33v
fit
:.

“m 4!
and???

‘4. :4

.l

,l
1',

?

iv. :3»:

it‘s”. “2:!
.. l. a ‘2 ‘5).
31..

- n I It I .
.06. 3...:nlfuPilrhr.
. {.41.}..“7 3?

$8155}! ..r
A. .5

It‘la..:~\H¥

.92 .

‘l. :Yl’...vll .-
in

 

 

 

. v.31; , .
[gnu-LL .III.

[AA-.559} .«
5|! $0.1

:1 aw? wimp £3... . ‘ “divufifi.

 

Al'l

 

THES'S

Z lllllllll’lllllll’llllllllllllllllllll'Illllllllllll

3 1293 01413 8832

LIBRARY

Michigan State
University

 

 

 

 

This is to certify that the

thesis entitled

A VOLUMETRIC QUANTIFICATION OF
IN-CYLINDER FLOW MOTION INSIDE
A FOUR-VALVE SI ENGINE USING

THREE DIMENSIONAL LASER DOPPLER

VELOCIMETRY
presented by

Kasser A. Jaffri

has been accepted towards fulfillment
of the requirements for

M. S . degree inMechanicaLEngineering

 

Mew

Major p fessor

Harold J. Schock

Date 7/3L196

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

PLACE W ETURN BOX to romovo this chookout from your rooord.
TO AVOI INES return on or before date duo.

DATEZDUE DATE DUE DATE DUE

     

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSUJo An Nflrmotivo AotioNEquol Opportunity lm
Wm:

A VOLUMETRIC QUANTIFICATION OF IN-CYLINDER FLOW
MOTION INSIDE A FOUR-VALVE SI ENGINE USING
THREE DIMENSIONAL LASER DOPPLER VELOCIMETRY

By

Kasser A. Jaffri

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Mechanical Engineering

1996

ABSTRACT
A VOLUMETRIC QUANTIFICATION OF IN-CYLINDER FLOW
MOTION INSIDE A FOUR-VALVE SI ENGINE USING
THREE DIMENSIONAL LASER DOPPLER VELOCIMETRY
By

Kasser A. J affri

The flow field contained within ten planes inside a cylinder of a 1998 Chrysler 3.5
liter, 24 valve, V-6 engine was mapped using a three component Laser Doppler
Velocimetry (LDV) system. A total of 1,548 LDV measurement locations were used to
construct the time history of the intake and compression driven flow fields.

Ensemble averaged velocity measurements allowed for a qualitative analysis of
the dynamic in-cylinder flow characteristics. IRIS EXPLORER, a data visualization
software running on a Silicon Graphics (SGI) Indigo2 workstation platform, was used to
view and animate the flow fields. Quantification of the volumetric flow features was
accomplished through calculations of tumble and swirl ratios, and circulation.
Turbulence levels were obtained by calculating the turbulent kinetic energy (TKE) of the
fluctuations, delivered by the LDV measurements, which produced low cycle-to-cycle
variations. FORTRAN code was generated to facilitate the calculations of TKE, tumble
and swirl, and circulation.

Symmetric tendencies were found to exist within the flow field throughout the
mapped intake and compression stroke. Cancellation effects produced by the four-valve
design produced small tumble and swirl ratios. The calculations of the volumetric TKE

produced results which could be used to accurately calibrate current turbulence models.

This Thesis is dedicated to my parents,

Kausar Ali Jaffi'i
&
Delfma F iorentino J affri

for their love and kindness
and their support during my years in academia.

iii

ACKNOWLEDGMENTS

I would like to thank the many people who have been involved with this work
which made it enjoyable even though it was quite demanding at times. Dr. Schock for
providing me the opportunity to do my graduate work in the ERL, as well as providing
the opportunity to give presentations. Dr. Keunchel Lee for passing along his knowledge
of physics and fluid mechanics, and his expertise in the LDV system; he is a kind man
and a true experimentalist. Mark Novak for his friendship and his constructive criticism
which almost always proved beneficial during the many good debates we had. Hans
Hascher for spending the long hours through the many nights acquiring the measurements
for this work. Dr. Brereton for being on my committee and always being willing and
helpful in answering any question I posed. Tom Stuecken, for his help in building the test
rig and exposing me to Beaver. Also I would like to thank some of the fellows who
worked with me at the ERL; JongUk Kim, Jon Darrow, Mikhail Ejakov, and Eric
Bertrand, they are all good people for which I have had the pleasure to become
acquainted with. Finally, I thank Heather Mckee for her love and support throughout the

time I spent at Michigan State.

hr:

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

NOMENCLATURE

INTRODUCTION

CHAPTER 2 - EXPERMENTAL FACILITIES
Test Rig Assembly

Laser Doppler Velocimetry System
CHAPTER 3 - EXPERIMENTAL WORK

Effective Cylinder Volume Mapping

Data Reduction and Validation

Computer Visualization Technique
CHAPTER 4 - FLOW DESCRIPTION

Planes l, 3, and 4

Flows Through the Center Seven Z-planes

Flows Along the Cylinder Liner and Through the Top
and Bottom Y-planes

F low Pattern Summary

vii

viii

xiii

10
15

15

18

26

30

31

38

49

CHAPTER 5 - TURBULENT KINETIC ENERGY
Definition of Turbulent Kinetic Energy
Turbulent Kinetic Energy Results
Volumetric Turbulent Kinetic Energy Calculations
Turbulent Kinetic Energy within the Vertical Planes
Turbulent Kinetic Energy in Horizontal Slices
CHAPTER 6 - TUMBLE AND SWIRL
Definition of Tumble and Swirl Ratios
Tumble Ratio Results for Vertical Planes
Swirl Ratio Results for Horizontal Slices

Tumble and Swirl Ratio Results for the Effective
Measurement Volume

CHAPTER 7 - CIRCULATION
Circulation Calculation Techniques for Slices and Planes
Circulation Results and Discussion

SUMMARY AND CONCLUSIONS

RECOMMENDATIONS

APPENDIX I

APPENDIX 11

LIST OF REFERENCES

vi

52

52

58

58

67

70

74

75

78

86

89

96

96

101

119

122

124

125

140

LIST OF TABLES

Table 1: Specifications of the AVL Single cylinder research
engine assembly

Table 2: 1998 Chrysler 3.5 L four-valve engine specifications

vii

Figure 1:
Figure 2:
Figure 3:
Figure 4:

Figure 5:

Figure 6:

Figure 7:

Figure 8:

Figure 9:

Figure 10:

Figure 11:

Figure 13:

Figure 14:

LIST OF FIGURES

V6 Head Mounted onto the AVL Research Engine
Seeding Conditions Imposed on the V6 Head
Actual LDV System

Schematic of the LDV System

Qllinder Coordinate System Displaying Cylinder
Volume with the Ten Measurement Planes

Single Component Fringe Pattern Constructed by a
Pair of 514.5 nm Beams

Single Component Fringe Pattern Constructed by a
Pair of 476,5 nm Beams

Cross Sectional Photograph of a Two Component
Fringe Pattern

Sample Iris Explorer Data Visualization Map

Flow Patterns in Planes 1, 3, and 4 at Crank Angle
67°. Vector Scale = 1.5

Flow Patterns in Planes 1, 3, and 4 at Crank Angle
89°. Vector Scale = 1.5

Flow Patterns in Planes 1, 3, and 4 at Crank Angle
135°. Vector Scale = 1.5

Flow Patterns in Planes 1, 3, and 4 at Crank Angle
148°. Vector Scale = 1.5

viii

10

14

17

18

19

19

29

31

32

33

33

Figure 15:

Figure 16:

Figure 17:

Figure 18:

Figure 19:

Figure 20:

Figure 21:

Figure 22:

Figure 23:

Figure 24:

Figure 25:

Figure 26:

Figure 27:

Figure 28:

Flow Patterns in Planes l, 3, and 4 at Crank Angle
169°. Vector Scale = 1.5

Flow Patterns in Planes 1, 3, and 4 at Crank Angle
181°. Vector Scale = 1.5

Flow Patterns in Planes 1, 3, and 4 at Crank Angle
207°. Vector Scale = 1.5

Flow Patterns in Planes 1, 3, and 4 at Crank Angle
241°. Vector Scale = 1.5

Flow Patterns in Planes 1, 3, and 4 at Crank Angle
271°. Vector Scale = 1.5

Flow Patterns through the Seven Center Z-Planes
at Crank Angle 67°. Vector Scale = 1.0

Flow Patterns through the Seven Center Z-Planes
at Crank Angle 89°. Vector Scale = 1.0

Flow Patterns through the Seven Center Z-Planes
at Crank Angle 140°. Vector Scale = 1.5

Flow Patterns through the Seven Center Z-Planes
at Crank Angle 181°. Vector Scale = 1.5

Flow Patterns through the Seven Center Z-Planes
at Crank Angle 210°. Vector Scale = 1.5

F low Patterns through the Seven Center Z-Planes
at Crank Angle 233°. Vector Scale = 1.5

Flow Patterns through the Seven Center Z-Planes
at Crank Angle 264°. Vector Scale = 1.5

Flow Patterns through the Top and Bottom Slice
Along the Cylinder Liner at Crank Angle 72°.
Vector Scale = 1.0

Flow Patterns through the Top and Bottom Slice
Along the Cylinder Liner at Crank Angle 94°.
Vector Scale = 1.0

34

35

36

37

37

38

39

40

41

42

43

43

45

Figure 29:

Figure 30:

Figure 31:

Figure 32:

Figure 33:

Figure 34:

Figure 35:

Figure 36:

Figure 37:

Figure 38:

Figure 39:

Figure 40:

Figure 41:

Figure 42:

Flow Patterns through the Top and Bottom Slice
Along the Cylinder Liner at Crank Angle 118°.
Vector Scale = 1.0

Flow Patterns through the Top and Bottom Slice
Along the Cylinder Liner at Crank Angle 148°.
Vector Scale = 1.0

Flow Patterns through the Top and Bottom Slice
Along the Cylinder Liner at Crank Angle 181°.
Vector Scale = 1.5

Flow Patterns through the Top and Bottom Slice
Along the Cylinder Liner at Crank Angle 210°.
Vector Scale = 1.5

Flow Patterns through the Top and Bottom Slice
Along the Cylinder Liner at Crank Angle 264°.
Vector Scale = 1.5

Grid to Calculate Turbulent Kinetic Energy in
Horizontal Slices

Turbulent Kinetic Energy per Unit Mass for the
Effective Cylinder Volume

Total Turbulent Kinetic Energy for the Effective
Cylinder Volume

Turbulent Kinetic Energy inside the Cylinder Volume
at CA of 61°

Turbulent Kinetic Energy at the Crank Angles of 81°,
100°, and 129°

Turbulent Kinetic Energy at the Crank Angles of 146°,
161°, and 179°

Turbulent Kinetic Energy at the Crank Angles of 199°,
240°, and 269°

Turbulent Kinetic Energy for all the Vertical Planes

Total Turbulent Kinetic for all the Vertical Planes

46

47

47

48

49

56

58

60

62

63

65

67

69

Figure 43:
Figure 44:
Figure 45:
Figure 46:
Figure 47:
Figure 48:
Figure 49:
Figure 50:
Figure 51:
Figure 52:

Figure 54:

Figure 55:
Figure 56:
Figure 57:

Figure 58:

Figure 59:

Figure 60:

Figure 61:

Figure 62:

Turbulent Kinetic Energy in all the Horizontal Slices
Total Turbulent Kinetic Energy in the Horizontal Slices
LDV Cylinder Coordinate Systems

Tumble Ratio around the X-axis for Fixed Origin
Tumble Ratio around the X-axis for Moving Origin
Tumble Ratio around the Z-axis for Fixed Origin
Tumble Ratio around the Z-axis for Moving Origin
Swirl Ratios per Horizontal Slice

Volume Tumble Ratio for Fixed Origin

Tumble Ratio for Moving Origin

Example of a Calculation Grid used for the
Circulation Analysis

Circulation in Planes 9 adnlO
Total Circulation in Planes 5, 6, 7, and 8
Total Circulation in Planes 1, 3, and 4

Flow Fields in Plane Pairs. Crank Angle 142°
Vector Scale = 2.0

Average Circulation for Slices at Y=65, 60, 55,
and 50 (mm)

Average Circulation in Slices at Y=45, 40, 35,
and 30 (mm)

Average Circulation in Slices at Y=25, 20, 15,
and 10 (mm)

Velocity Field in the Slice that Follows the Piston
Surface During Intake. Vector Scale = 1.5

71

72

75

79

81

84

85

88

91

92

98

101

102

103

105

107

108

112

Figure 63:

Figure 64:
Figure 65:
Figure 66:
Figure 67:

Figure 68:

Velocity Field in the Slice that Follows the Piston
Surface During Compression. Vector Scale = 2.0

RMS Circulation for all of the Planes 114
RMS Circulation for all of the Slices

RMS Circulation for the Upper Slices

RMS Circulation for the Lower Slices

Schematic of the Intake Port Used

113

115

116

116

124

Greek

F

F abs

ravg

F k

rplane 1,3-10

rims

rslice

F total (plane or slice)

V

0

0'

Q1

A9

Po

NOMENCLATURE

circulation

absolute value of circulation

average value of circulation

circulation in a calculation element k
circulation in planes 1 and 3 through 10
RMS value of circulation

circulation in a calculation element k
total circulation value in plane or slice
kinematic viscosity

general crank angle

specific crank angle

average crank angle

crank angle window

density of air mass at ambient conditions

density of air mass during compression

xiii

e:

mo

03x

032

Arabic

is

Hbi

Ibbi

standard deviation

variance in the x-component
variance in the y-component
variance in the z—component
vorticity vector

rotational vector

crank shaft speed

x-component of the rotational speed
y-component of the rotational speed

z-component of the rotational speed

Area
Area of circulation calculation element j
closed contour

bore diameter

angular momentum vector

angular momentum per unit mass about axis b for element i
index

unit vector

moment of inertia about axis b for calculation element i

xiv

Ncycles
n1

Nmeasurements

Ni 3)

'41

moment of inertia in the x-direction

moment of inertia in the y-direction

moment of inertia in the z-direction

index

unit vector

index

unit vector

air mass trapped in cylinder during compression

total number of data points, total number of calculation elements
number of cycles

number of measurements in one cycle

total number of measured velocities

unit normal

position vector

length vector

time

total measurement time

turbulent kinetic energy in Joules during intake
turbulent kinetic energy per unit mass during intake
turbulent kinetic energy in Joules during compression
turbulent kinetic energy per unit mass during compression

tumble ratio about axis b

XV

<1

chlinder
Vi

W

3|

xi

RI

instantaneous x—component of velocity

time averaged x-component of velocity
fluctuating part of the x-component of velocity
ensemble averaged velocity component
instantaneous y-component of velocity

time averaged y-component of velocity
fluctuating part of the y-component of velocity
velocity vector

volume of cylinder

calculation volume element I

instantaneous z-component of velocity

time averaged z-component of velocity
fluctuating part of the zocomponent of velocity
distance

single data point

mean average of a set of data points

distance

distance

xvi

INTRODUCTION

A homogeneous, stoichiometric air-fuel mixture in an Internal Combustion (I.C.)
engine can limit the formation of polluting emissions, reduce cycle-Mole combustion
variations, and improve fuel economy. This requires enhanced turbulent energy levels
throughout the mixing period which occurs during the intake and compression strokes.
For instance, it has been stated that one of the most effective means of improving the
lean- running ability of a combustion system is to increase the turbulent intensity within
the charge, so that the burning rate at a wide range of equivalence ratios is increased [1].
The current study involves an engine which has an induction design where fuel is injected
prior to Intake Valve Opening (1V0). Hence, the corresponding flow details over a wide
range of the intake and compression strokes are relevant and were thoroughly investigated
for this study.

Much debate has raged over what type of in-cylinder motion, tumble or swirl or a
combination of the two, is most effective for power, torque, emissions, and efficiency.
Consequently, much effort has been Spent to quantitatively correlate these motions
directly with combustion parameters such as burn rate. However, a concrete and
consistent correlation which links intake-generated flow motion to an optimum
combustion process has yet to be discovered. The Michigan State University Engine

Research Lab (MSUERL) group, over the past several years, has been conducting a series

2

of studies for the Chrysler Corporation with the goal of correlating in—cylinder flows to
engine performance. These studies are being conducted to provide a better physical
understanding of the in-cylinder flow phenomena and the effects of intake port and piston
top geometry on these flows. All the studies mentioned here were conducted with the
3.5L four-valve Spark Ignition (SI) engine, with variations in intake runner and port,
combustion chamber shape, and piston top. In 1993, Lee et al. [2] performed flow
visualizations on three different intake port configurations and two piston top geometries.
Two-component Laser Doppler Velocimetry (LDV) was used to compare a flat top piston
to a “pop-up” face piston. The visualizations showed that the different intake port
configurations played a significant role in the generation of the initial tumble motion in
the early stages of the intake stroke. This study also showed that a large pop-up on the
piston top can increase the tumbling motion intensity. In 1995, Yoo et a]. [3] conducted
the first simultaneous three-dimensional (3D) LDV measurements over a planar section
of a four-valve piston-cylinder assembly. The flow fields in one axial plane for two
different piston top geometries were examined: a pop-up piston head with valve pockets
cut into it versus a flat top piston surface. This study confirmed that a pop-up piston top
can increase the tumble motion and revealed that the TKE decay during the compression
process can be offset by the decay of flow structures. In 1995, Mueller [4] further
quantified the iii-cylinder flow using 3D LDV by mapping the flow field in two planes.
However, the Mueller study showed the existence of counter rotational motions, which
can sum to near zero values.

Unlike previous studies, the current work involved mapping ten planes within a

1998 Chrysler 3.5L four-valve V6. A total of 1,548 LDV measured points were taken to

3

map the 3D flow field that develops during the intake and compression processes. This
study is believed to be the first effective empirical study which maps 75% of the cylinder
volume.

A qualitative analysis of the flow patterns that develop during the intake and
compression strokes was done for this study. Animations of the flow patterns assembled
by the measured velocities were viewed on a Silicon Graphics Indigo2 (SGI). These
views gave the ability to extract dominating flow features which were shown to be
complex and fully three dimensional.

Quantification of the flow field was done by calculating three major parameters
which are believed to adequately characterize in—cylinder motion. Calculation of these
three major parameters were accomplished by using the measured velocities along with
their corresponding standard deviations. These quantities were TKE, tumble and swirl
numbers, and circulation. As stated above, TKE is a common parameter that engineers
and researchers have attempted to characterize in an engine since it is believed to play a
direct role in the combustion performance. Tumble and swirl numbers were calculated
since they are considered current standards for defining in-cylinder flow characteristics.
These values are defined in an attempt to quantify the bulk flow motion into two distinct
groups, i.e., swirl and tumble. Both numbers refer to solid body rotations in terms of net
angular momentum about specific axes. Circulation was also calculated since it too can
be used to identify bulk motion. Its quantity describes the net fluid flow around a closed
contour. In addition, small cell activity can also be evaluated using RMS circulation

which utilizes the net vorticity flux through cells constructing a closed contour.

4

Circulation, tumble, and swirl are all considered to be directly related to the large scale

mixing characteristics exhibited by in-cylinder flow motion.

Chapter 2

EXPERIMENTAL FACILITIES

Test Rig Assembly

The present study was conducted with a 1998 MY Chrysler 3.5 L four-valve, V6
engine. An AVL single-cylinder research engine type 530 was modified to adopt the
Chrysler engine. The AVL engine provided the reciprocating piston and a driving crank
for a cylinder of one bank of the V6 engine head. Table 1 gives the specifications of the

AVL single-cylinder research engine assembly.

Table 1: Specifications of the AVL single cylinder research engine assembly.

 

 

 

 

 

 

 

 

Model AVL (type 530)
Stroke 80 mm
Connecting Rod Length 230.3 mm
Maximum Speed 3500 rpm
Operating Motor Power 10 HP

 

The entire rig was mounted on a heavy steel bed plate along with a 10-hp variable
speed electric motor coupled to the AVL engine crankshaft. Rubber lattice mounts were
placed on each corner of the bed plate to reduce vibrations transferred to the floor from
the operating test rig assembly. A bank of the V6 engine head was mounted onto the
AVL engine. The cam shaft of the V6 head was belt driven by the cam of the AVL
engine. The AVL engine camshaft is adjustable so that the proper cam timing of the V6

engine head could be achieved.

6

A 10 mm thick quartz cylinder, with a 96 mm bore, was used to allow optical
access for LDV measurements. The cylinder was placed between the V6 engine head and
an adapter plate mounted on the AVL engine. Both ends of the cylinder were held into
place by Rulon-LD gaskets. The piston top geometry was a flat top with valve pockets
cut in it. The piston rings used were also made of Rulon-LD which required no
lubrication while withstanding high temperature and compressive loads. Shown in Figure

1 is the V6 engine head mounted onto the AVL engine.

 

Figure 1: V6 Head Mounted onto the AVL Research Engine

The intake manifold was cut so that only the runner that follows into the center
cylinder remained. This was necessary to make space available for the receiving optics

required for the LDV measurements.

7

To introduce seed particles for LDV measurements, a round hole was drilled into
the intake manifold so that a flexible rubber tube could be mounted. The tube sat flush to
the interior side of the intake manifold and perpendicular to the path of the intake runner.
Through this rubber tube, seeding into the intake runner was achieved by allowing a flow
of seed particles to be drawn into the cylinder via the intake induced flow. The seed
particles were generated by atomizing a mixture of propylene glycol and water. The
exhaust manifold consisted of a straight three inch pipe fasten to the head. The hose of an
exhaust fan was placed downstream from the exhaust runner to collect the mist of

particles in a fashion which would not disturb the flow, as shown in Figure 2.

 

Figure 2: Seeding Conditions Imposed on the V6 Head

8

The same modifications for previous in-cylinder flow studies for Chrysler [2, 4],
were done to the 1998 MY engine head for the actual LDV measurements. The
modifications included removal of rocker arms from front and rear cylinders and
connection of an external oil pump and vacuum unit.

Since the center cylinder of the head was used, the rocker arms were removed
from the front and rear cylinders. To maintain oil pressure, aluminum tubes were
machined with the same ID. as the rocker arms and were inserted on the rocker shafi in
their place. An external, closed-system oil and vacuum pump controlled the oil
circulation through the motored head. Since the head was mounted in an upright position
and the oil return lines were routed above the head face for greater optical access, oil
could not drain from the head properly. For this reason, the oil pump supplies the oil to
the engine head; and the vacuum pump draws the oil out of the head. It is also an
advantage to have a slight negative pressure under the valve cover. This reduces the
chances that oil would travel down the valve stems causing the piston to smear on the
interior of the quartz cylinder (smearing of oil on the cylinder distorts and reduces the
laser beam intensity tremendously). However, even using these precautions, we still
experienced the problem of oil leaking down the valve stems into the cylinder. This
problem was eventually corrected by installing a new set of valve seals and running
another set of vacuum lines (internally within the head) to positions near the seals of the
intake and compression valves.

lOW-30 engine oil at 2 x 105 Pa was supplied to the oil inlet port on the head face.
This port was tapped for a 3.175 mm pipe fitting where the supply line was attached.

Three 90 degree fittings were pressed into the oil drain passages, which were attached to

9

lines feeding the top half of the oil reservoir. The line from the bottom of the reservoir
feeds an oil pump that supplies filtered oil to the head. A line on top of the reservoir
leads to a flow control valve, then to a vacuum pump. The vacuum pump creates low
pressure in the reservoir (127 mm Hg), which draws oil and air from the oil drains. The
air and oil are separated in the reservoir, and the air is then drawn out from the top by the

vacuum pump.

Table 2: 1998 Chrysler 3.5 L four-valve V6 engine specifications

 

 

 

 

 

 

bore 96 mm inlet valve opening 3 ° BTDC
stroke 81 mm inlet valve closure 252 ° ATDC
compression ratio 10.0:1 motofing_e_n_gine speed 600 rpm
inlet valve diameter 36.5 mm Test Throat Condition WOT

 

 

 

 

 

Laser Doppler Velocimetry System

l

“:72?“ ,;
l—
'—:u' E41

* «an ,.

:f“

" ’ _. 'efin II": ...;

 

Figure 3: Actual LDV System

A simultaneous three-dimensional (3D) LDV measurement technique was
employed to quantify the in-eylinder flow field in the V6 engine. The 3D velocity
measurement technique was accomplished by placing three sets of orthogonal fringe
patterns at the same location in space (the measurement volume). The entire system
includes a multitude of components. The necessary laser beams are generated by an
Argon-Ion laser. The beams are then used to construct a measurement volume with the
aid of a beam collimator, steering mirrors, prisms, beam splitters, polarizers, Bragg cells,
and a focusing lens used in conjunction with a fiber optic probe. Movement of the

measurement volume within the cylinder was controlled by use of a transverse table.

11

Velocity measurements were obtained from the detection of the scattered light frequency
of particles passing through the measurement volume. This was accomplished with the
use of receiving optics with attached photomultipliers and frequency shifters. The
incoming velocity signals are processed with an Intelligent Flow Analyzer Model 750
(IFA 750), made by T81 and tagged with an incoming signal produced by a Rotary
Machine Resolver (RMR), connected to the motored engine. A Personal Computer (PC)
is used to control the measurement criteria.

To acquire the 3D LDV measurements, two distinct optical paths were required to
be setup. A schematic of the 3D LDV system is shown in Figure 4, with optical paths A
and B. Path A was set to construct a two-component fringe pattern, while path B was
used to construct a single—component fi'inge pattern. The multiwavelength line from the
Argon-Ion laser is initially separated into several Single wavelength lines by a pair of
rhomboid prisms. The green (514.5 nm) and blue (476.5 nm) lines were directed with
mirrors to run parallel to each other. They were passed through polarization rotators
which polarized the green line horizontally and the blue line vertically. The coherent and
polarized beams were each passed to a beam splitter. The beam splitter split each beam
into a pair of equal intensity beams that ran parallel with each other with a constant
separation of 50 mm. One beam of each pair was allowed to pass through a Bragg cell,
which shifted its frequency or changed its phase angle continuously with respect to its
counter beam running parallel to it. This has the effect of producing a fringe pattern that
moves or cycles a constant frequency thereby allowing a relative velocity measurement.
This makes it possible to measure the flow direction. Both pairs of beam$ were then

focused to a single point using a quartz focus lens with a focal length of 453 mm.

12

Optical path B, in Figure 4, displays a fiber optic system used to construct the
single component pattern. The fiber optic system utilizes the violet beam (476 nm) of the
Argon-ion laser. The violet line, separated by the rhomboid prism, is directed with
mirrors into the Colorburst. The Colorburst is an optical unit made by TSI, which
performs the functions of beam Splitter, polarizer, and Bragg cell. A pair of violet beams,
from the Colorburst, are fed into polarization-preserving fiber optic cables via beam
steering mechanisms (coupler) mounted on the Colorburst. This allows for fine optical
adjustment to transmit the violet beams into the fiber optic cables with minimal loss of
intensity. The beams are then transmitted to the measurement volume through a fiber
optic probe, which is mounted on the optic axis perpendicular to that in path A.

The LDV system was mounted on a traverse table (TSI model 9900-1) which is
capable of transitional movement in the three Cartesian directions within 0.001 mm.
Mounted on the sides of the traverse table were two rigid extension arms which enclose
the test-rig assembly . The arms are used to mount the fiber-optic probe and the receiving
optics. Thus, movement of the traverse table controls the movement of the measurement
volume within the cylinder. Four vibration isolators were installed under the traverse
table to prevent the vibrations from transferring from the engine test rig to the optical
equipment.

The receiving optics and the data acquisition elements are the other integral parts
of the system. Direct velocity measurements are obtained by detecting the frequency of
light scattered by particles passing through the measurement volume. This was
accomplished by three receiving optics each containing a narrow band filter of 514 nm,

488 nm, and 476.5 nm. Connected to each of these were photomultiplier tubes that

13

convert the light intensity fluctuations into an electrical signal. Each set of receiving
optics and photomultipilier tubes were mounted on the rigid arms extending from the
traverse table for forward scattering measurement. The signal from each photomultiplier
tube was then sent to the frequency shifter (one for each component). The frequency
Shifters allow the adjustment for a relative reference velocity depending on the range of
velocities that are desired.

The raw signals are passed to an IFA 750 correlator (TSI) , which separates good
signals from the noise. In addition, the IFA 750 also receives information of the crank
shaft position by use of an angle encoder mounted on the test rig assembly. The encoder
produces 1024 pulses per revolution which are resolved to 8192 pulses per engine cycle
via a RMR. This arrangement allows each component of velocity to be stamped with the
instantaneous rotational position of the crank shaft in the IFA 750. The IFA 750 is then
linked to a PC that is configured with TSI data acquisition software called PHASE.
PHASE was used to set measurement conditions and obtain the velocity from the signals
sent by the IFA 750. Furthermore, PHASE was used to calculate the information needed
to analyze the flow such as ensemble averaged velocity, turbulence intensity, standard
deviation, skewness coefficient, and flatness coefficient with crank angle information for

each measured location in the cylinder.

14

Susan >3 2: Ce €838 a. 8%:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1:838:58: c. 8.2m 2 Swim-893mg 6
Bauoicuau er Patagonia”. 6 239.38.. .3 533-8383 9
33.28 c. 532-82.08.38". E in 3 28.25 m
Swim-£8262“. cs 5o.Bv-Baeoa§aoo¢ 5 £563.80? 9. ...-E 8
5o.§-izm§oieu S canto-830%: E 5:84-385 c. 35225.6 6
ear-22.55%". 3 88cm? 6 Ends-bale“. c 53:38:38 6
$937.33.... A: 8.35-8.5 2
“- iiiiiiiiiiiii . . - .
u p o _ a: Errr
" /// \ N n
85388135.}. ..l'H iiiiiiiiiiiiiiiiiii
.
n / ink-Eggh-
. H
m / .............. m ......... .l./
. / u.. 2‘; n
i .- ... .- .. .. e .... .......... .. ..... m
4. ®.. in NH ........ e 1%; L _ E e _ :2... 25.1,.
.II" II I I «I t O 8 ........ “f H” g‘tm
, 3 > 90
_ _ u H W . .
_ . . e 1 m
_ _ n H _ < m m ..... Gil-2......“
_ _ m H _ . .
_ _ . , / .. I
_ _ . ll
_ _ h w/r/ 1r“//// " u
. IIIIIIIIIIIII

15

Chapter 3

EXPERIMENTAL WORK

Effective Cylinder Volume Mapping

The number of LDV measurements taken was assumed sufficiently large that we
could consider it a valid representation of the cylinder volume. From these
measurements, we were able to map the volume flow field. In addition, the mapping
provided an opportunity to calculate volumetric quantities such as a volumetric turbulent
kinetic energy versus crank angle which characterizes the turbulence levels in the
cylinder.

The time history of 1,548 LDV measurement locations were used to map the in-
cylinder flow field during much of the intake and compressive strokes. The location of
each LDV measurement was geometrically spaced 5 mm apart in a plane of ten
measurement planes. Each of the measurement planes were spaced 10 mm apart with the
exception of one plane that ran perpendicular to the other nine. The 3D velocity vector
was measured at each and every measurement location. This measured 3D velocity
vector was assumed to correspond to a volume element that spanned an equal distance
between it, at all the surrounding measurement points. A homogenous distribution of the
measured velocity was then considered to exist within the corresponding element. With
these assumptions, the construction of the mapped flow filed covers the center volume of

the cylinder spanned by all of these elements. Therefore, the mapping of the 1,548

16

measurement points was considered an effective cylinder volume mapping of the entire
cylinder’s volume. Hence, all the volumetric quantities, calculated in this report, were
based on this fundamental assumption. An important restriction to this is that we did not
include the flow field within the combustion chamber. This is due to the limited optical
access of the engine head that unfortunately does not allow for any measurements in this
configuration.

A coordinate system was set up within the cylinder used in identifying viewing x-
planes, z-planes, and y—planes. A schematic of the coordinate system is given in Figure 4.
The measurement planes are shown in Figure 5. The view illustrates that plane 1, and
planes 3 through 10 correspond to x-planes. Plane 2, and those that can be cut parallel to

it, are defined as the z-planes.

Plane] atx= 0mm
Plane2 atz= 0m
Plane3 atx=+10mm
P1ane4 atx=~10mm
Plane5 atx=~20mm
P1ane6 atx=+20mm
Plane7 atx=+30mm
Plane8 atx=-40mm
Plane9 atx=-45mm
Plane 10atx=+45mm

17

 

 

 

 

 

 

 

:\
\

 

 

 

 

 

 

 

 

 

 

Figure 5: Cylinder Coordinate System Displaying Cylinder Volume with Ten Measured

Planes

Data Reduction and Validation

A general description of the data acquisition and Signal processing methods used
in the 3D LDV measurements is given in this section. 3D LDV measurements begin with
the construction of the three component measurement volume. The measurement volume
consists of three sets of moving fringe patterns created by three pairs of single
wavelength beams. The fringe spacing of the three orthogonal fringe patterns can be
calculated with the beam wavelength and the half angle of each pair of beams. Shown in
Figures 6, 7, and 8 are actually fringe pattern images photographed during the coarse of
the experiment. Figures 6 and 7 display fringe patterns created by a pair of “green” and

“violet” beams, of 514.5 nm and 476.5 nm, respectively. A two component fi'inge pattern

constructed from both the “green” and “violet” beams is shown in Figure 8.

 

Figure 6: Single Component Fringe Pattern Constructed by a Pair of 5 14.5 nm Beams

 

 

. ml li‘“i

 

,,ll-:§ '
llll 1:12! li‘“

Figure 8: Cross Sectional Photograph of a Two Component Fringe Pattern

20

Velocity information is contained in the time variation of the light flux produced as a
particle(s) passes through the fringe patterns in the measurement volume. The
characteristics of the photo emission are considered the Doppler signal. The Doppler
signal is observed via three sets of receiving optics, including focus lenses and
photomultipler tubes (PM-tubes). Each detects the frequency of a particular scattered light
flux originating from the respective fringe pattern in the measurement volume. PM-tubes
operate on converting incident photons to a current characterized by “photo electrons” or
anode current.

Frequency shifting utilized in LDV applications allows measurements of flow
reversals. The effect of fi'equency shifting, one of the two beams used in construction of
the fringe pattern, can be thought of as causing a “moving fi'inge” system. A stationary
particle in the moving fringe pattern will produce a scattered light frequency equal to that
of the moving fringe pattern. If the velocity of the particle increases in the direction of
the fringe movement, then the scattered light frequency will decrease. If the particle
opposes the fringe movement, the frequency will increase. The observed scattered light
frequency can be used to calculate the particle’s true velocity based on the frequency of
the moving fringe system. The ability to frequency shift one beam, with respect to its
counterpart, is accomplished by the use of an acousto-optic Bragg cell. Allowing one of
the beams to pass through this optical device imposes a frequency shift or a phase angle
on it relative to its counterpart. The amount of frequency shifting imposed on the passing
beam becomes the frequency of the moving or cycling fringe system. It is this frequency

which is used in calculating the particle relative velocity.

21

The total Doppler light flux is a random superposition of the individual Doppler
burst, each occurring at a random arrival time with a random amplitude. The amplitude is
random even if the particles are evenly dispersed, because they pass through different
parts of the measurements volume. An analysis of the effective number of particles that
appears to be in the measurement volume is based on determining how much each
particle contributes to the total signal, i.e., a single good scatter plus many very weak
scatters may look like a single scatter. Using the assumption that the light flux is a non-
uniform Poisson process, or categorizing it according to the mean number of signals per
unit “burst width”, we can relate it to the mean number of particles in the measurement
volume. The total number of natural particles in the measurement volume may be very
large in most situations, however, only a fraction of them are usually large enough to
produce sensible signals.

Further analysis, using these ideas, concludes that there are two major signal
classifications that can be used for Doppler signals corresponding to the situation of high
photon density (high photon density just simply states that there is enough information
per second to permit following the history of a timeodependent velocity component ).
These domains can be stated as high- and low-burst density. Burst density is the sum of
(on average) N Doppler burst observed by the PM-tubes whose amplitudes and phases
depend on the arrival time. High-burst density implies a high-data density signal. In this
case a signal is present almost always, so velocity information can be obtained as a nearly
continuous function of time. However, a disadvantage of this situation is that the signal
often contains “phase” or “ambiguity” noise. The case when there is no data most of the

time, so called “drop out” periods, is classified as low-burst density. In this case multiple

22

events are very rare; so if there is a Signal, it is usually a single particle signal. Low-burst
density signals avoid phase noise because there is rarely more than one particle in the
measurement volume. Additionally, low-burst density signals can lead to both high- and
low—data density signals. Low-data density signals produce random, sparse data, which
require special data processing techniques to properly compute the statistical moments
from such data. If high-data density occurs due to low-burst density occurrence, the
result is nearly continuous data from single particles each producing an accurate reading
of velocity. This is the case that is sought in most LDV applications. With a single
Doppler burst from one particle, no phase noise is present; and there is no theoretical
limit on the measurement accuracy. However, in practice, accuracy depends on the actual
signal-to-noise ratio of the signal and the characteristics of the signal processor.

A signal processor (IF A 750) was employed to obtain true velocity information
from the complicated Doppler signals discussed above. When the Doppler signals enter
an IFA 750, they pass through bandpass filters. These filters are used to reduce wide-
band noise and eliminate the pedestal (basically the DC offset due to the PM-tube), which
in turn improves the signal-to-noise ratio. A high-bandpass filter reduces the pedestal,
while a low-pass filter reduces the amplitude of the background noise. There are 16
filters contained within the unit, which can be selected, based on the range of velocities
one is attempting to measure. The signal is then passed through a fixed-gain amplifier to
amplify the signal. This signal is then “clipped” to provide a signal (1 bit), which is then
fed into a burst detector and the sampler.

The burst detector uses the one bit signal to detect when the burst starts and stops

and also provides an estimate of the Doppler frequency. Since Doppler signals occur

23

more or less in random intervals, and continuous processing of the signal would result in
the processor processing mostly noise, the burst detector is used to discriminate between
the Doppler signal and the noise. The burst detector does this by continually monitoring
the quality of the signal-to-noise ratio of the incoming signal and by detecting the signal
when this value exceeds the present level. Once the signal exceeds the present level a
burst timer (in the burst detector) a measurement of the length of the signal burst takes
place. While this time measurement is occurring the signal is digitally sampled at
multiple frequencies by the burst detector. At the preclusion of the burst, the sampled
information is placed into buffer memory. The burst length time recorded by the timer is
used to calculate the center of the burst. The Doppler fi'equency, estimated by the burst
detector, is then used to determine which of the multiple sampling rates is optimum for
each individual burst of the incoming signal. Once the optimum sampling rate is
determined, 256 samples from the center portion of the signal burst are then transferred
to another buffer memory region for post processing.

The post processing of these samples takes place in the autocorrelator. The
autocorrelator performs a double-clipped autocorrelation on the 256 samples to extract an
exact Doppler frequency of the signal. The Doppler frequency is determined by this
autocorrelation firnction by knowing essentially the sampling frequencies. In addition,
several criteria are applied to the autocorrelation results. For a further explanation of the
autocorrelator, one should consult [5].

The IFA 750 also performs multi-channel operations, where it accommodates
three channels of processing, each with a filter, sample and burst detector. However,

each of these channels share the same post processor. The maximum number of

24

individual measurements the IFA 750 can handle is approximately 120,000 per second for
a single channel. Constructing a time history for the flow is accomplished because the
IFA 750 time stamps each data point sent to its buffer. Additionally, the IFA 750 is
controlled by a PC running software (PHASE) that can determine coincidence detection
of the three processors.

Coincidence detection is essential for three-component measurement, especially
for highly turbulent flows. The IFA 750 can be set to create a coincidence window via
PHASE. The coincidence window width can be set within a range from one microsecond
to one second. The measurements produced in this study used a coincidence width of 20
microseconds, which insured that all three channels obtained Doppler signals within this
time period before accepting the signal as valid.

Included with each three-component measurement is the transmission of non—
LDV data stamped to each measurement. This includes the crank-shaft position obtained
by an RMR which produces 1024 pulses per revolution which can be resolved to 8192
resolver points per cycle. The maximum number of 8192 resolver points per cycle is
achieved by combining the quadrature output Signals sent by the device.

Twenty kilobytes of data points were obtained for each measurement. This
corresponded to about 275-375 coincidence points (good points) per velocity component
per specific crank degree. Therefore, a total of approximately 1000 coincidence
measurements occurred. which was used to ensemble average a single three-component
velocity measurement at a given specific crank angle.

LDV measurements are suited for fluid studies in rotating machinery such as the

internal combustion engine studied for this report. The non-invasive measurement

25

technique can quickly record a large number of independent, virtually instantaneous
velocity measurements. These measurements are particularly useful for studies such as
this one which calculates turbulence intensity and other flow parameters used in design
optimization.

For a more detailed description of ideas briefly outlined in this section one should

consult references [5, 6].

26

Computer Visualization Technique

In order to effectively analyze this large LDV data set consisting of 1548 points
and 127 scenes, a new systematic means of organizing and visualizing the data was
employed. The platform used was a Silicon Graphics (SGI) Indigo2 Workstation running
the data visualization software Iris Explorer. Animations were created for 3D velocity
vectors and also for TKE contours. The size of the data set used to create the vector
animation was approximately 60 megabytes (Mb). In order to effectively animate and
store this large data set the SGI workstation was equipped with 128 Mb of RAM memory
and two gigabytes of hard disk memory.

The animations created by the SGI workstation were transferred to video through
the SGI Galileo Video interface. They can be taped in VHS or Hi-8 format. The SGI
graphics can also be saved in standard image formats, such as TIFF, and be transferred to
personal computer to be included in word processor documents.

Iris Explorer is a relatively flexible data visualization software which uses a large
number of built-in modules (or subroutines) and can incorporate FORTRAN, C, and C++
code into user-built modules. For the present study, the Explorer data type chosen was
the 3D curvilinear lattice due to its ability to be sliced and viewed from three directions.
A new coordinates file had to be created first that satisfied this 3D lattice and formed the
cylindrical shape of the measured volume. Some FORTRAN code was written to
reorganize the LDV velocity files into systematic 3D arrays of 9x12x19 elements which
correspond to the new coordinates file. The Explorer utility Datascribe was then used to

create reader modules to convert these ASCII files into curvilinear lattices. A three data

27

channel module was created for the velocity vector fields and a one channel module for
the TKE plots.

A Simple wireframe representation was created of the piston, cylinder, combustion
chamber, and two intake valves for the computer simulation. Simple wireframes work
well with animating large data sets because of the low rendering overhead, and they do
not block the view of the vectors and contours. But for future non-animated plots of in-
cylinder flows, it would be beneficial to render solid models of the intake runner, port,
valve, and combustion chamber to better visualize the flow boundaries with the vector
fields. The piston and valve movements were calculated and included in the simulation to
animate these components. A piston velocity vector was also included in the same scale
as the air flow velocity vectors, in order to compare the flow magnitudes with the piston
speed.

Separate Explorer application maps were created for animating velocity vectors,
animating TKE contours, creating black and white and color images, and for creating
plots of velocity vectors colored by TKE magnitude. These maps also contain the ability
to create isosurfaces of constant value TKE levels. These isosurfaces allow the user to
view where and when pockets of higher turbulence exist in the measured volume. The
data flow and module firing in Iris Explorer is controlled by the user via the map editor.
Figure 9 shows two typical Explorer maps that were used during this study. Streakline
and smoke visualization modules were applied to the time varying vector fields in an
attempt to produce streakline and flow volume plots of the cylinder flows. Unfortunately,
the data set measured does not work well with this type of module because of the

incomplete data regions along the sides of the cylinder and in the combustion chamber

28

where the optical access of the laser is limited. This type of visualization is intended for
Computational Fluid Dynamic (CFD) solution data sets where a dense grid encompassing
the entire flow volume is used to compute these flow paths.

Due to the complex nature of LDV measurements, a few velocity vectors seem to
have unusually high values such as two or three times larger than those at neighboring
points. These measurements usually occur when there is not enough good points to create
an accurate ensemble average. Hence, these points were zeroed out of the data set in the
case of velocity numbers. These same points usually correspond to unusually high TKE
values. For this case TKE values were averaged with its neighbors, so a large gradient to
zero does not occur in the contour plots. Out of the entire 133,595 total measurements,
about 40 of these measurements were found and changed accordingly. The method for
finding these bad points is to go through the animations scene by scene, using velocity
vectors colored by magnitude, and look for any sudden discontinuity of color in space or
time. These discontinuities also usually have vector directions that do not make sense
with the surrounding velocity field. These incorrect velocity points usually have incorrect
TKES also. These can be verified by viewing the TKE contour animations, noting when

and where any sudden discontinuity in the contours occurs.

29

W
ammo:

O D
EEC:
D 0 Cl
-—_-°:°.
D

 

O
Wet-mm
l:ll::3
Md
E::l
.
1::ZIE:1

O D
EEC:
LdToGomd) O C]
“Hoe 0 Cl i%£
eon
l:jl:l

...-...... ° °
L_,;_.
Lon-Gould» O D
L:ii:::j

torMog
mm o
L:ll:l
Che-3WD o D
l::ll:::l
o

 

 

Figure 9: Sample Iris Explorer Data Visualization Map

30

Chapter 4

FLOW DESCRIPTION

General descriptions of flow patterns that developed during the intake and
compression stokes are discussed in this section. Animations of the flow patterns
assembled by the measured velocities and viewed on the SGI are the basis and
justification for the descriptions given herein. The descriptions are given in three
segments. The first segment focuses on the descriptions of the flow in the center area
spanned by planes 1, 3, and 4. Secondly, the flow within plane 2, and those that lie
parallel to it on each side (three planes on each side), will be examined. A description of
the flows around and near the cylinder liner as well as the those that pass through the top
and bottom slices cut by y-planes constitutes the third segment. Lastly, the collective
information given by these segments is assembled to identify volumetric flow
characteristics. An additional note is that all the figures in this section give a
corresponding vector scale for the respective views. The vector scale was chosen to
optimize the views and is consistent in the views of multiple images. Furthermore, one
can view the animations recorded in VHS format which are included along with this

report for verification of the descriptions given within.

31
Planes l, 3, and 4

Given in all the figures listed in this section are views of planes 1, 3, and 4. These
planes are shown enclosed within a wire frame of the piston-cylinder assembly which
include the two intake valves. Plane 1 is the central bisecting plane between the two
intake valves and two exhaust valves. Plane 3 and Plane 4 are two parallel planes to
plane 1 (see Figure 5).

The relatively organized structure of the intake driven flow patterns can be seen
to develop as early as crank angle 67°. Referring to Figure 10, there are two regions of
high velocity magnitude that occur: along the cylinder bore opposite the two intake

valves and right between the two intake valves on plane 1.

W

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Plane 4 Plane 1 Plane 3

Figure 10: Flow Patterns in Planes 1, 3 and 4 at Crank angle 67°. Vector Scale=l .5

The creation of the first of two vortices can be seen in Figure 10. The part of the flow
that is initially sucked into the cylinder passes over the top side of the combustion
chamber to the opposite side of the cylinder and consequently deflects off the descending

piston surface creating this first rotating structure. In plane 4, another rotating structure

32

appears. The structure is not fully developed at this point since only half of the area can
be characterized as displaying rotational motion. The second vortex is being developed as
the piston moves downward. Shown in Figure 11 are views of planes 1, 3, and 4 at a
crank angle of 89°. Similar flow structures are observed at this crank angle which exhibit

a bit more organization in their patterns.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Plane4 Planel Plane3

Figure 11: Flow Patterns in Planes l, 3 and 4 at Crank angle 89°. Vector Scale=1.5

The origin of the second vortical structure can be attributed to the two dominating intake
flows colliding and forming the second counterclockwise rotational structure. The result

of this is shown in Figurel3.

 

 

 

 

 

 

 

 

 

 

 

 

Plane 4 Plane 1 Plane 3
Figure 13: Flow patterns in Planes l, 3 and 4. Crank Angle=135°, Vector Scale=1.5

The flow patterns in Figure 13 demonstrate that the two counter rotating vortices drive
the flow across the center of these planes in a diagonal fashion towards the far side of the
intake values. We find that the two counter rotating vortex structures are fully developed
at approximately 148° ATDC which is displayed below in Figure 14. In addition, the
large downward intake flow has shifted to the far side of the cylinder as noted by the

concentration of the red and yellow vectors in Figure 14.

0
J

 

 

 

 

1 I ’11
,/////’
/
{/5/ I 1’
{1’ fr“:
\\\._;///l
/

//"-"
”(fl
if,"
1 1’
\
uh

\\

i’
\\ '1

\\~4rr
\\\\\
\
I/

‘ /
,4

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l
J

Plane 4 Plane 1 Plane 3
Figure 14: View of Planes 1, 3 and 4. Crank Angle=l48°, Vector Scale=1.5

34

These two developing vortical structures become complete in the late intake stroke. The
center location of these structures may be dependent on the pathlengths of each air
passage. Also, these views suggest that the vortical structures may be connected in a
toroidal, or perhaps in a semi-circle style arch fashion, which will be discussed after the
views of the perpendicular vertical planes are shown.

Another 15° later, the vortical structures begin to move out of the area measured
as shown in Figure 15. The first vortex begins to move toward the combustion chamber,
while the lower, second vortex begins to follow the descending piston. An observed
effect of the changing positions of the vortical structures may be related to the location
and magnitude of the main cross flow. The main cross flow is located between the two
vortices. Notice how the large diagonal flow is most intense in plane 4 relative to planes

1 and 3 when comparing Figures 14 and 15.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\\\\\\ If, I: /

1 \1\1:4)\ “ \,'~'3\ \ ’l\,‘:-i ix‘f,‘
-, Np ...“ i>.m::...-\i§ ; ‘1.“
"k\\\ \F” \‘\\\\V\\“"i " \\ "’/
,.\ w, [ \\\\VN-.._ - - v

. ~ \\\~-.- / \\\\N~... ”/ \ .\\ \,
.’,\2~.~.\L”/ |~\\‘\ //// l’. 3\ \’
I/I'AE :‘\“://// ' ~\ 7// V'R ‘\ r:

\ ~\ ’ i
living/f :Eé>::~:=颒 1‘5 1‘, ’
§&\_r~ y//;/ ’ \ “WM [\§§1 ,/

C
c ‘i
‘1 a
11V \_ _/ \,
Plane4 Planel Plane3

Figure 15: Flow Patterns in Planes l, 3 and 4 at Crank Angle 168°. Vector Scale=1.5

35
Notice in Figure 15 that in plane 3 there is little cross flow in contrast to the amount that

plane 4 is experiencing. This is due to the inlet flow being biased to the left valve due to
the offset in the intake runner-port design (see APPENDIX I). Thus, the return flow
(main cross flow) favors plane 4. The vortical structures are shown to still be intact,
however, their rotational amplitude seems to begin to decrease at this point. Also, the
structures continue to migrate to different locations as seen by the changing locations of

the apparent centers of the cross-section of the two vortices. The view of the flow field at

BDC is given next in Figure 16.

 

 

l

 

 

 

 

t
/
x?
J
x?
J

 

\\\\\\\ t

I\\\\ \ r '\ . .\ ,0.

‘\ I \ /‘\\ I\\\\\\\.,\ \\ I‘I‘. \ \ \\\\

"\l\5‘i\\:§:\‘\‘:-"l\\§t ,\;},.u\::.\.\\\.-,;l\x ;}“,“.Z$\1i\1,‘,’,‘1\

iiix-ththggwu UH {waull’fl HI".'.\~‘-.:2‘///
ii' \\\\>~ "I/l ’1I"\\\\~4M ’// "\.' l\\“\. I//
\"\\\~”//// {\ u\\\\~ ////l f’;\\.\\\\ ”///
li'tttu‘r’wu ls“ .J ”/l '1’ .\ M» ”H!
rif\\\\:’:‘*"///// 1! \\\\\.;’//// I "\H \.. ’////
i’l\\\\\.;i':’/// 1' \\\\\~;" // (I l j\ ,\ \L’lll /
’l.\\\\\~.:’;/// \i ‘§§§“rv—y «11% "I ~,:,/// /,
\‘ \\\‘o‘1fl’ // \ \ \TIW// \\\‘e I W/;/

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I;
L
f
I

Plane 4 Plane 1 Plane 3
Figure 16: Flow Patterns in Planes 1, 3, and 4 at Crank Angle 181°. Vector Scale =1.5

Another feature, of the flow field in these planes at BDC, is that higher magnitude
flows, which originated from the fluid inducted into the cylinder across the top of the
combustion chamber (see Figure 13), have continued to dominate the exhaust valve side
of the cylinder. If we now view the flow another 26° later during the compression stroke,

we begin to see that the momentum of the dominating intake flow is transferred off the

36

piston surface and deflects towards the intake valves. The deflection of the main
induction flow has the effect of weakening the vortex structures that were initially setup
during the induction process. Whereas, the mechanism causing the defection of the main
induction flow may be linked to the slowing and reversing piston. The structures are seen

to begin to dissipate rapidly from this point shown in Figure 17 at crank angle 207°.

1
J

 

 

/

.,,,___
xxx
.1
II'J/
._—/l
11
-1
/

\\\.
/

I.
z/l/L, /

fM/IIM’”
K /

{rm—v
I] I l
\

§\\

\\\\ -
\\\\\\\
N\\\‘\
\\\\\\\\

\.
/
\\\

\

 

 

E E
fifi

Plane 4 Plane 1 I Plane 3 I

 

 

 

 

 

 

 

 

 

 

 

Figure 17: Flow Patterns in Planes 1, 3, and 4 at Crank Angle 207°, Vector Scale 2.0

The next point of interest occurs just before the intake valves close. At this point the
magnitude of the flow represented in these planes was found to decrease considerably.
Whereas, the region of substantial velocity magnitude is found above the ascending

piston surface as shown in Figure 18. The flow is driven by the ascending piston head.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\\I\\\\\1\\ I f’/'1’I¢_
!‘\’\\\\1\!\ “1 . ,‘ /\ ’I"’ as]
\\‘\\\‘~ \-~- ,1.) (:\1 /\/l‘, \ 1”]
:-ltttt:l(~v . .- ‘g 5 5., .- -/,
:\\\\‘~n.:: H {1.1. .::"~./
» \\ \ H\~,./;’; 1‘ \ L"::/’/l"/
1 ~\\\ \\ \ I W\\\ \ \
Li a La i—J hi i—J
Plane 4 Plane 1 Plane 3

Figure 18: Flow Patterns in Planes 1, 3, and 4 at Crank Angle 241°. Vector Scale =2.0

Further inspection of the flow patterns in Figure 18 give evidence of the continuation of
the original vortex structure that is spinning around the center of the remaining volume.
The flow is driven toward the intake valves.

The last sufficiently clear image of the velocity field occurs at crank angle 271°.
In Figure 19, the flow continues on its path toward the intake valves in a relatively

uniform motion, while the weakly circulating flow pattern is still observed at this point.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

will: {32* git-‘(Qiflgrp‘g‘sfil "'1’ Milt-5,29;

NW \\\‘\‘N ’ ‘Q‘SVNVM ‘1’ guilt

.. L.- d a ..J .1
Plane4 Planel Plane3

Figure 19: Flow Patterns in Planes 1, 3, and 4 at Crank Angle 271°. Vector Scale=2.0

38
Flows Through the Center Seven Z-planes

There are nineteen planes that can be cut parallel to plane 2 which are noted as the
z-planes. The flow field through the center seven is the focus of this section. Unlike the
views presented in the first section of this chapter, the views here display the flow field
through multiple planes in each image. Two or three different view orientations are
shown in each figure of the same flow field for the specified crank angle. These views
display a portion of plane 1 and those parallel to it which were analyzed in the first
section

The initial structure of the flow patterns through this center region of the cylinder
were shown to organize as early as crank angle 67°. In order to be consistent, the flow
field passing through the center seven z-planes is also given at crank angle 67° as shown

in Figure 20.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 20: Flow patterns through the seven center z-planes at Crank Angle 67°.
Vector Scale=1.0

39

During the start of the intake process the orientation of the valves affects the incoming
fluid. The similar flow patterns about the z-axis suggest that the dynamics of the flow
have strong tendencies toward symmetry. The direction of flow field through this region
is shown to exhibit both an upward and downward motion. Along the z-axis, or in the
direction of plane 1, the flow is basically downward and induced by the descending
piston. The flow along the bore is also seen to be downward and tends to follow the
curvature of the cylinder toward the positive 2 direction. The upward motion is seen
occurring approximately beneath the intake valves. The shifts in flow direction occur
symmetrically about the x-axis. The downward induced flows in this region occur
between the intake valves and symmetrically along the outer sides of both valves. Shown

in Figure 21 is the flow field in this region another 22° later at crank angle 89°.

 

 

 

 

 

 

 

 

 

 

 

 

Figure 21: Flow patterns through the seven center z-planes at Crank Angle 89°.
Vector Scale=1.0

As seen in Figure 21 the developing flow patterns maintain a large degree of symmetry.

One observation of these images may be misleading, namely that there exists two

40

rotational structures beneath both intake valves. However, this would be incorrect as
there are rotational structures which are centered primarily in the direction spanned by
these planes. The two counter-rotating vortices, outlined in section 4.1.1, are orientated
in the planes spanned by the y-z direction. These views, which span the x-y directions,
demonstrate the three dimensionality of the first counter rotating structure. The structure
is not connected in a toroidal fashion but rather has a non—connected counter rotating arch
shaped geometry. Thus, what we are seeing are the ends of a single arch or “horseshoe”

shaped structure as it extends into these regions. Figure 22 gives further evidence of this.

4

 

2;»?

r/’

ZZZ-4
-.. ‘

’21

.._'~

5.3.5;
1 r4964(
. —-l- _4 2‘-

 

 

 

 

 

 

Figure 22: Flow patterns through the seven center z-planes at Crank Angle 140°.
Vector Scale=1.5

The light blue regions depict the smaller velocities where the tips of the two currently
rotating structures end and become difi'used into the flow. An interesting observation of
Figure 22 is that the flow tends to be stronger along the bore at the positive x side of the
cylinder. An explanation of the cause of this effect is due to the intake runner. A slight

offset in the runner, just before the port, concentrates the flow to the left side of the

41

port producing this effect. Given in APPENDIX I a simple schematic of the intake port
design is shown.

The next view is the flow field at BDC, shown in Figure 23. Notice how the

 

 

 

 

 

Figure 23: Flow patterns through the seven center z-planes at Crank Angle 181°.
Vector Scale=1.5

larger, higher magnitude flows form a symmetric “half-hour-glass” shape about the y-
axis. The center region at about y=25 shows relatively small magnitudes, where the outer
portions wrap around and then form into a strong upward motion at the higher y
locations. Also, the two regions at the top of the cylinder and at the far sides of the x-axis
show two rotating areas. These, again, are the effects of the three dimensionality of the
first counter- rotating structure outlined in the first section of this chapter. In particular,
comparing Figure 16 with 23, one can see how this upper structure arches around the top
half of the cylinder. It seems to exist partially in the combustion chamber in the positive 2
side, and then bends inward toward the outer sides of the x-axis.

Further along into compression, we began to see how the ascending piston

influences the flow field. A net upward motion imposed by the ascending piston drives

42

the flow toward the intake valve side of the cylinder. Give in Figure 24 is the flow field

at crank angle 210°.

  

     
  
   
       
    
  

watt-a:
av-

.»~:“—.\K:

\: "MN“
h it“.

\P - "

~ *a’:.\
,, r 1:4
fig

“ C

ifs-v
n$u

\ S p

n

b

   

.'—-%—-_ Z
a».
maul.

     

 

 

 

 
      

 

 

 

 

 

Figure 24: Flow patterns through the seven center z-planes at Crank Angle 210°.
Vector=l .5

The majority of the stronger flow seems to lift off the ascending piston and orient itself
toward the outer sides of the bore. The center regions are seen becoming more “hollow”
in terms of relative velocity magnitudes in respect to the outer sides. The “half-hour-
glass” shape of the flow field, shown in Figure 23, begins to dissipate in terms of the
upper region of the cylinder. In addition, the strong flow about the y-axis no longer exists
in the same way as shown at BDC in Figure 23.

Continuing to track the in-cylinder flow motion we come to crank angle 233°.

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 25: Flow patterns through the seven center z-planes at Crank Angle 233°.

Vector—‘1 .5

The higher magnitude flows continue to dominate along the piston surface and move

toward the outer sides of the bore. The center region becomes even weaker as the strong

flows wrap around it. Next, in Figure 26, this trend is observed to continue at a crank

angle of 264°.

 

 

 

 

 

 

 

 

 

 

 

 

ector=1.5

Figure 26: Flow patterns through the seven center z-planes at Crank Angle 264°.
V

44

These are the last sufficiently full images of the flow through these seven center z-planes.
As one can easily see, the charge motion occurs along the bore and not in the center of the
remaining volume of the cylinder. The center region seems to be the calmest, or it is
where the fluid motion is the weakest (and the turbulent kinetic energy is the greatest).
In addition, the outlining symmetry along the z-axis is still evident at this point and will

most likely continue further into compression.

Flows Along the Cylinder Liner and Through the Top and Bottom Y-Planes

With the large number of LDV measurements, views of the flow images along the
liner as well as those which passed through the top and bottom slices or y-planes were
plotted. These flows are examined in this section. Much like the last section, the images
are of multiple planes given for two or three viewing angles at the specified crank angle.
In the same spirit of the early sections, the analysis begins near crank angle 67°. In

particular, shown in Figure 27, is the flow field at crank angle 72°.

 

fl

 

M

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 27: Flow patterns through the top and bottom slice and along the cylinder liner
at Crank Angle 72°. Vector Scale=l .O

 

.DescribedinChapter 5 and displayed in Figure 37

45

Here we are interested in the flow behavior at the entrance region, just above the piston,
and along the liner. From the images of Figure 27, the symmetric flow patterns follow
the curvature of the cylinder toward the positive 2 direction. This aspect of the flow starts
from the outer sides of the intake valves (the sides closest to the wall and opposite to each
other). From there it flows downward and diagonally, along each side of the cylinder
until the two symmetric portions meet. The area where these flows meet is at the far
positive side of the z-axis, or where plane 1 is experiencing a strong downward flow.

Given in Figure 28 is the flow field at crank angle 94°.

 

 

 

 

 

 

 

Figure 28: Flow patterns through the top and bottom slice and along the cylinder liner

at Crank Angle 94°. Vector Scale=l .0 .
In Figure 28, we again see large symmetric tendencies exhibited by in-cylinder fluid
motion. The flow seems to be accelerating onto the descending piston face. Notice how
the flow beneath the intake valves on the far negative 2 side of the cylinder hits the piston
face and then outward toward the sides of the cylinder. In addition, during intake, the

areas along the bore which can be seen by the measurements, almost all flow downwards.

46

Based on the image of Figure 28, one could speculate that if additional measurements
were done chiefly around the cylinder’s inner surface area, the results would show that
the flow motion would be consistently downward in these regions. In Figure 29 we see

how these regions finther evolve.

  
 

   

. l)"

3.
51.

V.

  
 

D.

, 3
—- u“. A .-
\7

 
     

- '5'”,-

 

  

u v
—.———. -a‘

 

 

 

 

Figure29: Flow patterns through the top and bottom slices and along the cylinder
liner at Crank Angle 118°. Vector Scale=l.0

In comparison to all the views shown thus far, there are two regions of the highest
amount of velocity inside the cylinder’s volume during the induction process. The first
extends from between the back side of the intake valves, down to the piston face, beneath
the intake valves. The other is just opposite on the side of the exhaust valves, as shown in
Figure 30. These strong flows are seen in the images of Figures 29 and 30 deflecting and

colliding into one another just off the top of the piston’s surface.

 

 

 

 

 

 

Figure 30: Flow patterns through the top and bottom slices and along the cylinder
liner at Crank Angle 148°. Vector Scale-=1 .0

At BDC we can note that the induction driving flows noticeably change. Shown in

Figure 31 is the flow field at crank angle 181°.

 

 

 

 

 

 

 

liner at Crank Angle 181°. Vector Scale=1.5.

The large induction driven flows, primarily situated beneath the intake valves (see Figure

27, 28,29), no longer dominate. Figure 31 displays how the larger downward flows are

48

now primarin located on the far positive z side of the cylinder. The origins, as described
before, stem from the fluid that is initially sucked into the cylinder which passes over the
top side of the combustion chamber to the opposite side of the cylinder. This trend of the
dominating stream shifting to this side of the cylinder is seen to occur as early as crank
angle 120°. This aspect of the flow begins to dominate the flow field for the remainder of
the compression stroke. Shown in Figure 4.1.22 is the consequent effect of this at crank

angle 210°.

 

 

Figure 32: Flow patterns through the top and bottom slices and along the cylinder

liner at Crank Angle 210°. Vector Scale=1.5.
Notice the slight shift in the region of high magnitude just above the piston surface in
Figures 3land 32. It seems to move from the far positive side of the z-axis to the center

of the piston. Figure 33 shows the evolution of these flows.

49

 

 

 

 

 

 

 

 

Figure 33: Flow patterns through the top and bottom slices along the cylinder

liner at Crank Angle 264°. Vector Scale=1.5 .
The bulk motion is mostly symmetric about the z—axis and moves mainly upward due to
the piston and may wrap around the top side of the combustion chamber. The flow is
observed passing upward through the top slices. This flow inevitably passes around the
topside of the combustion chamber and recirculates back into the measured volume. (see
Figure 5). Hence, a weak tumble motion probably occurs until the onset of the spark

ignition at this speed and most probably at higher speeds.

Flow Pattern Summary

An accumulation of the images provided by the animations, as well as the analysis
given in the previous sections of this chapter, were used to outline the dominating flow
features of the inocylinder fluid motion during the induction and compression strokes.

The four-valve design dominates the intake and compression flows. In particular,
symmetric flow patterns were observed about the z-axis. During the intake stroke, the

net fluid motion was observed to exist in both a downward and upward fashion. The two

50

main intake driven flows follow the descending piston head in a downward sense. Their
locations were found to be right in between the two intake valves, along the z-axis and
along the cylinder bore on the exhaust valve side of the cylinder (see Figures 11-15). The
origin of the large induction flow, orientated along the exhaust valve side of the cylinder,
originates from the fluid entering the cylinder by passing over the top side of the
combustion chamber and entering at the far positive 2 side of the cylinder. The fluid
which passes through the outer sides of the intake valves (those sides which are further
apart from one another), also moves downward. Hence, a tumble motion occurs primarily
orientated in the y-z direction due to the intake port and four-valve design. In addition,
very little swirl motion occurs due to the symmetry that exists in the intake runner, port,
and valve design (aside from the offset as pointed out in APPENDIX 1) .

A net upward motion results from the three dimensional effects of the large
downward flows accelerating toward the descending piston head. The strong induction
flows eventually collide with one another and deflect off the piston head. The result is
the evolution of two counter rotating vortical structures. The counter rotation of the
structures drive the flow in between them, resulting in a diagonally upwards crossflow
toward the intake valve side of the combustion chamber. Additionally, upward motion is
found to occur almost symmetrically beneath both intake valves. This is the result of the
first vortical structure whose geometry extends to these regions. The rotating areas are
seen in plane 2, and those parallel to it, which are the ends of the structure. The structure
has an arch shape whose ends are tilted slightly upward toward the combustion chamber
and point toward the intake valve side of the cylinder. The other vortical structure rotates

in the opposite direction of the first one positioned above it. Its ends seem to be tilted

51

slightly upward as well, but they point toward the exhaust valve side of the cylinder
opposite to that of the first structure. Furthermore, these structures were observed to
migrate. The migration of the structures are dependent on the intake flows as seen by the
shifting of the regions of large upward and downward velocities.

During compression, the two counter-rotating structures begin to dissipate. The
large tumble motion setup during the induction stroke continue to dominate. The tumble
motion exists in three dimensions. It seems to move upward, toward the intake valve
side of the combustion chamber originating fi'om the far x-axis sides of the cylinder. The
dominating flow then moves back into the cylinder mainly downward along the exhaust
side of the cylinder. The fluid motion is driven upward by the ascending piston. Regions
of higher velocity magnitudes were observed along and just above the piston surface
during compression. The tumble motion shows that the velocity magnitudes, through the
center of the remaining volume of the cylinder during the compression stroke, remain
weak. The stronger flows continue to symmetrically dominate to the outer sides of the
cylinder and pass around the under side of the combustion chamber. No swirl motion
was observed in the compression stroke. The weak tumble-like motion most likely exists

and will influence the development of the spark ignited flame kernel.

52

Chapter 5

TURBULENT KINETIC ENERGY

It is important to examine the turbulent nature of in-cylinder flow. A better
understanding of its nature, which strongly influences air-fire] mixing, helps to improve
engine efficiency, fuel economy, and emissions. If any single quantity is to be selected to
represent the fluctuating character of the turbulence, it is the turbulence kinetic energy
(TKE) of the fluctuations, which has been picked by virtually all researchers in the field

[12].
Definition of Turbulent Kinetic Energy

Turbulent motion is randomly time—dependent, strongly non-linear [7], and
generated in the presence of mean velocity gradients. Its characteristics are disorder,
randomness, and enhanced mixing under fully three-dimensional flow motion with high
Reynolds numbers [7]. With this random variation of quantities, statistically distinct
average values can be discerned [8]. The variations of the TKE, which is the kinetic
energy (per unit mass) of the three fluctuating velocity components is defined as [12]:

,2 _ u'2 +v'2 +w'2
=

2

 

q

:2 '2

‘73- u +v +w'2
2

 

q E
(5.1)

53

The fluctuating components of velocity are related to turbulent motion by

expressing them as a time-averaged and fluctuating parts:

u=§+u', v=r7+v', and w=v_v+w'

(5.2)

with the time-averaged part defined by

1 T
u = 3; !u(t)dt
(5.3)
where t is the time increment and T the total measured time span.
Since the LDV system generates ensemble-averaged data per specific crank angle

the ensemble-averaged velocity per component becomes

2’ [214631)]
m5) = ”‘ .

Measurements

 

(5.4)

where the specific crank angle, 79- , represents all angles, 6' , within the window
[(5 - A6) < 0' S (9- + A60]. nj represents the number of measurements in one of the

cycles within a specific crank angle window. NC”... is the number of cycles and
N1,,.,.,,,,u,.,,,,.,mll is the total number of measured velocities. One can calculate a mean part and
a fluctuating part from the instantaneous velocity measurement. The instantaneous

velocity at a measured point is given by

54

u(6, i) = as, (E) + 'zI(9',i) + u'(6',i)
(5.5)

where the cycle-to-cycle variation is assumed to be small; or 5(6', i) z 0 , when averaged

over a large number of cycles. The standard deviation is calculated fi'om the fluctuating

part. The definition of the standard deviation of a set of data points is found by:

0': ifXx —5c')2= ifixz-fz.
NpI ’ N ’

i=1

 

 

(5.6)
The LDV measurements are used to compute the standard deviation of the three

components from the variance as shown in equations (5.7), (5.8), and (5.9).

N

 

(5-7)

 

(5.8)

 

(5.9)
The turbulent kinetic energy per unit mass in terms of in-cylinder flow measurements

becomes:

55

u'(§)2 + v'(

 

 

)2 we“): .

 

4a
II
N %|

(5.10)

Turbulent Kinetic Energy Calculation Method

The calculation of TKE utilizes different calculation grids, which are composed of
volume elements. A volume element is the volume of fluid surrounding a measured
velocity vector. The entire air mass within the volume element is assumed to move at the
measured velocity. The size of a volume element, Vi , was determined by using
dimensions which were equally distanced to all its neighboring measurement points. For
instance when calculating TKE in a vertical plane, the calculation grid contained Vi’s
with the dimensions 10x5x5 mm3 were used. Volumetric calculations utilizing all of the
points within the changing cylinder volume during the intake and compression strokes
treated Vi’s in plane 2 with dimensions of 5x5x5 m3, and those outside of plane 2 with
10x5x5 mm3. Shown in Figure 34 is an emple of one of the girds that were adopted for
the calculations for one of the twelve horizontal planes (slices). Note, the different
volume element sizes in plane 2 can be seen in the center row of the grid along the x-axis

az=0mm.

56

K [mm]
.50 40 .30 -20 -1o 0 10 20 so 40 so

   
  

 

45 Calculation Volumes
-0
.35
so
as
.20
46
~10
5
Z 0
[mm] 5
10
15
20
as
so
as
40

‘5 Dots represent measurement

locations

Figure 34: Grid to Calculate Turbulent Kinetic Energy in Horizontal Slices

The calculations of TKE are based on the assumptions that the air density remains
constant at ambient conditions during the intake stroke and until intake valve closure.
The air mass inside the cylinder is assumed to remain constant during the rest of the
compression stroke, while the air density is assumed to be uniformly distributed over the
remaining cylinder volume. The total TKE inside the cylinder depends on the air mass
within the cylinder. Normalization by the air mass allows a calculation TKE, which
neglects air mass variations. Hence, the calculation of TKE per unit mass exposes
variations in turbulence solely from the fluctuating air motion.

The total TKE for the intake stroke is

57
N “'2‘
Tintakel “1.002 Viq'j
1:1
(5.11)

while the TKE per unit mass is given by

TM”: 19. [:2 V.-= “Z: V ’6‘]
(5.12)
where, po is the density of the air mass at ambient conditions. Subscript j indicates the
measured location and N is the total number of measured points.
During the compression stroke, after intake valve closure, between the crank angle
interval of 240° through 360°, the air density is a function of the crank angle. This

changes the calculation of total TKE to

N ....
Tcompressionl == pcz qu'j
jal
(5.13)

while the TKE per unit mass becomes

Tcompression2= pCEV WEI/lg?

(5.14)
where pc is the density of the air mass during compression. pc is calculated by:

pc _ chlmr(6) '
(5.15)

58

Turbulent Kinetic Energy Results

TKE was examined in three different ways: per entire measured volume, per
vertical plane, and per horizontal plane (slice). For each of these three calculations, total
TKE and TKE per unit mass were computed. The next three sections will discuss the

results of each method.

Volumetric Turbulent Kinetic Energy Calculations
TKE was calculated over the entire measured cylinder volume utilizing the sum of
the TKE contained within each volume element at every available measured location.

The result of the calculation is shown in Figure 35 .

 

 

 

 

 

 

 

 

 

30
E __ .,
a 25 0.39 1.
g o TKElJ/kg]
g 20 ,_ ——-—-K=-so.o*e:qo(-o.o19'CA)+1.3
E 15 «~
g 10 -r
g 5 db
...

O A. + 1r 4.

0 60 120 180 240 300

Crank Angle [Degrees]

Figure 35: Turbulent Kinetic Energy per Unit Mass for the Effective Cylinder Volume

S9

In-cylinder flow can be considered a type of turbulent shear flow. Shear stresses
exist due to the fi'iction on the cylinder walls, cylinder head, and piston top. In nearly all
turbulent flows, there is always production, convection, diffusion, and dissipation of
turbulent energy. Often the main contributors to the overall balance of these terms are a
result of the production and dissipation influences. These effects can be clearly seen in
the behavior of the volumetric TKE (per unit mass) curve in Figure 35. Acquired data
between crank angles 50° to 60° show a definite production or generation of TKE.
When the intake valves are open much of the turbulence is convected in having been
generated, and/or produced, by shear across the valves, as well as by production due to
the shearing motions within the cylinder volume. Furthermore, the reduction in TKE
after crank angle 60° is obviously noticeable. The curve smoothly decays approximately
exponentially over the remaining crank angles.

The decay of TKE was fitted with an exponential function as shown by the red

line passing through the data in Figure 35. The function was found to be

K = 90.0e(-0.019crankangle)+l.3
(5.16)

where K is the TKE per unit mass and crank angle the actual crank shaft position. The
function is valid between 60°<crank angle<280° and its curve is displayed in Figure 35.
During the intake stroke and compression stroke, the fitted curve demonstrates the
exponential decay of the turbulence. This fit to the measured decay may be usefull in the

calibration and validation process for computer-based turbulence models.

60
The total volumetric TKE calculation revealed results with some similarities to the

volumetric TKE per unit mass calculation. Shown in Figure 36 is the results of this

calculation.

 

0.0045

0.004 4

0.0035 ‘-

0.0025 0

0.0015 *

YWMMIJI
§

0.001 4

 

 

 

 

o t t :
60 120 180 240 300

0'1"" Anal- {m}
Figure 36: Total Turbulent Kinetic Energy for the Effective Cylinder Volume

The behavior of the curve demonstrates that the inlet boundary conditions plays a
dominant role in the of turbulence levels observed during the intake stroke. The rise of
TKE production is coupled to the piston motion; the peak level of TKE occurs shortly
after maximum piston speed. Following the peak is a decay much like the behavior of the
TKE per unit mass. At approximately at a crank angle of 100°, affect of increased mass
appears to contribute significantly to the TKE. At later crank angles, dissipation effects

dominate. One apparent difference can be noted in the decay trend. A small gradient

M... 13:.“ Hannah... ..r... ....d:

61

change occurs in the total TKE curve beginning at crank angle 240°, the time of intake
valve closure. Hence, the behavior change can be traced to the increase of the fluid
density during the compression stroke.

Two important questions regarding the turbulence in the engine cylinder are
whether it is homogenous (i.e., uniform) and whether it is isotropic (i.e., independent of
direction). The data already presented in a variety of past studies show that during intake
the turbulence is far from homogeneous, nor is it isotropic [11]. An ingenious method of
displaying turbulence levels within the cylinder volume was accomplished which further
helped to clarify this statement. The results showed that the turbulence levels are indeed
far from homogeneous and not at all isotropic.

The turbulent kinetic energy was computed for each measured point throughout
the cylinder volume. These values, together with a point locations file, were transferred
to a Silicon Graphics Indigo2 Extreme Workstation and rendered as contour maps. The
data visualization was created with IRIS EXPLORER sofiware. Figures 37 through 40
show the turbulent kinetic energy plotted at select horizontal slices and isosurfaces
displaying surfaces of equal energy. This function evaluates available TKE data
throughout the cylinder volume. These surfaces can be thought of as pockets of higher
turbulent energy. Their color, as well as the contour lines, corresponds to the color
legend scaled to maximum turbulent kinetic energy. The color legend relates to different
energy level scales for each of the crank angles shown. Same colors allow comparisons
between similar relative energy levels at different crank angles. Hence, for every crank

angle, a new color scale is generated according to the actual maximum energy level.

62

Figure 37 shows an example of two different isosurface levels at the beginning of the

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

intakestroke.
(1
D /‘ -= A E
f x / f \
9at61°CA,Sliceaty==65mm 0at61°CA,Sliceaty=65mm

Figure 37 : Turbulent Kinetic Energy inside the Cylinder Volume at a CA of 61°

High turbulence regions are located in the center line between the intake valves,
due to the collision of the intake flows between the intake valves (the “short side”) and
across the top of the valves (the “long side”). Figures 38 through 40 show results at
crank angles from 61° to 275°. The results are grouped in up to three figures for each
crank angle with different isosurface levels. The figure furthest to the left always
displays the high tmbulence level, while the figure furthest to the right exposes the lower

level. The center plots show intermediate isosurface levels.

 

 

0 atCA 81°, 9 atCA 81°, 0 atCA 81°,
Slicaatya-thndGSmm Sliceeaty=40and65mm Slicesaty=40md65mm

 

 

eatCA 100°, 0 atCA 100°, OatCA100°,
Slicessty=45and65mm Slicesaty==45md65mm

Sliceeaty=45md65mm

 

 

o atCA129°, e stCA129°,
Slicesaty=15,45md65rnm Slicesaty=15,45and65mm Slicesaty=15,45md65mm
Figure 38: Turbulent Kinetic Energy at the crank angles of 81°, 100° and 129°

0 atCA129°,

 

 

 

 

 

 

 

 

 

 

 

 

 

   

eatCA 146° OatCA 146°

eatCA146°
Slicesaty=10,45,and65mm Slieeeaty= 10, 45,3nd65mm

Slicesaty=10,45,and65mm

 

 

 

 

 

 

 

 

 

OatCA 161° OatCA 161°

GatCA161°
Slicesy-atlo,45,and65mm

Slicesaty=10,45.and65mm Slimaty=10,45,and6$mm

 

 

 

 

 

 

 

Bat CA 179° 9 atCA 179° eatCA 179°
Slicesaty=10.45,and65mm

Slimny=10,45,and65mm Slimaty‘lo,45,and65mm

Figure 39: Turbulent Kinetic Energy at the crank angles of 146°, 161°, and 179°

 

 

 

 

 

 

 

 

 

 

 

 

     

OatCA199" eatCA199" OatCAl99°
Slices a 10, 45, and 65 mm Slices at 10, 45, and 65 mm Slices at 10, 45, and 65 mm

 

 

 

 

 

 

 

 

 

 

 

 

 

     

OatCA240° OatCA240° 9atCA240°
Slices at 10, 45, and 65 mm Slices at 10, 45. and 65 mm Slices at 10, 45, and 65 mm

 

 

 

 

 

 

 

 

 

9 atCA 269°

Figure 40: Turbulent Kinetic Energy at the crank angles of 199°, 240°, and 269°

66

Figure 38 displays, besides the region in the top center of the cylinder, a high
turbulence region located in the bottom center of the cylinder. It is above the piston
surface, where two intake flows collide. They move downward along the cylinder liner,
opposite each other. Their flow patterns can be seen in Figures 13 and 14.

Figure 39 shows, for a crank angle of 161° (center plot), a “bone-shaped”, high
energy region along the z-axis located in the top center of the cylinder. This region is
produced by the collision of the downward oriented incoming air and the flow that moves
upward from the piston surface toward the intake valves. The generating flow pattern can
be seen in Figure 15 for the three center planes 1, 3, and 4. Together with the turbulence
of the two vorticities in the center of the cylinder a unique isosurface “ring” is created for
lower levels of TKE, as it can be seen in the right column of 39. A comparison with
Figure 15 at a crank angle of 168° reveals a correlation between the vortex structures,
mean flow, and TKE. The opening in the center of the turbulence structure (middle
column of Figure 39) is located at the same position of the two vortex structures in Figure
15. The small opening in the “ring” structure in the right column of Figure 39 is located
at the same position where, in Figure 15, the mean flow pattern moves diagonally upward
toward the intake valves.

Shortly after BDC, the TKE continuously decays. Figure 40 shows the dissipation
effect with the highest amount of TKE remaining in a region located in the center of the
cylinder. At a crank angle of 269°, Figure 40 shows a break-up into multiple, small
energy regions.

With the aid of isosurfaces which display regions of equal magnitude TKE, it is

clear that the turbulence levels are in fact dependent on position and were shown to have

67

characteristic geometric shapes. Thus, turbulence levels within an engine cylinder can be

considered to be both non-homogenous and non-isotropic

Turbulent Kinetic Energy within the Vertical planes

The turbulent kinetic energy was computed for the measurement planes.
Calculations of total TKE and TKE per unit mass were done. The results show similar
behavioral trends exhibited by the curves of the volumetric calculations. Given in Figure

41 are TKE per unit mass results.

 

 

    

—a— Plan. 1
35 -* ‘Hfim + Plane 2
‘ ..-, Plane 3
a Us” or Plane 4
3° " Q ”2 i ———P|Inc 5
j “a; ‘ +Ptano 6
25 0 H “:3, ~-Ptanc 7
" .-. "I, >t— Plane 8
5 -—-—- Plane 9

. “ + Plan. 10

 

 

 

 

 

0 e t : ¢
0 60 120 180 240 300
Crank Angle [Dogma]

 

Figure 41: Turbulent Kinetic Energy for all Vertical Planes

Figure 41 shows the calculated data for the ten measured planes. Plane 1 (the

center plane between the intake valves) displays, with its adjacent plane 3 at x = +10 mm,

68

the highest peek turbulent kinetic energy at about 38 J/kg. Plane 2, positioned between
the intake and the exhaust valves at z = 0 and orthogonal to all the other nine planes,
shows a lower turbulent kinetic energy than planes 1, 3, and 4. While planes 5 and 6,
located at x = :1: 20 mm right under each of the intake valves, display very similar
turbulent kinetic energy curves. The adjacent outer plane pairs, which are planes 7 and 8
at :1: 30 mm, and planes 9 and 10 at y i 40 mm, show increasingly different turbulent
kinetic energy curves towards the cylinder liner. The turbulent kinetic energy of plane 10
is about one third higher than that of plane 9 over the first measured 35° crank angles.
The distribution of TKE through these planes during intake indicates that the turbulence
levels are highest in the center region (that is were planes 1, 3, and 4 lay located between
the intake valves). In addition, the magnitudes of TKE through these planes demonstrates
that normalization of TKE with mass gives results that are truly independent of the
number of calculations volumes required in computation. For instance, plane 10, which
contains 45 calculation volumes yields a curve which has roughly the same order of
magnitude as plane 5, which uses has a total of 169 volume elements. Calculations of
total TKE for each of the ten planes shows the effect of the mass as well as turbulence.

Given in Figure 42 is the result of the total TKE computations

69

 

8.00E-04

7.00E-04 --

6.00E-04 .

T

5.00E-04 ‘-

 

4.00E-04 *-

3.00E-04 ‘*

TMMMM

2.005-04 ..

1.00E-04 ‘-

 

 

0.00E-t-00

 

 

Crank Angle [daunaa]

Figure 42: Total Turbulent Kinetic Energy for all Vertical Planes

Figure 42 shows the turbulent kinetic energy in Joules. All the curves start from very low
values and reach their maximum in the region of the largest piston speed. The curves can
be arranged in three groups, according to the different number of measurement points, or
the mass contained within the planes: planes 1 to 4 form one group with about 200 data
points for each plane, planes 5 to 8 another group with about 150 data points each, and
planes 9 and 10 form the last youp with 45 data points each. All planes on the positive
side ofthe x-axis, which are plane 3 at x = +10 mm, 6 at x = +20 mm, 7 at x = +30 mm,

and 10 at x = +40 mm, show a higher turbulent kinetic energy than their counterparts on

70

the negative side of the x-axis. This is consistent with the calculation of tumble and
circulation, where the planes located on the positive part of the x-axis show higher values.

The same production and dissipation effects seen in the behavior of the volumetric
TKE curves are also observed in Figure 41 and 42. Comparing the results of the different
types of calculations shows that using TKE per unit mass delivers more information on
the turbulence accept of the flow then total TKE computations, when analyzing the
vertical planes and the entire cylinder volume. Furthermore, TKE per unit mass analysis
provides a method which solely looks at the turbulence levels within the regions of

interest.

Turbulence Kinetic Energy in Horizontal Slices

TKE calculations in horizontal planes or slices located at constant vertical
locations, provides a unique means of examining turbulence levels within an engine
cylinder. Since each of the slices are located at constant vertical locations, a length
dependence can be acquired. One appropriate choice for the lengths are the distances
between the BDC position of the piston and the slices. This length dependence is seen in
the results of the TKE calculations for the twelve horizontal slices. Shown in Figure 43

are the TKE per unit mass results.

71

 

 

30
—— Yin-65
--— Y8460
25 ., -—YB+55

8

 

 

 

‘
O
L
‘T’

(I

WMMparwhapla‘

 

 

 

 

0 so 120 130 240 300
Crank Angle More“)

Figure 43: Turbulent Kinetic Energy in all the Horizontal Slices

The distance between the BDC piston position and each of the slices is shown in Figure
43. The location of Y=+65 (mm) is furthest from the BDC piston position and closest to
the top of the combustion chamber. Values of TKE in each slice become visible as soon
as the piston clears each horizontal slice. The effect of the measurements as the piston
clears each slice can be seen in the behavior of the tail end of each of the curves in Figure
43 (as well as Figure 44). The behavior of the curves during these periods are not a true
representation of the actual turbulence levels within each slice. The tail section shows a
quick drop off. The drop off is due to the lack of “reliable” data caused by the piston
cutting off the LDV signals as it passes through each of the horizontal planes. Besides

this effect, the behavior of the curves demonstrate that a consistent trend, during the

72

beginning and end of the intake and compressive processes, respectively, exists. The
higher the slice location (relative to BDC piston position), the greater the turbulence
level. The same statement can be made for the curves resulting from the total TKE

calculations.

0.(X)12

 

 

 

--—Y=+65

 

 

 

 

 

 

 

mmwml

Figure 44. Total Turbulent Kinetic Energy in the Horizontal Slices

The trend of the curves at the end of the compression stroke is of great interest.
The effect of mass in the calculations show the character of turbulence and its length
dependence toward the end of the compression stroke. As the slices traverse towards the
top of the cylinder an increase in the TKE level is evident. It is the character of

turbulence at the end of the compression process that is most important, as this is what is

73

believed to control the fuel-air mixing and burning rates. Heywood writes, “...During
the

compression stroke, and also during combustion while the cylinder pressure continues to
rise, the unburned mixture is compressed. Turbulent flow properties are changed
significantly by the large and rapidly imposed distortions that result fiom this
compression. Such distortions, in the absence of dissipation, would conserve the angular
momentum of the flow: rapid compression would lead to an increase in vorticity and
turbulence intensity...” [1 1].

Hence, the compression should lead to an increase in turbulence intensities toward the top
of the cylinder. The increase of the peak TKE at the tail section shows a length
dependence (higher the location of the slice the higher the TKE levels). The amount of
TKE contained within a lower slice is seen to be transferred by the ascending piston top
to the slice above it, non-linearly (i.e., it is not a pure superposition of TKE fiom the
lower slice from the lower to upper slice). However, a linear extrapolation from the peak
of the tail section of each of the curves, out to the spark ignition crank angle, may give an

estimate of the minimum TKE level during the onset of combustion (since the

compression process most likely produces an exponential increase in TKE).

74

Chapter 6

TUMBLE AND SWIRL

Tumble and swirl are defined as organized rotations of the charge about specific
cylinder axes. Tumble corresponds to rotation about either of the two axes in the plane of
the piston top. Whereas, swirl is the rotation about the axisymmetric cylinder axis.
Tumble and swirl are created by bringing intake flow into the cylinder with an initial
angular momentum [11]. The particular rotational motion set up during intake is
sometimes used to promote rapid mixing between the air charge and the injected fuel.
Additionally, swirl is used to speed up the combustion process in 8.1. engines [11].
Although, tumble and swirl are desired in many instances, the results obtained for this
four-valve intake design showed little of these types of motions exist toward the end of
the compression stroke. Tumble ratios were calculated for vertical planes in both the x
and 2 directions for volumes spanned to distances that were equally spaced to adjacent
planar sections. Swirl ratios were calculated for horizontal slices. Also, the points within
the entire measured volume were used to calculate both ratios. In comparisons of all the
results obtained, the tumble ratio about the x-axis displayed the largest degree of

rotational motion.

75
Definition of Tumble and Swirl Ratios

Tumble and swirl ratios were developed to express in-cylinder air motion by a
normalized number. Two origins were used to place the axis of rotation in certain areas.
Given in Figure 45 are the two cylinder coordinate systems used for these calculations.
One shows a cylinder coordinate system utilizing a moving origin about the instantaneous

center of volume, while the other displays a fixed origin located at TDC.

 

 

 

 

 

 

 

 

Cylinder Coordinate System Cylinder Coordinate System
with Moving Origin with Fixed Origin at TDC

Figure 45: LDV Cylinder Coordinate Systems

The position vector is a vector fi'om the origin to the measured location given by

F=xi+yj+zk

(6.1)

76

and

t7 = ui + v} + w];
(6.2)

gives the measured velocity vector. The rotational speed of the fluid around the axis

located at the particular origin is

60:wa +coyj+wzk

(6.3)
Equations (6.1), (6.2) and (6.3) are related by
§=éxF
(6.4)
The angular momentum per unit mass can be written as
fi=rxv=;x(axr)
(6.5)

Expanding the triple vector product of (6.5) gives the angular momentum around the

principal axes
f7 = (yzwx +zzwx)f+(x2wy +zzwy)f+(xzcoz +y2w,)l;
(6.6)
or
1‘7=(y2 +zz)wxf+(x2 +22)6oyj‘+(x2 +y2)a),l;
(6.7)
Since,
9 = Inwxf + {Wary} + [Raj
(6.8)

77

In =(y2 +22) , I», =(x2 +22) and 1:2 =(x2 +y2)
(6.9)

From the summation of the angular momentum in each measured volume, the tumble and
swirl ratios can be calculated. Their ratios are defined by the angular momentum of the
actual air motion divided by the product of the crankshaft rational speed and the moment
of inertia of a solid body of air centered along the particular axis of interest. The tumble

ratio about the x- and z- axes are given by [1]

gm).

 

 

 

Tx .-...- N
121(1)“), . we
(6.10)
N
Z (11,),
T2 = N i=1
)3 (1-), a».
(6.11)
and the swirl ratio is
N
Z (Hy).
Sy = Nl=l
2:1(11’3’), (00
(6.12)

N is the number of measured data points per crank angle. The tumble and swirl ratios are

normalized with the aid of the crankshaft angular velocity, (no; where (00 = 2691 , and n is

600 rpm.

 

78

Tumble Ratio Results for Vertical Planes

The tumble ratios around the x-axis (parallel to the cam shafi, donated by Tx) and the z-
axis (between the two intake valves, donated by T2), were calculated with the measured
velocities inside their planes. The location of the planes can be seen in Figure 5. The
measured velocities in plane 1, and 3 through 10, were used to calculate tumble about the
x-axis, while the velocities in plane 2 and those parallel to plane 2 were used to calculate
tumble about the z—axis. The tumble ratios were examined around a moving and a fixed

origin. The results of the tumble ratios for planes about the x—axis are given in Figure 46.

79

auto 85 an as: as 258“ 3.2 6385 “2. an
700.363 00.9.3 :55
9a 2: our 8

 

_ . p 1— _ - f .
_ . . J

 

 

 

 

 

 

 

t. {nugget 93:11:17

0

80

During approximately the first 135 crank degrees all the curves display a range of scatter.
After 135°, the behavior of all the curves becomes nearly constant. An explanation of
this trend is that the in-cylinder flows become more organized and less chaotic after a
crank angle of 135°. After this period the curves can be broken up into two distinct
groups. The first group consist of the inner five planes (planes 1, 3, 4, 5, and 6), while
the second is the outer four planes (planes 7, 8, 9 and 10). The inner group demonstrates
a positive ratio about the x-axis, while the outer group shows a negative ratio. Soon after
BDC, planes 7, 8, and 9 of the outer group, displays an increasing tumble ratio towards
the values held by the inner group. The behavior illustrated by Plane 10 also tends to
increase towards the values approached by the remaining planes, although it does so later
in compression. After crank angle 270°, all the planes show a positive tumble ratio of
about 0.25 illustrating that a rotational motion about the x-axis exist and should continue
if angular momentum is to be conserved.

The results of the tumble ratios utilizing the calculations based on a moving origin
for these planes gives a better description of the actual angular motion which takes place
within the cylinder. Since the origin is located at the instantaneous center of the cylinder
volume, the calculation can use the angular velocities completely around the axis as
opposed to those underneath it as the fixed origin based calculations do. The moving

origin calculation yields higher values during the intake stroke, as shown in Figure 47.

£95 352 he 82.x 2: e553 35 63:59 ”a. as»;

T8500“— aa.u:< .220

8a SN 09. our 8

~ p ~ . p _ _

 

< _ _ u

81

 

3th
mxhd
kao
hem—.0
wxho
mxkn
kad
mxho

 

 

 

 

vibr—

 

 

 

 

om “WM.

82

The curves of Figure 47 (as well as those of Figure 46) are broken up into plane pairs
noted by the color scheme (plane pairs are those who are symmetrically located about the
x-axis with one another; i.e., planes 9 and 10 located at x=~45 and x=+45, respectively,
and are plotted in green). Again, a degree of scatter exists in these curves, much like
those calculated using the stationary origin, and smoothes out after crank angle 135°.
Throughout most of the first 135° crank degrees the outer six planes (planes 5, 6, 7, 8, 9,
and 10) all show a positive tumble ratios. The remaining inner three planes (planes 1, 3,
and 4) display a different behavior during this period. Planes l and 4 give negative
tumble ratios, while plane 3 starts out negative followed by an increase which eventually
yields a positive tumble ratio after crank angle of 70°. Notice that plane 1 displays the
largest negative tumble ratio which peeks shortly before the maximum piston speed.
After crank angle of 135° the tumble ratios in all the planes, except planes 9 and 10,
approach and oscillate about the zero line.

Symmetric tendencies are very evident in the angular motion as demonstrated by
the tumble ratios in the plane pairs. Plane pairs 5 and 6 display the highest amount of
symmetry, followed by plane pair 7 and 8. Similarly, planes 9 and 10 behave much the
same. An interesting observation is seen in the behavior of plane pairs 3 and 4. The
behavior of their curves illustrates that they are experiencing opposite motions unlike the
other plane pairs, which tend to mirror one another. The cause for this asymmetric
behavior may be the effect of the small offset inside the intake runner, as shown in
APPENDIX 1. The left intake valve (see APPENDIX 1) experiences a higher mass flow
rate than the right valve. This larger flow rate may allow the flow to situate itself more

towards the long side of the left valve than the corresponding long side of the right valve.

83

Notice in Figure 47, plane 3 experience a positive tumble ratio, while planes 1 and 4
experience negative tumble ratios. This may be the effect of the higher flow passing over
the left valve (towards the long side) which causes plane 3 to have a positive tumble ratio
since it is located close to the left valve. This could be a legitimate explanation of the
asymmetric behavior of planes 3 and 4. Given in the next two figures, are the results of
the tumble ratio calculations for the planes around the z—axis. Figure 48 displays the

results using the fixed origin, while Figure 49 gives the moving origin results.

84

:35 85m com mifN 2: @983 2am 0383. ”we 233m
neon—ace no.9? xcflo
ova 09 on?

_ . r . _ _

 

 

a 1 .

om+©~ o flmuflm 0
2+9 n

N u» o
o w©~ o
8.9 <
8.9 o
are? n

 

 

 

 

11 m. Fl

01"! «WM.

1 m6

 

 

85

ice was: 5 “can 65 e393 82 6385 ”a. can

Henchman... 00.93 ...-20

com ovw one our om

~ _ _r . _ _ bl-

 

. u _

o1©~ o

8+©~ a
8+9 .9
o tan a

N a. x
9.9 o
8.9 <
om-©~ 0
9:9 0

 

 

 

 

 

 

013.8 .Iqmnl

 

1L-

..L 1 1 m ,1:- ;(.2 :1: 26:30

i

86

The tumble ratios about the z-axis (bisector of the intake values), show
considerable scatter during intake and show little angular motion during most of the
remaining crank angles. The scattered region, as seen in Figures 48 and 49, occurs till
about a crank angle of 90°, the point of maximum piston velocity. After this period the
curves become smoother and less chaotic much like the tumble ratios about the x-axis
(although it occurs early for the ratios about the z-axis). The magnitude of the ratios after
90° are quite small in both cases, demonstrating that there is very little rotational motion
about the z-axis. This is the effect of the four valve and intake design. Symmetric
tendencies of the flow about the x-axis produce this effect.

Comparing the results of the ratios utilizing both origins shows that the moving
origin produces higher values than calculations based on a stationary origin. During both
the intake and compression strokes the twnble ratio calculations using the moving origin
demonstrates lager magnitudes. This is quite evident in the behavior of the plane located
at z=-40 (mm). This plane (as well as plane 9) displays increasing rotational motion
particularly towards the end of combustion as shown in Figure 49. Hence, this motion
orientated in the outer sides of the z-axis displays the a small degree of rotation which

may continue till the onset of spark ignition.

Swirl Ratio Results for Horizontal Slices

The swirl ratios calculated about the y-axis via both origins, produced smaller

magnitudes in comparison to the tumble ratios about the x- and z- axes. The relatively

small magnitudes of the swirl ratios can be attributed to the symmetry within the four

87

valve engine. The results of the swirl ratios for slices based on a stationary origin are

given in Figure 50.

88

3N

8% 3:88: 8 83 25m "863mm

.8289 292 .226
on.

8?

 

 

 

 

 

m6.

 

@83th
5583....
E5334
EEomao
5893.4
5533.0
2583

$2.83.

56330
EEoN>h<
8E9»;
5.593..”—

 

 

-- Nd

 

 

m6

was was

89

The curves representing each of the twelve slices are paired by color (i.e., the top two
slice at Y=65 and Y=60 (mm) are in red). This demonstrates how the flow within
horizontal areas adjacent to one another evolve during intake. Matching adjacent pairs
one can see how the swirl ratio tends to increase towards negative values during intake
and peeks about the Y=45(mm) slice. After the peek the swirl ratios begin to approach
zero values. The behavior of the changes in each of the curves can be used as an
understanding of how the flow develops. From the behavior of these curves, the flow can
be seen to develop in a fashion as the previous slice did with a bit smaller rotational
magnitude. This is particularly evident when examining the slices located between Y=30
(mm) and Y=55 (mm) during the intake stroke. Although some swirl motion is evident
by these curves the actual amount of swirl is quite small. Note the largest magnitude in
all the curves of Figure 50 is about -0.3, much less then some of the values produced by
tumble ratios about the x axis (recall that during intake Tx could be in the range of 1.0 to
2.0). Given in chapter 7, and in Figures 62 - 63, are top views of the flow field though
slices which track the descending and ascending piston top. These views demonstrate the
symmetric tendencies in the flow fields which do not permit large swirling motion to exist

as evident in the small swirl ratio values obtained.

Tumble and Swirl Ratio Results for the Effective Measurement Volume

Tumble and swirl ratios utilizing the cylinders volume are the quantities that such

calculation were intended for. Typically, the swirl ratio is a quantity that is measured on

flow benches. However, the experimental techniques which is commonly implemented in

90

swirl measurements are not as accurate as the calculation of the swirl ratio using a large
number of 3D LDV in a motored piston cylinder assembly. Heywood [11] explains a
common approach in the measurement of swirl;

“...in rig tests the flow and the valve open area are fixed and the angular momentum
passes down the cylinder continuously, in the engine intake process the momentum
produced under corresponding conditions of the flow and valve lift remains in the
cylinder. Steady-state impulse torque-meter flow rig data can be used to estimate engine
swirl in the following manner. . .”.

The dynamic influence of the time dependent intake flows are lost in this method, since
the measurements occur during in a steady state condition. Furthermore, swirl
measurements cannot be accomplished during the compression stroke. The results of the
calculations of the swirl and tumble ratios using all available 1,538 measurement points
are given in Figures 51 and 52. Figure 51 displays the results of using a fixed origin,
while Figure 52 displays the results based on an instantaneous moving origin These

results provide a genuine and accurate quantification of the net angular motion exhibited

by the flows within the cylinder.

91

euro 85 as 63 63:5. 25.5 an 28E

.8282 6.92 .296
8» 2a 8. our

 

 

 

 

 

 

(march; 5911c:

92

auto 9:62 as gas 6383. an 8am

"39509 0.92 3:20
own 8F 89

 

 

_ ~ _
«

 

 

 

-- no.

-- «o.

-1 N6

-. v.0

-- 9o

-- 9°

 

 

01138 W].

)nznngc 512m;

93

The volumetric tumble and swirl ratios results, utilizing both origins, demonstrate
that there is not a momentous amount of bulk angular motion contained within the
cylinder during intake and compression. The results of each ratio are explained below.

For the moving origin calculations, tumble about the z-axis, Tz, appears to be the
weakest quantity throughout most of the intake and compression stroke. T2 is most active
early during intake during a crank angle range of 55° through 90° . After the point of
maximum piston speed, the angular motion about the z-axis remains almost constant as
characterized by a small positive slope leading it towards a zero value. When evaluating
the stationary origin results, Tz shows a very similar behavior. This ratio is basically
zero, indicating no angular motion about the z-axis.

The motion around the x-axis, as suggested by the moving origin results of Tx,
displays a degree of unorganized motion until a crank angle of 135°. This is evidenced
by the amount of scatter that exists in the data during this period. After crank angle 135°
the curve of Tx remains constant until approximately 210°. At this time, Tx dips and
oscillates about the zero line until approximately 270°. The curve then begins increasing
toward negative values. In comparison with the stationary origin curve, at approximately
210°, the curve begins increasing towards positive values, and does not dip and head
towards the zero line, as does the corresponding moving origin curve. Furthermore, at
approximately 27 0°, an increase toward higher positive values occurs. The different
behavior exhibited by these curves is the effect of the center of the angular motion
occurring in the top half of the cylinder, in the region of the stationary origin. The
stationary origin is located at the top of the combustion chamber at Y=80 (mm) ,whereas,

the moving origin situates itself in the center of the remaining volume depending on the

94

crank position. Its placement varies between 0 and 80 (mm). Since the stationary origin
is located in the highest region of the cylinder, it observes the rotational motion that is
occurring, whereas the moving origin does not.

The swirl motion within the cylinder volume is largest during intake and decreases
towards a zero value through the remaining crank angles. The results of the swirl
calculation is shown in Figures 51 and 52. The results are the same for both origins,
since the axis is centered vertically in the center of the cylinder. Using a stationary and
moving origin in the placement of the axis of revolution is irrelevant in this case. Placing
the axis at the top, or in the center of the stroke, does not change its direction. The curve
displays a region of scatter, again indicating unorganized motion, till about 135°. During
the measured crank angles, the swirl holds negative values, and displays a positive slope
which leads it near the zero line. The range of ratio magnitudes held by this curve
demonstrates that the swirl motion is small.

Measurements of angular motion, such as tumble and swirl ratios, provide some
insight into the motion of fluid within a engine cylinder, however they do not render a
good representation of the entire fluid activity. Tumble and swirl ratio measurements
were intended to give an understanding of the mixing characteristic exhibit by the flow.
The results of the tumble and swirl calculations indicate that little bulk motion exists
within the cylinder through most of the intake and compression stroke. However, a
conclusion based on these results may be misleading; i.e., that the flow features do not
promote rapid mixing. This conclusion is not warranted. In fact, cancellation effects
produced by the symmetric tendencies exhibited by the flow fields do not allow strong

angular motions to exist within the cylinder. For example, the motion of the two counter

9S

rotating vortex structures, observed during intake, are canceled out in the tumble ratio
calculations about the x-axis (using either origin). Their motions contribute to the mixing
ability of the flow, yet they are not represented in the magnitude of the tumble calculation.
Hence, the use of these quantities in interpreting mixing characteristics should be done
when examining designs which are intended to produce one type of angular motion (such
as, a swirl port design used in some diesels). Furthermore, the use of these quantities
when analyzing the mixing ability exhibit by the in-cylinder flow motion should be
carefully considered when using a four valve intake port design such as the one used in

this study.

96

Chapter 7

CIRCULATION

Circulation calculations provide a means of quantitatively characterizing large and
small cell motions of in-cylinder flow. Total (TOT) and average (AVG) values of
circulation can be used to analyze the bulk or net motion of the fluid around a defined
plane or contour constructed by a grid of cells. Root Mean Squared (RMS) circulation
calculation is a parameter that quantifies the small cell activity enclosed within a defined
area or an enclosed contour. All of these quantities were calculated for this study around

contours enclosing vertical planes and horizontal slices within the cylinder.

Circulation Calculation Techniques for Slices and Planes

Circulation is defined by fluid flowing about an enclosed contour which is stated
mathematically by the following definition
I‘ =17 0 d5
(7- 1)
Vorticity is defined as

&=Vx7
(7.2)

Circulation is related to the flux of vorticity via Stokes Theorem

97
1‘: (11700? = [(17 x I7)-adA = [a ofidA
A A

(7.3)

The vorticity flux produced by a velocity field through an area defined by closed contour
is used to calculate circulation values.

Calculation grids were set up to evaluate the vorticity flux though the vertical
planes and the horizontal slices so that circulation values could be produced. These grids
allow calculations of circulation around each of the cells which construct it (which is
equal to the vorticity flux of the fluid passing through them). Summing the cells
contained within the area yields values which are considered to be representative of the
TOT and AVG circulation about the analyzed area. Calculation of the RMS circulation
for each of the cells and then summing, yields a quantity that is considered representative
of the small cell activity. This idea is based on the fact that RMS values strip away the
cancellation effects created by symmetric flow patterns. Figure 54 illustrates the grid
used for calculating circulations for horizontal slices. Similar grids were also constructed

for each of the vertical planes.

98

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

X (mm)
40 30 20 10 0 10 20 30 40
4° - : Measurement
-35 P08111008
I
.30 O C
a x : Center of Circulation
'25 , elements used for
-20 I. calculations.
-15
I
-10
I
.5 a
Z ° I
(ml) 5 .
10 a
15 a
2° ' The shaded area surrounding
25 I each measurement position
I indicates the calculated
3" vorticity flux in that region.
35 The circulation of the
‘0 element centered at x is the
. . . . ' sum of the four vorticity
C 1rculation Calculatlon Grld for Horizontal Slices regions enclosed within the
inner rectangle.

Figure 54: Example of a calculation grid used for circulation analysis.

The circulation of single element (indicated by and ‘x’ in Figure 54) located on a
horizontal slice is calculated by,

4 A.
r, z 2W) (a) H

1=1 & j ‘3‘ 1 4 k

(7.4)

where the total circulation is:

gr.=%Z§[(%J,-(%ll

(7.5)

 

99

The k index ranges from 1 to 92 that corresponded to the total number of cells or

elements in a slice. For an element on plane 1, as well as planes 3 through 10, we have

vale-1,4611%)

(7.6)
with the total circulation given by
N A N 4 a” a
- ... r ....L K...) H]
p11,310 2:17; 4;”; 4’; 531,
(7.7)

Here 11: index will differ, depending on which plane is being evaluated, and N is the total
number of cells in an appropriate calculation. The net circulation for the elements located

on plane 2, with 198 calculation centers, is given by;

veneer)

(7.8)
which yields a net circulation of
19s A 198 4
I‘m2 = Zr, _ .../.2 [(2%) {9}.) ]
k=1 4k=1j=l dc j é’, k
(7.9)

General expressions used in calculating AVG, RMS, and ABS of circulation in

the slices and planes are given by,

and

100

 

 

_ Frow.,(plaw.nr..xlrcc)
ravg - n
(7.10)
rmsz
(7.11)
n
E irkl
I°abs _ n
(7.12)

respectively. The partial spatial derivative seen in these expressions were found using 3

and 5 point finite difference approximations [9, 10]. The FORTRAN code written to

accomplish these calculations is given in APPENDIX 11.

101

Circulation Results and Discussion

Symmetry is a characteristic observed throughout the analysis of all the circulation
data produced. An examination of the total circulation vs. crank angle curves for equally
distant y-z planes from plane 1, shows the symmetry that occurs in these regions. Shown

below in Figure 55 is the total circulation for planes 9 and 10.

0.8

 

0.6 <~

0.4 4»

 

 

 

 

 

 

 

 

 

 

0 -
1) so
a —0.2 r
P V.. . .
-o.4~» <93“? 0 01318199
Q9311” an10
43.61 :5!
5’6
-o.e 1:1

 

Crank Angle [Rona]

Figure 55: Circulation in Planes 9 and 10

Similarities to the curves displayed in Figure 55 are also evident in other pairs of
symmetric planes. These are shown in Figure 56 for planes 7 and 8 as well as planes 5

and 6.

 

102

 

0.7 ~-

.0
ca

.0
..a

 

 

 

-0.1 (l 50

—0.3 ..

 

To“ m W30)

 

.o_7 .. 131318107

 

 

 

 

 

 

-0.5 -~ [if
0' 01918108
31

Crank Ami—[Dacron]

 

 

 

—o.2 ._

Tullm W)
o
a

-o.4 4-

 

-0.6 «»

 

 

 

 

 

 

 

-0.8
Crank Angle [Danna]

Figure 56: Total Circulation in planes 5, 6, 7, and 8.

Each pair of curves in these figures suggests that these plane pairs are
experiencing the same net bulk motion during the intake and compression process. The
similarity tends to be highest for planes 5 and 6, which are located directly beneath the
center of the intake valves. It tends to decrease slightly for plane pairs that lie from the

center of the intake valves to the outer walls of the cylinder. Additionally, it also

103

decreases for the pairs between the two intake valves (planes 3 & 4). This trend is shown
in Figure 57 by the TOT circulation curves of planes 1, 3, and 4.

A

 

 
   

 

 

 

 

 

 

 

0.30 $
0 ‘1 ._ A
g M -- . ‘ a a
11 - we .
‘ ‘ " : a ' r J,
. 3' . o 3&0
-o.1 , , .. é’ o
% 93,8 9 o o
3 ° 8198 9
(I a
a)?
a -0.3 40‘Q A
1" o
4) o
0.5” A o Olee1
Q30 0 ale93
0 0° oPlane4
-0.7
CMIWHMMI

Figure 57: Total circulation in planes 1, 3, and 4

Total circulation results for planes 1, 3, and 4 are shown in Figure 4. The total circulation
is fairly uniform in the crank angle range between 130° to 230°. The total circulation in
planes 1 and 3 is about 0.1 m2/s , while in plane 4 it is about -0.5 m2/s. The circulation
direction of plane 1 and 3 is opposite of that on plane 4 after crank angle 130°. The
velocity vector flow fields in plane pairs at crank angle 142° are shown in Figure 58.
These flow fields show the similarities that are found in the TOT circulation calculation
curves, specifically, the almost identical nature of the curves exhibited by plane pairs.
These plane pairs are planes which are equally distant from the z-y plane and parallel to
plane 1. Examining the flow fields in Figure 58 one will note that the flow fields are not

identical, nor are they exact mirror images of one another, but do have distinct

 

104

similarities. Additionally, we note that planes 5 and 6 seem to exhibit the highest degree

of similarity with one another.

105

 

 

 

3; \Tx}

1 -.
/x, \\ VV

/

Planes 8 and 9

PlanesSand6

\

 

 

 

       

   

L

  

 

     

 

A
1 alllllI

.\ /
.w . - w.
‘ Deng ,
5...... ,
‘ .lrm .. 7 v ..
,. .I \.

  

4

 

  

I .1 4
, wit/Irv . . .3 t. V
. 0 1

\xNfir. \

Planes 3 and 4

  

Planes 7 and 8

    

 

 

 

Figure 58: Flow fields in Plane Pairs. Crank Angle 142°, Vector Scale=2.0

106

Circulation calculations for horizontal slices (x-z planes) also exhibit some amount of
similar behavior. Shown in Figures 59 through 60 are the AVG circulations for pairs of
slices next to each other. The curves in the following figures demonstrate how the
inducted and then compressed fluid rotates about the y-axis. The trend of each of these
curves is that the rotational direction quickly increases and then decreases followed by a
change in rotational direction from which it then levels out and ceases to rotate further

into the compression stroke.

107

 

 

 

 

E 41.002 7 0
Est:

 

 

0Y=55
DY=50

 

Crank Angle [Deana]

Figure 59: Average Circulation for Slices at Y=65, 60, 55, and 50 mm.

 

 

'h
u'l

 

108

0.002

 

0.0015 ‘-
0.001 --

 

 

 

 

cm!“ "I. W!

0.0025

 

0.002 --
0.0015 --
0.001 4»
.0005 4
o _

0 50

‘ -0.001 -~

 

0.0015 «- '3
DU oY=35
0002 J— 830 DY=30

0.0025

 

 

 

Crank Angie [Degrees]

Figure 60: Average Circulation in Slices at Y=45, 40, 35, and 30 (mm).

 

 

40.001 <-

 

-0.0015

o
o 1 i
100 Elisa '3 £01: 250 300

 

o? :1 oY=25

0° [5) Cl Y=20

 

 

 

 

 

cnnk Anet- MM]

 

 

 

00018

100 250 300

 

 

 

Crank Angle [Dachau

Figure 61: Average Circulation in Slices at Y=25, 20, 15, and 10 (mm).

110

Consistent trends in the curves for each of the slices, with respect to one another,
show a level of confidence or certainty in the calculation techniques implemented.
Variations fiom one slice to another change slightly and in a smooth sense.

The largest circulation occurs primarily during intake in the upper regions of the
cylinder (between y=30 and y=65). A noticeable trend during intake is that the
circulation peak occurs at about the y=65 slice and drops as the slices descend. Figures
59 - 61 show how the circulation increases and then decreases in the slices during
approximately the same crank interval during intake. Also, a continuous shift of the
maximum circulation in each slice toward increasing crank angles, demonstrates how the
flow develops during the intake stroke. The flow can be thought to be developing in the
same way as it does previously in the slice that is located directly above it with a bit
smaller rotational magnitude. It begins swirling in a conical fashion and then just as
quickly begins to decrease until it turns direction and levels out. The change in direction
may be attributed to the dissipation of the secondary vortex structure observed in the flow
animations and shown in chapter 4.

A comparison of the circulation values with the velocity vector plots reveals that
the overall circulation around these slices is small. Symmetric flow patterns about the y-
axis produce a cancellation effect. This suggests that the characteristics of the curves in
Figures 59 - 61 are basically due to consistent, small variations in the slightly swirling
flows around each slice. Hence, the range of circulation values that these curves fall
within may be considered negligible in a circulation sense since they are not observed at
all in the velocity fields that develop in the cylinder shown in this study. To further prove

this point the velocity vector flow fields in the slices which track or follow the piston

111

head during intake and compression strokes were plotted. These plots demonstrate the
cancellation effects which produced the ‘small’ scale circulations in the slices. These
plots were chosen since they correspond to the higher circulation values found in the
calculations during intake. In particular, each plot represents the data used to calculate
the point near the fust large peek in Figures 59 - 61 In addition, the compression plots
represent the data used to calculate near the last point of the curves seen in Figures59 -

61. The plots are given in Figures 62 - 63.

 

 

 

 

       

   
   

Slice Y=35mm, Crank Angle=96°
l

ill /

Slice Y=20nnn, Crank Angle=122° Slice Y=15mm, Crank Angle=l3l° Slice Y=10mm, Crank Angle=l42°
Figure 62: Velocity Field in the Slice that Follows the Piston Surface During Intake.

Vector Scale=1.5

 

 

 

Figure 63: Velocity Field in the Slice which Track the Piston Surface During

Compression. Vector Scale=2.0

114

Comparing the RMS values, from all ten planes, gives a volumetric perspective of
the flow activity in terms of circulation in the y-z direction. RMS circulation quantities,

calculated in all ten planes, are given below in Figure 64.

 

 

 

 

 

0.005 --

 

 

 

 

CHM! ~19li
Figure 64: RMS Circulation for all of the Planes

The trend of the curve in Figure 64 illustrates the amount of fluid activity in each plane.
RMS calculations strip away the cancellation effects, thus leaving a curve that is
representative of the amount of activity in the planes. The curve shows an exponentially
decaying activity until a crank angle of about 250°, where it then begins exponentially
increasing. The overall RMS circulation, evaluated in the planes, shows that similar
degrees of small complex motions are exhibited in each of these regions. (The
complexity of motion is basically originated from the vorticity flux passing through each
of the small cells which construct the plane.) The only noticeable curve that deviates
from the norm comes from plane 9. This is a result of the low average net flow that plane

9 experiences as confirmed by the SGI flow visualizations. Plane 2 is displayed in

115

Figure 64 which shows approximately the same amount of RMS circulation as the others.
An interesting conclusion fiom this figure can be drawn. There are two locations in
Figure 64 where the RMS values of all the plane are approximately the same. During the
crank interval of 110° to 115°, where a RMS value of 0.01252t0.0025 (mz/s) is found.
Between the crank interval of 230° to 240°, the RMS values of all the planes is 0.005
$0.001 (mz/s). This shows that during these periods the RMS circulation in the y-z
direction is consistent though the cylinder volume.

Next the RMS values of circulation though the slices is considered. Displayed in

Figure 65 is the RMS values for each of slice.

0.025

 

0.015 <—

0.01 '-

"MW!

   

 

 

 

0.005 <- “3m
.- «3% A
Ewen-"'-
0 i... u I i ‘ 'm
0 50 150 200 250 300
Crank Highwaymen)

Figure 65: RMS Circulation for all of the Slices

Two distinct patterns emerge when examining Figure 65. The behaviors of the upper six
slices and the lower slices each follow the same trend. Shown in Figures 66 and 67 are

the RMS values for these sets of slices.

116

0.025

 

0.015 --

0.01 ~-

“WW2”

 

 

 

 

o 50 160 150 260 2.4.0
Crank Anglo [Dean-q
Figure 66: RMS Circulation for the Upper Slices

0.016

 

0.014 +-

0.012 «-

 

°.008 1- .‘r 1.5-0‘; .

0.008 1-

WWW]
gt.
a 3‘
2:12.;
I
_>
P

0.004 -. or . a

0.002 --

 

 

0 so 100 130 260 250 300
Crank Anglo W]
Figure 67: RMS Circulation for the Lower Slices

 

117

Cancellation effects of the symmetric flow patterns are stripped away in the RMS
calculations. An apparent effect of this is the relative magnitude of these values in
comparison to the AVG calculations. The difi‘erence gives an order of magnitude of
about ten.

The RMS values shown in Figures 65 - 67 show that higher circulation activity is
in the slices located in the center of the cylinder, and generally decreases in the slices that
increase upward and downward from the center region (slices y==35, y=40, & y=45). This
behavior can be related to the driving piston motion, which is maximum at the center of
the while slowest at the upper and lower regions.

Much like the RMS in the planes, there are two locations where the RMS values,
within all the slices, are approximately the same. Within the crank interval of 120°
through 130°, a RMS value of 0.0011 i0.0025 (mz/s) is found. During angles 230°
through 240°, RMS values of all the existing slices is at 0.004 i 0.001 (mz/s). In
comparison to the similar regions identified in the planes, the second crank interval (230-
240°), as well as the magnitudes of the RMS (0.004 i0.001 (mZ/s) ), match. The
identifiable match of the RMS values, as well as the crank intervals for all the slices and
planes, suggest that the overall activity of the fluid contained within the volume is
spatially homogeneous. In other words, the net small cell activity is distributed evenly
throughout the volume. Physically, this corresponds to a range of crank angles just prior
to intake valve closure occurring at 240°.

In comparing the results of RMS, AVG, and TOT circulations obtained within the
planes to those of the slices we can summarize the following results. During intake the

net bulk motion moves along the y-z directions as indicated by the TOT circulation plots.

118

Additionally, the larger amount of small cell activity is shown to be orientated within the
same planes which span the y-z directions. During compression, the net large scale
circulation type motion is also contained primarily in these planes. Whereas, the RMS
values contained within both planes and slices show that the net small cell activity in the
compression stroke is of the same order of magnitude and is uniformly distributed

between the crank interval of 230° to 240°.

 

119

SUMMARY AND CONCLUSIONS

Several methods of quantifying in-cylinder flows have been investigated. They
include the evaluation of ensemble—averaged mean velocity flow patterns, turbulent
kinetic energy, tumble and swirl ratio, and circulation. Summary statements regarding
these various methods are given. Each method provides a different insight into the flow
within the cylinder. A comparison of these results with firing engine data will reveal the

importance of each.

Concluding Statements

The results of this study support the following conclusions:

1) The volumetric flow characters exhibit a large degree of symmetry about the z-axis

for the engine studied in this report.

2) During the intake stroke, the largest induction flows are first seen to favor the intake
side of the cylinder and then the exhaust side towards BDC.

3) Two counter—rotating 3D vortical structures are revealed by the dynamics of the
symmetric induction flows. Both non-connected structures have “arch-shaped”
geometries and exhibit ends which point opposite one another.

4) The symmetric tendencies exhibited by the flow about the z-axis continue late into the

compression stroke.

120

5) During compression, regions of the highest velocity magnitudes are located just above
the ascending piston head and symmetrically along the outer intake sides of the cylinder.
The weakest flows were organized in the center of the changing cylinder volume,
producing a weak rotational pattern relative to those produced during induction.

6) The flow analysis showed that the dominating intake and compression flows were all
tumble like, exhibiting very little swirl motion.

7) The slightly offset interior of the intake runner produced higher in-cylinder velocities
on the +x-axis side of the cylinder which caused some asymmetric flow patterns about the
z-axis.

8) Turbulent Kinetic Energy (TKE) isosurfaces depict comparable regions of turbulence
within the cylinder volume. High regions of turbulence were located in regions where
organized structures met in collision.

9) A “ring-shape” turbulent kinetic energy isosurface was observed to exist in the
cylinder during the end of the intake stroke. During the compression stroke, the highest
turbulent regions were found to be centered in the remaining volume.

10) Curve fits to these volumetric TKE data may be used to calibrate current CFD
turbulent models.

11) Tumble and swirl ratio calculations were comparably larger when using a fixed
origin for the calculation rather than a moving origin located at TDC.

12) The swirl ratio calculation, was found to be nearly zero, demonstrating the
cancellation effects of the nearly symmetric features of the in—cylinder flow motion.

13). The tumble ratio about the x-axis for the fixed origin shows a small increase toward

the end of the compression stroke.

121

14) The tumble ratio does not accurately represent the complex air motion inside the
cylinder during the intake stroke, due to the appearance of counter-rotating vortices.

15) The calculation of total and average circulation yielded little information into the
mixing characteristics of the in-cylinder motion.

16) RMS circulation adequately describes the fluid activity during the intake and
compression strokes. The RMS values were found to be uniformly distributed within the

cylinder’s volume between the crank angle interval of 230° to 240°.

l.

122

RECOMMENDATIONS

The SI engine used in this study with regular dual intake ports supports the generation
of high turbulence regions resulting from the intake generated flow collisions. The
slight offset in the intake runner produced flow patterns that were asymmetric. A
straight intake runner may equally distribute the in-cylinder flow motion.
Furthermore, a straight runner—port design, can be used as a baseline geometry for in-
cylinder motion. Once established, different piston head geometries could be
experimented with via the aid of flow visualizations to determine an accurate cause
and effect dependence on the major flow characteristics.

Within the near future, a correlation between the mean kinetic energy and the
turbulent kinetic energy should be investigated. The influence of the flow patterns on
burn rate data should also be evaluated. The burn rate results are already available
from Chrysler engine test stands.

A brief investigation was conducted to determine the effect of higher engine speed on
some inecylinder velocity measurements. For this, velocities at five different
locations throughout the cylinder volume were measured. The results showed an
increase in the velocity magnitude but no major changes in direction. A more detailed
investigation utilizing a larger number of measurement locations should be conducted.
The behavior of the induction and the compression flow field can be used in

combination with a CFD simulation model. The simulation can be used to investigate

 

123

the flow behavior outside of LDV mapped range. Furthermore, the simulation can be
used to track the propagating flame kernel after the onset of spark ignition. The
calculation of mass fraction burned and the comparison to experimental data with
simulated results will give one the information necessary to evaluate chamber and

model performance.

APPENDIX I

 

 

 

 

“Long Side”
Cylinder Head“.
with Intake
Valves “Short SidC”
Duct l

 

 

 

 

Duct 2 I '
l I

Plenum

 

Figure 68: Schematic of the Intake Port Used

The offset in duct 1 may result in a higher mass flow rate entering the left valve. The
higher fluid velocities were observed to be highest in plane 1 and tended to slightly favor
the left valve side of the cylinder where planes 3, 6, 7, and 10 are located. The result of
the slight ofl'set in the intake port design of duct 1 is a feasible explanation of the larger
fluid velocities which appear in all of the flow field images shown in this study.

APPENDIX II

125

program CIRCUL

 

C

c This program is being developed to calculate circulations for slices which
c involve vorticity calculations that require spatial derivatives. These

c derivatives are found be using 3 and S finite point difference methods.

c The 3 and 5 point formulas are located at the end of this program The

c conventions used in this code refers to the circulation grid which gives

c validation to all the measurement locations.

C

0000

C

Start date:l/3l/96 Print Date:3/5/96 of Final Copy

By: Kasser Jafi‘ri

 

real val(9)

real strk,Vis,Ms,Vs

real xl,x2,x3,x4

real zl,z.2,z.3,z4

real tk,avgcir

real xcomp(10,23l,130)

real ycomp(10,23l,130)

real zcomp(10,23l,130)

real u(90,65,130)

real w(90,65,130)

real dudz(90,65,130)

real dwdx(90,65,130)

integer xposa(10,231)

integer yposa(10,23l)

integer zposa(10,23 l)

integer xpos, ypos, zpos
integer ncel

integer ix1,ix2,ix3,ix4

integer izl,izZ,iz3,iz4

real f(5),totalrmc,nnstk,rmscir
real totalcz, tk2, totalca,tka
real RR, Vi, Vcc, rhoamb, Mo, Vo,Ak
real delz

ml kprrzp

integer slice

integer jend(10), 1(10)

integer swin, ewin, sbin, bin, plane
integer xvsznikdi

integer zlocp,zloc

character fname(10)"8, foutv‘12, coordn(10)"l3
character srg'3

integer stu

integer stdeal,zloce

 

C

C

Engine Configuration & Constants

connecting rod length in meters
crl = 230.32/1000.

no

126

crank shafi radius in meters
crd = 4011000.

connecting rod ratio
RR = crd / crl

engine stroke
strk = 2. " crd

engine speed in rpm:
sen = 600.

engine angluar velocity in rad/sec:
pi = 4.0‘atan(1.0)
aen = 2.0‘pi‘sen/60.

piston diameter or bore in meters:
pd = 96.0/1000.

Comb chamber volume (65 cm"3) in cubic meters:
Vcc = 65J(100.0"'100.0*100.0)

Cylinder Volume at crank angle 240 degrees (intake valve closes)
Vo = (pd”2"pi)/4.0"(62.5/1000.)
Rho air at 20"C and 1013 bar [kg/mA3]
rhoamb = 1.2

Mass of air at BDC (Mo), including volumetric efficiency
Mo=rhoamb*(Vo+Vcc)

----Cylinder Volume only around a measured point in cubic meters--
Vi=(5.0“5.0‘10.0)/(1000."1000."1000.)

---~- in the special case of plane 2 Vi is 5‘5‘5 mm3 --------
Vis=(5.0"5.0‘5.0)/(1000."100031000.)

Volume of a slice (intake)
Vs==(110."‘Vi +19.‘Vis)

Mass of a slice (intake)
Ms=rhoamb°Vs
VFVo+Vcc

Area of a kth element in square meters

Ak=50/(1000.“2.)

 

Dataconstants
sbin==l
swin=50

 

127

ebin= 130
ewin=310

 

Read data filename and open the file.
input data filename should be BP'0.001 or "".xxx

input coordinate files should be COORDBP(1-10).TXT

000000000

 

 

====== —"—*—Input for all the different Planes -—-—- "' " —- ”—
c--—------Planel
fname(l)=’BPlO'
jend(l) = 231
1(1) = 4
coordn( 1 F'COORDBP] .TXT‘
c---------P1ane2----—
fnarne(2)='BP20’
jend(2) = 218
1(2) = 4
coordn(2)=-'COORDBP2.TXT’
P1ane3--------
fname(3)='BP30'
jend(3) == 193
1(3) = 4
coordn(3)='COORDBP3.TXT'
~-Plane4—-----—
fname(4)='BP40'
jend(4) = 193
1(4) = 4
coordn(4)='COORDBP4.TX'1"
c--—--—---P1ane5-------
fname(5)=='BP50'
jend(5) = 169
1(5) = 4
coordn(5)='COORDBPS.TXT'
P1ane6-------
fname(6)=='BP60'
jend(6) = 169
1(6) == 4
coordn(6)='COORDBP6.TXT'
c----~~—--P1ane7----~
fname(7)='BP70'
jend(7) = 145
1(7) = 4
coordn(7)='COORDBP7.TXT'
P1ane8
fnamo(8)-'='BP80'
jend(8) =3 145
1(8) = 4
coordn(8)='COORDBP8.TXT‘
c Plane9
fname(9)='BP90'

 

Co...-

 

 

C"

c---

 

on...

 

 

 

 

128

jend(9) = 48

1(9) = 4

coordn(9)='COORDBP9.TXT‘
c-«-----—Plane10.-—---

fname(10)='BPlOO'

jend(lO) = 48

1(10) = 5

coordn( 10)='COORDBP10.TXT

c‘#####fitfitiifittlttttfittiititfit¢¥0000000#ttttttttttt

c
c#######################READ IN VELOVITY COMPONETES AND COORDS

Wfitd‘f)
wfitdi’t)'ttfittttttOO*STARTmG READ SEQUENCEtfiittttttttl

C ---- read though all 10 Planes ---------
D0105 plane = 1,10

c ---open coordinate file for the given plane and read the header»-
open(19,file=coordn(plane),status=’old')
read(19.")
“364093)

C --- Startfile number for all planes always 001 --------
DO 101 ichrt =1,jend(plane)

c wread in the 3 positions of the COOOrdsfile ---

read( 19,165) xpos, ypos, zpos
165 format(13x,l3,5x,l3,5x,l3)

c ——----obtain the correct plane-point file format ~~~~~
srg(l :1) == char-(48+ichrt/ 100)
srg(2:2) = char(48+ichrt/10-ichrt/100‘ 10)
srg(3:3) = char(48+ic1ut-lchrt/10‘ 10)

foutv(1 :1(plane))=fi1ame(plane)
foutv(l(plane)+l :1(plane)t1)=-".'
foutv(l(plane)+2:l(plane)+4) == are

c "open the current plane-point file and read first two header lines--
open(20, file=foutv(1 :1(plane)+5), status=’old')
M203)
rea<1(20,‘)

c -~--array that stores the COORDS position information----—-
xposa(plane,ichrt)=xpos

yposa(plane,ichrt)=ypos
zposa(plane,ichrt)=zpos

 

129
Read the turbulence velocities-std deviation (prol, pr02, pro3 in order)

 

val(l) = mean velocity for procl.
val(2) = standard deviation for procl.
val(3) = turbulence intencity for procl.
val(4) = mean velocity for proc2.
val(5) = standard deviation for proc2.
val(6) = turbulence intencity for proc2.
val(7) = mean velocity for proc3.
val(8) = standard deviation for proc3.
val(9) = turbulence intencity for proc3.

000000000000

 

DO 102 bin=sbin,ebin

read(20,260) va1( 1 ),va1(4),val(7)
260 format(5x,f6.2,20x,f6.2,20x,f6.2)

ycomp(plane,ichrt,bin)=—val(1 )
zcomp(plane,ichrt,bin)==-val(7)

1F (plane.le.6) THEN
C all points still with shift direction u,u,u
xcomp(p1ane,ichrt,bin) = + val(4)
ENDIF

IF (plane.eq.7) THEN

IF (ichrt.le.53) THEN
C points with shift direction u,u,u
xcomp(p1ane,ichrt,bin) = + val(4)
ELSE
C points with shift direction u,d,u
xcomp(plane,ichrt,bin) = - val(4)
ENDIF

ENDIF
IF (p1ane.ge.8) THEN
C all points now with shift direction u,d,u

xcomp(p1ane,ichrt,bin) = - val(4)
ENDIF

c ---ARRAY vel. com. cooresponding to plane, point location amd bin

102 CONTINUE

C - before going to the next measured point in selected Plane, close Input -
close (20)

130
101 CONTINUE
write(*,*)'Done with Plane', plane

c --—---before going the the next plane, close coordinate input file -----

close(l9)

105 CONTINUE
write("',')'FINISHED! ARRAY INPUTS COMPLETE'

c WWW#WWWW#W#WW##W#WW
slice==5

5000 CONTINUE
slice==slice+5

c WWW ####

 

c SET UP VELOCITY COMPONETS INTO AN ARRAY ACCORING TO POSITION
write("',"') 'Setting up velocity arrays according to position’
DO 710 bin=sbin,ebin

DO 720 plane =1,10
DO 730 ichrt=l,jend(plane)
IF (yposa(plane,ichrt).eq.slice) THEN

xp=xposa(plane,ichrt)
zp—rzposa(plane,ichrt)
u(xp,zp,bin)=xcomp(plane,ichrt,bin)
w(xp,zp,bin)=zcomp(plane,ichrt,bin)
ENDIF
730 CONTINUE

720 CONTINUE

7 10 CONTINUE

c#########################################################
write(",") ' DONE inputing arrays'

cfiWWWWWWWWfiWfii
a
C
C

 

--~calculate and array dudzm-mwm

 

 

dclz=5.0/ 1000.0

 

c---------outer rows side rows
DO 400 bin=sbin,ebin
DO 405 xp=—40,40,80
DO 410 zp=-10,10,5

IF(zp.eq.-10) THEN
DO 60 ik=1,3
ii=(5’ik)-15

60

61

62

410
405

131

f(ik)=u(xp,ii,bin)
CONTINUE
dudz(xp,zp,bin)=drv3 pt(f,delz, 1 )
ENDIF

IF (zp.eq.10) THEN
DO 61 ik=l,3
ii=(-5"ik) +15
f(ik)=u(xp,ii,bin)
CONTINUE
dudz(xp,zp,bin)=drv3 pt(f,delz,- l )
ENDIF

IF((zp.ge.-5).and.(zp.1e.5)) THEN

DO 62 ik=l,3
ii=(zp-10)+(5‘ik)
fiik)=u(xp,ii,bin)
CONTINUE
dudz(xp,zp,bin)=drv3pt(f,delz,0)

ENDIF

CONTINUE
CONTINUE

 

C

c—----—--inner seven rows

63

 

DO 905 xp=-30,30,10

1F (xp.1t.0) THEN

2100P=(-1 ‘(XP))-90
zloc=zlocpl2

ELSEIF (xp.gt.0) THEN
zlocp=xp-9O
zloc=zlocp/2

ELSEIF (xp.eq.0) THEN
zloc==-40

ENDIF

zloce==-1"'zloc

DO 910 zp=zloc,zloce,5
top: three point END spline-«um
1F (zp.eq.zloc) THEN
DO 63 ik=l,3
ii=(5"'ik)+(zp-5)
flik)=u(xp,ii,bin)
CONTINUE
pr.zr>.bin)=drv3pt(flde12.1)

132

c-- ----- --bottom: three point END spline-mu-

64

ELSEIF (zp.eq.zloce) THEN
DO 64 ik=l,3
ii=(-5‘ik)-l(5+zp)
f(ik)=u(xp,ii,bin)
CONTINUE
dudz(xp,zp,bin)=drv3pt(f,delz,-l)

c---~------top: three point CENTER spline—---—----

65

ELSEIF (2p.eq.(zloc+5)) THEN
DO 65 ik=l,3
ii=(5‘ik)+(zp-10)
flikH(xp,ii,bin)
CONTINUE
duddxprzpsbinf‘drflplmdcmo)

c-—-------bottom:three point CENTER spline-mm-

ELSEIF (zp.eq.(zloce-5)) THEN
DO 66 ik=l ,3
ii=(zp-10)+(5"ik)
f(ik)=U(XP.ii.bin)
CONTINUE
dudz(xp,zp,bin)=drv3pt(f,delz,0)

FIVE POINT CENTER SPLINE-«mnm-

 

67

910
905

ELSEIF( (zp.ge.(zloc+10)).and.
(zp.1e.(zloce-10))) THEN

DO 67 ik=1,5
ii=(zp-(3-ik)*5)
f(ik)=u(xp,ii,bin)

CONTINUE

dudz(xp,zp,bin)=drv5pt(f,delz,0)

ENDIF

CONTNUE
CONTINUE

 

400

C

CONTINUE

write(",’) 'Finished arraying du/dz !! '
writd‘.") ‘ ’

 

n
V

c--—-—--—ca1culate and array dwdx-m-n

delx=10./1000.
no 300 bin=sbin,ebin
D0 305 zp='~40,40,5
DO 310 xp=-40,40,10

133

ngroup I pointsWW
IF((zp.ge.-10).and.(zp.1e. 10)) THEN

c----------outer left most two columns
IF((xp.eq.-40).or.(xp.eq.—30))TI-IEN
IF(xp.eq.-40) THEN
ipFl
ELSE
ipt=0
ENDIF
DO 80 ik=l,3
ii=(10"ik)—50
flik)=w(ii,zp,bin)
80 CONTINUE
dwdx(xp,zp,bin)==drv3pt(fidelx,ipt)

 

c—mm-m-m-outer right most columns-
ELSEIF((xp.eq.40).or.(xp.eq.+30))THEN
IF(xp.eq.40)THEN
ipt=-1
ELSE
ipt=0
ENDIF
DO 81 ik=1 ,3
ii=50—(10"ik)
flikFMfilprbin)
81 CONTINUE
dwdx(xp,zp,bin)=drv3pt(f,delx,ipt)

 

c inner points of group I---—---
ELSEIF((xp.ge.-20).and(xp.le.20)) THEN
D0 82 ik=l,5
lm=xp-30
ii=(10“ik)+1m
flik)=W(ii.2psbin)
82 CONTINUE
dwdx(xp,zp,bin)==drv5pt(f,delx,0)
ENDIF

ENDIF

 

C

ngroup II pointsr~r~—~-~--~-~v~~~m~-~-~v~--~~~~~~~~
IF((abs(zp).le.3 0).and.(abs(zp).ge. 15))THEN

c-----—---outer left most two columns
IF((xp.eq.-30).or.(xp.eq..20))TI-IEN
IF (xp.eq.-30) THEN
ipt=l
ELSE
ipt==0
ENDIF
DO 83 ik=l,3
ii=(10‘ik)—40
Kikwiiszphin)

 

134

83 CONTINUE
dwdx(xp,zp,bin)==drv3pt(f,de1x,ipt)

c------------outer right most columns-
ELSEIF((xp.eq.30).or.(xp.eq.20))THEN
IF(xp.eq.30)THEN
ipt=~l
ELSE
ipFO
ENDIF
DO 84 ik=l,3
ii=40-(10‘ik)
flikFMiianin)
84 CONTINUE
dwdx(xp,2p,bianrv3pt(f,delx,ipt)

 

c inner points of group II mmmmm
ELSEIF ((xp.ge.-10).and.(xp.le. 10)) THEN
DO 85 ik=1,5
lrn=xp—30
ii=(10‘ik)+lm
f(ik)=w(ii,zp,bin)
85 CONTINUE
dwdx(xp,zp,bin)=drv5pt(f,delx,0)
ENDIF
ENDIF

 

WM» III pointsnnnmm
IF (abs(zp).eq.35) THEN
c----------outer lefi most column
IF ((xp.eq.-20).or.(xp.eq.-10)) THEN
IF(xp.eq.-20) THEN
ipt=1

 

ELSE
ipt=0
ENDIF

D0 86 ik=l,3
ii=(10‘ ik)-3O
flikFWOismbin)
86 CONTINUE
dwdx(xp,zp,bin)=drv3pt(f,delx,ipt)

c-—--------outer right most column-mum
ELSEIF((xp.eq.20).or.(xp.eq.10)) THEN
IF(xp.eq.20)THEN
ipF-l

ipt=0
ENDIF

ELSE

DO 87 ik=l,3

135

ii=3 O-( 1 0‘ ik)
filk)=w(ii,zp,bin)
87 CONTINUE
dwdx(xp,zp,bin)=drv3pt(f,de1x,ipt)

 

c inner point of group III-«m»-
ELSEIF (xp.eq.0) THEN
DO 88 ik=l,5

lm=xp—30

ii=(10‘ik)+lm

flikFMiismbin)
88 CONTINUE
dwdx(xp,zp,bin)=drv5pt(f.delx,0)

IF

CWvawGroup IV pointsz-vw
If(abs(zp).eq.40) THEN
c all poinrsm-

 

IF ((xp.ge.-10).and.(xp.1e.10)) THEN
DO 89 ik=l,3

IF(xp.eq.10) THEN
ii=20—(10‘ik)
ipt=-l

ELSEIF(xp.eq.0) THEN

ii=(lO‘ik)-20
ipt=0

ELSEIF(xp.eq.-10) THEN
ii=-20+(10‘ik)
ipt=1

ENDIF

flik)=w(ii.zp.bin)
89 CONTINUE

dwdx(xp,zp,bin)=drv3pt(f,delx,ipt)

0;- A AA _‘ ‘MA‘A A‘A A .. ‘ A.‘ ‘A A .-

310 CONTINUE
305 CONTINUE
300 CONTINUE
write(‘,‘) 'Finished arraying dw/dx 1! '
writers)"
0..0.00000.‘0COOCfitfitiiitifi¢0.0030000000000#00.¢¢¢00¢!

 

136

cfifittii......tifititCI¢CQ¢CCCCCCCOOCCC¢ttttfittfifititt
c calculate vorticity and circulations

write(‘,") 'Writing circulation file for Slice', slice

srg(lzl) = char(48+slice/100)
srg(2:2) == char(48+slice/10-slice/100° 10)
srg(3z3) = cha.r(48+slice-slice/10"' 10)

foutv( 1 :5)'='CIRSL'
foutv(6:6)=‘.'
foutv(7z9Fsrg

open(44,fi1$foutv(l :9),status=’unknown')
write(44,*)’Circulation for slice,',slice
write(44,‘)'Bin Crank Ang Ncel TotalCirl
>Avgcir RmsCir ABScir'

DO 2000 bin=sbin,ebin

em = real(50.+260./140.‘(bin-1)+O.5‘260./l40.)
err = cra’pi/180.

avgcir=0.0
totalc=0.0
ncel==0
tk==0.0
tk2=0.0

totalrmc==0.0
rmstk=0.0
rmscir=0.0

totalc2=0.0
totalca==0.0
tka=0.0

0 open the circulation element position file
open (43 ,file='cirpos.txt’,status='unknown')
c readthefirsttwoheaderlinesandskipthem
read (433‘)
read (43,"‘)
c start calculation loop
DO 361 i=l,92
m4(43.189) k,kxp,kzp
189 format(3x,iZ,5x,13.0,4x,f5.2)

c DEFINE POSITIONS FOR THE CALCULATION FOR THE CIRULATION ELEMENT
xl=kxp-5.
x2=kxp+5.
x3=xl
x4=x2

137

21 =kzp-2.S
22==zl
z3=kzp+2.5
z4=z3

ixl=int(x1)
ix2=int(x2)
ix3=irrt(x3)
ix4m'nt(x4)

izl=int(zl)
izz=im(zz)
iz3=int(z3)
iz4=int(z4)

IF ((dudz(ixl ,izl ,bin).ne.0).and.
(dwdx(ix1,izl,bin).ne.0).and.
(dudz(ix2,i22,bin).ne.0).and.
(dwdx(ix2,i12,bin).ne.0).and.
(dudz(ix3,iz3,bin).ne.0).and.
(dwdx(ix3,i23,bin).ne.0).and.
(dudz(ix4,iz4,bin).ne.0).and.
(dwdx(ix4,iz4,bin).ne.0)) THEN

VVVVVVV

ncel=ncel+l
tk=((dudz(ix1,izl,bin)'dex(iXIaiszin))+
> (dudz(ix2,i22,bin)-dwdx(ix2.i22.bin))+
> (duddix3,iz3,bin)—dwdx(iX3,il3ebin))+
> (dudz(ix4,iz4,bin)-dwdx(iX4siz4.bin)))'(W4-)
tk2==tk*rk

tka=abs(tk)
ENDIF

totalca=totalca+tka
totalc2=totalc2+tk2
totalc=totalc+tk

361 CONTINUE
avgcir=totalclncel
abscir=tota1calncel
rmscir=sqrt(totalc2/ncel)

write(44,669) bin, cra, nce1,tota1c,avgcir,rmscir,abscir
669 format(i3,3x,f9.4,3x,i4,3x,fl5.6,2x,f15.6,2x,f15.6,2x,f15.6)

138
close(43)

2000 CONTINUE

close(44)

c00000000tfitt$titfitfiltfitt000*.00000000.

IF (slice.eq.65) THEN
goto 4999

ELSE
goto 5000

EN DIF

4999 CONTINUE
Write?!)
write(*,") 'Program DONE l!!!‘

 

 

801 stop
end
FUNCTION DRVSPT(y,h,ipt)

c calculate derivative using a 5-point spline
dimension y(5)
real y,h,drv5pt,temp
if (ipt.eq.0) then

0 not an endpoint---.--...~....-

temp=(y(l)-8.0‘y(2)+8.0‘y(4)-y(5))/(12."'h)

else

C at an endpoint --------

tcmH-ZS.‘y(1H48.’y(2)—36."y(3)+16.")'(4)
> -3.’y(5))/(12."h)
endif

drv5pt=~temp
return
end

FUNCTION DRV3PT(y,h,ipt)
c calculate derivatives using a 3-point spline
dimension y(3)
real y.h.drv3pt,temp
integer ipt

if (ipt.eq.0) then
c not an endpoint mmmmm
temp-1 lr'(1)’(°1 .)+ Y(3) )/(2.‘h)
elseif (ipt.eq.1) then
c atlefi or bottom endpoint-mm-

 

 

139

temt>=(-3-")’(1)r‘4-“‘y(2)')'(3))/(2-‘11)
e1seif(ipt.eq.—l) then

temp=-l .‘(-3."'y(1)+4.’y(2)-y(3))/(2.*h)
endif

drv3pt=temp
return
end

REFERENCES

140

LIST OF REFERENCES

Potter, M. , Foss, J. , “Fluid Mechanics”, Great Lakes Press, 1982

Hughes, W. ,Brighton, J. , “ Fluid Dynamics”, 2nd Ed., McGraw-Hill, 1991

Mueller, J .D., “Quantification of In-Cylinder Flow Characteristics with Laser
Doppler Velocimetry”, Masters Thesis, Michigan State University, 1995

Lee, K., Yoo, S.-C., Stuecken, T., McCarrick, D., Schock, H., “An Experimental
Study of In-Cylinder Air Flow in a 3.5L Four Valve SI Engine by High Speed
Flow Visualization and Two-Component LDV Measurement”, SAE 930478

StephensomP.W., Claybaker, PJ., Rutland, C.J., “Modeling the Effects of Intake
Turbulence and Resolved Flow Structures on Combustion Diesel Engines”, SAE
960634

Haworth, D. , Sherif, H. , Huebler, M. , Chang, S. , “Multidimensional Port-
and Cylinder Flow Calculations for Two- and Four Valve-Per-Cylinder Engines:
Influence of Intake Configuration on Flow Structure”, SAE 900257

Goldstein, R. J. , Adrian, J. R. , “Fluid Mechanics Measurements”, Hemisphere
Publishing Corporation, 1983

Yoo, S. , Lee, K. , Novak, M. , Schock, H. , “3D LDV Measurement of In-
Cylinder Air Flow in a 3.5L Four-Valve SI Engine”, SAE 95648

Camahan, B., Luther , Wilkens, J.O., “Applied Numerical Methods”, John Wiley
& Sons, Inc. 1969

10.

ll

12

141

Burden, R. L. , F aires, J. D. , Reynolds, A. C. ,”Numerical Analysis”, Second
Edition, Prindle, Weber & Schmidt, 1981

Heywood, B. John, “Internal Combustion Engine Fundamentals: ,McGraw-Hill
Publishing Company, 1988

Schetz, J. A., Boundary Layer Analysis, McGraw-Hill Publishing Company, 1982