 

 

 

 

 

 

 

.. .
, ., .., . .1...
I is.
5

8.9.9... ;.
ﬁmﬂm. .-

:1:
3...}:
:9

9. .13. ..
“31.5.5.

«:31: x n...

.0 .1 .2i!

#2.:
31.1.... Avl
it 1

:1: .3.
52.13;

, 3514...

.r...

.1.

. {5313.

at
(Q! «‘1

13635;!
3 “1-21,
Iggy}

!

.s
It» 3v. 2111320.
0..-.‘5 k .l

.vﬂﬂg

nt .10..

‘2

 

may.

MW“!iﬂiilﬁiliiﬂﬁiillﬁiliﬁiiil

3 1293 01399 6677

This is to certify that the

thesis entitled

Dynamic Flow Study in 3 Catalytic Converter
using Laser Doppler Velocimetry and
High Speed Flow Visualization

presented by

Kwang Sup Hwang r: a"

has been accepted towards fulﬁllment
of the requirements for

M.S . Mechanical Eng.
degree in

 

 

 

Jor professor

Date June 16 1995

 

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

 

—-A r s,_‘__ www—

 

LI BRARY
Michigan State
Unlverslty

 

 

 

.PLACE Ill RETURN BOXto monthl- checkout mom your record.
TO AVOID FINES mum on or baton date duo.

DATE DUE DATE DUE DATE DUE.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU I: An Afﬁrmative Action/Equal Oppommlty Instltwon
Wynn-9.1

 

 

 

 

 

 

 

 

 

 

DYNAMIC FLOW STUDY IN A CATALYTIC CONVERTER USING
LASER DOPPLER VELOCIMETRY AND
HIGH SPEED FLOW VISUALIZATION

by

Kwang Sup Hwang

A THESIS

Submitted to

Michigan State University

inpartial fulﬁllment of the requirements

for the degree of

MASTER OF SCIENCE

Department of Mechanical Engineering

1995

ABSTRACT

DYNAMIC FLOW STUDY IN A CATALYTIC CONVERTER USING
LASER DOPPLER VELOCIMETRY AND
HIGH SPEED FLOW VISUALIZATION
by

Kwang Sup Hwang

Internal ﬂow characteristics of a close-coupled, catalytic converter
were examined by LDV measurements and high speed ﬂow visualization.
Although previous studies have been done on catalytic converters, most were
conducted at steady state, using water ﬂow that was seeded with a small
quantity of tracer particles. The purpose of this study was to develop a better
understanding of dynamic ﬂows inside catalytic converters.

A Hyundai 4-cylinder engine head was ﬁtted with standard exhaust
valves, intake and exhaust manifolds, constant volume cylinders with no
pistons, and full-open intake ports to simulate actual engine operation. Air
and seed particles were forced through the system by a blower while an

electric motor drove the cam. A clear- plastic, catalytic converter assembly

was connected to the exhaust manifold of the engine head, allowing optical
access for the ﬂow visualization and LDV. Measurements were conducted at
two air ﬂow/engine speed conditions: 120 standard cubic meters per hour
(scmh) at 1000 rpm and 205 scmh at 2000 rpm. Two locations were chosen
for the LDV measurements: 2.5 cm above the ﬁrst catalyst brick and in-
between the ﬁrst and second catalyst bricks. Flow visualization was also
taken on two planes: along the major axis and parallel to the minor axis of the
converter. Both planes are located at the inlet plenum, right above the ﬁrst
catalyst brick.

The high speed ﬂow visualization ﬁlms and LDV results showed that
areas of separation and circulation were present in the inlet region of the
converter. Backﬂows into the neck of the converter were also observed.
Each cylinder exhausted into a different region of the converter, with the
front-middle region having the heaviest amount of ﬂow. High shear ﬂows
were created by each cylinder, while other regions of the inlet region showed
backﬂows or recirculation. The midsection of the converter had a more

uniform overall ﬂow pattern than did the inlet section of the converter.

This thesis is dedicated to my brother,
Yong-Joon Hwang

( 1961-1986)

ACKNOWLEDGMENTS

I would like to take this opportunity to thank several people whose
help and guidance made this work possible. First, my advisor Dr. Harold
Schock for his support and guidance; to Dr. Keunchul Lee for his patience of
teaching me the operation of LDV and also for making working at the MSU
Engine Research Lab more enjoyable; John Mueller for having someone to
work with the LDV, Dr. Craig Somerton and Dr. Mei Zhuang for being on
my committee and giving me helpful comments; and to Tom Stuecken for his
help in designing and building the test rig.

I also would like to express my gratitude to Hyundai Motor
Corporation for their ﬁnancial support.

Finally I would like to thank my parents, Byungtai and Moonhwa

Hwang for their continuous support during my graduate studies.

TABLE OF CONTENTS

Page
LIST OF FIGURES ................................................................................ viii
LIST OF SYMBOLS ............................................................................... x
CHAPTER 1 - INTRODUCTION ........................................................... 1
1.1 Current Situation ....................................................... 1
1.2 Research Motivation ................................................. 3
CHAPTER 2 - EXPERIMENTAL FACILITIES ...................................... 5
2.1 Test Rig .................................................................... 5
A. Engine Head and Cylinder Assembly ................... 5
B. Plastic Catalytic Converter ................................... 6
C. Flow System ........................................................ 7
D. Electric Motor and Angle Encoder ....................... 8
E. External Lubrication System ................................. 8
2.2 High Speed Flow Visualization(HSFL) ..................... 10
A. Seeding Particles for Flow Visualization .............. 10
B. HSFV Hardware ................................................. 14
2.3 Laser Doppler Velocimetry (LDV) ............................ 16

vi

A. Seeding Particles for LDV .................................... 16

B. LDV Hardware .................................................... 17
2.4 Computer Animation ................................................. 21
CHAPTER 3 - EXPERIMENTAL PROCEDURE .................................... 23
3.1 Coordinate System ..................................................... 23
3.2 Experimental Procedure ............................................. 24
3.3 Quantitative Measurement ......................................... 24
3.4 Flow Visualization ..................................................... 28
CHAPTER 4 - RESULTS AND DISCUSSIONS ...................................... 31
4.1 Exhaust Manifold Consideration ................................. 31
4.2 Flow Visualization Results ......................................... 32
4.3 LDV Results .............................................................. 33
CHAPTER 5 — CONCLUSION AND RECOMMENDATION ................. 41
LIST OF REFERENCES .......................................................................... 44

vii

LIST OF FIGURES

Page
Figure 1 Photograph of the test bench ....................................................... 46
Figure 2 Schematic of the test system ....................................................... 47
Figure 3 Photograph of the clear plastic assembly ................................. 48
Figure 4 Photograph of the plastic cylinder assembly on the test stand ...... 48
Figure 5 Photograph of the two piece converter ........................................ 49
Figure 6 Photograph of the assembled converter ....................................... 49
Figure 7 Photograph of the 3 hp DC electric motor ................................... 50
Figure 8 Photograph of the angle encoder ................................................. 50
Figure 9 Schematic of external oil supply system ...................................... 51
Figure 10 Schematic of the microballoon seeder ....................................... 51

Figure 11 Microballoon seeder illustrating (a) bypass ﬂow, (b) ﬂow through

the container and (c) throttled ﬂow containg air and

seed ......................................................................................... 52
Figure 12 Schematic of the flow visualization system ............................... 53
Figure 13 Schematic of the LDV system ................................................... 54

Figure 14 Upper one component two beam system .................................... 55

viii

Figure 15 Lower two component 4 beam system ....................................... 55
Figure 16 Photograph of the ﬁberoptic probe mounted vertically on the test
rig ............................................................................................ 56

Figure 17 Coordinate systmem and axes description for the catalytic

converter .................................................................................. 57
Figure 18 Schematic of four ﬂow visualization planes ............................... 58
Figure 19 Schematic of intake and exhaust valve timing ............................ 59
Figure 20 Sketches of ﬂow patterns from the ﬂow visualization ﬁlms ........ 60

Figure 21 Photograph of the catalytic converter and the wire mesh model..61
Figure 22 The location of six points in the inlet plenum ............................. 62
Figure 23 W- velocity vs. crank angle at low ﬂow rate .............................. 63
Figure 24 Animated inlet plane velocity proﬁle with 120 scmh ﬂow rate at
engine speed 1000 rpm ............................................................. 64
Figure 25 Animated midsection plane velocity proﬁle with 120 scmh ﬂow
rate at engine speed 1000 rpm ................................................... 66
Figure 26 Animated inlet plane velocity proﬁle with 205 scmh ﬂow rate at
engine speed 2000 rpm .............................................................. 68
Figure 27 Animated midsection plane velocity proﬁle with 205 scmh ﬂow

rate at engine speed 2000 rpm ................................................... 7O

ix

U(t)
U(t)
UEA(9)

u’ea (9)

LIST OF SYMBOLS

particle relaxation time

Lagrangian macro time scale of turbulence
particle density

particle diameter

average gas viscosity

momentum eddy diffusivity

turbulence intensity

turbulence viscosity

ﬁnge spacing

time taken for the particles to cross the ﬁnge
Doppler shift frequency of scattered light
wavelength of the laser light

half angle between beams

Instantaneous velocity

mean velocity

ensamble averaged mean velocity

ensemble average ﬂuctuation

Nc

number of cycle
crank angle

cycle number

xi

CHAPTER 1
INTRODUCTION

1.1 Current Situation

Monolithic catalytic converter systems have been demonstrated to
enable automobile manufacturers to comply with the Federal emission
standards as speciﬁed in the Clean Air Act Amendment of 1970. However,
the emission standards have been rising; not only the United States but the
governments around the world as well Automotive manufacturers now face a
problem of further reducing emissions and keeping the cost down as well.
The catalytic converter has pronunced to be an excellent device to remove
harmful exhaust gases before they reach the atmosphere. The catalytic
converter used in spark-ignition engines consist of an active catalytic material
in a specially designed metal enclosures that direct the exhaust gases
throughout the catalyst bed. The active material employed for CO and HC
oxidation or N 0 reduction must be distributed over a large surface area so
that the mass-transfer characteristics between the gas phase and the active
catalyst surface are sufﬁcient to allow close to 100 percent conversion with

high catalytic activity.

Little has been published relating to catalytic converter velocity
distributions and practical means to improve ﬂow mal-distributions. Previous
studies done by Wendland and Matthes [1-3] were conducted on a steady
ﬂow-ﬁeld using water ﬂow seeded with a small quantity of tracer particles
Lemme and Givens [4] measured the velocity with a small diameter pitot tube
near the monolith exit face. Howitt and Sekella [5] measured ﬂow velocity in
round cross-section monolith converter using hot-wire anemometer and air as
the working ﬂuid. The ﬂow ﬁeld was found to be extremely maldistributed,
with velocities decreasing with increasing distance from centerline.

Normally, the working ﬂuid has been air or water. The advantage of
using air is that it allows one to perform studies over a wide speed range.
However, difﬁculties arise with seeding air because of its poor light scattering
abilities makes it difﬁcult for use with LDV and ﬂow visualization.

Water analog models allow for the use of a larger seed than when using
air. Larger particles scatter more light, which greatly facilitates ﬂow
visualization and LDV measurements. When a water analog is used, one
must ensure that dynamic similitude is preserved. If boundary conditions are

not exactly preserved, the ﬂow results can be altered.[6] By creating a

uniform ﬂow distribution, the lifetime of the converter would be increased

and/or a smaller converter could be used.

1.2 Research Motivation

The amount of exhaust ﬂow produced by an engine determines the size
of the catalyst brick needed in the catalytic converter. Maximum use of the
catalyst volume is needed to maintain a high catalytic converter efﬁciency.
To increase the active surface area, a uniform ﬂow distribution across the
active element is important. This results in the ability to use a smaller
catalytic converter with uniform ﬂow as opposed to a larger converter
required for non-uniform ﬂow. Use of non-uniform ﬂow will result in an
overuse of the areas that receive high ﬂow volumes and underuse of low ﬂow
volumes. Thus, having uniform ﬂow with a small converter will be more cost
effective than the larger one with non-uniformity.

The purpose of this study was to characterize the ﬂow patterns in a
catalytic converter under normal engine operating conditions. This will lead
to a better understanding of the internal ﬂows in the catalytic converter. The
effects of the exhaust manifold geometry and the angle of the diffuser at the

inlet of the converter have a signiﬁcant effect on the ﬂow characteristics.

To gain complete optical access of the inside of the converter, a clear
plastic converter was built. In understanding the complex ﬂow ﬁelds, high
speed ﬂow visualization was employed in various planes of the converter. An
LDV measurement is then applied to quantify the velocity ﬁeld of interest.
Measurements were conducted at two regions: the inlet plenum (2.5 cm in
front of the ﬁrst catalyst brick) and the rrridsection (between the catalyst
bricks) of the catalytic converter. The ﬂow rates used for the studies were
120 standard cubic meters per hour (scmh) at an engine speed of 1000 rpm
and 205 scmh at an engine speed of 2000 rpm. The ﬂow visualization ﬁlms
were taken at two perpendicular planes in the inlet plenum, for two different
ﬂow rates at a ﬁlm speed of 5000 frames per second (fps). Velocity proﬁle
animation was created from the measured velocity vectors to examine the
ﬂow patterns through the catalytic converter. This technique allows for the
comparison of the ensemble-averaged velocity proﬁles from LDV
measurements to the cycle—resolved velocity patterns from the ﬂow
visualization ﬁlms.[7] The results will be used for future designs of an

improved converter.

CHAPTER 2
EXPERIMENTAL FACILITIES
2.1Test Rig
In this experiment the converter was placed on a test bench consisting
of a blower, laminar ﬂow meter, engine head with production intake and
exhaust manifolds, and exhaust system. The picture of the test bench is shown

in Figurel and the schematic of the test system on Figure 2.

A. Engine Head and Cylinder Assembly

The 16 valve 4 cylinder engine head used in this experiment was
modiﬁed to simulate the normal operating exhaust ﬂow with steady air ﬂow
supplied by a blower. Modiﬁcations were minimized to preserve the
characteristics of the engine operation. To reduce the complexity of having
an actual engine and to gain better optical access, clear -plastic ﬁxed volume
cylinders were fabricated and installed into the engine head in place of the
standard piston/cylinder setup. The bore of cylinder block was simulated
using plexi-glass cylinders. The diameter and stroke of the plexi-glass

cylinders was identical to that of the original. Intake valves were replaced

with 2.5 cm tubes to prevent the ﬂows crossing to the exhaust valves directly.
The exhaust valves were operated by a cam driven by a DC electric motor.
Photographs of the engine head and cylinder assembly are shown in Figure 3

and Figure 4.

B. Plastic Catalytic Converter

The converter housing was fabricated from clear plastic to provide optical
access for the LDV measurements and for the ﬂow visualization. The plastic
was molded into two pieces so that the catalyst brick could be inserted with
ease. Figure 5 shows the two pieces of plastic model that were molded from
the actual converter.

The two pieces were bolted together using 4 cm steel brackets around
the edge. The steel brackets provides the necessary rigidity for the converter
from the vibration from the engine. The putty used in the constructing of the
converter into one piece, as a sealant. This prevents air leaking out of the
converter. A picture of the assembled converter is Figure 6. Although the
housing was fabricated of clear plastic for optical access, in the initial stage of
the experiment many problems were found. Some natural properties of the

plastic had signiﬁcantly altered laser beam polarization when it passed

through the converter. This problem was later intensiﬁed when the laser
beams passed through some areas of the converter containing sharp bends.
Thus, when LDV measurements were taken along the major axes near the
comers, the data rates were too low to obtain the needed information. To
overcome these difﬁculties, windows were installed at the corner of the sharp

bend areas.

C. Flow System

The ﬂow system consists of a blower, laminar ﬂow meter with differential
pressure inclined manometer, and hoses to dispense the atomized di-octyl
phthalate (DOP) which was used as laser light scatter for LDV
measurements. The laminar ﬂow meter was used to provide the means of
measuring the ﬂow rate by a connection to a differential pressure inclined
manometer. For the proper air ﬂow rate, a blower was throttled to control
ﬂow rate. A pipe (115 cm) was installed at the approach to the assembly to
ensure that the approach ﬂow to the element was unifrom. The relative length

of the pipe is about ten times the diameter of the ﬂow meter entrance.

D. Electric Motor and Angle Encoder

The engine head was driven by a 3 hp DC electric motor that was
connected to the cam shaft of the engine. A photograph of the 3hp DC motor
is given in Figure 7. A generator pickup feedback controller was installed on
the shaft of the electric motor to maintain uniform speed. Flow was examined
based on the cam angle, which is detected by an angle encoder installed
directly on the cam shaft. Figure 8 is a close up view of the angle encoder.
This encoder generates two series of 1024 square pulses per revolution.
These two series of pulses are offset a half pulse length from each other. The
computer can count 4 times the number of pulses from these two series of
pulses. For this study, 4096 pulses per revolution were used and this

represents 0.088 degree of angular resolution.

E. External Lubrication System

The engine head was lubricated in the same fashion as the actual
engine, except that it used an external, closed system with an oil pump and a
vacuum pump to control oil circulation. 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 negative pressure under the valve cover, as this reduces the
chances that oil will travel down the valve stems. It is important to prevent
the oil from going down the valve stems because the smearing oil could cause
the refraction of the laser beam or cause microballoons to adhere to oily parts
in the converter. A 10W-30 engine oil was supplied to the head through
0.635 mm tubing to the inlet port on the head face. After the lubrication of
cams, the oil reservoir creates low pressure which draws oil and air from the
engine head. The returned oil is then fed to the oil pump, which supplies
ﬁltered oil to the engine head. The schematic of the lubrication system is in
Figure 9.

To accommodate all this equipment, a steel channel structure was built
to include the engine head, intake and exhaust systems, electric motor control
system, and engine oil supply system. The base measurements for the steel
channel structure are 120 cm by 120 cm and the height is 225 cm. This steel
structure was aligned to the LDV optical table for the catalytic converter to
stand vertically. To prevent the vibration from the engine operation to the
LDV optical table, rubber isolators were located under the 4 adjustable legs

of the structure base.

10

2.2 High Speed Flow Visualization

The ﬂow visualization technique has been facilitated through the use of
a specially designed particle mixer. This device is designed so that it
completely mixes the seed before entering the test area. Before the
development of the mixer, particles were directed into the ﬂow in small
amounts through the intake port. With the new mixing device, particles can
be completely mixed and stored in a large reservoir for continuous use during
testing. This device is different from conventional particle seeders in which
the entire mixture of ﬂuid and seeding particles is stored in a chamber prior to

entering the test section. A schematic of the mixer is in Figure 10.

A. Seeding Particles for Flow Visualization

The seeding particles for ﬂow visualization are phenolic microballoon
particles supplied by Union Carbide and designated as BIO-0930 ( density of
0.23 g/cm3). In earlier study, conducted by Schock et.al [6] it was discovered
that these particles showed signiﬁcant improvement in visual contrast over

other seeding materials such as titanium dioxide or aluminum particles. This

ll

improved light seeding particles for ﬂow scattering ability provided more
detail which allowed for a greater understanding of the major ﬂow features.
The microballoons must be dried before entering the test area to
prevent agglomeration. In the seeder, the mixing of the microballoons takes
place in a 60-liter cylindrical enclosure. There are two mechanisms that are
used to mix the particles once they are dried and placed onto the wire screen
in the container. The ﬁrst is a fan with a 3000 rpm motor that serves to
agitate the microballoons and keep them swirling in the container. The fan is
located just below the center of the container, therefore, any particles not
entrained in the swirling motion of the fan will fall to a wire screen mesh
located about 5 inches below the fan. The screen is a 40 x 40 lines per inch
mesh, and the wire diameter is 0.01 inch. This mesh prevents clumps of
microballoons from falling to the base of the container where the air inlet is
located. Just above the mesh sits a two arm mixer driven by a 28 rpm gear
reduction motor which serves to move and lift any microballoons that settle
on the wire screen into the air stream of the fan. Once the inlet air is ﬂowing
through the screen, the microballoons are lifted off and tend to hover just
above the screen. The rotating arms of the mixer then force the hovering

microballoons into the air stream of the fan.

12

The air inlet and outlet for the seeder is made of 4-inch PVC tubing.
The direction of intake air ﬂow is governed by a throttle valve. The air will
bypass the microballoons and the cylindrical enclosure when the valve is in
the down position. The path of the air ﬂow for the valve in the shown
position is Figure 11 (a). When the throttle valve is up, air is redirected into
the cylindrical enclosure and entrains the microballoons. The entrained air
containing the microballoons is then sent to the test area through the outlet of
the seeder. The outlet was placed at the top of the mixing chamber in order
to minimize the number of heavy microballoon clumps that enter into the air
ﬂow. These clumps would tend to hover or swirl in the lower portion of the
chamber.

There are various methods to control the number of rrricroballoons.
The ﬁrst method deals with the amount of rrricroballoons placed in the seeder.
An accurate representation of the ﬂow ﬁeld could be altered if there are too
many microballoons are used. The other method used to control the density
concerns the position of the throttle. In the Figure 11 (c), if the throttle is
positioned in the middle of the inlet pipe, the mixture entering the test area
will contain pure air and air mixed with microballoons. This mixture of air

and microballoons occurs because air travels through both the seeder bypass

l3

and the seeder itself. Either one of the above mentioned methods can be used
to control particle saturation.

The ﬁrst step in Operating the seeder is to lead the microballoons onto
the wire screen and seal the lid on the container. The throttle should be in the
down position to start with so that only pure air is sent to the test area which
completely bypasses the wire screen covered with microballoons. It is
important to avoid the entrainment of microballoons through the seeder and
into the test area before they are actually needed. The next step is to turn on
the agitating fan to stir up the particles. After this step, the throttle of ﬂow
valve is opened and mixing rotor is turned on. This ensures that the mixer
will push hovering microballoons into the swirling motion of the fan. Once
everything is turned on and the test area is ready, the throttling valve is then
opened to redirect the ﬂow into the seeder. When this happens, air travels
through the inlet, up through the bottom wire mesh, and into the container.
The swirling motion entrains the microballoons, and they are pulled through

the outlet and into the test area.

14

B. HSFV Hardware

The high speed ﬂow visualization system consists of a 40 watt copper
vapor laser (CVL), a high speed camera, a synchronization system, a particle
generator and optics. The CVL Model 451 is a gas discharge device of 40-
watt average output power manufactured by Metallaser Technologies. It
emits short pulses at 5000 pulses per second (pps) and a pulse energy of 8 m]
in the green and yellow regions of the visible spectrum, at wavelengths of 510
and 578 nm (2/3 power at 510.5 nm, 1/3 at 578.2 nm). The rate of the pulse
can vary between 4000-6000pps. The pulse duration is approximately 30 ns
and pulse jitter is 3 ns. The main components of the laser are a plasma tube,
high voltage DC power supply and a pulsed discharge system (thyratron
driver and thyratron). The thyratron driver is a high voltage, high repetition
pulse generator which provides trigger pulses to the thyratron. The laser
operator could use the internal 5 kHz oscillator or apply an external pulse
trigger select.

The laser beam, which has a beam diameter of 5.08 cm and has a
Gaussian power distribution, is directed toward the converter by a set of three

circular mirrors. A square cylindrical lens and a cylindrical rrrirror were used

15

to focus the beam into a light sheet approximately 1.5-mm thick. These
mirrors together provide high resolution angular control with coplanar-
orthogonal adjustments. This laser light sheet provides enough exposure to
the ﬁlms used for the high speed ﬂow visualization.

A 16 mm Nac E-lO/EE high speed rotating prism camera with a fast f
2.5 optical system is used to take ﬂow images on a Kodak ﬁlm (7292
tungsten, 320 ASA, 2-stops push processed to 1600 ASA). The camera is
equipped with a trigger pulse generator and optical pick-up to trigger the
CVL laser at 5 kHz and synchronize the laser pulses with the ﬁlm frame rate
of 5000 frames per second (fps) .[8]

An electronic timing system was designed and built speciﬁcally to
switch thytatron driver of the CVL from internal to external mode. In the
external mode the pulse generator of the camera triggers the laser during the
ﬁlming process. At the end of the frlm the timing control switches the laser
back to the internal mode. Also, the synchronization system can be used to
control the operation of external parameters for a preset number of crank

angles. The schematic is shown in Figure 12.

2.3 Laser Doppler Velocimetery
A. Seeding Particles for LDV

An important consideration in LDV measurements is the selection of
seeding material. The scattering particles are the basic source of the Doppler
signal, and their importance in the overall performance of an LDV system
should not be underestimated. By increasing the particle diameter from
several tenths of a micron to several microns, the signal factor increases from
102 to 104.

The particles used for seeding must be small enough to follow the ﬂuid
and not cross more than one fringe at once. On the other hand, if the seed
particle size is too big, the particle will have inertial effects which prohibit it
from following the ﬂow. The practical limit which determines the particle
following a turbulent ﬂow is, t / T _<_ 0.02 where t is the relaxation time of the
particle and T is the Lagrangian macro time scale of the turbulence.[8] The
particle relaxation time is given by

t=p,d,2/ 1811 (1)

where pp is the particle density, tip, is the particle diameter, and 1] is the
average gas viscosity. The Lagrangian macro-time scale is given by

T: e / v’2 (2)

17

where e is the momentum eddy diffusivity and v’2 is the turbulence intensity.
These quantities can be written in more convenient form such as
8 =11 / pair (3)
v’2 = (2/3) KB (4)
where it is the turbulence viscosity and KB. is the turbulence kinetic energy.
Initially, water was atomized and used as the scatter particles in the air
ﬂow. This proved ineffective, though, because the water evaporated too
quickly, Next, a mixture of propylene glycol and water was used. This
combination also proved ineffective so pure DOP (Di-Octyl Phthalate) was
employed. The DOP was able to scatter enough light to take the

measurement.

B. LDV Hardware

One of the most important facts in describing ﬂow phenomena is to
conduct quantitative analysis. LDV is one of the best ways of conducting
quantitative ﬂow analysis. Figure 13 shows a schematic of the LDV system
used for this study. LDV was employed to measure the velocity proﬁles
inside the converter The three-component LDV system is composed of a 4-

Watt argon-ion laser, beam collimator, polarization rotators, beam splitter,

l8

beam steering lens, focusing and receiving lenses, digital burst correlator,
traverse mechanism, and a computer for data acquisition. In order to measure
three—dimensional velocity, the system is composed of an upper, one-
component system for the ﬁber optics, and a lower two-color, four beam
system. The one component ﬁber optic system is shown in Figure 14. Figure
15 shows the two component four beam system. The multi-wavelength beam
is separated into several, single-wavelength beams by a prism. Two
wavelengths ( green, 488.0 nm and blue, 514.5 nm) are selected to measure
the two component velocity on the lower system. For the upper component
another wavelength (violet, 476.5 nm) was directed with the ﬁber optic
system.

The LDV conﬁguration used in this experiment requires the laser
beams to be transmitted through a beam splitter and Bragg cell. The beam
splitter separates the laser beam into two equal intensity beams that are then
focused onto the measurement location. Two blue beams (488 nm) were used
for the v-component while the two green beams (514.5 nm) were focused for
w-component measurements. A pair of violet beams (476.5 nm) was
transmitted through ﬁber optics to measure the u-component velocity. A

beam spacing of 50 mm and focal lengths of 453 mm ( for green and blue)

l9

and 350 mm (for violet beams) resulted in the half angle values of 5.206°
and 6.6810.
The two pair of beams (green and blue )from the lower component were setup
so that the beams could transmit horizontally along the traverse. The violet
beam from the upper component is split by Colorburst system which performs
the function of a beam splitter, and a polarization rotor also directs the beams
into ﬁbers. A single violet beam is directed to the Colorburst, where it outputs
a pair of violet beams. The beam exits vertically through a ﬁberoptic probe
where it is used to measure the velocity component perpendicular to x and y.
Figure 16 shows the ﬁberoptic tube mounted perpendicularly to the converter.

The LDV measures ﬂuid velocities by detecting the Doppler frequency
shift of laser light that has been scattered by small particles (0.1 to 10 um)
ﬂowing with the ﬂuid. It is also important to select the particles that follow
the ﬂuid ﬂow and have a large refractive index at the wavelength used for the
measurement. To accomplish this, a Six-Jet Atomizer (TSI Model 9306)was
used to atomize DOP for laser light scattering particles.

The atomized DOP joined the air stream just prior to the ﬂow entrance
into the intake manifold. A long, narrow leading tube was used to connect

the atomized ﬂuid with the air stream at the center of the air supply tube. The

20

scattered light is collected by a photomultiplier detector and converted into
voltage signals. The velocity of the particles in the ﬂuid is proportional to the
Doppler frequency and can be determined by the relationship as

V: df/t] =df*F ................................................................................. (l)
where

df= fringe spacing,

t1 = time taken for the particle to cross a pair of fringes,

F = Doppler shift frequency of the scattered light.

The relationship used to determine the velocity of the particles
traveling with the ﬂuid ﬂow is shown in Equation (1). The fringe spacing, df,
can be readily determined by

df= A/(2 sinK) .............................................................................. (2)
where

A = wavelength of the laser light,

K = half angle between the beams.

The Doppler shift frequency, F, is measured in Hz and is equal to the
number of fringe patterns a particle crosses in one second. A frequency

shifter is used to measure ﬂow reversals. The system works by having the

21

Bragg cell use the interaction of laser light and high frequency ultrasound in a
transparent medium to shift the frequency. Three frequency shifters were
employed, one for each color. The IFA 750 Digital Burst Correlator was
used to measure the frequency of the burst signal. The three-channel
correlator analyzes the Doppler signals so that three velocity components at
the measured position can be calculated separately [7-11].

One of the greatest advantages of LDV as opposed to other ﬂow measuring
techniques such as pitot tubes and hot-wire anemometers is its characteristics
of being non-intrusive. Unlike pitot tubes, LDV has no effect on the ﬂow
stream because there is no physical measuring device to restrict the ﬂow.

Also one can measure ﬂow reversals with the LDV system.

2.4 Computer Animation

LDV measurement is a technique to measure ensemble-averaged
velocity vectors at a measurement volume, while ﬂow visualization shows the
cycle-resolved velocities in a plane. Since LDV is a point measurement
technique, it does not show the ﬂow patterns directly. It is useful if a set of
the LDV measurement data can be animated in time. This animated LDV

result can be easily compared with the ﬂow visualization results. PV-WAVE

22

visualization software was employed to animate LDV measurement data on a
SUN workstation. The software creates 40 or 50 images and runs them

continuously.

23

CHAPTER 3
EXPERIMENTAL PROCEDURE

3.1 Coordinate System

An explanation of the terminology and coordinate system used for this
study is now given. Cartesian coordinates were employed for the ﬂow
measurements and analysis. Figure 17 (a) shows the coordinate system in the
catalytic converter. Figure 17 (b) is the view from the top. The streamwise
air ﬂow direction was chosen as positive 2 direction, and the plane
perpendicular to the z-axis was deﬁned as the x-y plane. The x-axis is
perpendicular to the engine crankshaft, while the y-axis is parallel to the
engine length. The positive x-value represents the direction away from the
engine and the positive y-value stands for the direction toward the ﬁrst
cylinder (see Figure l). The origin of the x-y coordinate system is at the

center of the measured plane.

24

3.2 Experimental Sequence

The test system was initiated by a speciﬁc sequence of steps. First, the
vacuum pump was turned on to drain residual oil in the head. The oil pump
was started to supply ﬁltered engine oil to the engine head.
Oil ﬂow was checked by observing the tygon tubing for oil ﬂow and by
monitoring the vacuum gauge. The DC motor was then activated and set to
the appropriate engine speed. The blower system was turned on and allowed
to run for 10 to 20 minutes before throttling to the correct ﬂow rate in order to

provide steady ﬂow to the system.

3.3 Quantitative Measurement

Although, the ﬂow visualization technique can determine the bulk
feature, the differences of these features are often subtle and cannot be
described in a quantitative way. Therefore quantitative analysis is important
to understand the ﬂow characteristics in a highly ﬂuctuating turbulent ﬂow
like the one in the catalytic converter. To measure the local velocities and
ﬂuctuations, a three-component LDV system was employed. The three

velocity component are crucial to understanding the nature of the turbulent

25

ﬂow inside the catalytic converter which cannot be determined by
measurements of individual components.

One of the important characteristics of turbulent ﬂow is that of its
irregularity or randomness. Statistical methods are used to deﬁne such ﬂow
ﬁeld. The quantities that characterize turbulence are usually the mean velocity
and the ﬂuctuation velocity about the mean, and length and time scale. The

local instantaneous velocity can be deﬁned by
U(t) = U + u’(t) (1)

Here, the mean velocity is the time average of U( t)

_ . 1 (to+t)
U = 11m — U(t)dt (2)

I —> 0° ‘C 10
The turbulent portion of the velocity is u'(t). For a steady ﬂow, the mean
stands for the simple time averaged value. In the converter, the ﬂow pattern
changes during the engine cycle. The mean value can vary signiﬁcantly from
one engine cycle to the next. Instantaneous velocity can be written as a

function of cam angle 0 and cycle number instead of a function of time .
U(e,t)=r7"(e,i)+u'(e,i) (3)

The ensemble averaged velocity U M (0) over Nc cycles can be deﬁned as

26

 

(4)

Nc
Uea(9) 2 NC 12:] U(9,l)

where the Nc is the number of cycles for which data is available. The
ensemble average velocity is only a function of cam angle since the cyclic
variation has been averaged out. The difference between the mean velocity in.
a particular cycle and the ensemble-averaged mean velocity over many cycles

is deﬁned as the cycle-by cycle variation in mean velocity
U(G,i)=l7(0,i)—17ea(0) (5)
Thus the instantaneous velocity can be split into three components
U(0,i) = (7.49) + U(0,i) + u' (9,1) (6)
This is the most desirable equation to be used for the measuring velocity in
the converter. By maintaining constant engine speed the variation from one

cycle to another (U (0, t)).

The ensemble averaged ﬂuctuation “ea (0) is given by

1 NC
' ea 6 Z ‘— ' 6 ,i
u ( ) NC 2, u ( ) (7)

27

It includes all ﬂuctuations about the ensemble averaged mean velocity.[12]
Based upon equation(6) and (7), ensemble averaged mean velocity and mean
ﬂuctuation were calculated.

The LDV measurements were conducted in two planes: the inlet plane
and the midsection plane. The inlet plane is located 2.5 cm in front of the
ﬁrst catalyst brick in the inlet plenum, and the midsection plane is the center
plane between the catalyst bricks (see Figure 17 (a) ). Ensemble-averaged
velocity vectors [4,5] were measured by LDV for a total of 300 points for two
different ﬂow rates. LDV measurements were conducted along two major
axes and two minor axes in each plane. The two major axes’ locations are x
= 12.5 mm and -16.5 mm, while y = + 30 and - 30 mm were chosen for two
minor axes (see Figure 5). Three components of velocity were measured at
each point along the major axes. Because of problems with laser beam access
due to the seam between the two halves of the converter, two components of
velocity vectors (u and w) were measured along the minor axes.
Measurements were taken every 5 mm along the major axes and every 3 mm

along the minor axes.

28

3.4 Flow Visualization

The CVL is used as a light source for the high speed ﬂow visualization
process. Once the CVL beam is on, the optics are adjusted to form a light
sheet through the area of interest. When a laser beam with a power of 25-30
watts is achieved, a test run of the camera is performed to verify that the
shutter is synchronized with the laser pulses at 5 kHz. The ﬁlm is then
loaded, and the camera positioned and focused on the light sheet in the
cylinder, which was ﬁlled with incense smoke to scatter light for focusing.
The seed generator is loaded, sealed, and connected to the intake manifold.
The exhaust fan and lubrication system are turned on while the laboratory
lights are turned off to reduce unwanted light intensity to the camera. The rig
is then motored at the desired speed. The seeding particles are phenolic
microballoons with an average diameter of 40. The rrricroballoons are
introduced into the converter through the intake manifold, pass into the ﬁxed
volume cylinder, and ﬁnally to the exhaust manifold. The ﬂow in the system
is very complicated, and high-frequency periodic ﬂows are present. It is
necessary to take the images at high speed to obtain a satisfactory level of

clarity of the unstable ﬂow features. For this visualization, the camera was

29

run at a ﬁlm speed of 5000 fps. At this speed 120 m of movie ﬁlm is exposed
in approximately in 3 seconds. Four ﬂow-visualization ﬁlms were taken at
planes in the inlet plenum of the converter. Flow was examined at two
different engine speeds (2000 and 4000 rpm) with two different ﬂow rates
(120 and 205 scmh).

There are four planes chosen for ﬂow visualization ﬁlms. Figure 17
shows the four planes in which the ﬂow visualization ﬁlms were taken. The
ﬁrst character of the planes (H or L) stands for the high or low ﬂow rates.
The second character (1 or 2) means a plane parallel to the major (I) or minor
(2) axis. The H1 and L1 planes are deﬁned as planes parallel to the major axis
of the catalyst brick and the H2 and L2 planes are parallel to the brick's minor
axis with high and low ﬂow rate, respectively.

Planes H1 and L1 are identical planes in different operating conditions.
To take ﬁlms for these planes, a laser sheet was provided from the 4th
cylinder side. The back face of the converter was painted black to prevent
interference from reﬂections. The camera was positioned at 90 degrees to the
laser sheet.

Planes H2 and L2 were located slightly apart. The H2 plane is offset

approximately 1.5 cm from the center of the converter to observe the effects

30

of the oxygen sensor on the ﬂow in the inlet of the converter. Plane L2 was
located at the center of the converter to observe the ﬂow on the minor axis.
Flow visualizations at 1000 and 2000 rpm were found to be similar. Due to
the ﬂow similarity it was valuable to examine the ﬂow around the oxygen

sensor to check the viability of the sensor location.

31

CHAPTER 4

RESULTS AND DISCUSSION

4.1 Exhaust Manifold Consideration

Interpreting the results from LDV measurement and ﬂow visualization
in a catalytic converter under normal engine operating conditions requires
evaluating a large amount of data. Even though LDV measurements are
different from the ﬂow visualization results, it is easier to understand the ﬂow
characteristics when the results are compared in the same manner. It is much
easier to understand when the ﬂow visualization ﬁlms have been seen.

Before examining the ﬂow patterns, it is important to understand the
exhaust manifold conﬁguration. Four exhaust paths from the four separate
cylinders are merged into two passages at the end of the exhaust manifold.
Exhaust paths from cylinder 1 and cylinder 4 are combined and go through an
inside partition (near the engine), while exhaust paths from cylinders 2 and 3
go through an outer partition (away from the engine). It is also necessary to

discuss engine ﬁring order because it relates to the exhaust valve opening and

32

closing orders. Figure 19 shows the typical valve timing for a four—cylinder,

four-stroke engine with a 1-3-4-2 ﬁring order.

4.2 Flow Visualization Results

Figure 20 shows the sketches of ﬂow patterns from the high speed ﬂow
visualization ﬁlms. However, the arrows in the sketches do not represent the
velocity magnitude. Figures 20 (a) and 20 (b) are the side view from the
cylinder 1 side (H2 and L2 in Figure 17), while Figures 20 (c) and 20 (d) are
the front view from the opposite side of the engine location (H1 and L1 in
Figure 17).

Figure 20 (a) is a ﬂow pattern produced by the exhaust from cylinder 2
or 3. Figure 20 (b) shows that from cylinder 1 or 4. The downward pulse on
one side and the backﬂow on the other side in Figure 20 (a) are stronger than
those in Figure 20 (b). This phenomenon can be explained by the
conﬁguration of the exhaust manifold. The exhaust passages in the manifold
from the cylinder 1 and 4 are connected at the converter ﬂange with a 90°
bend, while the passages from the cylinder 2 and 3 have a straight connection.
So, the exhaust ﬂow from the cylinder 2 or 3 is faster and uses a smaller area

Of the converter than the ﬂow from cylinder 1 or 4.

33

Figure 20 (c) represents the ﬂow pattern of exhaust from cylinder 3 or
4, while Figure 20 (d) shows that from cylinder 1 or 2. The exhaust manifold
used for this study is slightly asymmetrical about its center. Even with this
asymmetry of exhaust passages, the ﬂow patterns are similar to each other
(Figures 20 (c) and 20 (d)). The effect of the shape of the exhaust manifold
can be observed in the direction of the ﬂows in the converter. The exhaust
from cylinders l and 2 travels from the left to the right, creating a pulse in the
converter to the right. The exhaust from cylinders 3 and 4 travels from right
to left, creating a pulse on the left side of the converter. In both cases regions
of recirculation form in the far corners of the converter. For both cases, the
ﬂow speed from cylinder 2 or 3 is faster than from cylinder 1 or 4. The
exhaust from the cylinder 2 or 3 ﬂows mainly through the middle portion of
the converter, and the exhaust from cylinder 1 or 4 directed to the Opposite

side of the manifold passage.

4.3 LDV Results
Figure 21 contains pictures of the catalytic converter used for this study
and the wire mesh model used in the animation of the LDV data. Figures 21

(a) and 21 (c) are the front view of converter (from the exhaust side of the

34

engine) and Figures 20 (b) and 20 (d) are the side view from the cylinder 1
side. In order to simplify the presentation of the results, the wire frame was
removed, and the two planes of interest were printed together. Figures 24 to
25 show the animated velocity vectors at speciﬁc cam angles. The pictures
on the left side is the viewpoint from the exhaust side of the engine, the same
as Figure 20 (c). The pictures on the right is observing the from the side of
cylinder 1. This view point is same as the one shown in Figure 20 (d). For
the pictures on the left, cylinder 1 is to the left and cylinder 4 is to the right.
The main ﬂow direction (positive z-direction) is toward the bottom of the
converter, where the exhaust gases exit. The cam angle for the velocity
profile is provided for each set of pictures.

Eight lines of vectors are drawn for each angle: four parallel to the
major axis (left to right in the ﬁgures) and four parallel to the minor axis (top
to bottom in the ﬁgures). The top of the catalytic converter is rotated around
the y-axis counter clock wise ( see Figure 17 (b)) to ease the View of the four
major and minor axes. The upper four pair of vectors are the measurement
from the inlet portion while the lower four pair are the midsection. For the
major axes, the upper row of vectors are at x = -16.5 mm, while the bottom

row is at x = 12.5 mm. In reference with the view point of Figure 20 (c), the

35

column of vectors for the minor section is located on the left side at y = 30
mm, while the right side column is at y = -30 mm.

Regular vector arrowheads from PV-WAVE did not show up well in
printouts of the three-dimensional vector ﬁelds. Asterisks proportional to the
vector magnitude were placed at the heads of the vectors in place of the
arrowheads. Since the rows of vectors parallel to the major axis are three-
dimensional, their 2-D lengths may not be prOportional to their magnitudes.
A scale is provided to estimate the magnitudes.

Figure 24 is the results from the low ﬂow rate set-up (engine speed of
1000 rpm and flow rate of 120 scmh). Figure 25 is the high ﬂow rate set-up
(2000 rpm and 205 scmh).

For the low ﬂow rate (Figure 24), the results show that the velocity
distributions are quite complicated and the ﬂows are periodic with a high
frequency. The ﬁgures show that the exhaust manifold conﬁguration has an
effect on the ﬂow direction in the inlet plenum. In early cam angles, the
exhaust gas ﬂows to the right side. This indicates that the air from cylinder 1
is still affecting the ﬂow. At a cam angle of 43.4 degrees, most of the exhaust
ﬂows through the left front section of the converter, as the exhaust from

cylinder 3 reaches the converter plenum. At the right side and back of the

36

plenum at this angle, a ﬂow reversal was observed. The left side ﬂow stays
developed until the exhaust from cylinder 4 arrives. At cam angle 101.3, this
ﬂow from cylinder 4 becomes the main ﬂow, and a strong ﬂow on the back
right side is developed. Around a cam angle of 230 one can observe a strong
ﬂow in the front of the converter from cylinder 2. This ﬂow pattern is a
mirror image of the ﬂow at 43 degrees. At a cam angle of 307 the ﬂow from
cylinder 1 becomes the main stream in the back of the converter. The exhaust
from cylinders 2 and 3 mainly ﬂows through the front section (positive x), and
the exhaust from cylinders l and 4 through the back section (negative x). The
exhausts from cylinders l and 2 ﬂow to the right side of the plenum, while
those from 3 and 4 ﬂow to the left side. This shows that the ﬂow in the inlet
plenum is affected mostly by the conﬁguration of the exhaust manifold.

The periodic phenomenon in the midsection appears later in the engine
cycle than in the inlet plenum. At around 60 degrees Of cam angle, ﬂow
velocity from cylinder 3 becomes high in the front section. About 90 degrees
later (159.2 degrees) one can see a strong ﬂow in back of the converter. At
cam angles of 250 and 350 degrees, flow patterns which are mirror images of
those at 60 and 159.2 degrees, respectively, are observed. The two major

differences between the ﬂows in the inlet plenum and the midsection are the

37

direction and magnitude of the velocity vectors. In the inlet region, peak
velocities for the two ﬂow rates were around 30 m/s and 23 m/s; while
between the two bricks, peak velocities were around 14 m/s and 6 m/s. Also,
in the inlet plenum the difference between the maximum and minimum
velocities is much bigger than in midsection, because ﬂow reversals are
present in the inlet region but absent in the midsection.

For the case of 2000 rpm and 205 scmh, general ﬂow patterns were
similar to those under 1000 rpm and 120 scmh conﬁguration (see Figure 24).
For both cases, the ﬂow from cylinders 2 or 3 is directly injected to the
converter, with most of the exhaust gas ﬂowing through the front section of
the converter. Exhaust from cylinder 1 or 4 are injected into the far side of
the converter.

The ﬂow ﬁeld inside the converter is highly complex. Due to its
complexity, the velocity proﬁles are not easy to understand. To aid in the
understanding, separate points from the inlet plenum were plotted with the
velocity magnitude corresponding to the engine crank angle. As with the PV-
WAVE velocity proﬁle, P1,P2,P3 is located at the lower horizontal line. P4,
P5, P6 is located at the top horizontal line. Figure 22 is the location of the six

points that were picked. As mentioned earlier, the main ﬂow direction is the

38

positive w- velocity. Figure 23 is the graph of the velocity magnitudes with
respect to the crank angle.

In early cam angle stages, gas exhaust is to the right side of the
converter in the page. It could be seen from the graphs at x = —l6.5 mm, the
+ w velocity is a weak ﬂow, the ﬂow at x: 12.5 mm has the dominance. This
is an indication that cylinder 1 is still dominating the ﬂow.

At a cam angle of 43 degrees, the exhaust from cylinder 3 reaches the
converter plenum. Most of the exhaust ﬂows through the left front section of
the converter. At this stage, Pl at x: 12.5 mm starts to show a rise in the
ﬂow magnitude as well as the upper region. However, the ﬂow velocity
starts to drop ( back ﬂow starts to form ) starting at 43 degrees for P3.
Around a cam angle of 230 one can observe strong ﬂow in the front( X: 12.
5) of the converter from cylinder 2. Because the ﬂow magnitude is relatively
greater at the lower right portion, the velocity of P1 becomes the greatest. In
looking at the graphs, ﬂow ﬂuctuates heavily at the lower line ( X: 12.5 mm)
than the upper line ( x: -l6.5). It can be observed that from the velocity
proﬁle, the ﬂow is concentrated more at the front portion of the converter.

Regions of circulation in the inlet plenum were Observed from the ﬂow

visualization ﬁlms. The ﬂow visualization ﬁhns showed the same results as

39

LDV measurements. On the minor plane ﬁlms (H2 and L2), 3 weak
circulation was observed just before the ﬂow changed from the back section
to the front of the exhaust manifold. Strong separations in the major planes
(H1 and L1) occur at the entrance region of the converter. When the exhaust
from cylinder 2 or 3 starts to ﬂow, weak circulation regions have been
observed on both sides of the converter. For the ﬂows from cylinders l and
4, a strong circulation on the opposite side and a weak circulation on the
same side of the exhaust manifold passage tube were created.

The most used area of the ﬁrst catalyst brick is the middle front region.
The back section of the middle region is partially exposed to the exhaust gas
when the ﬂows from cylinder 3 or 4 are fully developed. The magnitudes of
the velocity vectors at the major axes were much smaller at X: -16.5 mm than
at x=12.5 mm. and more recirculation is present in the back ( i.e. x= -l6.5
rmn) than in the front ( Le. x: 12.5 mm). Both ends of the major axes
experienced circulation for large amounts of time. This indicated underuse of
the end regions of the catalyst and overuse of the central region, which will
reduce the life time of the catalytic converter in actual engine operation. The
plane between the two bricks has a more uniform ﬂow distribution overall,

but for the high ﬂow rate, pulses in the ﬂow were observed. The front section

40

Of the converter had higher velocities, on average, than the back. The middle
region of the two major axes also had much higher peak velocities than the
end points. The second brick had a fairly uniform ﬂow distribution for the
low ﬂow rate.

Flow around the oxygen sensor was studied by careful examination of
the ﬁlm for the plane H2. This showed that the oxygen sensor had little effect
on the ﬂow. It also showed that the oxygen sensor does not monitor the
exhaust gases for a whole cycle. The trajectory of the exhaust from cylinders
2 and 3 was such that the majority of the ﬂow did not touch the oxygen
sensor directly. Also, the gas from cylinder 4 ﬂows toward the side opposite
the oxygen sensor without reacting with the oxygen sensor. The location of
the oxygen sensor seems to be too close to the exit of the exhaust manifold to

insure constant monitoring of the exhaust gases.

41

CHAPTER 5
CONCLUSIONS AND RECOMMENDATION
After careful examination of the experimental results, the following

conclusions have been reached:

1. The ﬂow visualization techinque developed offers an effective method of
examining the effect of the highly complex ﬂow inside the catalytic
converter. The combination of ﬂow visualization and LDV measurements

makes it possible to interpret the internal ﬂow.

2. Dynamic ﬂow characteristics were different from those under steady ﬂow
conditions in the catalytic converter. Therefore, to design the best
converter possible, it is necessary to study ﬂow characteristics in the

catalytic converter under normal engine operating conditions.

3. Redesign of the angle, diameter and length of the entrance region is
needed to prevent the separation and circulation present in the converter,

thereby increasing the amount of active surface area used.

42

4. To reduce the periodic effect on the ﬂow pattern, the size and conﬁguration
of the exhaust manifold must be reconsidered. A new design should direct
the flow from each cylinder straight into the neck Of the converter assembly

as much as possible.

5.With the current exhaust system, the oxygen sensor is not continually
exposed to direct exhaust emissions from each cylinder. The position of
the sensor should be changed or the ﬂow should be altered as

recommended above.

6. The oxygen sensor generates a small wake, but does not create a

signiﬁcant disruption of the ﬂow.

7. Velocities are greatly reduced and are more uniform after the ﬁrst catalyst
brick. The space between the catalyst bricks of present design seems to

work properly.

43

8 The catalytic converter used for this study is relatively well designed for

uniform ﬂow distribution, and hence to have a long life.

REFERENCES

. Wendland, D.W., and Matthes, W.R., “Visualization of Automotive
Catalytic Converter Internal Flows,” SAE Paper No. 861554, 1986.

. Wendland, D.W., Sorrell, PL. and Dreucher, J.E., “Sources of Monolith
Catalytic Converter Pressure Loss,” SAE Paper NO. 912372, 1991.

. Wendland, D.W., Matthes, W.R., and Sorrell, P.L. “Effect of Header
Truncation on Monolith Converter Emission-Control Performance,” SAE
Paper No. 922340, 1992.

. Lemme, CC, and Givens, W.E., “Flow through Catalytic Converters - An
Analytic and Experimental Treatment,” SAE Paper No. 740243, Feb 1974
. Howitt IS, and Sekella, T.C., “Flow Effects in a Monolithic Catalytic
Converters,” SAE Paper No 740244, Feb 1974

Schock, H.J., Hamady, F.J., Deﬁlippis,M.S., Stuecken, T., and LaPointe,
L.A., “High frame Rate Flow Visualization and LDV measurement in a
Steady Flow Cylinder Head Assembly,” SAE Paper No 910473, Feb 1991
. Lee, K., Yoo, S.-C., Stuecken, T., McCarrick, D., and Schock, H.J., “An

Experimental Study of ln-Cylinder Air Flow in a 3.5L Four-Valve SI

45

Engine by High Speed Flow Visualization and Two-Component LDV
Measurement,” SAE Paper No. 930478, 1993

8. Regan, C.A., Chun, KS. and Schock, H.J., “Engine Flow Visualization
Using COpper Vapor Laser,” Proceedings of SPIE, Vol. 737, pp. 17-27,
1987.

9. Lee, K., Yoo, S.-C., Weber, M., and Schock, H.J., “Flow Characterization
through a Catalytic Converter Using Laser Doppler Velocimetry,”
Technical Report MSUERL-93-2, Submitted to Ford Motor Co., 1993.

10 Instruction Manual for Digital Burst Correlator, TSI Inc., April, 1991.

1 1 .Adrian, R., “High Speed Correlation Techniques,” TSI Quarterly, Vol
VIII, Issue 2, April-June, 1982.

12 Heywood, J.B., “ lntemal Combustion Fundamentals,” McGraw-Hill,

New York, 1988

,. _ w . ,_ ,

 

Figure 1. Photograph of the test bench

47

ESE? cow—3:8 £558 05 mo org—26m .N 0.5mm"—

:ﬂ 3:35

82— nae—~09” “mg :0
a

852.8

£558

\r
C aim
a .

 

 

 

 

 

 

 

 

_|._ a 3 8a.... hues... -
_.OEOOO_ 1H _ t...

_ .. he i l... __H .N,

 

 

 

 

 

 

 

48

 

Figure 4. Plastic cylinder assembly on the test stand

49

 

Figure 5. Photograph of the two piece converter

 

Figure 6. The assembled converter

50

 

Figure 7. Picture of the 3hp DC electric motor

 

Figure 8. Picture of the angle encoder

51

Vacuum _ -

pump

r—> Engine heed m

Filter

 

 

 

 

 

Oil
L ................. 1 Filter reservoir

 

 

 

 

 

Oil pump

Figure 9. Schematic of external oil supply system

E D

Agitation
fan
\

Wire Screen

 

 

 

\

    

   

  

 

 

 

 

 

 

 

 

 

 

Figure 10. Schematic of microballoon seeder

52

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 11. Microballoon seeder illustrating a) bypass flow, b)ﬂow through the

container and c) throttled flow containing air and seed.

“—38 SEES—mm.» Boa 05 .«o ouansom .N— 0.53"—

 

..83 .893 eon—gov £2.52 P

couﬁmegoim ,

 

 

 

 

 

53

 

 

 

 

 

 

 

 

 

 

_

_

L
332m

 

 

 

 

 

 

 

 

e8: oﬁwem

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A) Argon -Ion Laser

B) Beam Collirnator

C) Steering Mirror

D) Prism

E) Steering Mirror

F) Beam Splitter - 488.0 nm
G) Beam Splitter - 514.5 11111
H) Polarizer - 488.0 um

I) Polarizer - 514.5 um

I) Bragg Cell - 488.0 nm

K) Bragg Cell - 514.5 nm
L) Beam Stop

M) Focusing Lens

Figure 13 LDV system.

N) Catalytic Converter

0) Shaft Encoder

P) Receiving Optics - 514.5 nm
Q) Receiving Optics - 488.0 nm
R) Receiving Optics - 476.5 nm
S) IPA 750 Signal Processor

T) Rotating Machinery Resolver
U) Master Interface Unit

V) Frequency Shifter - 476.5 nm
W) Frequency Shifter - 488.0 nm
X) Frequency Shifter - 514.5 nm
Y) Colorburst

2) Fiberoptic Probe

 

 

 

Figure 15. Lower two component 4 beam system

56

 

Figure 16. Picture of the ﬁberoptic probe mounted vertically on the converter.

57

cow—0250 och—gay ecu Eggnog we: 035280 23 683m Suﬁ—:80 .2 853...—

A8
33> mom.

Em.N—HK EEW.©—:HK

 

 

\

«23 SEE

/

 

 

T

.L

 

r\.

3.8 8.3.:

 

 

3
02m 2:ch Eat 30$ 02m

0

            

£5 BEE nee—a

BE

not:
3.258

EE WVHN
Ea 8an 05—A—

Seus .25

.8=.ulm_zow>xo

  

Ego—u 8.5
N...

.L

58

£5“ BEE 3 3:88 ween—n 05 Be «4 e5 N: .25 38 SEE 05 2
3E ween—a 05 2a 3 E8 _= does—a Steam—«83 BoE .w- Bawﬁ

x35
“:33 oh

 

 
 

 

 

 

— - - - - - - - - - - - — - - - -
- - - - - — — - — - - - - ‘D - O

 

 

 

29:52: 6.558 2 .225;

59

Ace—ohm 835 .«o De.
.8 mega came—o c v mew—E o>_a> ems—ﬁne e5 38.: no ouanaom .2 0.52"—

8—

8 cine-68

 

 

if

 

 

 

O—

 

 

 

 

 

 

E

 

 

-1,

 

3%

$335

8%

Ines—e

Top of Top of
Converter Converter

 

 

h

engine side T l I engine side 1

 

 

J
k
a. flow from cyl. 2 or 3 b. flow from cyl. l or 4
Minor axis ﬂow visualization
top of converter top of converter

oxygen sensor oxygen sensor

 

 

 

 

c. flow from cyl. l or2 d. flow from cyl. 3 or4

Figure 20. Sketches of ﬂow patterns from the ﬂow visualization ﬁlms

61

 

Figure 21. Photograph of the catalytic converter and the wire mesh model
a. Front view of prodecution converter ( top left)

b. Side view of production converter (top right)

c. Front view of wire mesh mode (bottom left)

(1. Side view of wire mesh mode (bottom right)

62

 

 

 

 

P2 x=12.5mm
................... ---
’ =-l6mm
.................. --p5 MW...
l l
p.30... MW , .0...

Figure 22. The location of the six points in the inlet plenum

 

63

 

 

v‘v‘ '

W-Velocity (mls)
a V
r

w-Veloclty vs. Cnnk Angle
Low Flow Rate
(x=12.5mm)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

o L
P ”.-
.5 1.
I p1 0 p2 A p3
.10 l : ¢ 4. ; ‘ﬁ : . 4
0.0 00.0 100.0 270.0 300.0 400.0 540.0 000.0 120.0
Crank Angle (Degree)
(20
W-Veloclty vs. Crank Angle
Low Flow Rate
20 - XI-16.5 o
)
0%
15 ., a O 0.‘
.v. : .
a 10 T O ‘ t ‘
a 3 4 .
S o 0‘ . o ’ Q d h
g 5 v x ? ~ 0 0 o ' .9
a 'g - ' ° 5’ “ r“
>' 0 O ‘vv’k,“. ‘ ' J _) ‘
g o .‘Lhr‘u-n ‘ .._ 0‘“ ."JLL‘... ,‘n
x I ﬂ . I.M
O. . *
. _
.5 4.
I p4 0 p5 A p6
.10 ; r : 4% c ; . a
0.0 00.0 100.0 210.0 000.0 400.0 540.0 030.0 720.0
Crank Angle (degree)

 

 

(b)

Figure 23. W-velocity vs. crank angle (a) low ﬂow rate (b) high ﬂow rate

 

 

camangle :71“, “macaw/S)
7%))»; m 71 11W e

’ 1111mmlu§w W5

   

// Z» W“ 41 W Mfg/l

1111/1111 )1:

”1 11/111” ’5: 1111mm“

Figure 24. Anima tedinle etplanev veiloc ttyproﬁle ewith 120 socmhfl wrate at
engine speed 1000 rpm.

cam angle

//// /././1 11 253,, kg \

J “W U i\,\§/\\_\.\\‘L<

-/ .7/1/1 111/ // ' 287.6 @411;th \

1 11h \\\ -

‘ﬁjjlﬁﬁ g ”‘3 ergkx

1% 171112111?

igure 24. continued

4.8

24.1

43.4

62.7

62.0

\11
\111

\ll
\Hl

cam angle

11111111111111

1111111111/(11
112/11111111111 111

111 1111111111! 111

lllll\l\\\l lll
l\llllllll ll\

\\\ \xxxxxxxuyiu

1L1 lx\\\xxx>x\ xxx

\\

ll\

\\\\\\\\\ll \ll

\\\\\\\\\\ \\\

101.3

120.6

139.9

159.2

178.6

rate at engine speed 1000 rpm

( 1cm 8 8.5 m/s)
\\\ \\\\\\\\ll)rlll
111 11111\\\\\\ xx»
lll \l\\\\\}\\\ 11\
1\\ lllll\\l\\\/(xl

1H \\\\,\\\\\\7 \ H

111 l\\.\\\\.\\\.\ \u

in \\\\\\\\\\71U

ill \\\\\\\\\\\ 111

\\\ \\\\\\\\\1l lll

111 l\\\\\\\\\\ tit

Figure 25. Animated midsection plane velocity proﬁle with 120 scmh ﬂow

67

aunangh
xxx xxxxxx11111 111
151‘ mm;
111 111\\\\\\\\ 111

\\\ \\\\\\l\ll\ 111

191.1 287.5

111 11xxxx\\\\\ 1\1
111 11111111111 111
mo; “”3
111 11111111111 111
111 11111111111 111
229.7 326.2
111 11111111111 111
11 11111111111 111
zwo
111 111/1111111 111

(Figure 25. continued)

' ll)/lllll/////l Ill
(14 /////1111111\'\1

111 1111111111
“‘/(/111111111

1 111

11;/11111111111 111
111 x1111111111 111

1x1 11111111111 1\l

111/(1111111111

111 11111111111
111/(1111111111

111

111

111

  
 
  

4.5 151-3

 

cam angle

Figure 26. Animated inlet plane velocity proﬁle with 205 scmh ﬂow rate
at engine speed 2000 rpm.

181.4

210.!»

229.7

 

l 11

\- 287.6

 

(Figure 26. continued)

  

(1cm: 8.5 m/s)

 

4.8

24.1

43.4

62.7

82.0

70

V

(1 cm I 8.5 this)
n
W
cam angle

“7111111111“ 111 101.3 \\ \\.\\\.\\.1111111
11111111111 111 “1 111111111611

1171111111111” 1“ 120.5 \\ \\\.\.\.\.\.1111 111
m 11111111111111 111 111111111!“

1121111111111 111 139.9 \\ \\\\\\\\\x 111
111 1111111111111 111 \\.\\>-\\\\

i

  

\\I\\\\\\\\\2\:1 159.2 \1 \\\\\\\\\\x 111
111 U in.» \ ll 1§\\\_\§_\\\\\’ 11

\\

 

\\.\\.\.\.\\1 11111 3.5 \\ “11111117111

111 xix-Bxpygm 111 A,\_\.\,\.\.\.\.\.\.x 111

Figure 27. Animated midsection plane velocity proﬁle with 205 scmh ﬂow

rate at engine speed 2000 rpm

181.4

191.1

210.‘

229.7

249.0

\\\.\.\\.1111111 111
111P1111111111‘ 11
\\ \.\.\.\.\.111111 111
111 1_\\,\_\,\\.\\.\.1{111

\1 11111111111 111
111l1\.\.\.\.\.\.\1\(111 '

11y11111111111'111
111 11111111111l111
WIN/[1111111111

u/ yﬂ/{dﬂﬂ \\\

(Figure 27. continued)

71

268.3

287.6

306.9

326.2

345.5

1711//////// //

1H 1.1/1.71.11! H
111 11111///././ //
11)£\ll.//_/j/jj [11.
11211111111111 111
111 1111111111151]
1171/1/1/1111 W
111 11/////////)11

1711111111111
111 1111/1/11]! ll 1.

   

"7'1111111111’11111