

33

Figure 2-12. Ensemble-averaged velocity vectors with activated CMCV at 121 CAD

considering (a) 1-100, (b) 101-200, and (c) 1-500 consecutive undelayed frames; swirl
measurement plane at 1500 rpm

 

 

 

 

 

Cylinder
/WaII_qH -\.\
+ ,a-‘V g-N‘s‘yh.
/)"‘_~NM\\\£
20F(,r"..\\\\\ki/
rt\\1/~.\\\11I"
1l*\‘d- ”—“1\tl'

A x \

a It '1]:\1\ 1\\\

E0“ ‘c’ttr't‘ ‘\\\\x

N .r-x/ “\illl" .«xxxﬁ

iﬂfl/d-“tlt‘i’f’ d.__,‘_.
L le-‘ﬁ‘5‘5x1‘t‘ ”

-20- )1“..le '
xxxltlll( ,-
ettjjjfjji 2031/5

\L;14J+J¥LL1A4/ ,
20 40 60
_ x(mm)
F1g.2-12(a)
/ -."rI_ﬁ\
( ,#’~,.~#-\\.\
P I’,,,.~-*~\,\\S
20 (ygr'-’>\\\o\\1/
.xi“’~.\\\\[f’
frl\‘ ”'""‘~\\l
A \‘c
a " tllll"\\‘\\\
EC” ’rt_. il‘ ‘ \\\-
N f/fr'm (It. r~~e~q-~
“"""'s111‘ (III—“F
r Iﬁﬂdhi‘ttlii ft!
-20- ”:’""“tll" "
1 "\\\illlll «
k ““1““! ZO—fn/s
2 9‘40 60
x(mm)

Fig. 2-12 (b)

34

Figure 2-12, continued

 

 

...—.H \
i ”_~ -\\\\
20_ ,,/.--~‘~\\\\l
(,;(° I5\\\\\1f
I 1\‘\"“\\\Xif’
,|\\\ td—ﬁ\\\if
E " tittt““\“
E0“ It.....’l\‘ ‘\\\\\“
N ,,// ‘5I‘\ .~\\\-.
”Egg- ill‘ :1 ,,,,,
...,»ﬂ—a‘ hll1\ f:
-2()_ /q"“lllll
\\\li!iil -
\1llr ff —0
\ Hi 20m/s
11 1 1 ./
20 40 60
x(mm)

Fig. 2-12 (c)

Figures 2.13 (a), (h) and (c) show the instantaneous velocity vector ﬁelds for
three consecutive engine cycles, i.e. 99, 100 and 101. Ensemble-averaged velocity ﬁeld
(Fig. 2.13, d) is also presented for the comparison. The instantaneous ﬂow ﬁelds show
large cycle-to-cycle variations with signiﬁcant difference from the ensemble-averaged
velocity ﬁeld. Mean subtracted flow ﬁeld for 101 engine cycle is shown in Figure 2.14 b.

The high ﬂuctuations in velocity are evident from the plot.

35

 

 

/’- \k‘ ..«\“N
"/'-\\~~ 1 \x"" I
20 ”‘1’ ”‘V’ .. ..t
A ...A’IIt‘ - \\\l’ J I, _ iiif~
g ‘_I\ ‘(~i‘ it'” ,,//I (l ”’f x/
v ..H \\‘« ~”\\\\‘ ,,/\t . ~ {1"'
N (111,»-..4__.\~wF-\\\i‘ /Il!\1:it /./r "'
0”! a-o’I/"/' "7 \‘s "" w x... ”Ml/[1f ”'\
"""'"’T"‘“"""‘/’ \l_ If! ,//;z "”"\\u
'/-~~~--~v~”"”-’ w~.~//¢~”"' ..
/t1‘~‘\\\‘ xttt Il’ryf/ﬁlﬁ‘x //’l~\
-201. "' " t\\t‘ \ “ L‘l ‘t't///’:"'*/
“\" ’ " r/ / "/
l/ H /

 

 

 

 

x(mm) x(mm)
»m\\\ﬂtd‘-‘ \/ ’...\.~""‘"\ \
l \1\ ”/ﬂi/ -_._ .~~\\\
20' \“-/‘l/ \ r . .. ~~~\\‘tl
A ‘1“' \3’ I, t\\“~\\i/
E I \.//--/()’L\ ,,\\lr~~.\\itf’
ElZi, /«//~—-\-t‘~"" ..\~ .~~~xx1i
N ////\-v-‘(xtbt,\"‘ I. \lil't\ \\'\,\"
O-l////JM“\\/~f-Al\gr """fl‘ 1\\\\\
-" I‘l' ’ I“! V ' ‘\\“'
Iran—M.“ \\f .- {‘\
\\~d\‘\/5}Stt”’\( «Ail-\ll‘t ’ “‘-
\x\\. ‘ ArA-\\ is I!
t~-~~"I/‘t\ “1
20 "it"lllglit‘-“ ”““t({t' ‘
- " r-
“"r’/{'! /\‘ ’~~\\(]!th
"\J/’/- I. ~ ,‘\h‘l’flll _
\(C) H 20 m/s , \(d) 20 m/s ,
__Laax¥_L_L.. lgnnrj LngLgkllA—A ./
20

40 x (mm) 60 20 40x (mm) 60

Figure 2-13. Instantaneous velocity ﬁelds from consecutive engine cycles 990, 10017 and

101C, and ensemble-averaged velocity ﬁeldd with activated CMCV at 121 CAD; swirl
measurement plane at 1500 rpm

36

 

 

 

 

 

 

 

 

d/s‘h /
1\ f/f‘i’ 1" /// "
x a. \
I- M .t- "‘
20- \“ 1’ \ . ..,.t~i \ /‘\\
A r ‘1'” ‘\‘ "\"’ “ it
E I '1 //"'"f!"‘ ll. t /"-’a-”/\
I l ‘1’. \
E .1, ’///”-~\ \"” ..zx//»-.~t ""
N ’///\AM‘s\Lt‘\d\\l /,,//-f/-‘~1‘l'/"'"
0'! "/"*'~-'\,\/“"""\‘-- t..//.»z...a-\\/""/’ ‘
I/far‘~*'\\\t"" " IIIII /\\\l/ I . \
sW'RN‘V/[l\r".’ \h ‘~_\"’ I‘ " ‘\
~M\\ Illl\\\\\ \ \\‘., // l“§\K/‘
20- "‘1"I//11w‘—‘~ _ '!”~’//"‘t"—\
M'“’/J/fl'/\I ‘ ’w’I /\.
\l
I \\J/f,.l f. - \\;p, ‘f" _.
‘\(a) l 20 m/s/, \(b) 20 m/s /
44gg4L1+144 LP4AAL¥JJLL14
20 4 60 20 40 0
x(mm) x(mm)

Figure 2-14. Instantaneous velocity ﬁeld“ and mean subtracted ﬂow ﬁeldb for lOlst
cycle at 121 CAD with activated CMCV; swirl measurement plane at 1500 rpm

The circulation of the ﬂow on a swirl or tumble measurement plane can be used

. . . . 42
as a primary parameter to charactenze the rotat1onal motion of the ﬂow, Schock et al.

(2003). Circulation, F, is deﬁned as follows:
F=§CFodI=jjA50d2 (2.2)

where C is a closed curve on a tumble or swirl measurement plane and A is the area
enclosed by the curve C.

It is to be noticed that when the ﬂow on swirl or tumble measurement plane is studied at
different CADs, the area of each plane is not the same. Therefore, to compare the results,

the circulations of ﬂow on different planes have to be normalized by their respective

areas, Schock et al. 42 (2003), which is

37

l -- - 1 -— —
F =-—— VOdl=-—— coed/1 2.3
N Asst ASHAS ( >

where A S

is the studied area of the swirl or tumble measurement plane. Note that F N ,
by deﬁnition, is the averaged vorticity. To be consistent with the literature, this averaged

vorticity over the computational plane is referred as the normalized circulation (NC). In

this work, e. g., over the tumble measurement plane, vorticity (oz, Equation 2.4, for two-

component velocity ﬁeld is computed using a second order accurate ﬁnite difference

technique (Cohn and Koochesfahani“, 2000), and then the spatial mean is calculated to

obtain the averaged vorticity (or normalized circulation). Each frame has different NC
because of the cyclic variability. 500 frames are considered in this work to obtain the

PDF plot of the normalized circulation for each case.

6v au
(0—

Z —0;—5y— (2.4)

Figure 2.15 shows the probability density function (PDF) plot of the normalized
circulation at different CADs with deactivated CMCV at 1500 rpm engine speed. The
mean value of normalized circulations gives the ensemble-averaged information of the
ﬂow inside an engine cylinder. The range of NC variation quantiﬁes the cycle-to-cycle
variability of the ﬂow. A more representative measure to quantify the cycle-to-cycle
variation inside the engine cylinder is the standard deviation (0') of NC PDF. Almost all
the change in PDF occurs within 3:3 standard deviations. As shown in the ﬁgure,

standard deviation of NC decreases from 121 CAD (01205.33 Hz) to 257 CAD

(0' =96.76 Hz). The shift in peak shows that the negative circulations are more probable

38

at the later crank angle. It then increases slightly at 300 CAD (UZIOIJS Hz). These
results show that cycle-to-cycle variation decreases from 121 CAD to 25 7 CAD, and then
increases slightly to CAD 300. Figure 2.16 shows the PDF plot of normalized circulation
with activated CMCV. As shown in the ﬁgure, standard deviation decreases from 121
CAD (0:30437 Hz) to 300 CAD (02109.17 Hz). It is interesting to note that CMCV
effects are not signiﬁcant on cycle-to-cycle variation at 300 CAD when compared to the
results of deactivated CMCV; see Figure 2.15 for comparison. This shows that CMCV
has a considerable effect on fuel-air mixing at this engine speed, i.e. 1500 rpm, for the
swirl measurement plane; however, it has less contribution to enhance the ﬂame speed for

faster combustion.

Swirl measurement plane with CMCV deactivated, 1500 rpm

 

0.0040

i 0.0044
i PDF

0.0036 J

,' 300 CAD
0.0032 ~ 2 ‘/

1 0.0028 I
i 0.0024 4 ——121CAD 3'
i 0.0020 4 ------ 257 CAD .- 1.

0.0016 - “"300 CAD
0.0012

  

1 0.0008 ~
i 0.0004 . ,

\ 0.0000 . 4‘
1 4200 -700 NC (1/8) -200 300

 

 

 

 

 

 

Figure 2-15. PDFs of normalized circulation at different CADs for swirl measurement
plane with CMCV deactivated at 1500 rpm engine speed

39

 

 

0.0044
*‘ 0.0040
,~ 0.0036
1 0.0032
0.0028
0.0024
0.0020
0.0016 ~
1 0.0012 -
0.0008 «
0.00044

PDF

 

vairl measurement plane with CMCV activated, 1500 rpm

—— 121CAD
300 CAD """" 257 CAD
- . - . - 300 CAD

257 CAD

121 CAD

 

 

0.0000
-1200

Figure 2-16.

-700 NC (1/8) —200 300

PDFs of normalized circulation at different CADs for swirl measurement

plane with CMCV activated at 1500 rpm engine speed

The ﬂow ﬁeld inside an engine cylinder is highly turbulent; and, thus, the ﬂow
characteristics at any point are different from cycle-to-cycle. The velocity ﬁelds are

further processed to evaluate the turbulent kinetic energy. For swirl measurement plane,

the turbulent kinetic energy is calculated as follows;

k=-et<w2>+<vz>+<w2>i80040)

where <v'2> is estimated as (<u'2>+<w'2>)/2 in equation 2.5, Miles et al. 67 (2003).

Similarly

<w’2> is estimated as [(102 >+ <v’2>)/ 2 when the TKE is evaluated for the

tumble measurement plane. The local TKE, at each grid node, is then used to evaluate the

40

 

averaged TKE over the studied plane. With N points over the plane and the local TKE, k,-,

the averaged TKE, km, , over the plane is;

1 N k
kavg’W £1 i (2.6)

Figure 2.17 shows the change in cycle-to-cycle variability characterized by averaged

turbulent kinetic energy over the swirl measurement plane at 1500 rpm engine speed with

. 2 .
respect to the crank angle. The values are normahzed by S p where S p is the mean

piston speed, Equation 2.7 (Heywoodz, 1988),
Sp = 2LN (2.7)

L is the stroke and N is the rotational speed. Note that the mean piston speeds at 1500 and
2500 rpm of engine speeds in this work are 5.3 and 8.8 m/s, respectively. The results are
qualitatively in agreement with the PDF of NC results as shown in Figures 2.15 and 2.16.
The CMCV effect on cycle-to-cycle variation is signiﬁcant during the intake stroke;

however, the effect reduces traversing during the compression stroke.

41

 

 

 

 

 

 

 

 

 

i850 q . 6.50
u \\ Swrrl measurement plane, ,__,
z \ 1500 rpm N a.
300 1 «.5 ‘ re 4 5.50
> 1...:
m E
250 E p, 4.50
01 "c
0.)
DD
200 ~ g 1. 3.50
>
<
150 2.50
100 l a 1.50
Crank angle (degrees) \3 ’ ‘:‘*e
50 r r 0.50
90 140 190 240 290 340
—-+— STDEV of NC, deactivated CMCV -—-I——— STDEV of NC, activated CMCV

 

— —Av— -— Averaged TKE. deactivated Cit/CV — 4:}— — Averaged TKE. activated CMCV

Figure 2-17. Standard deviation of normalized circulation and turbulent kinetic energy at
different CADs for swirl measurement plane at 1500 rpm engine speed

2.3.2 Swirl measurement plane, 2500 rpm

For the swirl plane at 2500 rpm engine speed, measurements are performed at 138
and 171 crank angle degrees during the intake stroke and at 221 and 253 CADs during
the compression stroke with wide open throttle (WOT). Figure 2.18 shows the ensemble-
averaged velocity vectors and the contours of ms velocity when the engine was operated
with deactivated (left) and activated (right) CMCV at 171 (top) and 221 (bottom) CADs,

respectively. The vectors in the plots show the magnitude of ensemble-averaged velocity

vectors (u) and (v). The velocity magnitudes show that CMCV enhances the ﬂow ﬁeld

at 171 CAD during the intake stroke and its effects relatively decay during the
compression stroke at 221 CAD. However, the presence of swirl motion is still evident

with activated CMCV; see some counterclockwise and clockwise rotations toward the

42

right half and lower left half of the plane, respectively, with activated CMCV. The rrns
values with activated CMCV at 171 CAD are higher than 17 m/s toward the center of the
measurement plane. Some clockwise rotations can also be observed at this crank angle
toward the lower left half of the plane. It is to be noticed that the rrns values decay during
the compression stroke for both the cases. However, the rms cyclic variations are still
higher with activated CMCV during the compression stroke at 221 CAD (note the
maximum value of about 13 m/s).

Figures 2.19 and 2.20 show the PDF plots of normalized circulation at different
CADs for swirl plane at 2500 rpm engine speed with deactivated and activated CMCV,
respectively. As shown in the ﬁgures, the standard deviation of NC decreases from 138
CAD (a =305.53 Hz) to 253 CAD (a =117.11 Hz) for deactivated CMCV. These results
show that cycle-tovcycle variation decreases from 138 to 253 CAD with deactivated
CMCV. With activated CMCV, the standard deviation decreases from 138 CAD to 221
CAD, and then increases slightly at 253 CAD. The CMCV effect shows signiﬁcant
enhancement in cycle-to-cycle variation at 171 CAD during the intake stroke. However,
it is interesting to note that the effect is not signiﬁcant at 138 CAD at this higher engine

speed.

43

       

 

 

 

 

 

rms contour \ rms contour \
THIS
20 (m/s)
17

El" 12

T: 7

-20
-—o i .' \_—

20 m/s - hi” , 20m/s "
\1 1 I J ""L 1 1 I
20 40 x(mm) 60 80 20 40,,(mm) 60 8O
CMCV deactivated (a) 171 CAD CMCV activated
/ rms contour l/ rrns contour
-.rv"‘*"'"‘-- L1.

20* :‘:;:t§f,‘,‘i..i rms ,f‘fi.’"‘f?‘.7.....7',\
'. testis/1"' I (m/s) ~~~~~ l” ' F"//
...-titl/ltlf"' 13 .—»~" "”//

t11lllfliﬂl|t ’i "”"WH’IIIH

0' tlt],f]{/If"”' 10 .../iriritlrfffll}

A -‘-rrr(r::r"" ~-rlrlllt\ttlllll"

E 'Illillff""" :“‘r\'t\'ll\\\\l.l\\i\

T: IIIII/I """"" 7 f""‘\\\i\\\\\\,\\\'\

20_ "’///”"" "f--\\\\1\\\s\\yi

- \ ' ”/lr""' ’ 4 (.'\\\_\\(\\~\\y<
‘_' a 'II;,;’-" -' X‘s-111111" 5

15m/s I 15m/s 2,};77r“, 2’
J I l /I :1; Al I /l
20 0 0 60 8

40x(m)6 8 20 40x(mm) 0
(b)221CAD

Figure 2-18. Ensemble-averaged velocity vectors, contours of velocity rrns for
deactivated (left) and activated (right) CMCV at (a) 171 (top), and (b) 221 (bottom)
CADs; swirl measurement plane at 2500 rpm engine speed

 

Swirl measurement plane with CMCV deactivated, 2500 rpm “i

 

0.0044
0.0040

1 PDF
. 0.0036 4

00032 ~
___ ,‘x 253 CAD
0.0028 138 CAD Kidd/f»

1 --—-171CAD ,. t 2210“)
00024 i ------- 221 CAD 1,7ie/
0.0020 , ,, , ..
-'--'253CAD / \‘.\
L 1.
0.0012 ff , ‘3 \
0.0008 4 ,' .
0.0004 J

. 0.0000 . .
4200 -700 -200 NC (“8) 300 800

 

   

138 CAD

 

 

 

 

Figure 2-19. PDFs of normalized circulation at different CADs for swirl measurement
plane with CMCV deactivated at 2500 rpm engine speed

Swirl measurement plane with CMCV activated, 2500 rpm
0.0044
0.0040 a
PDF
0.0036 _— 138 CAD
0.0032 — — — -171 CAD

0.0028 253 CAD ------- 2210210
0.00244 221cm

00020 \ ‘
0.0016 4 t' 'x

171 CAD /-~\ v.
0.0012 4 \ ’ .- i

138 CAD / A, .
0.0008 1 , .’ \.

00004 ‘i I, "x / \‘ “‘3 \'\§
0.0000 —’ 1.") . ‘44- ‘~ .
L -1200 -700 -200 NC (“5) 300 800

Figure 2-20. PDFs of normalized circulation at different CADs for swirl measurement
plane with CMCV activated at 2500 rpm engine speed

 

--—--253 CAD

 

 

 

 

 

 

45

Figure 2.21 shows the change in cycle-to-cycle variability characterized by the
standard deviation of NC and averaged TKE over the plane with respect to the crank
angle. The TKE with activated CMCV is only slightly higher than that of deactivated
CMCV at 138 CAD, as also seen in the PDF plots. The cycle-to-cycle variations are

higher with activated CMCV at 17] CAD and reduce during the compression stroke.

 

 

‘ 350 3.50 i

 

 

 

 

 

 

 

U Swirl measurement plane, i
E 2500 rpm 05—9“
300 o 13 »2.80
> hi
m 1.0
Q >4
1— 1—
250 m 1, 2.10
0
DD
8
‘3
200 < < 1.40
150 0.70
Crank angle (degrees)
100 H 0.00
90 140 190 240 290 340
l

 

+ smET/Tomcfdeguvaiedimlncv’ 7 44— sﬁdiﬁcchvgd 014ch —
| - —cr— — Averaged TKE. deactivated Cit/CV — ~0— — Averaged TKE, activated Cit/CV
Figure 2-21. Standard deviation of normalized circulation and turbulent kinetic energy at
different CADs for swirl measurement plane at 2500 rpm engine speed
2.3.3 Tumble measurement plane, 1500 rpm

Measurements are performed during the intake stroke at 121 crank angle degree
and during the compression stroke at 192, 257 and 300 CADs. Similar to the results as

observed in swirl measurement plane (at 1500 rpm), the rrns values for the tumble plane

at this engine speed are higher at 121 CAD and decay during the compression stroke,

Mittal et 31.68 (2008).

46

Figure 2.22 shows the PDF plots of normalized circulation at different CADs with
deactivated CMCV. As shown in the ﬁgure, the standard deviation of NC decreases from
121 CAD (a=219.76 Hz) to 257 CAD (0:86.62 Hz). It then increases to 210.69 Hz at
300 CAD. A similar pattern is observed when measurements are performed with
activated CMCV with higher cycle-to-cycle variations compared to the deactivated
CMCV; see Figure 2.23. These results show that cycle-to-cycle variation decreases from
121 CAD to 257 CAD, and then increases to CAD 300. Note that the CMCV effect is not
signiﬁcant on cycle-to-cycle variation at 300 CAD. Therefore, similar to the swirl plane
measurements, CMCV has a profound effect on ﬂow enhancement during the intake
stroke; however, its effect decays during the late compression when compared to the

deactivated CMCV.

 

 

  

 

 

 

 

 

Tumble measurement plane with CMCV deactivated, 1500 rpm
0.0044
0.0040
PDF
0.0036 ‘ —— 121 CAD ,. .
§ 0.0032
— - — - 192 CAD 300 CAD
0-0028 7 ------- 257 CAD ' '
0.0024 — . — - - 300 CAD \ 257 CAD
1 0.0020 /
.4 '. ._ ,/ 192 CAD
.. / 'x ' '
0.0016 " 1‘ 121 CAD
0.0012 .,",/‘~\ \\
0. '
0008 .4/ / \,\‘ \\
0.0004 . g, y.- ‘-g.\ \\
1 0.0000 A"/ - ’I' ' \' i
L -1200 -700_ #HNCQ/S) -29_0_*___ ’39922 -

 

Figure 2-22. PDFs of normalized circulation at different CADs for tumble measurement
plane with CMCV deactivated at 1500 rpm engine speed

47

 

 

  
  

 

 

 

0 0044 Tumble measurement plane with CMCV activated, 71500 rpm
0.0040 . PDF
. 0.0036 —121CAD
0.0032 - ---192 CAD
0.0028 a 300 CAD ------- 257 CAD
i 0.0024 - - — -- 300 CAD
0.0020 257 CAD ,/,?'\\
0.0016 - 192 CAD ” '~. .\\
0.0012 '
121CAD
0.0008
‘ 0.0004
~ 0.0000 ‘ ' ’ . r
-1200 -700 NC (“3) -200 300

 

_-.___ __.l

 

Figure 2-23. PDFs of normalized circulation at different CADs for tumble measurement
plane with CMCV activated at 1500 rpm engine speed

Figure 2.24 shows the cycle-to-cycle variations characterized by the standard
deviation of NC and averaged TKE for this tumble measurement plane at 1500 rpm

engine speed with respect to the crank angle. The TKE decays from 121 to 257 CAD and
69
then increases slightly to 300 CAD for both the cases. Joo et al. (2004) also

investigated the fact that the standard deviation of large- and small-scale variations was

greater during the early intake and late compression when compared to the late intake and

early compression process.

48

 

 

 

 

 

 

 

 

% 12$ ble measurement plane, F
Ck rpm 9.
30° ‘ ‘8 ‘ g . 5.50
5 1.1.1
0 ii
250 E7; ,3 4.50
0
DD
8
200 3: « 3.50
150 2.50
100 1.50
\ \m_ g _ _ _ 4:1
Crank angle (degrees) \ \ g , - .A
50 e ’ 0.50
L 90 140 190 240 290 340
TI; _ sirnev" T731 Edegmatgmi—I— T T STDEVTWECEateTJ/tc?’
. — —A-— — Averaged TKE. deactivated Cit/CV — —cr— — Averaged TKE activated CMCV

Figure 2-24. Standard deviation of normalized circulation and turbulent kinetic energy at
different CADs for tumble measurement plane at 1500 rpm engine speed
2.3.4 Tumble measurement plane, 2500 rpm

Measurements are performed during the intake stroke at 138 and 171 crank angle
degrees and during the compression stroke at 221 and 253 CADs with WOT. Figures
2.25 and 2.26 show the PDF plots of normalized circulation at these different crank angle
degrees. For deactivated CMCV, the standard deviation of NC decreases from 138 CAD
(0:264 Hz) to 221 CAD (0:10215 Hz). It then increases to 253 CAD with the value
of 139.38 Hz. These results show that cycle-to-cycle variation decreases from 138 to 221
CAD, and then increases to 253 CAD for deactivated CMCV. With activated CMCV, the
standard deviation decreases from 138 to 253 CADs. The cycle-to-cycle variations show
signiﬁcant difference at 138, 171 and 221 CADs (higher with activated CMCV);

however, it is only slightly higher at 253 CAD with activated CMCV. Therefore it can be

49

assumed that the CMCV effect reduces during the late compression to enhance the ﬂow

ﬁeld inside an engine cylinder.

 

0 0044 Tumble measurement plane with CMCV deactivated, 2500 rpm

‘ 0.0040 «
0.0036 , ._
0.0032 —— 138 CAD 253 CAD
0.0028 — — - - 171 CAD Z" ’

0.0024 ------- 221 CAD L,’ ‘\.

.' I'\ /
(l \

 

PDF

      

221 CAD

10.0020 —-—--253 CAD l.’ [0‘ \ V171 CAD
, ' . \
0.0016 , ,5 1‘ ‘.
‘ t ’; - r 138CAD
0.0012 - _. .

0.0008
0.0004

0.0000 . .
-2000 -1500 -1000 -500 NC (1,5)0 500

  

 

 

 

Figure 2-25. PDFs of normalized circulation at different CADs for tumble measurement
plane with CMCV deactivated at 2500 rpm engine speed

50

Tumble measurement plane with CMCV activated, 2500 rpm

0.0044
F

~000M)

1 0.0036
0.0032
0.0028 —
0.0024 ~

0.0016
0.0012

1 0.0008 ~
0.0004

 

PDF

253 CAD
\ o\.
./ \ --—--2530AD

00020 221 CAD =’~-. 1

 

——138CAD
-—--171 CAD L
------- 221CAD

 

 

0.0000
-2000

 

 

 

 

-1500 -1000 —500 NC (1/3) 0 500

 

Figure 2-26. PDFs of normalized circulation at different CADs for tumble measurement
plane with CMCV activated at 2500 rpm engine speed

51

Figure 2.27, when characterized by averaged TKE, shows consistently that cycle-
to-cycle variations are higher with activated CMCV during the intake stroke and its effect

decays during the late compression.

 

 

 

 

 

 

i
L 350 3.50
Tumble measurement plane,
U f-‘ﬁ
1 (I)
L 300 a :5 e: .. 2.80
0.1
l 3 M
L_ E—r
250 L W '3 -L 2.10
on
L a
Q)
L 3:
. 200 L 1.40
L
L 150 0.70
l
L Crank angle (degrees)
L 100 a 0.00
L 90 140 190 240 290 340
L L+ sway of NC, deactivated CMCV _+— erev of NC, activated CMCV
L . — -£x- _ Averaged TKE. deactivated CMCV - .0, -— Averaged TKE. activated (MN .

 

 

Figure 2-27. Standard deviation of normalized circulation and turbulent kinetic energy at
different CADs for tumble measurement plane at 2500 rpm engine speed

2.4 Summary
The effects of charge motion control are presented on ﬂow inside an IC engine
cylinder. PDF plots of normalized circulation are computed for both tumble and swirl
measurement planes at various CADs to characterize the cycle-to-cycle variations for
deactivated and activated CMCVs. The effects of CMCV on in-cylinder ﬂow

measurements are summarized below.

52

Swirl measurement plane at 1500 rpm: Cycle-to-cycle variation decreases from 121
CAD to 300 CAD with activated CMCV. Results show that CMCV has a
considerable effect to enhance the ﬂow ﬁeld during the intake stroke; however, the
effects are not signiﬁcant during the compression stroke for the swirl measurement
plane at 1500 rpm engine speed.
Swirl measurement plane at 2500 rpm: With activated CMCV, the cycle-to-cycle
variation decreases from 138 to 221 CAD, and then increases, only slightly, at 253
CAD.
Tumble measurement plane at 1500 rpm: Results show that cycle-to-cycle variation
decreases from 121 to 257 CADs, and then increases at 300 CAD with both the
conditions, CMCV deactivated and CMCV activated. The effects of CMCV are
signiﬁcant to enhance the cycle-to-cycle variations except at 300 CAD.
Tumble measurement plane at 2500 rpm: Cycle-to-cycle variation decreases from 138
to 221 CADs, and then increases at 253 CAD for deactivated CMCV. With activated
CMCV, the cycle-to-cycle variation decreases from 138 CAD to 253 CAD. Similar to
the results in the tumble measurement plane at 1500 rpm, it is found that the CMCV
effect reduces during late compression to enhance the ﬂow ﬁeld inside the engine
cylinder.

Insight on cycle-to-cycle variation, based on instantaneous measurements, is

needed to optimize the engine design. The results of this study have particular

signiﬁcance on fuel-air mixing, and, therefore, on the combustion process. Results show

that CMCV considerably enhances the ﬂow ﬁeld during the intake and compression

strokes; however, the effects are not signiﬁcant during the late compression. Therefore,

53

for this conﬁguration, it can be assumed that CMCV enhances the fuel-air mixing more

than the ﬂame speed.

54

CHAPTER-3

IN-CYLINDER FLOW MEASUREMENT USING STEREOSCOPIC
MOLECULAR TAGGING VELOCIMETRY

3.1 Introduction

The stereoscopic molecular tagging velocimetry technique is used to obtain the
multiple point measurement of an instantaneous three—component velocity ﬁeld inside the
optical cylinder of an internal combustion engine assembly. The stereo imaging approach

is commonly used earlier in the ﬁeld of machine vision and also in PIV to obtain the

three components of the displacement ﬁeld, Bohl et al. 33 (2001). Two basic approaches:

(i) lens translation and (ii) angular displacement have been utilized for this purpose. In
the lens translation method, two detectors are placed side by side with the image planes

parallel to the object plane; see Figure 3.1 [70]. More details of this method can be found

in Arroyo and Greated71 (1991). The authors demonstrated the technique in the study of

Rayleigh streaming ﬂow in a rectangular tube. Bohl et al. 33 (2001) and Bohl72 (2002)

used MTV with angular displacement method to measure the three-component velocity
ﬁeld generated by a propeller. The method is capable to provide nearly an entire overlap
between the two images. The present work also takes the advantage of angular
displacement method to measure an instantaneous three—component velocity ﬁeld inside
the optical engine cylinder. In this approach, the two cameras (left and right) view the

same region of interest from different angles (see Figure 3.2 [70]), and the undelayed and

55

delayed images are captured at different crank angle positions using both the cameras. A
new image processing technique is implemented to obtain the velocity data. The new
algorithm is computationally less expensive and eliminates the need to specify the
geometric details, as in the earlier works, to obtain the three-components of velocity. The
geometric distortions are corrected by in-situ calibration. The procedure involves two
major steps: (i) calibration process and (ii) data acquisition and reduction. In the
calibration process, two images of a calibration target are used to obtain the mapping
functions, which deﬁne the relationship between the positions in three-dimensional

physical space (x, y, z) and the positions in two-dimensional image planes, (X l , Y1) and
(X r , Yr ). The data reduction program ﬁnds the displacement of each grid point (laser

intersection) between the undelayed and delayed images for each camera. Mapping
functions and the two-dimensional image displacements are then used to obtain the three
components of the displacement vectors, and hence the three components of the velocity
ﬁeld are obtained in a physical space coordinate system. RMS cyclic variations of the
three-component velocity ﬁeld and out-of-plane mean vorticity are presented inside an
engine cylinder. Preliminary results show that cycle-to-cycle variations are more
prominent in the velocity component perpendicular to the tumble plane, as opposed to the
in-plane components. Such new insights will help better understand the details of these

ﬂows and further improve CFD models for IC engines.

56

 

........ .' ._ - - 7 ' Object Plane

Lens Plane
-- - O -------------------- O - - —-
Camera #1 Camera #2
-‘ ' Image Plane -

I’----—------—---—-------‘----.l.wl

 

Figure 3-1. Basic conﬁguration of lens translation method for stereoscopic imaging
(Raffel et al., 1998)

 

 

r '.

.‘3'3 .°. I
..Camera #1 ,' ' Camera #2 r’
. I . a

Figure 3-2. Basic conﬁguration of angular displacement method for stereOSCOpic
imaging (Raffel et al., 1998)

3.2 Experimental setup
The optical engine assembly used for the experiments represents a Ford 5.4 l 3-

valve engine. It has a 90 mm bore diameter, 105.6 mm stroke, 169 mm rod length, and

57

two intake and one exhaust valves. Figure 3.3 shows the engine cylinder with the
locations of tumble measurement plane and grid nodes, as well as the positions of intake
and exhaust valves. As shown in the ﬁgure, the measurement plane is located more
toward the right half of the engine cylinder. The top edge (of the plane) is located close to
the Head Deck level. The grid spacing is about 4.3 mm (~ 64 pixels). The experiments
are performed at the engine speed of 600 rpm. Measurements are taken at three different
crank angle degrees positions, i.e., 90 CAD, 180 CAD, and 270 CAD. At 90 CAD, the
piston is halfway to the stroke length and moving downwards to the BDC position during
the intake stroke. 180 CAD represents the piston position at BDC during the intake
stroke. Similarly, at 270 CAD the piston is halfway to the stroke length and moving
upwards to the TDC position during the compression stroke. The working ﬂuid in this
experiment is nitrogen, and it was deoxygenated since the presence of oxygen quenches

the phosphorescence of biacetyl.

58

   
      
  
 

LHead Deck
40 Level
20 III I I I i 1 “Measurement
A Tumble Plane
Ea 52223222
A Measurement
-20 Grids
Piston Top Level
“‘0 at 180 CAD

 

0

I x (mm) 50

' Bore Diameter = 90.2 m I

Exhaust Valve

~ Tumble Plane
Location

Figure 3-3. Tumble measurement plane inside an IC engine cylinder

59

Horizontal /
Beam Laser Beams
Front

Min” blocker
View Mirror

Tilted Lens \ LLLL / exr
. Sensor
Q @TXI Q Array

  
  

Optical
Piston

, Vertical
Laser Beams

  
 

C lindrical
C ALE/n5 Left Right ~
Camera Camera
Beam
Splitter % G > f
Mirror
Incoming
Laser Beam

Figure 3-4. Schematic of SMTV optical setup

The optical setup for SMTV measurements inside the engine cylinder is shown in
Figure 3.4. It consists of a pulsed ultraviolet laser (wavelength 308 nm, output energy of
250 mJ/pulse) whose light beam is used to create the grid pattern in the cylinder. The
laser beam passes through numerous optical elements including a beam splitter and
cylindrical lenses, which generate two light sheets propagating in the horizontal and
vertical directions. As shown in the ﬁgure, the vertical laser beams are obtained using a
mirror placed below the Optical piston. Therefore, the horizontal laser beams are
introduced into the cylinder through the optical cylinder and the vertical beams through
the optical piston. Beam perforated blockers are used to transform the light sheets into

sets of laser beams. As shown in Figure 3.3, the undelayed laser grid is formed in the x-y

60

plane and displaced with time because of ﬂuid motion in the three-dimensional (x, y, 2)

physical space. Coordinate systems (X l , 1’1) and (X r , Yr) are associated with the sensor

planes (or image planes) of the left and right cameras (see Figure 3.4), respectively. The
images at two successive time stamps are captured using LaVision’s SprayMaster and
Cooke Corporation’s Sensicam camera systems. Each image is an 8-bit image with the
size of 1024x1280 pixels. The imaging lens in front of each camera is mounted in a
ﬂexible arrangement that allows the lens axis to be tilted relative to the axis of the
respective camera. This allows us to meet the Scheimpﬂug condition for optimal imaging
at an oblique angle. For SMTV measurements, in this work, the angle between the two
cameras was about 100 degrees and the exposure time was set to 30 us. Note that the

delayed frame is referenced (in time) with respect to the undelayed frame by delay time.

3.3 Measurement Procedure

Fluid velocity measurements are acquired using stereoscopic molecular tagging
velocimetry. Calibration and data reduction algorithms are used to obtain the three
components of displacement vectors (Ar, Ay, and A2) at each grid point in physical space
using the recorded images. In the ﬁrst step, a calibration target (see Figure 3.5) is used to
determine the mapping functions. The target has a sufﬁcient number of characteristic
marks (typically circular) in all spatial coordinates (x, y, z for SMTV) with the known
centroids in physical space. The calibration object is then positioned inside the engine
cylinder. Images are taken using the camera system shown in Figure 3.4. The calibration
program ﬁlters the two calibration images using threshold approach, and then captures

the ﬁltered edge data for each bright spot and ﬁts the data into the circle equation to

61

obtain the actual centroids in the two image planes. A least-square ﬁt algorithm with

iterative procedure is used for this purpose, Spath73 (1995).

‘x
I

 

 

 

 

 

 

 

 

 

 

LY;
e ‘ «-.., f
. F L i- ": T“.
. . f ' " ""x-
e e e r 1"».
. . . L ‘n T
. . ‘ . LI'.‘~'
Q g
Q Q
Q Q Q 9 0
e ‘ e 9
0 Q 0 9 g
9 0
Q 0
Q g. S 9 e
e ‘ e 9
\ Q Q 9 g
9 o
0 0
§ Q C 9 Q
.. e ‘ e 1 9
W- t 9 l 9 \
9 e 9
i" ,L\ 11.‘ . . §
1 9 L . .
-~ \ e
L e
..L:

 

 

 

Figure 3-5. Calibration Target

Details of Spath [73] algorithm are presented below for completeness. In this method, the

starting values 0(0),b(0) and ’10) of the unknown center (a, b) and the unknown radius r

are determined for the iterative procedure considering the circle equation in its

nonparametric form;

(JC--a)2+(y-b)2=r2 13“

62

where (a, b) is the unknown center and r is the unknown radius of the circle. The

objective function considered to be minimized is as follows:

- ” 2 2
S(a.b,r)= Z r -r (3.2)
LLkLZ L

 

where, r szLk—m (k)! —bL2

Note that (x k , y k) is the ﬁltered edge data in this work. The necessary conditions for a

minimum of Equation 3.2 are,

6_S _6S _65

_____ —0
6r 6a 617 (3'3)

2_b2

result in the following system of equations by setting C = r2 — a

L2 E rfLa+L2 E xkykLb+L E rkLe= E karlf+yfj (3.4)

k=l k=1 k=l k=1

L2" 2: ."rkykLath z yngL" z rkLc = ‘3 y,(x,f+y,f) (3.5)
k: l k: I k=l k=l

L2" 2 kaa+L2n Z ykLb+nc= E Lxﬁ+y£j (3.6)
k—l k=l k=l

For nontrivial cases the desired starting values are obtained by solving Equations 3.4, 3.5

and 3.6 and hence,

 

a10) =0, 0(0) 2 b, rm) :\/C+02 +02 (3.7)

63

The iterative solution is obtained by considering the circle in its parametric form; see

Spath73 (1995),

x=a+rcosz

y=b+rsinz (3'8)
and the objective function to be minimized is:
n
S(a,b,r,zl,...,zn) =k§1LLyk —b—rsin 2k)2+ (xk —a—rcoszk)2L (3.9)
where z 1, , 2,, are the additional unknowns. The ﬁrst part of the necessary conditions,
6S_6S_6S_ (310)
6a 617 6r .
yields,

2 n n n ’7 ’7 .
r= n 2 kaCOSZ/tﬂ’ksmzki‘” Z xk Z coszk“ Z yk Z smk M (3'11)
k=1 k=l k=l k=l k=1

2 2
n n
where dz”2 "—1 L X coszk] +[ 2 sinsz
n k=l k=I

(3.12)
l k
n n
b=lL Z yk—r Z sinzk] (3-13)
" k =1 k =1
Whereas the second part of the necessary condrtron, 67— ’ , gives
k

64

J’k‘b
tanzk - xk _a (3.14)

 

10) ()

(0) (0) . . t
a b , as descr1bed earlrer, let a ,

With the known starting values a and r

(t) () ()

t . . . .
b , and r , are determined at t =0, 1, 2, Equatlon 3.14 is used to determine 2,, t

and

then 21.“) is inserted into Equation 3.11 to get rm” (0 (1+1)

, and ﬁnally 2;, and r are inserted

(1+1)

into Equations 3.12 and 3.13 to get a and b0“). Once the locations of the

characteristic marks in each image are determined, mapping functions are found such that
any given point in the object space (x, y, z) is related to its corresponding location X, Y in
the image plane of each camera. The obtained mapping functions are then used in a data

reduction algorithm to obtain the velocity data.

 

Figure 3-6. Undelayed (left) and delayed (right) images at 90 CAD SMTV
measurements for right camera

Figure 3.6 shows the undelayed (averaged of 45 frames) and delayed images for

the Fight camera at 90 CAD SMTV measurements. Similar images are obtained for the

65

left camera. A spatial correlation technique is used to obtain the grid displacements in the

image planes, Gendrich and Koochesfahani29 (1996). In this technique, a small window,

referred to as the source window, is selected about a grid point in the tagged region in the
undelayed image, and it is spatially correlated with a larger roam window in the delayed
image. The two windows are centered at the same spatial location, and the roam window

is large enough to encompass the grid displacement. A spatial correlation coefﬁcient, R,

is calculated between the two intensity ﬁelds of the source window (11) and the roam

window (12) as a function of pixel displacement between them. The location of the peak

in R(X,Y) deﬁnes the displacement vector within one pixel, and calculated as follows,

Gendrich and Koochesfahani29 (1996):

 

Is'lr _I—s'l—7:
R(X,Y)= (3.15)

0"] .01
S r

 

where over bar refers to the expected value and 0 denotes the standard deviation. The
two subscripts, s and r, are used for the source and roam windows. The windows are

selected to be squared shape. The various terms used in Equation 3.15 are as follows in

terms of the pixel values;

 

M N
1.1 =—1—— z z 15(i,j)lr(i+X,j+Y), (3.16)

s r M.N-=Lj___-L

T ‘ Eli/It")
=— 19])

(3.17)

66

 

 

“ 1 15%/”X Y) (318)

I =-——--- i+ ,j-t- , .
" M-Ni=1j=1r

a, = 43:4 Ellstimi—ZLZ. (3.19)
s M-Ni=1j=1

0' — LE E (I (i+X '+Y)-TL2 (320)
1r M'Nizljzzl r ’J r ‘

Sub-pixel accuracy is obtained by using two-dimensional higher order polynomial

ﬁt. The 2-D polynomial function of p’h order is as follows;

10 _
R(X,Y)= z x"Y°+a" Lx" ‘Y‘+ ..... +a ”XOY") (3.21)
n: ’

(077,0 -—1 n,

0

where the minimum number of required coefﬁcients are %( p +1)(p + 2). A fourth order

polynomial is used in this work to obtain the sub-pixel accuracy which requires a
minimum of 15 data points from the correlation ﬁeld. Since the sides of the square
window have odd numbers of pixels to position the region symmetrically relative to the
correlation peak, 5x5 pixels are considered for the ﬁtting window. The location
corresponding to the maximum correlation coefﬁcient can be found by setting the ﬁrst

order partial derivatives of Equation 3.21 (with respect to X and Y) to be zero and

subsequently solving the two algebraic equations; see Zheng and Klewicki65 (2000),

a p n 0 77-1 1 0 n _

aXnEOmmox Y +an_L,lX Y + ..... +an,nX Y )_0 (3.22)
0 p n 0 n—l 1 0 n

6YnE0(an,0X Y +an_mx Y + ..... +an’nX Y )_0 (3.23)

67

The above two equations are highly coupled nonlinear algebraic equations that have to be

solved iteratively. This work utilizes a decoupled technique, as presented by Zheng and

Klewicki65 (2000), to solve Equations 3.22 and 3.23. This decoupled technique is an

effort to seek a balance between accuracy and efﬁciency. As suggested by the authors,
this decoupling results in about an order of magnitude decrease in the required CPU time.
In this technique, the iterative solution of the maximum correlation point in two-
dimensional distribution is replaced by solving two decoupled one—dimensional
distributions. The assumption made here is that the maximum correlation point has been
determined to within pixel accuracy. This is obtained using spatial correlation technique,
as discussed earlier.

Three components of displacement in physical space at each grid node are

obtained using the following system of equations, Soloff et a1. 74 ( 1997).

 

 

 

 

L
(Ml) F201,1 F20,2 FXI,3\
F F F W
A"; __ Yl,1 Yl,2 Yl,3 Ay
— 3.24
AXr FXr,l FXr,2 FXr,3 A2 ( )
M F F F \
K r) ( Yr,l Yr,2 Yr,3)

where

5F. 5F. 5F!”
F-- = ~41- ,F.. = ——U- ,andF.. = J
’1’] 6x 11,2 6y y,3 z

and i=X,Y;j=l,r

 

The mapping functions derivatives need to be evaluated at each grid point (undelayed) in

the physical space coordinate system (x. y, z). The code calculates the laser grids

68

(undelayed) precisely for the two image planes and transforms the grids in physical space
using cubic dependence on X and Y. Three-dimensional displacements are then obtained,

solving Equation 3.24, using the concept of measurement plane pointers (MPPs),
Naqwi75 (2000).
Artiﬁcial images are used to obtain the accuracy of the mapping algorithm and

data reduction program (Mittal et al. 76 (2009), Sadr et al. 77 (2008)). This is a common

approach to establish and test the accuracy of the data reduction methods in optical ﬂow

velocimetry; for example see Raffel et al. 70 (1998). Gendrich and Koochesfahani29

(1996), and Zheng and Klewicki65 (2000), used this approach with a known spatial

distribution of displacements in their MTV work. Initially laser grids (8x13) in physical
space (x, y, 2) are generated and then transformed through the mapping functions to
obtain the corresponding grid nodes in the left and right camera planes. A three-
component velocity ﬁeld is then considered and the initial grid points are transformed by
this velocity ﬁeld to a new set of coordinates (x+Ax, y+Ay, z+Az). These images are
processed by calibration and data reduction algorithms to obtain the displacements. The
obtained values are then used to estimate the accuracy of the measurement technique. The
uncertainty (rms level) in determining the displacements for in-plane measurements is
about 0.08 pixels. This is higher than the planar MTV (less than 0.05 pixels) due to the
uncertainty associated with the calibration program. The uncertainty in the out-of—plane

component is about 0.14 pixels. The errors in estimated mean and rms for 95%

conﬁdence bound (Benedict and Gould78 (1996)) are about 2 pm and 1 pm for in-plane

components, and 4 pm and 2 pm for out-of—plane component, respectively. Note that this

69

uncertainty (in run) reduces with the improvement in spatial resolution of the cameras.
Bohl et a1. 33 (2001) reported the uncertainty (rrns level) 0.05 pixels for in-plane

measurements, about the same level for planar MTV, and 0.05-0.15 pixels in the out-of-

plane component.

3.4 Experimental Results

SMTV tumble plane measurements, inside an internal combustion engine
assembly, are presented for three different crank angle positions, i.e. 90 CAD, 180 CAD
and 270 CAD. The averaged undelayed image, considered for the correlation to obtain
the two-dimensional image displacements, is an average of 45 frames. 47 delayed frames
are considered to obtain the instantaneous velocity ﬁelds at each crank angle position.
Then, velocity statistics are calculated. The delay time between undelayed and delayed
images is 10 us, 15 us and 30 us for 90 CAD, 180 CAD and 270 CAD measurements,
respectively. A large delay is needed for slower ﬂows and vice versa. The piston is
located at y = -7.7 mm for 90 and 270 CAD, while the position at 180 CAD is at y = -52.3
mm.
3.4.1 Flow field at 90 CAD

At 90 CAD, the piston is halfway to the stroke length and moving downward to
the BDC position during the intake stroke. Figure 3.7 shows the ensemble-averaged

velocity data (left), (u),(v) and(w), and out-of—plane mean vorticity (right) at 90 CAD.
The velocity vectors in the plot represent the magnitude of (u) and (1) components, while
color contours show the magnitude of the (w) component. The values of velocity

magnitude vary from 6.7 to 53.1 m/s. They show the ﬂow dominance in positive 1:

70

direction due to the fact that both the intake valves are positioned toward the left side of
the cylinder and are open at 90 CAD. Also, it is to be noticed that the measurement plane
is located more toward the right half of the cylinder. At 90 CAD, the piston is at its
maximum speed, and because the piston area is substantially larger than the valve’s open
area, the velocity magnitudes are higher toward the upper section of the measurement
plane. The out-of—plane velocity component changes the direction at several nodes, as
expected. The out-of-plane mean vorticity at each grid node is calculated from the
ensemble—averaged velocity ﬁeld data. As shown in the ﬁgure, there are no signiﬁcant

rotations observed at 90 CAD measurements.

  

 

 

 

 

 

50 a
‘ (w) velocity LE
‘ (m/s) 40 ‘ . Out-of—plane
6 '53, . mean vorticity
-. * 1000 l/
‘21 30 ’5 ( s)
E 2
02 g 20
‘4 . a 0
_6 10 1‘ ~ _2
-3 . :‘
0 a g ‘4‘
__.~ *2; 33 m/s ‘ l
20 30 40 50 20 30 40 50
x(mm) x(mm)

Figure 3-7. Ensemble—averaged velocity (left) and out-of-plane mean vorticity (right) at
90 CAD

The ms of u, v and w velocity components are shown in Figure 3.8. High
variations are depicted for the out-of— plane velocity component due to the change in
direction between several consecutive frames. Cycle-to—cycle‘ variations are less for the v
velocity component. In general, over the entire plane, the rms values in x direction are

about a factor of 1-2 higher when compared to the rms values in y direction except

71

toward the right edge of the bottom half of the measurement plane. The slightly higher
(rms) values in x direction are due to the higher velocity magnitudes observed in x
direction. It is found that toward the right edge of the plane the velocity component, in
the x direction, changes direction between several consecutive frames due to proximity to
the cylinder wall. The ﬂow is more uniform, temporally, in the y direction, and the
velocity magnitude decreases from top to bottom due to reduced suction effects through
the intake valves. This leads to higher rms values in the x direction compared to the rrns

values in the y direction toward the right edge of the bottom half of the plane. The rrns

plots of u, v and w velocity components also show that the estimation of <v'2> as

L<u'2> + <w‘2>L/ 2 in equation 2.5 may not be a reasonable assumption. However, more

detailed and higher order statistical information is required to obtain further insights on
the nature of turbulence in this ﬂow.

The high values of rms velocities show the turbulent nature of ﬂow at this crank
angle (90 CAD). The ﬂow inside an engine cylinder is affected by several factors

including engine geometry, speed, load, valve timing and lift. The obtained high nns

. . . . . . . 43
values 1nsrde an IC engine cylinder are consrstent wrth the work of Ismarlov et al.

(2006) for both swirl and tumble measurement planes using well established two-

. . 79
component molecular tagging veloc1metry technrque. Zhang et a1. (1995) presented the

high nns values under with/without swirl conditions. Jarvis et al. 24 (2006) reported the

high values of rms cyclic variation during the intake stroke at 149 crank angle degree

using digital particle image velocimetry technique. They showed relative decay in the

72

cyclic variations at further crank angle degrees during the intake stroke. Their results

showed that the cyclic variations are less during the compression stroke (compared to the
intake stroke). Bicen et al. 9 (1985) found high level of ﬂuctuations in a reciprocating

engine at several locations using the laser doppler velocimetry technique.

 

 

 

 

_.!§§h '
25 35 45 25 35 45
x(mm) x(mm)

   

10 20 30 4 6 8 10 1214
u rms (m/s) v rms (m/s) w nns (m/s)

Figure 3-8. RMS of u, v and w velocity components at 90 CAD

3.4.2 Flow ﬁeld at 180 CAD

180 CAD represents the piston location at BDC position during the intake stroke.
Figure 3.9 shows the ensemble-averaged velocity ﬁeld (left) and out-of-plane vorticity
(right) data at 180 CAD measurements. The vectors in the ensemble-averaged velocity
plot represent the magnitude of (u) and (0) components, while color contours show the

magnitude of (H) component. The velocity magnitudes at this crank angle are smaller

compared to the 90 CAD measurements and vary from 3 to 18 m/s. The smaller velocity

magnitudes are due to reduced piston speed, momentarily at rest, at this crank angle (180

73

CAD). The mean values of (u), (v) and (w) velocity components over the plane are 8.2

m/s, -2.1 m/s and -0.5 m/s, respectively. The out-of—plane mean vorticity shows smaller

magnitudes compared to 90 CAD measurements, as expected.

 

 

 

 

 

 

 

 

 

50 <4 . x , . - i r
T ‘ ‘ , . + Out-of— lane
L“), velocrty i 1 mean vgrticity
(”l/5) * 1000(1/5)
2 . .
1"“.
0 ,3 ‘
1
0
-2 s
-l
—4
_ ‘ ‘ , . _ 15 m/s
20 30 40 50 20 30 40 50
x (mm) x(mm)

Figure 3-9. Ensemble-averaged velocity (left) and out-of—plane mean vorticity (right) at
180 CAD

Figure 3.10 shows the cycle-to-cycle variations for u, v and w rms velocity
componentsﬁThe values of u velocity nns vary from 3.6 to 16.2 m/s. Similar to 90 CAD
measurements; high variations are observed for w velocity component and least for v

velocity component. Results show that over the intake phase the relative cyclic variation

decays. This is in agreement with the results of Jarvis et al. 24 (2006).

74

              

 

 

 

200 40 20 40 200 40
x(mm) x(mm) x(mm)
5 10 15 1.5 2.5 3.5 4 6 s

u rms (m/s) v rrns (m/s) W rrns (m/s)
Figure 3-10. RMS of u, v and w velocity components at 180 CAD

3.4.3 Flow ﬁeld at 270 CAD

At 270 CAD, the piston is halfway to the stroke length and moving upward to the
TDC position during the compression stroke. Figure 3.11 shows the ensemble-averaged
velocity ﬁeld (left) and out-of-plane mean vorticity (right) data. The vectors in the
ensemble-averaged velocity plot represent the magnitude of (u) and (v) components,
while color contours show the magnitude of the (w) component. The mean velocity
magnitudes vary from 2.9 to 9.9 m/s and are much smaller than the 90 CAD and 180
CAD measurements. The mean values of (u),(v)and(w) velocity components over the
plane are 6.2 m/s, 1.3 m/s and -0.8 m/s, respectively. There are some rotations observed
in the bottom half of the plane, centered toward the bottom-right comer. Figure 3.12, the
rms plots of the three velocity components, shows that at this stage ﬂow is less turbulent.
The values of u velocity rrns vary from about 2 to 4 m/s except at the right edge of the

plane. Similarly, the ms of v velocity varies from about 0.6 to 1.6 m/s and shows more

75

cycle-to-cycle variations at the left edge. The w velocity rms varies from about 1.5 to 2.5

m/s except at the bottom right edge of the plane.

 

 

. <W> velocity ‘ LOut-of—plane
__ (m/s) mean vorticity
, * 1000 ( l/sl

. 06

0.4
0.2

0
-0.2
< -0.4

 

 

 

 

  

           

1'3. $— , h" i I
20 30 40 20 30 40 20 30 40
x (mm) x(mm) x(mm)
2 6 1 1.5 2 3
u rms (m/s) v rrns (m/s) W rrns (m/s)

Figure 3-12. RMS of u, v and w velocity components at 270 CAD

76

3.5 Summary

The application of the Stereoscopic Molecular Tagging Velocimetry technique for
the measurements in highly transient, three-dimensional ﬂows inside an internal
combustion engine is presented. In this technique images of a tagged region obtained
from two cameras, positioned at an angle relative to each other, are used to obtain the
three components of the velocity ﬁeld at each image instance. The images are analyzed to
obtain the three components of the velocity ﬁeld inside an IC engine cylinder. The
present implementation of SMTV has the advantage that it does not require detailed
modeling of the system geometry. The uncertainties in determining the displacements for
in-plane components and the out-of-plane component are about 0.08 pixels and 0.14
pixels, respectively. The three-component velocity ﬁeld, out—of-plane vorticity and cycle-
to-cycle variations are presented for tumble plane measurements at three different crank
angle positions (90, 180 and 270 CAD).

Results show high cycle-to-cycle variations in the out-of—plane velocity
component; however, less variation is observed in the velocity component along the
cylinder axis. The ﬂow has fully three-dimensional unsteady behavior during the intake
stroke; however the variations are less during the compression stroke. At 90 CAD the
cycle-to-cycle variations are higher for all the three velocity components (u, v and w),
compared to the measurements at 180 CAD when the piston is at BDC position during
the intake stroke. In this work, the SMTV images are captured using LaVision’s
SprayMaster and Cooke Corporation’s Sensicam camera systems. Each image is an 8-bit

resolution image with the size of 1024x1280 pixels. In order to meet strict exhaust

77

emissions and fuel efﬁciency standards, improved numerical simulations, supported by

this type of spatially and temporally resolved experimental studies, are desirable.

78

CHAPTER-4

INTAKE MANIFOLD AND IN-CYLINDER FLOW MEASUREMENTS INSIDE
AN INTERNAL COMBUSTION ENGINE ASSEMBLY

4.1 Introduction
It has been acknowledged for a long time that the ﬂow ﬁeld inside an IC engine

cylinder is highly complex, turbulent, and three-dimensional. It is the most important

factor controlling the combustion process, Heywoodl (1987). Therefore a thorough

understanding of the highly unsteady in-cylinder ﬂow is necessary to optimize the engine
design. In general, there are two approaches to investigate the in-cylinder ﬂows in an IC
engine assembly: (i) laser diagnostic methods and (ii) multi-dimensional numerical
simulations.

The studies of laser diagnostic in-cylinder ﬂow measurements are performed

using laser doppler velocimetry, particle image velocimetry and molecular tagging

velocimetry. Recently, Mittal et al. 76 (2009) used a stereoscopic molecular tagging

velocimetry technique to obtain the three-component velocity ﬁeld inside an IC engine
cylinder. However, it is to be noticed that the experimental techniques (either PIV or
MTV) of in-cylinder ﬂow analysis are limited to velocity ﬁeld measurements over a
plane and do not provide detailed ﬂow information over the complete volume of the
engine cylinder. It is the multi—dimensional numerical simulations that provide this
information, and offer signiﬁcant time and cost savings to design the engine with

improved performance. With the advent of powerful computers, several studies of in-

79

cylinder ﬂow have been reported using multi-dimensional numerical simulations, i.e. Luo

et al. 44 (2003), Shojaeefard and Noorpoor45 (2008). However, boundary conditions are

assumed in the modeling efforts. Experimental information that can provide the real-time
boundary conditions is necessary to perform more accurate multi-dimensional numerical
simulations of complex in-cylinder ﬂows. This may also require a validation with
experimental results, i.e. PIV or MTV/SMTV in-cylinder ﬂow measurements.

In this work, an experimental study is performed to measure the velocity and
pressure inside the intake manifold of a motored internal combustion engine assembly.
The aim of this work is to provide the data of real—time boundary conditions for more
accurate multi-dimensional numerical simulations of in-cylinder ﬂows in an IC engine.
The geometry of the intake manifold is simpliﬁed for this purpose. A hot-wire
anemometer and a piezoresistive absolute pressure transducer are used to measure the
velocity and pressure, respectively, over a plane inside the circular section of the intake
manifold. In addition, pressure measurements are performed at two locations over an
elliptical plane near the intake port. In-cylinder pressure is measured using a piezoelectric
type pressure transducer. Experiments are performed at 900 and 1200 rpm engine speeds.
Phase-averaged velocity and pressure proﬁles are then calculated from the instantaneous
measurements. Results of velocity and pressure measurements show the oscillatory
nature of ﬂow inside the intake manifold. In addition, in-cylinder ﬂow measurements are
performed to validate the modeling efforts. The stereoscopic molecular tagging
velocimetry technique is used for this purpose to obtain the multiple point measurement

of an instantaneous three-component velocity ﬁeld inside the engine cylinder.

80

4.2 Experimental setup

The internal combustion engine assembly used for the experiments is a three-
valve, two intake and one exhaust, 0.675 1 single-cylinder engine. It has a 90 mm bore
diameter, 105.6 mm stroke length and 169 mm rod length. Figure 4.1 shows the
experimental setup, which is used to measure the instantaneous velocity and pressure
proﬁles inside the intake manifold. Experiments are performed at 900 and 1200 rpm
engine Speeds. This work is aimed to provide the necessary boundary conditions for
multi-dimensional numerical simulations of in-cylinder ﬂows. Due to this, the geometry
of the intake manifold is simpliﬁed, i.e. intake plenum is removed and the circular
geometry is considered. However, near the intake port, there is a gradual transition from
elliptical to circular section; see Fig. 4.1. Surface models of this elliptical part and the
intake port section are under development using 3D laser scanning. Figure 4.2 shows the

geometry of the intake port with two intake valves.

81

,,———~—~-
/’ \_
-\
\
\.

/ Locations ofvelocity \\

1/ measurements
1/ l 2 3 4 5
Section LAA * * t *

    
  

  

Figure 4-1. Experimental rig showing measurement planes and locations inside the
intake manifold for ﬂow measurements

 

Figure 4-2. Intake port geometry

82

Velocity measurements inside the intake manifold may be performed using two
broad approaches: (i) methods involving the use of a sensor within the ﬂow under
investigation and (ii) optical methods. An earlier approach is considered in this work. A
constant temperature hot-wire anemometer is used for this purpose to measure the axial-
velocity proﬁles. Note that the hot-wire anemometer is a purely analog device and offers
a very high frequency response compared to the optical methods. Due to this, it can
provide a continuous boundary condition data for numerical simulations of complex in-
cylinder ﬂows. Also, it is simple to use and inexpensive compared to the optical methods.
Measurements are performed over a plane in circular section (radius 26 mm) at few radial
locations; see Fig. 4.1. As shown in the ﬁgure, location 2 is at the center of the plane.
Locations 1 and 3 are each 3 mm apart from the center. Similarly, locations 4 and 5 are at
21 and 24 mm from the center, respectively. A honey comb section is placed between the
intake manifold and the tank which can be used to premix the MTV molecules into the
working ﬂuid. This is to make the ﬂow more uniform and one-dimensional (inside the
intake manifold). Working ﬂuid in this experiment is nitrogen due to the fact that similar
setup is used for the in-cylinder ﬂow measurements (using stereo MTV). Note that the
presence of oxygen quenches the MTV chemicals and hence nitrogen is used.

Hot-wire anemometer is extensively used by several researchers in earlier in-

cylinder ﬂow studies; see Hassan and Dent5 (1971) and Parsi and Daneshyar80 (1989).

. 7
Wrtze (1980) compared the in-cylinder ﬂow measurement results of HWA and LDV in

a motored reciprocating engine. The author mentioned that accurate in—cylinder hot-wire
measurements are only possible for the intake and exhaust strokes. Therefore, inside the

intake manifold, HWA can be used more comfortably and accurately where the property

83

variations are even less signiﬁcant. One of the limitations using a hot-wire anemometer is
that it does not resolve the ﬂow direction. However, it can often be resolved by intuition
and geometric considerations. Fortunately, in engine intake manifold, the direction of
ﬂow during the intake stroke is known. It is then reversed when velocity magnitude
reaches to zero and so on.
The calibration of hot-wire used in this work is performed in air due to
unavailability of calibration facility in nitrogen. Note that the ﬂuid mechanic properties

of air are in close agreement with nitrogen. Fluid velocity is determined using King’s law

which states that

32 = A + 8110‘45 (4.1)

where A and B are the calibration constants, e is the hot-wire response and u is the ﬂuid
velocity.

Pressure measurements over the same plane (which is considered for velocity
measurements) are performed using a Kistler piezoresistive type absolute pressure
transducer. In addition, pressure measurements over an elliptical plane near the intake
port are also performed at couple of locations (centers of major and minor axes; locations
a and b, respectively); see Fig. 4.1. This is due to the complex transition of intake
manifold geometry from elliptical to circular section near the intake port and hence to
verify the pressure proﬁles in numerical simulations where boundary conditions will be
provided at the circular measurement plane. Data acquisition is performed using a high
speed combustion analysis system (CAS) with two crank angle degree resolution; see

Fig. 4.3 for C AS setup. Sampling was triggered by a TTL pulse generated by the engine

encoder.

84

 

Figure 4-3. Combustion analysis system (CAS) used for intake manifold ﬂow
measurement

Fluid velocity measurements inside the engine cylinder are performed using
stereoscopic molecular tagging velocimetry. The experimental setup is shown in Fig. 4.4.
Figure 4.5 shows the schematic of the engine cylinder with the locations of tumble
measurement plane and laser grids. The measurement plane symmetrically bisects the
two intake valves. The grid spacing is about 9 mm in the x direction and 5.7 mm in the y
direction. Experiments are performed at 900 and 1200 rpm engine speeds with similar
conditions used for the intake manifold ﬂow measurements. As discussed earlier, the

working ﬂuid in this experiment is nitrogen.

85

 

Figure 4-4. Experimental rig for in-cylinder ﬂow measurement using stereo MTV

 

 

\Piston top

12‘ O C‘ O O C’
e e o c o e at TDC
o o e o e o
A e o o e e o
40 7 E o o o o o o
E e e e o o O
K o e e c- c '3
20 _ o (t o O o 9
c o r;- O ‘1‘ 0
c o '3' Q 9 C‘
O c. g. c 0 0
0 — o o O O O 0

Measurement plane
and laser grids

'20 ‘ Piston top

Lat BDC \ I

0 x (mm) 50

 

 

 

#51
Bore diameter = 90.2 mm

Figure 4-5. Tumble measurement plane inside the engine cylinder

86

The optical setup used for the SMTV measurements inside the engine cylinder is
shown in Figure 4.6. As discussed in Chapter 3, it consists of a pulsed ultraviolet laser
with wavelength 308 nm and output energy 300 mJ/pulse whose light beam is used to
create the grid pattern in the cylinder. The laser beam passes through numerous optical
elements including a beam splitter and cylindrical lenses, which generate two light sheets
propagating in the horizontal and vertical directions. Beam perforated blockers are used
to transform the light sheets into sets of laser beams. The undelayed laser grid is formed
in the x-y plane and displaced with time because of ﬂuid motion in the three-dimensional

(x, y, 2) physical space. Coordinate systems (X I’Yl) and (X r , Yr) are associated with

the sensor planes of the left and right cameras (see Figure 4.6), respectively. The images
at two successive time stamps are captured using LaVision’s Imager Intense CCD
cameras (both for the left and right cameras) with intensiﬁed relay optics. Each image is a
lZ-bit resolution image with the size of l376x 1040 pixels. The imaging lens in front of
each camera is mounted in a flexible arrangement that allows the lens axis to be tilted
relative to the axis of the respective camera. This allows us to meet the Scheimpﬂug
condition for optimal imaging at an oblique angle. The angle between the two cameras

was set to 90 degrees.

87

Laser Beams

  
  
 
 
  

  

 

Beam
Mirror blocker Engine . _
Cylinder
.,\
. f ‘\
Tilted Lens \ i ﬁx
r
:1" sax """"" “*"wSensor
Cylindrical Left I Array
6 Lengs \. Camera nght
Camera
1 ‘\.
Beam ‘
Splitter
Mirror
Incoming
Laser Beam

Figure 4-6. Schematic of SMTV optical setup

4.3 Experimental Results

Velocity and pressure measurements are presented inside the intake manifold of a
motored internal combustion engine assembly at 900 and 1200 rpm engine speeds. 150
consecutive cycles are recorded; then, velocity and pressure statistics are calculated. In all
the figures, zero crank angle degree represents the start of the intake stroke, i.e. piston at
the top dead center of the intake. At 180 CAD piston is at the bottom dead center during
the intake stroke. Similarly, 360 CAD represents the TDC of the compression stroke.
SMTV in-cylinder flow measurements are presented at various crank angle degrees

during the intake and compression strokes at 900 and 1200 rpm engine speeds. The

88

averaged undelayed image, considered for the correlation to obtain the two-dimensional
image displacements, is an average of 50 frames. 100 delayed frames are considered to
obtain the instantaneous velocity fields at each crank angle position. Then, velocity
statistics are calculated. The delay time between undelayed and delayed images vary

between 12—26 us. A large delay is needed for slower ﬂows and vice versa.

4.3.1 Flow measurements at 900 rpm engine speed

Flow measurements are presented at 900 rpm engine speed. Figure 4.7 shows the
averaged in-cylinder pressure trace considering 150 consecutive cycles. The peak in-
cylinder pressure is about 21.2 bars and located at the top dead center of the compression

stroke due to the fact that engine is operated under motoring conditions, i.e. without

 

 

 

 

 

 

 

 

combustion.

25 1 i T ﬁ j I 7

20 r .3“: {W 1
Q.
l
15 l" .i x ‘l
/ .
\ 4
5 . , \ i
f \x
_ . JJ/f. \x‘n _._-—~—-—— .

0 j i 1 1 T _L- —
0 l 00 200 300 400 500 600 700

CA (degrees)

Figure 4-7. Averaged in-cylinder pressure trace at 900 rpm engine speed

89

 

1.03'1
1.02“

A
S
.0
v
Cu

1.01:

 

1.00-
0.99—
0.98 -
0.97 -

 

CA (degrees)

0.96 ' T 7 m r
O l 00 200 300 400 500 600 700

 

 

 

Figure 4-8. Averaged pressure trace at the circular section of the intake manifold at 900
rpm engine speed
Figure 4.8 shows the averaged pressure profile at the circular section (as
discussed earlier) inside the intake manifold of the engine assembly. The pressure profile
clearly shows the oscillatory nature of ﬂow inside the intake manifold. The minimum
pressure is about 100 kPa and it is observed during the intake stroke, as expected. The
peak pressure during the cycle is about 101.6 kPa.
Figure 4.9 shows the velocity profiles over the circular section at radial locations
2, 3, 4 and 5 inside the intake manifold. The peak velocity is about 9 m/s during the
intake stroke at 50 crank angle degrees, and then decreases to 5.6 m/s at 70 CAD. Due to
maximum piston speed at 90 CAD, it then increases again to 9 m/s at this crank angle.
Note that the velocity magnitudes are higher at the center of the plane during the forward
ﬂow; however, during the back ﬂow magnitudes are relatively higher near the wall. This
shows that the flow characteristics inside the intake manifold are strongly depend on

piston motion and intake valve timing and lift; see Figure 4.10 for intake valve liﬁ proﬁle

90

and its ﬁrst derivative. During the complete engine cycle, oscillating peak of the velocity
magnitude decreases traversing from intake to exhaust stroke and it is close to zero at the
end of the exhaust stroke. In addition, Figure 4.1] shows the averaged velocity data
during the intake stroke (at various CADs) with respect to the radial distance from the
center of the plane. Note that the radial positions in the ﬁgure at 0, 3, 21 and 24 mm
represent the measurement locations 2, 3, 4, and 5, respectively. The velocity at the wall

(26 mm from the center), which is assumed to be zero due to no-slip condition, is also

plotted in the ﬁgure.

 

10‘

81

-- at location, 2
at location, 3

Velocity (tn/s)

   
 

 

 

 

 

41 at location, 4
" ‘ at location. 5
24
01
-2 .
4 CA (degrees)
0 100 200 300 400 500 600 763

 

 

Figure 4—9. Averaged velocity proﬁle at different radial locations in the circular section
of the intake manifold at 900 rpm engine speed

91

 

 

  

—— Intake lift (mm)

 

 

 

 

 

 

Figure 4-10. Intake valve lift and its ﬁrst derivative

10
I I
8
A o 0
ﬂ
3 6
9% 4 2 Q
2
§ 2 ° ' . SOCAD
o SOCAD
0 - - - 900m:
A 120 CAD
'2 x 150 CAD
- 130 CAD Radial locationfrom center(mm)
.4 . , . . .
0 2 4 6 8

Figure 4-11. Averaged velocity at different radial locations 2 (0 mm), 3 (3 mm), 4 (21
mm), 5 (24 mm) and at the wall (26 mm) during the intake stroke at various crank angle

degrees; 900 rpm engine speed

Figure 4.12 shows the averaged pressure proﬁles over the elliptical section, near

the intake port, at locations a and b. Averaged pressure proﬁle over the circular plane is

92

E 8 ,' - 1st derivative of intake lift
5 .4»
st: 6 .'
—= /
l a r“
l 7; 4
>
I ..
x
J :22
E.
l o /
I CA (degrees)
-2
i 0 100 200 300 400 500 600 700

First derivative of intake lift

10712 14 16 7718 20 22 24 26

 

also shown in the ﬁgure for comparison. As expected, during the forward ﬂow, pressure
is relatively higher at the circular plane compared to the locations a and b over the
elliptical plane near the intake port However, it is to be noticed that pressure is slightly
higher at location a compared to location b. This may be due to the intake port geometry
(see Fig. 4.2) which shows that ﬂow is relatively stagnant at location 0 compared to
location b. Similarly, during the backﬂow, due to the reversal of ﬂow direction, pressure

magnitudes are higher at elliptical plane locations (a and b) compared to the values at

circular plane.

 

 

 

1.03"1 A
33
2—t e I“ 4
LO ‘3‘ A‘ 1' ° A "3 "t
1.01 ’ = - , .- ' ' A\ A
q‘V- VVV'VVVVVAEV
1.004 y
099“ V
0.98 '1’ ‘°“ at circular section
° at location a, elliptical section
0'97 " —- at location b, elliptical section
CA (degrees)
0.96 T T l T l l T—
0 l 00 200 300 400 500 600 700

 

 

 

Figure 4-12. Averaged pressure trace at the circular and elliptical sections of the intake
manifold at 900 rpm engine speed

In-cylinder flow measurement at 900 rpm engine speed are performed at 120 and
140 crank angle degrees during the intake stroke and 260 and 280 crank angle degrees
during the compression stroke. In the ﬁrst step, calibration images of the calibration
target are obtained using both left and right cameras. 150 images are recorded and then

the averaged calibration images are obtained; see Figure 4.13 for the left and right

93

calibration images, respectively. The centroids of each bright spot are determined using a
least-square ﬁt algorithm with an iterative procedure, and then the mapping functions are
obtained such that any given point x in the object space is related to its corresponding
location X, Y in the image plane. In the second step, undelayed and delayed MTV images
are captured at different crank angle positions inside the optical engine assembly. 50
undelayed frames are considered to obtain the averaged undelayed image at each crank
angle position for both the cameras. 100 delayed frames are recorded to obtain the
instantaneous velocity ﬁelds at each crank angle position. The delay time between
undelayed and delayed images vary between 12—26 us for the measurements at 900 rpm
engine speed. Figure 4.14 shows the averaged undelayed and a delayed frame for both the
cameras at 120 crank angle degree. The two-dimensional image displacements are then
calculated and thus the three—components of velocity (or displacement) are obtained in
physical space; see Chapter 3 for more details about the technique. Then the velocity

statistics are calculated.

94

 

Figure 4-13. Calibration images of left (left) and right (right) cameras

95

 

Figure 4-14. Undelayed (left) and delayed (right) images at 120 crank angle degree for
left (upper) and right (lower) cameras at 900 rpm engine speed

Figure 4.15 shows the ensemble-averaged velocity data (left), (11>,(v)and (w), and

ms of velocity (right) at 120 crank angle degree during the intake stroke at 900 rpm. The

velocity vectors in the ensemble-averaged plot represent the magnitude of (u) and (v)
components, while color contours show the magnitude of the (w) component. Due to the

fact that both the intake valves are open, the velocity magnitudes are higher toward the

Upper section of the measurement plane. The axial component of velocity, (v), show the

ﬂow dominance in negative y direction due to the downward piston motion at this crank

angle. The high nns values show that the ﬂow is highly turbulent at this crank angle

96

during the intake stroke. It is as high as 26 m/s toward the upper right section of the

measurement plane.
Figure 4.16 shows the ensemble-averaged velocity data (left), <z(>,<v> and(w) , and

ms of velocity (right) at 140 crank angle degree during the intake stroke. Similar to the
measurements at 120 CAD, the axial component of velocity shows the ﬂow dominance in
negative y direction and its magnitude decays traversing from upper to lower section of
the measurement plane. The rms values are higher than 18 m/s toward the upper section
of the plane. However, note that it decreases compared to the measurements at 120 CAD.

This shows that the cycle-to-cycle variations decay during the intake stroke.

(W ) velocity

 

 

    

 

 

 

 

 

 

(m/s) rrnstm/s)
70 4 70 . . . . .
rmscontour
60* , 4 3 60- re: _
/ a‘=“"-:£-._ 2
50—/ . i 50.
‘ ‘ 1 «er»
r- . A _
540.3/ a 0 E40
:30-r'. I ’. ‘ I _} _1 K 30_ §
20-‘ " ‘ ’ ’ ' 20-
~ - ' r z 1 _3
10' ~ ’ ’ _4 10
.- 4 I I
or? _ / l ‘5 O
-10 20m/s -6 _0
10 20 30 40 50 60 70 10 20 30 40 50 60
x(mm) x(mm)

Figure 4-15. Ensemble-averaged velocity ﬁled (left) and velocity rms (right) at 120
crank angle degree and 900 rpm engine speed

97

(w > velocity
(m/S) rrns (m/s)

       

4 70
ms contour 18
3 60 ,. , ._, ,
2 " ' l7
1 5 16
0 E4 15
-1 K 3 14
-2 l3
-3 20 12
,'410 11
-5 10
6 -10 9
10 20 30 40 10 20 30 40 50 60 70
x(mm) x(mm)

Figure 4-16. Ensemble-averaged velocity ﬁled (left) and velocity rms (right) at 140
crank angle degree and 900 rpm engine speed

Figure 4.17 shows the ensemble-averaged velocity data (left), (14), (v) and (w) , and

rrns of velocity (right) during the compression stroke at 260 crank angle degree. Due to
the upward piston motion at this crank angle, the axial component of velocity shows the
ﬂow dominance in positive y direction. As expected, the velocity magnitudes are smaller
at this crank angle compared to the measurements at 120 and 140 CADs during the intake
stroke. Similarly, rms values are lower, which shows that the cycle-to-cycle variations
decay traversing from intake to compression stroke. The peak of rrns values is about 4.5
m/s at this crank angle. The ensemble-averaged velocity (left), (u), (v)and(w), and rms of
velocity (right) during the late compression at 280 CAD are shown in Figure 4.18. It is to

be noticed that the rms values increases during the late compression. It is as high as 5.5

m/s at this CAD, 100 et al. 69 (2004) also investigated the fact that the standard deviation

98

of large— and small-scale variations was greater during the early intake and late

compression compared to the late intake and early compression.

 

 

 

 

 

 

 

 

 

 

(w ) velocity
(m/s) rmstm/s)
/U r Y 70 - w . .
. 1.5 ms contour
60' ‘ ‘ ‘ . .
\ - a \ \ \ l
50> \ I \ \ \ ‘
K \ . . ~ g 0.5
40' x? 1 1 1 x a
K i \ i x \ 0
- - ‘1‘ \ x
_ \ \ 3! -0.5
"' x
-l
—l.5
-2
-10 -10 - - - - .
It) 20 30 40 50 ()1) 7t) 10 20 30 4O 50 60 70
x(mm) x(mm)

Figure 4-17. Ensemble-averaged velocity ﬁled (left) and velocity rrns (right) at 260
crank angle degree and 900 rpm engine speed

99

(w > velocity

 

 

 

 

 

 

 

 

 

 

(m/s) m/ )
70 ,0 ' . ' . . rmsf s
601‘ W ’ "03' \ ‘ 4 60‘
\ \ \ \ \
50>- \\ \ \ ‘ \' J 50-
M \ \ \ \
40* \ \ 1 % 40r
E; x \ \
30*x\ \ \ \ \ \ 30- .. .
20 r 20 _ rms contour
1mg 57.7... 1 111-5
E A
0. '.._ OL
10 Mﬁ,47 7‘, 1‘7 ,. ,L ,7 7,L7, .2J, , / ,1 !0 J I A L J
10 20 30 40 50 60 70 10 20 30 40 50 60 70
x(mm) x(mm)

Figure 4-18. Ensemble-averaged velocity ﬁled (left) and velocity rms (right) at 280
crank angle degree and 900 rpm engine speed

4.3.2 Flow measurements at 1200 rpm engine speed

Velocity and pressure measurements are presented at 1200 rpm engine speed.
Figure 4.19 shows the averaged in-cylinder pressure trace. 150 consecutive cycles are
considered to obtain this averaged pressure proﬁle. The peak pressure is about 21.5 bars
and located at TDC of compression. No signiﬁcant difference can be observed in pressure
values compared to the in-cylinder pressure values recorded at the lower engine speed of

900 rpm due to motoring conditions; see Fig. 4.7 for comparison.

100

 

25 . , , , 1 . 1

 

 

 

20 1 3 ,
an

15 . -*
10 . .
.. ‘ ..l

5 / K.
1-—__-——~——J—--’ W” k\\-—_.__,W~—~——.M
0 0 100 200 300 400 500 600 700

CA (degrees)

Figure 4-19. Averaged in-cylinder pressure trace at 1200 rpm engine speed

i
«If ,
<4

-o— at circular section
. at location a, elliptical section .
-— at location b, elliptical section _

 

 

CA (degrees)

 

l T— l l—

0 100 200 300 400 500 600 700

Figure 4-20. Averaged pressure trace at the circular and elliptical sections of the intake
manifold at 1200 rpm engine speed

101

Figure 4.20 shows the averaged pressure proﬁles over the circular section and at
locations, a and b, over the elliptical plane near the intake port. The minimum pressure at
the circular section is about 98.7 kPa, and it is observed during the intake stroke, as
expected. The peak pressure during the cycle is slightly less than 102 kPa. Similar to the
measurements at 900 rpm, during the forward flow, pressure is higher at the circular
plane compared to the locations over the elliptical plane. Similarly, during the backﬂow,
due to the reversal of flow direction, pressure magnitudes are higher at elliptical plane
locations compared to the values at circular plane.

Figure 4.21 shows the velocity proﬁles at various radial locations. The peak
velocity magnitude is about 15 m/s during the intake stroke at 60 crank angle degrees.
This may be due to high piston acceleration at this point. It then diminishes and increases
again at about 120 CAD. Also, as shown in the ﬁgure, the velocity magnitudes are higher
at the center of the plane during the forward ﬂow when compared to the velocity
magnitudes near the wall; however, during the back flow, magnitudes are relatively
higher near the wall. Also, note that the velocity magnitudes at locations 1 and 3 are in
close agreement with each other. This shows the axis symmetry nature of flow for the
geometry considered in this work. As expected, the oscillating peak of velocity

magnitude decreases traversing from intake to exhaust stroke during the cycle.

102

‘ ‘ ' ‘at location, I
-“—at location, 2

’13
Velocity (m/s)

   
 

 
   

 

 

j 8 ‘ at location, 3
, 6 ‘ ‘ * at location, 4
. 4 , ‘1“ 7 at location, 5
2 , ‘1“
0 - \ a; _ . ._ , ' ,,
l ‘2 ‘ 33%;? V 1 CA (degrees)
0 100 200 300 400 500 600 700 ‘

Figure 4-21. Averaged velocity proﬁle at different radial locations in circular section of
the intake manifold at 1200 rpm engine speed

 

 

 

 

 

 

 

 

 

(w )velocity
.m (m/S) -.. rrns(m/si
IV 7 [u r r r ' '
rms contour
60r/ 1 x 1 (,0- W- 4
, 1 , ' a
50” / x I I 50‘ . i 1
x r I \ ‘ ‘ i, ,
A40r / x h ' A401 1
E _, / as. .
E30r - - Ezo- ,, 4
20’ ' ' ' 20“
10 ﬁw - 10' . ﬂ
0;; V, , &s 0-
-10 0

 

10 20 30 40 10 20 30 40 50 60 70
x(mm) x(mm)

Figure 4-22. Ensemble-averaged velocity ﬁled (left) and velocity rms (right) at 170
crank angle degree and 1200 rpm engine speed

Figure 4.22 shows the ensemble-averaged velocity data (left), (u),(v)and(w) , and

rrns of velocity (right) at 170 crank angle degree during the intake stroke at 1200 rpm

103

engine speed. The velocity vectors in the ensemble-averaged plot represent the magnitude
of (u) and (1.) components, while color contours show the magnitude of the (w)
component. As expected, the axial component of velocity, (v), has higher magnitudes
toward the upper section of the plane and it diminishes toward the lower section of the
plane. The rms values are as high as 16 m/s toward the central upper right section of the
measurement plane.

Figure 4.23 shows the ensemble-averaged velocity, (u) (v) and (w), and rrns of
velocity at 210 CAD during the compression stroke. Due to upward piston motion at this
crank angle, the flow is dominant in positive y direction. The rms vales are as high as 11

m/s, however, lower than the rms values observed during the intake stroke at 170 CAD.

 

 

   

 

 

 

 

 

(W > velocity
.m (m/S) .... rrns (m/si
’V . ' /u j r r . r 4
3
6 \\ ‘1 ...,t, 60? 1 10.5
5 ‘ ‘ 41$ 2 501 10
\- t \ . 9.5
E4 “if?“ \ 11:340. 9
E} \ \ \ E .
X \ \~ g. ‘ E30 8.5
\ \ \ .-.' 0
2 -
\ 1% 20 8
l ‘ _l 10> 7.5
N .. . . .. 7
2 0' nns contour
-10 . n . . 15 "V5 . _1n m m . . . 1 6'5
1o 20 30 4o 50 60 70 ”10 20 30 4o 50 60 70
X(mm) x(mm)

Figure 4-23. Ensemble-averaged velocity ﬁled (left) and velocity rms (right) at 210
crank angle degree and 1200 rpm engine speed

104

4.4 Summary

Experiments are performed to measure the velocity and pressure inside the intake
manifold of a motored internal combustion engine assembly. A hot-wire anemometer and
piezoresistive absolute pressure transducers are used to measure the instantaneous
velocity and pressure, respectively. Measurements are performed at 900 and 1200 rpm
engine speeds. Results showed the oscillatory nature of ﬂow inside the intake manifold.
The peak velocities, inside the intake manifold, are about 9 and 15 m/s during the intake
stroke at 900 and 1200 rpm engine speeds, respectively. In-cylinder flow measurements
are also performed to validate the modeling efforts, under development. Stereoscopic
molecular tagging velocimetry is used for this purpose.

Results of this study have particular signiﬁcance on the development of more
accurate multi-dimensional numerical simulations of in-cylinder ﬂows. This study
provides the necessary boundary conditions for numerical simulations. Also, in-cylinder

velocity and pressure data can be used to validation the modeling efforts.

105

 

.. ...r.) 7.

CHAPTER-5

IN-CYLINDER COMBUSTION DIAGNOSIS IN AN ETHANOL-GASOLINE,
DUAL-FUELED, SPARK IGNITION ENGINE

5.1 Introduction

Improvement in fuel efﬁciency and reduction in exhaust emissions are the two
main driving forces behind the new developments in internal combustion engines. In
addition, the increasing interest in using renewable energy sources as alternative fuels has
prompted research into the nature of combustion when these fuels are utilized. Ethanol is

one of these fuels. Ethanol has a higher octane number than gasoline and thus the

. . . . . . . . . 81
potential for higher efﬁCIency, by increasmg compressmn ratio, ex1sts, Mittal et al.

(2008). However, its lower heating value leads to fewer driving miles per gallon when it
is used as a direct replacement in current gasoline engines. Also, vehicle modiﬁcations
are required to operate the gasoline vehicles on high ethanol content blended fuels due to
its poor startability in cold weather and the requirement of ethanol corrosion-resistant
materials for the fuel system. Therefore, the blend of ethanol (in gasoline) is more widely
used in the vehicles that are designed to operate on gasoline fuel. This reduces the
consumption of fossil fuels. Also, due to the high latent heat of evaporation of ethanol,
combustion temperature decreases compared to vehicles operating on pure gasoline. This

enables higher torque and higher thermal efﬁciency due to reduced cooling heat loss,

Nakata et al. 82 (2006). In the United States, 10 percent ethanol in gasoline is widely used

in many fuel blends. In Brazil, the ethanol percentage is even higher and this has been in

106

use for several years, Brinkman83 (1981). Ethanol can be produced from renewable

energy sources such as corn, wheat, and sugar cane. Table 5.1 shows the physical

properties of ethanol and gasoline for comparison, Srivastava et al. 84 (2009). Note the

higher octane rating of ethanol than gasoline. This overcomes the disadvantage of a lower

heating value of ethanol compared to gasoline.

 

 

 

 

 

 

 

 

 

 

Parameter Ethanol Gasoline
Molecular weight (g/mol) 46.070 1 13.228
Average octane number (RON+MON)/2 104 91
Lower heating value (MI/Kg) 28 43
Latent heat of vaporization at 300 K (KI/Kg) 1025 317
Heat of formation (KI/mole) -217 -221
Liquid density at 293 K and 1 atm (Kg/m3) 789 751
Boiling point (K) 351 411
Ignition temperature (K) 695 553

 

 

 

 

Table 5-1. Properties of ethanol and gasoline

Several studies have been performed to investigate the effect of ethanol fuel in

. . . 85 . .
internal combust1on engines. Starkman et al. (1964) prov1ded the comparison between

various alcohol and hydrocarbon fuels. Brinkman83 (1981) studied the effect of

equivalence ratio and compression ratio on efﬁciency and exhaust emissions of an IC

107

engine operating on ethanol fuel. Salih and Andrews86 (1992) studied the inﬂuence of

gasoline-ethanol blends on emissions and fuel economy in a spark ignition engine. Their
results showed signiﬁcant reduction in nitrogen oxides (N O.) and carbon monoxide (CO)

emissions using ethanol-blended fuels. The unburned hydrocarbon. emissions with

ethanol blends were found to be higher than for 100 % gasoline. Nakata et al. 82 (2006)

also investigated the inﬂuence of ethanol fuel on SI engine performance, thermal

efﬁciency, and emissions. Bromberg et al. 87 (2006) studied the case with Direct-

Injection (DI) ethanol and Port-Fuel—Injection (PFI) gasoline where ethanol was only
used to reduce the engine knock; however, the study of ethanol fuel in IC engines was not
extended to optimize the combination of both PH and DI fUCI injection systems.

The objective of this work is to investigate the combustion characteristics of an
ethanol-gasoline, dual fueled, single cylinder spark ignition engine. A dual fuel injection
system with both direct-injection and port-fuel-injection is used in this work. The
performance of PFI—E85 and DI-gasoline, and PFI-gasoline and DI-E85 systems is
presented. E85 is a blend of 85 % ethanol and 15 % gasoline by volume. In each test, the
percentage of E85 is varied from 100 (0 % gasoline) to 0 (100 % gasoline) to compare
the various cases. PFI-gasoline and DI-gasoline (PFl and DI-gasoline) results are also
presented to provide a baseline for comparison. The cycle-to—cycle variability is
presented using coefﬁcient of variation in indicated mean effective pressure. Mass
fraction burned and burn duration are determined from the analysis of measured in-
cylinder pressure data. The well-known Rassweiler and Withrow method (Model 1), with

a new linear model for the polytropic index, is used to obtain the MFB curves. The

108

differences are presented for the net pressure method (Model 2) to evaluate the burn
rates. It is found that combustion is faster with the increase in PF I percentage for all the
three setups with dual ﬁiel injection. The PFI-E85 and DI-gasoline system showed that
the burn duration decreases signiﬁcantly with the increase in PFI percentage; however,
the PF I-gasoline and D1—E85 system showed only slight differences with the increase in
PFI percentage. Model 2 showed good agreement with Model 1 at high load conditions;
however, it predicts slower combustion at light load conditions. Model 2 has an
advantage that the data processing time is short enough to allow for online processing.
Therefore further investigation is performed to improve Model 2 calculations to evaluate

the mass fraction burned at light load conditions.

5.2 Experimental setup
Figure 5.1 shows the experimental setup of the dual fueled spark ignition engine
used for this study. The engine used for the experiments is a single cylinder Ford 5.4 l 3-
valve engine with two intake and one exhaust valves. Both port-fuel-injection and direct-
injection are used for each test measurement. Details of seven different tests performed
for PFI-E85 and DI-gasoline, PFI-gasoline and DI-E85, and PFI and Dl-gasoline systems
are shown in Table 5.2. Experiments are performed at different loads, 3.3 bar IMEP and
wide open throttle; and engine speeds of 1500 and 2500 rpm. For each test measurement,
ﬁve different cases are considered with PFI percentages of 100, 70, 50, 30, and O. The
percentage of D1 is correspondingly increased to maintain a constant relative air-to-fuel

ratio (It), inverse of fuel-to-air equivalence ratio (¢). For each case, the experiments

started at zero DI fuel injection, and then the D1 fuel injection was increased to the

109

desired percentage while maintaining a constant relative air-to-fuel ratio. Note that when
a certain percentage of E85 is used, the airvto-ﬁiel ratio/i, which is to be constant, is
measured by the oxygen sensor. The uncertainties due to environmental variations are
minimized by taking measurements for different tests in a single day with constant engine
coolant temperature. Pressure measurements inside the engine cylinder are performed
using a piezoelectric pressure transducer. 300 consecutive cycles are recorded for each
case with one degree of crank angle resolution. The measurement range for the pressure
transducer was from 0 to 250 bars. Data acquisition is performed using a high speed

combustion analysis system of A&D technologies.

 

A. PFI - E85 and DI - gasoline

 

 

 

 

1:1? Load Eggge Speed
A1. 3.3 bar IMEP 1500
A2. WOT 1500
A3. WOT 2500

 

 

 

B. PFI - gasoline and D1 - E85
Bl. 3.3 bar IMEP 1500
B2. WOT 1500
C. PFI and DI - gasoline

CI. 3.3 bar IMEP 1500
LC}. WOT 1500

 

 

 

 

 

 

 

 

 

 

 

 

Table 5-2. Test matrix

110

 

Figure 5-1. Experimental rig

5.3 Measurement Procedure

The primary purpose of combustion in a piston engine is to generate the pressure
for shifting the expansion process away from the compression process and produce the
work cycle. IMEP is an important parameter to measure the performance of this work-
producing cycle that transforms heat from the combustion process into mechanical work.
The measured pressure proﬁle is used to calculate the IMEP for each cycle (Equation
5.1), and thus the mean is obtained over N consecutive cycles (300 cycles in this work)

for each case.

_ ipdV
IMEP_ —V— (5,)

'd

111

where Vd is the displaced volume. Work delivered to the piston over the entire cycle is
obtained using the trapezoid rule. Cyclic variability is presented using the coefﬁcient of
variation in indicated mean effective pressure, covimep. It deﬁnes the cyclic variability

in terms of indicated work per cycle. This is obtained using Equation 5.2, where 0' is the
standard deviation.
0'.
tmep

0V. = x 100 (5.2)
zmep IMEP

 

5.3.1 Mass fraction burned analysis

MF B is a measure of fraction of energy released, due to combustion of fuel inside
an engine cylinder, with respect to the total energy released at the end of combustion
during a cycle. Several methods have been suggested to evaluate the MFB in gasoline
engines from the measured pressure data; see Section 1.2.2. This work utilizes the
Rassweiler and Withrow method, denoted as Model 1, with a new linear model for
polytropic index to evaluate the MFB curves. Start of combustion (SOC) and end of
combustion (EOC) are uniquely determined from the logarithmic pressure indicator
diagram using a least-square ﬁt algorithm. Comparison is presented for another analysis
model, denoted as Model 2, which utilizes the net pressure to evaluate the MFB curves as
discussed in Section 5.3.1.2. Model 2 has an advantage that the data processing time is

short enough to allow for online processing.

5.3.1.1 Model 1, Rassweiler and Withrow method

112

The implementation details of this method by Rassweiler and Withrow51 (1938)

vary in literature. The data depicted in Figure 5.2 are provided for the measured pressure
data in Test C2 with 100 % PF l-gasoline. In this ﬁgure, zero crank angle degree
represents the piston position at top dead center. The initial state, i, and the ﬁnal state, f,
are the essential singularities of the combustion process which represent the start and end

of combustion, respectively. The transition from i to f is referred to as the dynamic stage

. . 88 . . .
of combustion, Bitar et al. (2006). In this model, it is assumed that for any crank angle
interval, A6 , the actual pressure change, Ap , inside an engine cylinder is composed of a

pressure rise due to combustion, Apr , and a pressure change due to the volume change,

AP, , therefore AP = Ap, + App . The pressure change due to volume change can be

obtained using a polytropic index. A constant index value was chosen originally by

Rassweiler-Withrow where the magnitude was the average of the polytropic index before

and after the combustion process. Brunt and Emtage53 (1997) used the compression
index, nc, up to TDC and the expansion index, ne, thereafter. A new linear model for
polytropic index, nd, during the combustion process is introduced in this work; however,
the results are similar when the compression index, nc, was used up to TDC and the

expansion index, ne, thereafter (Brunt and Emtage53 (1997)); see Mittal et al. 89 (2009).

Therefore the pressure change due to volume change in any interval,A0 , in the absence of

combustion, can be obtained using Equation 5.3.

113

n

V d

—- — lik_
[MDV“FM: ‘pk-l—ipk-1 [ ] l 613)

 

50 T" ‘ ‘ 7 ’ f ' *
45L
40'
35 y- .0 ‘ "o. -<
30 l 0' ‘. 1
25 ' ... ... 4
20 l ’4’. .5. 1

t .-
15 ,0". CA (degrees)

102" 4 4 4 4 4
-30 -20 -10 n 10 20 30 40 50 60

 

 

 

Figure 5-2. Measured pressure data, Test C2 (PFI-gasoline 100 %)

An indicator diagram in logarithmic scales, Figure 5.3, is used to evaluate the

polytropic indexes, tie and 119. V, in the ﬁgure represents the normalized volume with

respect to the clearance volume. The two states, 1' and jf are identiﬁed by the departure of

the curve from the state lines representing the compression and expansion processes

using a least-square ﬁt algorithm. Now, with the known Ape :Ap —Apv, and
assuming that the mass of charge burned in the interval,A6, is proportional to the

. . th.
pressure rise due to combust1on, the MFB at the end of the k 1nterval can be evaluated

using Equation 5.4,

114

m k
bk _ ZOApc

__ (5.4)
23’ Ape

 

m
bio ta]

where N is the total number of crank angle intervals during the combustion process.

 

 

 

 

4 /'""" . . . 4
"Tie,
3.5‘! f “‘5‘. 1
\. “a.
3 i - ’ "o. 1
\ \s
2 5 1- r—ﬁ .\ \ 1
. Ln ‘0
a
2‘ 80:0 ““4“”. 1
.9. a".
1.5 P ”o... 1
1 ‘ ‘1
0.5 ' J
0 _I J
0 0.5 1

 

Figure 5-3. Indicator diagram in logarithmic scales, Test C2 (PFI-gasoline 100 %)

5.3.1.2 Model 2, Net Pressure method

This model evaluates the mass fraction burned by normalizing the net pressure

. . . 9
With respect to the overall net pressure increase at the end of combustion, Zhu et al. 0

(2003). The net pressure change, Ap(k) , between the two crank angles is:

 

n=l 3
V(k) ' V(k)
k = k+1 — k ——
Apt ) pt ) p( )1V(k+l)j V1 (5.5)
g
and the net pressure at each crank angle is
pNET (k) = pNET (k - 11+ Ap(k) (5.6)

115

where p is the pressure, V is the volume, and V [g is the chamber volume at the ignition

point. The simplicity of Model 2 is apparent from Equations 5.5 and 5.6. It is
computationally efﬁcient and due to the short data processing time, it can be used for

online processing.

5.4 Experimental Results

Combustion characteristics of an ethanol-gasoline dual fueled injection system in
a single cylinder SI engine are presented. Figure 5.4 shows the mean IMEP (top) and
COV,mcp (bottom) for the various test data. As shown in the ﬁgure, the mean IMEP is
almost constant for DI-E85 and PFI-gasoline system for different cases with the decrease
in DI-E85 (or increase in PFI-gasoline); however, it increases with the increase in PFI-
E85 for PFI-E85 and Dl-gasoline system at similar load conditions. The mean IMEP is
about 7.5 bar for PFI-E85 (100 %) and DI-gasoline (0 %) system when engine is Operated
at 1500 rpm with WOT. At similar Operating conditions, the mean IMEP for DI-E85 and
PFI-gasoline is about 9 bar. At 2500 rpm engine speed with WOT, the mean IMEP
increases from about 10 to 10.92 bar with the increase in E85 percentage (0 to 100) for
PFI-E85 and DI- gasoline system. The COV of IMEP decreases with the increase in PFI
percent (beyond 50 %) for each test point (Tests A and B in Table 5.2). The cyclic
variability has the peak either at 30 or 50 % of PFI. 100 % PFI shows the minimum
cyclic variability for each test, and the COV of IMEP is even less than 1 for the PFI-E85

and DI-gasoline system when the engine is operated at 1500 rpm with WOT condition.

116

 

 

 

 

 

 

 

 

 

 

 

 

 

 

: E j, ............... . .......... a ---------- t --------------- 11
5 a 8.1
1 ﬁ .r ____.___.___+—-—~~"“+“‘—~4—.4———~4——-—-1
. E 61
f g 4,
g 11*.” ~ """"""""" t ;;-_- ----- t ---------- — «44m ...”...
2* *‘r‘ PFI-E85, 3.3 bar imep
. . *‘P" PFI-E85, WOT
10 + PFI-E85, WOT, 2500 rpm
11;? ‘ 1 “""’ DI-E85, 3.3 bar imep
3 8 ~ ' . DI-E85, wor
.g 6 ,'-'V’ ' xix- ‘ .* _--._--.s
> \* -.\_\ \\--. ' - c - _ _‘ - - ‘
U “Q“ ~___ 3 ' ——___N_.
2 ‘I’ - ..... _: it- ~~JWN
PFI ("/o)0
60 80 .109 ..

Figure 5-4. Mean IMEP (top) and COV,mcp (bottom) for PF l-E85 and DI-gasoline, and
DI-E85 and PF I-gasoline systems

5.4.1 PF I-E85 and DI-Gasoline system

This section describes the combustion characteristics of an ethanol—gasoline, dual
fueled injection system in a single cylinder SI engine where PFI system is used for E85
injection and D1 system is used for gasoline injection (PFI-E85 and Dl-gasoline). Tests
Al, A2 and A3 in Table 5.2 represent these measurements at 3.3 bar IMEP and WOT
load conditions with engine speeds of 1500 and 2500 rpm. Five different cases are
considered for each test with PFI E85 percentages of 100, 70, 50, 30, and 0. The
percentage of D1 gasoline is correspondingly increased to maintain a constant relative air-
to-fuel ratio (xi) for each case.

Figure 5.5 shows the averaged pressure proﬁles (top) and MFB curves (bottom)

for Test A1 (3.3 bar IMEP, 1500 rpm). Three different cases are shown with PFI-E85

117

percentages of 0, 70 and 100. As shown in the ﬁgure, the peak pressure increases with the
increase in PF I-E85. This is expected due to lower combustion gas temperature with
ethanol than with gasoline. Based on this, it can be assumed that the cooling heat loss
from the combustion chamber decreases with ethanol. Similarly, the burn duration

decreases with the increase in PFI—E85. Ethanol has an advantage of high combustion

speed for SI engines, Brinkman83 (1981). The combustion process starts at about -20

CADs for all the cases. There is no signiﬁcant difference in 10 % burn durations;
however, 50 % and 90 % burn durations are much faster with 100 % PFI-E85 when
compared to the other cases. The burn duration is about 51 CADs for 100 % PFI-E85 and
increases with the decrease in PFI-E85. The difference of Model 2 results when
compared to Model 1 is presented using vertical bars in the ﬁgure. The vertical bars in
the downward direction indicate that Model 2 predicts slower combustion when
compared to Model 1 and the difference increases with the decrease in PFI percentage. It
was found that the proper selection of polytropic index (n) and the determination of end
of combustion are the important parameters for calculating the MFB curves using Model
2. Model 2 determination of EOC is signiﬁcantly later than the actual CAD value, at this

light load condition, and therefore it predicts slower combustion compared to Model 1.

118

 

21

193%
I71:
15"
131*

  
 
   

PFI—£85 and Dl-Gasol—ine" 1_
, 3.3 baI_IMEP. 1590 rpm '

  

 

 

 

0.8 1
0.7 ii
0.6 t
0.5 ‘
0.4 it

 

PFI-E85 100%
-—- —PFl-E85 70%
...... PFI-E85 0%

 

0.3 1* /.

 

 

0.2 x4 .

0-1 i H CA de rees
0.01 — 7 1 r r( g ')

 

 

 

-20 - 10 0 IO 20 30

j’

40 50

Figure 5-5. Averaged Pressure (top) and MFB curves (bottom) for PFI-E85 and DI-

gasoline system at 3.3 bar IMEP and 1500 rpm

Figure 5.6 shows the averaged pressure proﬁles (top) and MFB curves (bottom)

for Test A2 (WOT, 1500 rpm) with PFI 585 percentages of 0, 70 and 100. Similar to Test

Al results, the peak pressure increases with the increase in PFI E85. It is interesting to

note that the peak pressure is about 44.5 bar with 100 % PFI-E85 and then decreases

sharply with the decrease in PFI-E85. Also, there is some phase shift in the peak pressure

locations (CADs) due to ﬁxed spark timing considered for the test data. As expected, the

burn duration increases with the decrease in PFI-E85. At this load condition, combustion

starts at about ~12 CADs for all the cases and signiﬁcant difference can be observed even

at earlier stages (10 % burn) with much faster combustion for 100 °/o PFI-E85. Similarly,

ll9

50 % and 90 % burn durations are also less for 100 % PF I—E85 when compared to the
other cases. The burn duration is about 39 CADs for 100 % PF I-E85 and increases with
the decrease in PFI E85. Model 2, at 100 % PFI-E85, predicts slightly faster combustion
than Model 1. It is in good agreement when PFI-E85 is 70 % and assumes slightly slower
combustion when PFI-E85 is 0 %. Overall, Model 2 is in good agreement with Model 1

at this higher load condition.

 

45
401
35‘
30*
251
204
151
10-""'

5 s . r

1.0
0.9 i
0.8 1
0.7 t
0.6 v
0.54
0.4~
0.3 r
0.2 4
0.1 i

P (bar)

     

PFI-E85 and Dl-Gasoline
WOT, 1500 rpm

  
 

 

 

 

 

PFI-E85 100%
-— —PFI-E85 70%
...... PFI-E85 0%

 

- CA (degrees)
-20 ~10 0 10 20 30 40 50

 

Figure 5-6. Averaged Pressure (top) and MFB curves (bottom) for PFI-E85 and D1-
gasoline system at WOT and 1500 rpm

Figure 5.7 shows the averaged pressure proﬁles (top) and MFB curves (bottom)
for Test A3 (WOT, 2500 rpm) with PFI E85 percentages of 0, 70 and 100. Similar to

Tests A1 and A2, the peak pressure increases with the increase in PFI, and it is as high as

120

55 bar with 100 % PFI-E85. The burn duration is about 37 CADs for 100 % PFI. It
increases with the decrease in PFI-E85; however, no signiﬁcant difference can be
observed when PF I-E85 is 70 and O %. Also, Model 2 shows good agreement with Model

1 at this high load condition and higher engine speed.

 

 

55
50‘
45"
40‘
35‘
30'
25-
20-
15‘
10-

5

1.0

0.9 ‘ MFB
0.8 ‘

0.7 ‘
0.6 ‘
0.5 ‘

0.4 ~
0.3 ‘
0.2 -

0'1. CA de rees
0.0 _' . ( 'g )

-20 -10 O 10 20 30 40 50

P (bar)

  

 

 
 
 

PFI-E85 and Dl-Gasoline
WOT, 2500 rpm

 

 

 

 

Li

 

—PFl-E85 100%
— —PFI-E85 70%
------- PFI-E85 0%

1

 

Figure 5-7. Averaged Pressure (top) and MFB curves (bottom) for PFI-E85 and DI-
gasoline system at WOT and 2500 rpm

5.4.2 DI-E85 and PFI-Gasoline system

This section describes the combustion characteristics of an ethanol-gasoline, dual
fueled injection system in a single cylinder SI engine where a D1 system is used for E85
injection and a PFI system is used for gasoline injection, (DI-E85 and PFI-gasoline).

Tests BI and 82 represent these measurements at 3.3 bar IMEP and WOT load

121

conditions, respectively. Five different cases are considered for each test with PFI
gasoline percentages of 100, 70, 50, 30, and O. The percentage of D1 E85 is
correspondingly increased to maintain a constant relative air-to-fuel ratio for each case.
Figure 5.8 shows the averaged pressure proﬁles (top) and MFB curves (bottom)
for Test Bl (3.3 bar IMEP, 1500 rpm). Only two cases with D] E85 percentages of 0 and
100 are shown due to less variation in measured pressure and MFB curves. As shown in
the ﬁgure, there is no signiﬁcant difference in measured pressure proﬁles at this part load
condition; however, it is interesting to note that the peak pressure increases slightly with
the decrease in DI-E85. Similar to the pressure proﬁles, no signiﬁcant difference is
observed in burn durations. The burn duration is about 58 CADs for 0 % DI-E85 and
increases only slightly with the increase in Dl-E85. Similar to the results shown in Figure
5.5 (PFI-E85 and DI-gasoline system with part load condition), Model 2 predicts slower

combustion and the difference with respect to Model 1 increases with the decrease in PFI

percentage.

122

17'
15'
13-
11-
9 .

______

P (bar)

PFI-Gasoline and 131-1385.l
3.3 bar IMEP, 1500 rpm .
5

1.0'
0.9‘

0.8-
0.7-
0.6~
0.5-
0.4-
0.3‘
0.2-
“ CA (de

00 1' ' . . grees) . a
-20 -10 0 10 20 30 40 50

 

 

 

MFB

 

—" D1-E85 0%
DI-ESS 100%

 

 

 

 

 

 

 

Figure 5-8. Averaged Pressure (top) and MFB curves (bottom) for DI-E85 and PFI-
gasoline system at 3.3 bar IMEP and 1500 rpm

Figure 5.9 shows the averaged pressure proﬁles (top) and MFB curves (bottom) for
Test B2 (WOT, 1500 rpm) with DI-E85 percentages of 0 and 100. This ﬁgure can be
compared with Figure 5.6 (PFI-E85 and DI-gasoline system) with similar load
conditions. The peak pressure is slightly higher for 0 % DI-E85, and it is as high as 46
bar. Similar to the part load condition, Figure 5.8, there is no signiﬁcant difference in
bum durations, and it is about 40 CADs for 0 % DI-E85. Model 2 results at this high load

condition are in good agreement with Model 1 due to accurate determination of EOC.

123

 

 

A
U!
P (bar)

   

pheasant.“amiss:
wor, 1500 rpm

    

 

 

 

j MFB

 

"_ DI-E85 0%
-- -- D1-E85 100%

 

 

 

 

 

CA (degrees)
-20 - l O 0 IO 20 30 4O 50

 

Figure 5-9. Averaged Pressure (top) and MFB curves (bottom) for DI-E85 and PF 1-
gasoline system at WOT and 1500 rpm

5.4.3 PFI-Gasoline and Dl-Gasoline system

This section describes the combustion characteristics of a dual injection system
where gasoline is used for both PFI and DI. Tests Cl and C2 represent these
measurements at 3.3 bar IMEP and WOT load conditions, respectively. Five different
cases are considered for each test with PFI gasoline percentages of 100, 70, 50, 30, and 0.
The percentage of D1 gasoline is correspondingly increased to maintain a constant
relative air-to-fuel ratio for each case.

Figure 5.10 shows the averaged pressure proﬁles (top) and MF B curves (bottom) for
Test C1 (3.3 bar IMEP, 1500 rpm). Three different cases with PFI gasoline percentages

of 0, 50 and 100 are shown in the ﬁgure. The peak pressure increases with the increase in

124

 

PF 1 percentage, and it is about 15 bar for 100 % PF 1. However, no signiﬁcant difference
can be observed in burn durations. The combustion process starts at about -18 CADs (all
cases) and the burn durations are 61, 63 and 65 CADs for 100, 50 and 0 % PF I-gasoline,
respectively. As seen earlier, Model 2 has slower combustion, compared to Model 1,

when engine is operated at 3.3 bar IMEP.

 

      

PF 1 and DI-Gasoline
3.3 bar IMEP, 1500 rpm

 

  
  
 
 

 

 

——PF1-Gasoline 100%
-— -PF1-Gasoline 50%
------- PF l-Gasoline 0%

 

CA (degrees) 8 l

1— ﬁ

10 20 30 40 50

1L:
O
. II
O
c; ‘1

Figure 5-10. Averaged Pressure (top) and MFB curves (bottom) for PF 1 and DI-gasoline
system at 3.3 bar IMEP and 1500 rpm

Figure 5.1 1 shows the averaged pressure proﬁles (top) and MFB curves (bottom) for
Test C2 (WOT, 1500 rpm) with PFI gasoline percentages of 0, 50 and 100. Similar to
Test C1, the peak pressure increases with the increase in PF 1, and it is as high as 48 bar

with 100 % PFI. The burn duration is slightly shorter, and about 45 CADs for 100 % PFI.

125

No signiﬁcant difference is observed in further cases with the decrease in PF I percentage.

Again, Model 2 shows good agreement with Model 1 when the engine is operated at wide

open throttle.

 

  

PF 1 and DI-Gasoline
WOT, 1500 rpm

    
    
 

 

 

——PFl-Gasoline 100%
—- —-PF1-Gasoline 50%
------ PF l—Gasoline 0%

 

CA (degrees)
-20 -10 0 10 20 30 40 50

Figure 5-11. Averaged Pressure (top) and MFB curves (bottom) for PFI and DI-gasoline
system at WOT and 1500 rpm

5.5 Improving the MFB calculation using net pressure method

Combustion characteristics of an ethanol-gasoline, dual fueled, single cylinder SI
engine are presented in Section 5.4. The results of mass fraction burned calculation, using
the net pressure method (Model 2), are compared with the well established Rassweiler
and Withrow method (Model 1). Both the methods showed good agreement at high load

conditions; however, the net pressure method showed slower combustion than that of the

126

Rassweiler and Withrow method at part load conditions. Model 2 has an advantage that
the data processing time is short enough to allow for online processing. Therefore, further
investigation is performed to improve Model 2 calculations to evaluate the mass fraction
burned at light load conditions. In-cylinder pressure data is considered for the PFI-
gasoline and DI-gasoline system with 100 % PFI. It is found that the proper selection of

polytropic index, n, and the determination of end of combustion are the important

parameters for calculating the MFB curves using Model 2, Mittal et al. 89 (2009). The

EOC in Model 1 is determined accurately using a least-square ﬁt algorithm; however, in
Model 2, it is determined by locating the peak of the net pressure. At high load
conditions, Model 2 determines the EOC with a reasonable accuracy, and therefore the
results are in close agreement with Model 1. At part load conditions, this method (Model
2) is less ideal to determine the EOC than that at high loads. It predicts EOC signiﬁcantly
later than the actual crank angle value, and therefore Model 2 has slower combustion
compared to Model 1. The accurate determination of EOC in Model 2 is affected by
selection of polytropic index, n. Note that the constant polytropic index, n = 1.3, is used
in Equation 5.5. The effect of the polytropic index on the EOC determination, and
therefore its inﬂuence on the MF B calculation, at part load conditions, is shown in Figure

5.12 (upper graph). The difference in calculating the MFB (using Model 2) reduces

compared to Model 1 when the index value n = ne = 1.20 is used in Equation 5.5. This is

due to the determination of the EOC with improved accuracy; however, some slight

differences can still be observed with this index value (n = ne). The index value n = nc =

1.165, in Equation 5.5, determines the EOC with reasonable accuracy compared to Model

127

1 determination using a least—square ﬁt algorithm. Therefore the MFB proﬁle also shows

good agreement with Model 1 when n = no is used in Equation 5.5. The model 2 MFB
calculation, with n = nc , also shows good agreement with model 1 results at the WOT

load condition (lower graph). Note that the polytropic index of expansion (ne), SOC and

EOC determinations using a least-square ﬁt algorithm make Model 1 computationally

more expensive than that of Model 2. Therefore, Model 2 is simple to implement, and the

. . . . . . . 89
data processmg time IS short enough to allow for real-time applications. Mittal et al.

(2009) discussed the robustness of the net pressure method at different engine speeds and

also with the variation in PFI and DI percentages of the dual fuel system in 81 engine.

128

 

 

  
  
    

0.9 ‘1 MFB

0.71L ' ‘ ,
1 PFI and DI-Gasolme

(PFI 100%)
3.3 bar IMEP, 1500 rpm

  
   
   

Model 2, n = ne
_— “Model 2, n = 1.3

 

 

 

 

 

MFB

0.6 4 PF 1 and Dl-Gasoline
(PFI 100%)
WOT, 1500 rpm

 

CA (degrees)

 

-30 -10 10 20 A 50
Model 1, linear polytmpic index, nd 0 Model 2, n = nC

 

 

 

 

Figure 5—12. MFBs for PFI-gasoline and DI-gasoline system at 3.3 bar IMEP (top) and
WOT (bottom)

5.6 Summary

Combustion characteristics of a dual fuel injection system in a single cylinder SI
engine are presented. Variations including PFI—E85 and DI-gasoline, PFI-gasoline and
DI-E85, and PF 1 and DI-gasoline systems are compared at different load conditions. In

each test study, the percentage of D1 system is varied from 0 to 100 %. The cycle-to-

129

cycle variability is presented using COV in IMEP. MF B and burn angles are determined

from the analysis of measured in-cylinder pressure data.

PFI-E85 and DI-gasoline: Results show that mean IMEP increases with the increase
in PFI-E85 due to reduced charge cooling effect and fast ethanol combustion.
Similarly, the combustion is faster with the increase in ethanol percentage (reduced
D1 charge cooling). The burn durations are 51, 39 and 37 CADs with 100 % PFI-E85
when the engine is operated at 3.3 bar IMEP (with 1500 rpm), WOT (1500 rpm) and
WOT (2500 rpm) load conditions, respectively.

PF I-gasoline and DI-E85: Results show that there is no signiﬁcant difference in the
mean IMEP with the increase in DI-E85. The effects of fast ethanol combustion due
to increase in DI-E85 are compensated by the increase in D1 charge cooling effects.
Similarly, combustion duration decreases only slightly with the decrease in DI-E85.
The burn durations with 0 % D1-E85 are 58 and 40 CADs when engine is operated at
3.3 bar IMEP and WOT load conditions, respectively.

PH and DI-gasoline: At part load condition, there is no signiﬁcant difference in burn
durations with the variation in PFI; however, at WOT it decreases only slightly with
the increase in PF 1. The burn durations with 100 % PFI are 61 and 45 CADs when
engine is operated at 3.3 bar IMEP and WOT load conditions, respectively.

The comparison to evaluate the mass fraction burned from Model 2 (Net Pressure

method) is presented with respect to Model 1 (Rassweiler and Withrow method with a

new linear model for polytropic index). Results show that Model 2 is in good agreement

with Model 1 at high load conditions; however, it predicts slower combustion at light

load conditions. It is found that the net pressure method determines the EOC with

130

reasonable accuracy at WOT; however, its prediction of EOC at part load condition is

signiﬁcantly later than the actual value. The prOposed modiﬁcation of the net pressure

method using a compression polytropic index (n = nc) provides good agreement with the

Rassweiler and Withrow method at part— and full- load conditions for calculating the

mass fraction burned.

131.

CHAPTER-6

FINAL REMARKS

In the studies of gaseous highly transient and three-dimensional flows that
developed inside an IC engine, the MTV-based flow technique is superior to other non-
intrusive measurement techniques, i.e. PIV, due to the ability of laser-tagged
phosphorescent molecules that follow the flow perfectly. In this dissertation, MTV is
used to obtain the multiple point measurement of an instantaneous velocity ﬁeld. The
effects of charge motion control on ﬂow measurement inside an engine cylinder are
studied. It is found that charge motion control has a profound effect on fuel-air mixing;
however, its inﬂuence is not as signiﬁcant during the late compression. In-cylinder
engine flow measurements are extended to obtain an instantaneous three-component
velocity ﬁeld using stereoscopic MTV. The image-processing technique, implemented to
obtain the velocity data, is computationally less expensive and eliminates the need to
specify the geometric details, as in the earlier works, to obtain the three-components of
velocity. Preliminary results show that cycle-to-cycle variations are more prominent in
the velocity component perpendicular to the tumble plane, as opposed to the in-plane
components. This type of experimental information will be valuable in the development
of numerical simulations and advanced combustion systems under development. This
work can be utilized as a tool for engine developers where only the logistics of operation

need to be described.

132

In order to meet strict exhaust emissions and fuel efﬁciency standards, improved
numerical simulations, supported by this type of experimental studies with
experimentally proven boundary conditions, are desirable. Note that experimental
measurements provide useful information of the ﬂow ﬁelds inside the engine cylinder;
however, it is multi-dimensional numerical simulations that offer the potential of
signiﬁcant time and cost savings to design the engine with improved performance. Real-
time boundary conditions are necessary to perform more accurate multi-dimensional
numerical simulations of complex in-cylinder ﬂows. Therefore, velocity and pressure
measurements are performed inside the intake manifold of an engine assembly that can
provide the necessary boundary conditions for more accurate multi-dimensional
numerical simulations. The geometry of the intake manifold is simpliﬁed for this
purpose. A hot-wire anemometer and piezoresistive type absolute pressure transducers
are used to measure the velocity and pressure, respectively. In-cylinder ﬂow
measurements are also performed to validate the modeling efforts. Stereoscopic
molecular tagging velocimetry is used for this purpose.

In the second part of this work, combustion characteristics of an ethanol-gasoline,
dual fueled, single cylinder spark ignition engine are investigated. The cycle-to-cycle
variability is presented using the coefﬁcient of variation in indicated mean effective
pressure. Mass fraction burned and burn duration are determined from the analysis of
measured in-cylinder pressure data. The well known Rassweiler and Withrow method
(Model 1), with a new linear model for the polytropic index, is used to obtain the MF B
curves. The differences are presented for the net pressure method (Model 2) to evaluate

the burn rates. Model 2 showed good agreement with Model 1 at high load conditions;

133

however, it predicts slower combustion at light load conditions. Model 2 has an
advantage that the data processing time is short enough to allow for online processing.
Therefore, further investigation is performed to improve Model 2 calculations to evaluate
the mass fraction burned at light load conditions.

In the future, experimental results of in-cylinder ﬂow measurements will be
compared with multi-dimensional numerical simulations. Also, in the future,
measurements of a highly transient three-component velocity ﬁeld will be taken at
different crank angle positions inside the cylinder of Mahle 4-valve engine. The cycle-to-
cycle ﬂow variations will then be studied to develop a correlation for mass fraction
burned during combustion process. The results will be compared with the calculated

MFB curves using Rassweiler and Withrow method at the same load and engine speeds.

134

REFERENCES

135

REFERENCES

1. Heywood, J.B. (1987) “Fluid motion within the cylinder of internal combustion
engines - The 1986 Freeman Scholar Lecture,” Journal of Fluids Engineering, 109,
pp. 3—35.

2. Heywood, J.B. (1988) Internal combustion engine fundamentals, McGraw —— Hill
Book Company, ISBN 0-07-028637-X.

L»)

. Shen, Y. (2003) Development of a basis for control of combustion in an internal
combustion engine, PhD Thesis, Michigan State University, East Lansing, MI.

4. Farrell, P.V. (2007) “Examples of in-cylinder velocity measurements for internal
combustion engines,” Proceedings of the Institution of Mechanical Engineers. Part
D, Journal of automobile engineering, 221, pp. 675-697.

5. Hassan, H., and Dent, J.C. (1971) “The measurement of air velocity in a motored
internal combustion engine using a hot-wire anemometer,” Proc. Instn. Mech. Engrs.
(IMechE), 185, pp. 583-591

6. Bahram, K., Tahry, SH, and Kuziak, W.R. (1986) “Measured steady ﬂow velocity
distributions around a valve/seat annulus,” SAE Paper No. 860462.

7. Witze, PD. (1980) “A critical comparison of hot-wire anemometry and laser doppler
velocimetry for IC engine applications,” SAE Paper No. 800132.

8. Morse, AP, and Whitelaw, J.H. (1981) “Measurements of the in—cylinder ﬂow of a
motored four-stroke reciprocating engine,” Proceedings of the Royal Society of
London. Series A, Mathematical and Physical Sciences, 377, pp. 309-329.

)0

Bicen, A.F., Vaﬁdis, C., and Whitelaw, J.H. (1985) “Steady and unsteady airﬂow
through the intake valve of a reciprocating engine,” Journal of Fluids Engineering,
107, pp. 413-420.

10. Hall, M.J., and Bracco, F.V. (1987) “A study of velocities and turbulence intensities
measured in ﬁring and motored engines,” SAE Paper No. 870453.

11. Kent, J.C., Haghgooie, M., Mikulec, A., Davis, GO, and Tabaczynski, RJ. (1987)
“Effects of intake port design and valve liﬁ on in-cylinder ﬂow and bumrate,” SAE
Paper No. 872153.

12. Fraser, R.A., and Bracco, F.V. (1989) “Cycle-resolved LDV integral length scale

measurements investigating clearance height scaling, isotropy, and homogeneity in an
IC engine,” SAE Paper No. 890615.

136

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

Lee, K., Yoo, S.C., Stuecken, T., McCarrick, D., Schock, H., Hamady, F ., LaPointe,
L.A., Keller, P., and Hartman, P. (1993) “An experimental study of in—cylinder air
ﬂow in a 3.51 four-valve SI engine by high speed ﬂow visualization and two-
component LDV measurement,” SAE Paper No. 930478..

Yoo, S.C., Lee, K., Novak, M., Schock, H., and Keller, P. (1995) “3-D LDV
measurement of in-cylinder air flow in a 3.51 four-valve SI engine,” SAE Paper No.
950648.

Hascher, H.G., Jaffri, K., Novak, M., Lee, K., Schock, H., Bonne, M., and Keller, P.
(1997) “An evaluation of turbulent kinetic energy for the in-cylinder ﬂow of a four-
valve 3.51 SI engine using 3-D LDV measurements,” SAE Paper No. 970793.

Miles, P., Megerle, M., Nagel, Z., Reitz, R.D., Lai, M.C.D., and Sick, V. (2003) “An
experimental assessment of turbulence production, Reynolds stress and length scale
(dissipation) modeling in a swirl-supported DI diesel engine,” SAE transactions, 112
(3): pp. 1470-1499.

Reuss, D.L., Adrian, R.J., Landreth, C.C., French, D.T., and Fansler, TD. (1989)
“Instantaneous planar measurements of velocity and large-scale vorticity and strain
rate in an engine using particle image velocimetry,” SAE Paper No. 890616.

Reuss, D.L. (2000) “Cyclic variability of large-scale turbulent structures in directed
and undirected 1C engine ﬂows,” SAE Paper No. 2000-01-0246.

Li, Y., Zhao, H., Peng, Z., and Ladommatos, N. (2001) “Analysis of tumble and swirl
motions in a four-valve SI engine,” SAE Paper No. 2001-01-3555.

Funk, C., Sick, V., Reuss, D.L., and Dahm, W.J.A. (2002) “Turbulence properties of
high- and low-swirl in-cylinder ﬂows,” SAE Paper No. 2002-01-2841.

Bevan, K.E., and Ghandhi, J.B. (2004) “PIV measurements of in-cylinder ﬂow in a

four-stroke utility engine and correlation with steady ﬂow results,” SAE Paper No.
2004-32—0005.

Ghandhi, J.B., Herold, R.E., Shakal, 1.3., and Strand, TE. (2005) “Time-resolved
particle image velocimetry measurements in an internal combustion engine,” SAE
Paper No. 2005-01-3868.

Towers, DP, and Towers, CE. (2004) “Cyclic variability measurements of in-

cylinder engine ﬂows using high-speed particle image velocimetry,” Meas. Sci.
Technol., 15, pp. 1917-1925.

Jarvis, S., Justham, T., Clarke, A., Garner, C.P., Hargrave, GK, and Richardson, D.
(2006) “Motored SI IC engine in-cylinder ﬂow ﬁeld measurement using time

137

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

resolved digital PIV for characterization of cyclic variation,” SAE Paper No. 2006-
01-1044.

Jarvis, S., Justham, T., Clarke, A., Garner, C.P., Hargrave, GK, and Halliwell, N.A.
(2006) “Time resolved digital PIV measurements of ﬂow ﬁeld cyclic variation in an
optical IC engine,” J. Phys. Conf Ser., 45, pp. 38-45.

Stansﬁeld, P., Wigley, G., Justham, T., Catto, J., and Pitcher, G. (2007) “PIV analysis
of in-cylinder ﬂow structures over a range of realistic engine speeds,” Exp. Fluids,
43, pp. 135—146.

Calendini, P.O., Duverger, T., Lecerf, A., and Trinite, M. (2000) “In-cylinder velocity

measurements with stereoscopic particle image velocimetry in a SI engine,” SAE
Paper No. 2000-01-1798.

Koochesfahani, M. M., Cohn, R. K., Gendrich, C. P., and Nocera, D. G. (1996)
“Molecular tagging diagnostics for the study of kinematics and mixing in liquid phase
ﬂows,” Proceedings of the Eighth International Symposium on Applications of Laser
Techniques to Fluids Mechanics, July 8 — 11, 1996, Lisbon, Portugal, vol. I, 1.2.]-
1.2.12; Also in Developments in Laser Techniques and Fluid Mechanics, Chapter 2,

section 1, p. 125, Eds. Adrian, Durao, Durst, Maeda, and Whitelaw; Springer-Verlag,
Berlin, 1997.

Gendrich, C.P., and Koochesfahani, M.M. (1996) “A spatial correlation technique for

estimating velocity ﬁelds using molecular tagging velocimetry (MTV),” Exp. Fluids,
22 (1), pp 67-77.

Hill, RB, and Klewicki, J.C. (1996) “Data reduction methods for ﬂow tagging
velocity measurements,” Exp. Fluids, 20, pp 142-152.

Sadr, R., and Klewicki, J.C. (2003) “A spline-based technique for estimating flow
velocities using two-camera multi-line MTV,” Exp. Fluids, 35, pp 257-261.

Koochesfahani, M. M. (1999) “Molecular tagging velocimetry (MTV): Progress and
applications” AIAA Paper No. AIAA-1999-3786.

Bohl, D.G., Koochesfahani, M.M., and Olson BJ. (2001) “Development of
stereoscopic molecular tagging velocimetry," Exp. Fluids, 30, pp 302-308.

Koochesfahani, M.M., and Nocera, D.G. (2001) “Molecular tagging velocimetry
maps ﬂuid ﬂows,” Laser Focus World, 37 (6), pp. 103-108.

Gendrich, C. P., Koochesfahani, M. M., and Nocera, D. G. (1997) “Molecular tagging

velocimetry and other novel applications of a new phosphorescent supramolecule,
Exp. Fluids, 23 (5), pp. 361-372.

138

36. Lempert, W.R., Magee, K., Ronney, P., Gee, K.R., and Haughland, RP. (1995)
“Flow tagging velocimetry in incompressible ﬂow using photo-activated nonintrusive
tracking of molecular motion (PHANTOMM),” Exp. Fluids, 18, 249-257.

37. Lum, C., Koochesfahani, M. M., and McGrath, J. J. (2001) “Measurements of the
velocity ﬁeld with MTV during the solidiﬁcation of an alloy analog with mushy
region,” ASME/IMECE2001 Paper No. HTD-24222.

38. Koochesfahani, M.M., Cohn, R., and MacKinnon, C. (2000) “Simultaneous whole-
ﬁeld measurements of velocity and concentration ﬁelds using combined MTV and
LIF,” Meas. Sci. T echnol., ll (9), pp. 1289- 1300.

39. Stier, B., and Koochesfahani, M.M. (1998) “Whole ﬁeld MTV measurements in a
steady ﬂow rig model of an IC engine,” SAE Paper No. 980481.

40. Goh, A. C. H. (2001) Active ﬂow control for maximizing performance of spark-
ignited stratiﬁed charge engines, MS thesis, Michigan State University, East Lansing,
MI.

41. Koochesfahani, M. M., Goh, A. C., and Schock, H. J. (2004) “Molecular tagging
velocimetry (MTV) and its automotive applications,” In The Aerodynamics of Heavy
Vehicles: Trucks, Busses, and Trains, Lecture Notes in Applied and Computational
Mechanics, 19, pp. 143-155, Eds. McCallen, R., Browand, F., and Ross, J. (Springer-
Verlag).

42. Schock, H.J., Shen, Y., Timm, E., Stuecken, T., Fedewa, A.M., and Keller, P. (2003)
“The measurement and control of cyclic variations of ﬂow in a piston cylinder
assembly,” SAE Paper No. 2003-01-1357.

43. lsmailov, M.M., Schock, H.J., and Fedewa, A.M. (2006) “Gaseous ﬂow
measurements in an internal combustion engine assembly using molecular tagging
velocimetry,” Exp. Fluids, 41, pp 57-65.

44. Luo, M., Chen, G., and Ma, Y. (2003) “Three-dimensional transient numerical

simulation for intake process in the engine intake port-valve-cylinder system,”
Journal of Zhejiang University - Science, 4 (3), pp. 309-316.

45. Shojaeefard, M.H., and Noorpoor, AR. (2008) “Flow simulation in engine cylinder
with spring mesh,” American Journal of Applied Sciences, 5 (10): pp. 1336-1343.

46. Shen, Y., Schock, H.J., and Oppenheim, A.K. (2002) “Dynamic stage of combustion
in a direct injection methanol fueled engine,” SAE Paper No. 2002-01-0998.

47. Shen, Y., Schock, H.J., and Oppenheim, A.K. (2003) “Pressure diagnostics of closed
system in a direct injection spark ignition engine,” SAE Paper No. 2003-01-0723.

139

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

Oppenheim, A.K. (2004) Combustion in piston engines, Springer, ISBN 3-540-
20104-1.

Burgdorf, K., and Denbratt, I. (1997) “Comparison of cylinder pressure based knock
detection methods,” SAE Paper No. 972932.

Mittal, M., Zhu, G., Stuecken, T., and Schock H.J. (2009) “Effects of pre-injection on
combustion characteristics of a single-cylinder diesel engine,” ASME International
Mechanical Engineering Congress and Exposition, Paper No. IMECE2009-10493.

Rassweiler, G.M., and Withrow, L. (1938) “Motion pictures of engine ﬂames
correlated with pressure cards,” SAE Transactions, 42, pp 185-204.

Stone, CR, and Green-Armytage, DJ. (1987) “Comparison of methods for the
calculation of mass fraction burnt from engine pressure-time diagrams,” Proc. of
IMechE, Vol. 201, No. D1, pp 61-67.

Brunt, M.F.J., and Emtage, AL. (1997) “Evaluation of burn rate routines and analysis
errors,” SAE Paper No. 970037.

Shayler, P.J., Wiseman, M.W., and Ma, T. (1990) “Improving the determination of
mass fraction burnt,” SAE Paper 900351.

Clarke, J.R., and Stein, RA. (1999) “Internal combustion engine with variable
camshaft timing, charge motion control valve, and variable air/fuel ratio,” US patent,

US: 5957096.

Stein, R.A., Galietti, K.M., and Leone, TO. (1995) “Dual equal VCT - A variable
camshaﬁ timing strategy for improved fuel economy and emissions,” SAE Paper
950975.

Li, Y., Liu, S., Shi, S., and Xu, Z. (2000) “Effect of the swirl control valve on the in-
cylinder air motion in a four—valve SI engine,” SAE Paper 2000—01-2058.

Kim, H., Yoon, S., Xie, X.B., Lai, M.C., Quelhas, S., Boyd, R., Kumar, N., and
Moran, C. (2005) “Effects of injection timings and intake port ﬂow control on the in-
cylinder wetted fuel footprints during PFI engine startup process,” SAE Paper 2005-
01-2082.

Lee, D., and Heywood, J.B. (2006) “Effects of charge motion control during cold
start of SI engines,” SAE Paper 2006-01-3399.

Mittal, M., and Schock, H.J. (2009) “A study of cycle-to-cycle variations and the
inﬂuence of charge motion control on in-cylinder ﬂow in an IC engine,” submitted to
Journal of Fluids Engineering.

140

61. Koochesfahani, M.M., and Nocera, D.G. (2007) “Molecular tagging velocimetry,”
Handbook of Experimental Fluid Dynamics, chapter 5.4, Editors: J. Foss, C. Tropea
and A. Yarin, Springer—Verlag.

62. http://msds.chem.ox.ac.uk/BU/2,3-butanedione.html

63. Stier, B., and Koochesfahani, M.M. (1999) “Molecular tagging velocimetry (MTV)
measurements in gas phase ﬂows,” Exp. Fluids, 26, pp 297-304.

64. Sidebottom, H.W., Badcock, C.C., Calvert, J.G., Rabe, BR, and Damon, E.K. (1972)
“Lifetime studies of the biacetyl excited singlet and triplet states in the gas phase at
25°,”J. Am. Chem. Soc., 94, pp 13-19.

65. Zheng, Q., and Klewicki, J .C. (2000) “A fast data reduction algorithm for molecular
tagging velocimetry: the decoupled spatial correlation technique,” Meas. Sci.
Technol. 11, pp 1282-1288.

66. Cohn, R. K., and Koochesfahani, M. M. (2000) “The accuracy of remapping
irregularly spaced velocity data onto a regular grid and the computation of vorticity,”
Exp. Fluids, 29(7), 8061- SO69.

67. Miles, P., Megerle, M., Nagel, Z., Reitz, R.D., Lai, M.C.D., and Sick, V. (2003) “An
experimental assessment of turbulence production, Reynolds stress and length scale

(dissipation) modeling in a swirl-supported DI diesel engine,” SAE transactions, 112
(3): 1470-1499.

68. Mittal, M., Schock, H.J., and Sadr, R. (2008) “Evaluating the inﬂuence of charge
motion control on in-cylinder ﬂow using MTV,” ASME International Mechanical
Engineering Congress and Exposition, Paper No. IMECE2008-66490.

69. J00, S.H., Srinivasan, K.K., Lee, KC, and Bell, SR. (2004) “The behavior of small-
and large-scale variations of in-cylinder ﬂow during intake and compression strokes
in a motored four-valve spark ignition engine,” Int. Journal of Engine Research, 5
(4): pp. 317-328.

70. Raffel, M., Willert, C., and Kompenhans, J. (1998) Particle image velocimetry: A
practical guide, Springer, ISBN — 3540636838.

71. Arroyo, MP, and Greated, CA. (1991) “Stereoscopic particle image velocimetry,”
Meas. Sci. Technol., 2, pp. 1181-1186.

72. Bohl, D.G. (2002) Experimental study of the 2-D and 3-D structure of a concentrated
line vortex array, PhD Thesis, Michigan State University, East Lansing, MI.

73. Spath, H. (1996) “Least-squares ﬁtting by circles,” Computing, 57: 179—185.

141

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

Soloff, S.M., Adrian, R.J., and Liu, Z.C. (1997) “Distortion compensation for

generalized stereoscopic particle image velocimetry,” Meas. Sci. T echnol. 8, pp 1441-
1454.

Naqwi, A. (2000) “Distortion compensation for PIV systems,” Proc. 10th International

Symposium on Applications of Laser Techniques to Fluid Mechanics, Lisbon, paper
6.2.

Mittal, M., Sadr, R., Schock, H.J., Fedewa, A., and Naqwi, A. (2009) “In-cylinder

engine ﬂow measurement using stereoscopic molecular tagging velocimetry
(SMTV),” Exp. Fluids, 46 (2), 277-284.

Sadr, R., Mittal, M., and Schock, H.J. (2008) “In-cylinder IC engine velocity
measurements using stereoscopic molecular tagging velocimetry,” 61St Annual

Meeting of American Physical Society, Division of Fluid Dynamics, San Antonio,
Texas.

Benedict, L.H., and Gould, RD. (1996) “Towards better uncertainty estimates for
turbulence statistics,” Exp. Fluids, 22, pp. 129—136.

Zhang, L., Takahiro, U., Toshiaki, T., and Katsuhiko, Y. (1995) “A study of the
cycle-to-cycle variation of in—cylinder ﬂow in a motored engine through digital image
processing of visualized images,” SAE Paper 950727.

Parsi, M., and Daneshyar, H. (1989) “A study of cyclic variations in gas velocities in
an engine cylinder,” SAE Paper 892100.

Mittal, M., Zhu, G., Schock, H.J., Stuecken, T., and Hung, D.L.S. (2008) “Burn rate
analysis of an ethanol-gasoline, dual fueled, spark ignition engine,” ASME

International Mechanical Engineering Congress and Exposition, Paper No.
IMECE2008-66 l 39.

Nakata, K., Utsumi, S., Ota, A., Kawatake, K., Kawai, T., and Tsunooka, T. (2006)
“The effect of ethanol fuel on a spark ignition engine,” SAE Paper 2006-01-3380.

Brinkman, ND. (1981) “Ethanol Fuel — A single-cylinder study of efﬁciency and
exhaust emissions,” SAE Paper 810345.

Srivastava, S., Schock, H.J., Jaberi, F., and Hung, D.L.S. (2009) “Numerical

simulation of a direct—injection spark-ignition engine with different fuels,” SAE paper
2009-01-0325.

Starkman, E.S., Newhall, H.K., and Sutton, RD. (1964) “Comparative performance
of alcohol and hydrocarbon fuels,” SAE Paper 640649.

142

86.

87.

88.

89.

90.

Salih, F.M., and Andrews, GE. (1992) “The inﬂuence of gasoline/ethanol blends on
emissions and fuel economy,” SAE Paper 922378.

Bromberg, L., Cohn, D.R., and Heywood, J.B. (2006) “Optimized fuel management
system for direct injection ethanol enhancement of gasoline engines,” US patent, US:
26102136.

Bitar, E.Y., Oppenheim, A.K., and Schock, H.J. (2006) “Model for control of
combustion in a piston engine,” SAE Paper 2006-01-0401.

Mittal, M., Zhu, G., and Schock, H.J. (2008) “Fast mass fraction burned calculation
using net pressure method for real-time applications,” Proc. IMechE, Part D: J.
Automobile Engineering, 2009, 223 (3), 389-394.

Zhu, G., Daniels, C.F., and Winkelman, J. (2003) “MBT timing detection and its
closed-loop control using in-cylinder pressure signal,” SAE Paper 2003-01-3266.

143