 

 

THEEB

lllllllllllllllllllllllllllllllllllllllllHlllllllllll

3 1293 01688 55

This is to certify that the

thesis entitled

A QUANTITATIVE STUDY OF PRIMARY FLOW FIELDS WITHIN
AN AUTOMOTIVE TORQUE CONVERTER USING LASER DOPPLER
VELOCIMETRY

presented by
Matthew Glenn Foster

has been accepted towards fulﬁllment
of the requirements for

Masters degree in Mechanical Engineering

 

 

 

M or professor

Date August 24, 1998

 

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

 

 

 

 

 

 

 

LIBRARY

Michigan State

University

 

 

PLACE IN RETURN BOX

to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/96 gamma-p14

A QUANTITATIVE STUDY OF PRIMARY FLOW FIELDS WITHIN AN
AUTOMOTIVE TORQUE CONVERTER USING LASER DOPPLER
VELOCIMETRY

By

Matthew G. Foster

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

MASTER OF SCIENCE
MECHANICAL ENGINEERING

1998

ABSTRACT

A QUANTITATIVE STUDY OF PRIMARY FLOW FIELDS WITHIN AN
AUTOMOTIVE TORQUE CONVERTER USING LASER DOPPLER
VELOCIMETRY

By

Matthew G. Foster

All modern automotive automatic transmissions require the use of a torque
converter to allow for the transmission of torque from the engine to the drive train.
Although they are commonly used throughout the automotive industry, there is little
understanding of the primary or streamwise ﬂows within the torque converter. This is the
ﬁrst experimental study of streamwise ﬂows through the impeller passages of an
automotive torque converter under operating conditions.

An automotive torque converter with windows inserted between blades of the
impeller allowed for the optical access necessary to conduct Laser Doppler Velocimetry
flow measurements. These measurements were used to produce animations of velocity
profiles of streamwise flow through six planes in the impeller and gap region under two
different operating conditions. Transient mass ﬂow rates that are dependent on the turbine
position were also observed and documented. The measurements have led to a better
understanding of flow conditions within the torque converter under operating conditions
and provide the guidance needed for the advancement of improved computational ﬂuid

dynamics models.

ACKNOWLEDGMENTS

There are many people whose efforts and support allowed my completion in these
endeavors. My wife, Maureen, and children, Alex and Ethan, have provided continual
support and motivation that allows me to look forward to every day regardless of how
difﬁcult the task at hand may be. I would like to thank my parents for the tremendous
amount patience and support given to myself and my family over the years that they
continue to give to this day. I would like to take the opportunity to thank many people who
made this possible; Tom Stuecken for his assistance with everything from repairing my test
rig in an emergency to just helping me ﬁnd that one tool that is always missing, Dr.
Keunchul Lee for always giving me that one hint that made everything clear and for his
expertise with the LDV system and the torque converter which made this project possible,
to Mark Novak for his help in developing the software used to generate the data ﬁles for
animation and for dropping what ever he was doing to help me out when I needed his help,
to Bobbie Slider, for keeping the lab and everyone in it, in order, to Dr. Brereton, Larry
Dalimonte, Jon Darrow, Mahmood Rahi, Mikhail Ejakov, and Hans Hacher for their help,
humor and friendship through it all, and last but not least, to Dr. Harold Schock who gave
me a chance, a chance to work, a chance to learn and a chance to grow as a person. Thank

you.

iii

TABLE OF CONTENTS

LIST OF TABLES ..................................................... vi
LIST OF FIGURES .................................................... vii
CHAPTER 1. INTRODUCTION .......................................... 1
1 . 1 Motivation ........................................................ l
1.2 Design Considerations ............................................... 3
1.3 Problem Deﬁnition ................................................. 5
1.4 Literature Review .................................................. 6
CHAPTER 2. EXPERIMENTAL SETUP ................................. 12
2.1 Laser Doppler Velocimetry .......................................... 12
2.2 Test Rig, Drive motor, Dynamometer motor and Controller ................ 18
2.3 Transmission Oil Circulation and Temperature Control System ............. 21
2.4 Torque Converter Modiﬁcations ...................................... 21
CHAPTER 3. EXPERIMENTAL PROCEDURE ........................... 27
3.1 Performance Testing ............................................... 27
3.2 LDV Torque Converter Test Conditions ................................ 28
3.3 Coordinate Systems ................................................ 28
3.4 Front Window Measurements ........................................ 30
3.5 Side Window Measurements ......................................... 34
3.6 Post Processing of Data ............................................. 36
CHAPTER 4. RESULTS AND DISCUSSION .............................. 46
4.1 Performance Test .................................................. 46
4.2 Front Window Velocity Measurements (Planes l and 2) ................... 49
4.2.1 Plane 1 ....................................................... 51
4.2.2 Plane 2 ....................................................... 60
4.2.3 Summary of Mass Flow rate results for Planes l and 2 .................. 69
4.3 Side Widow Velocity Measurements: Plane 3 ........................... 70
4.3.1 Plane 4 ....................................................... 75
4.3.2 Plane 5 ....................................................... 84
4.3.3 Plane 6 ....................................................... 90
4.3.4 Summary of Mass Flow rate results for Planes 3 through 6 ............... 98

iv

CHAPTER 5. CONCLUSIONS ......................................... 100

5.1 Summary and Conclusions ......................................... 100
CHAPTER 6. RECOMENDATIONS .................................... 104
6.1 Recommendations ................................................ 104
LIST OF REFERENCES ............................................... 106

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

Table 7.

Table 8.

Table 9.

Table 10.

Table 1 1.

Table 12.

Table 13.

LIST OF TABLES

Speciﬁcations of torque converter used .................................................. 23
Conditions used for the flow measurements ........................................... 28
Location of Plane 1 in Global cartesian space ........................................ 51
Measurement area calculated from probe locations in planes 1a and 1b.58

Location of Plane 2 in Global cartesian space ........................................ 61

Measurement area calculated from probe locations in planes 2a and 2b.67

Location of Plane 3 in Global cartesian space ........................................ 71
Measurement area calculated from probe locations in Plane 3 .............. 75
Location of Plane 4 in Global cartesian space ........................................ 77
Measurement area calculated from probe locations in Plane 4 ............ 80
Location of Plane 5 in Global cartesian space ...................................... 84
Measurement area calculated from probe locations in Plane 5 ............ 88
Location of Plane 6 in Global cartesian space ...................................... 90

vi

Figure 1.
Figure 2.
Figure 3.
Figure 4.
Figure 5.
Figure 6.
Figure 7.
Figure 8.

Figure 9.

Figure 10.
Figure 11.
Figure 12.
Figure 13.
Figure 14.
Figure 15.
Figure 16.
Figure 17.
Figure 18.
Figure 19.

Figure 20.

LIST OF FIGURES

A four element torque converter design .................................................. 2
Flow path within converter to be studied ................................................ 3
Laser Doppler Velocimetry system atop of LDV traverse table ........... 13
Back Scatter Schematic ......................................................................... 14
Measurement volume and fringe pattern generated at focal point ........ 16
Test stand with torque converter drive and dynamometer motors ........ 19
LabVIEW VI used by Powertek controller ........................................... 20
Oil cart used to supply bearing and charge oil to test stand .................. 22
Front window torque converter assembly ............................................. 24
Plexiglas window used in front window assembly ............................. 25
Side window impeller assembly and the window used ....................... 26
Coordinate systems used as viewed from the turbine side .................. 3O
Traverse table zero reference point for front window ......................... 31

Orientations of Planes 1 and 2 with respect to the impeller passage ..34

Location of Planes 3 through 6 in relation to the impeller passage ....36

Post processing ﬂowchart .................................................................... 37
Location of impeller passages relative to Global coordinates ............. 39
The three type of cell geometries used in calculating area ................. 43
Orientation of areas used for mass flow rate calculations ................... 44
Comparison of Ford data vs. the front and side window data ............. 47

vii

Figure 21.
Figure 22.
Figure 23.
Figure 24.
Figure 25.
Figure 26.
Figure 27.
Figure 28.
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.

K-factor data from Ford tests and modified converters ...................... 48

Impeller passage and related terminology ........................................... 50
Velocity proﬁle for plane 1a.4, lower mid-chord measurement .......... 53
Velocity proﬁle for Plane 1b.4, lower mid-chord measurement ......... 54
Velocity proﬁle for plane 1a.8, lower mid-chord measurement .......... 56
Velocity proﬁle for plane 1b.8, lower mid-chord measurement ......... 57

Mass ﬂow rate data from planes 1a and 1b for the 0.4 speed ratio ..... 59

Mass ﬂow rate data from planes 1a and 1b for the 0.8 speed ratio ..... 60

Velocity proﬁle for plane 2a.4, upper mid—chord measurement .......... 62
Velocity proﬁle for plane 2b.4, upper mid-chord measurement ......... 63
Velocity proﬁle for plane 2a.8, upper mid-chord measurement .......... 65
Velocity proﬁle for plane 2b.8, upper mid-chord measurement ......... 66

Mass flow rate data from planes 2a and 2b for the 0.4 speed ratio ..... 68
Mass flow rate data from planes 2a and 2b for the 0.8 speed ratio ..... 69
Comparison of all mass flow rate data over all 31 impeller passages.70
Velocity proﬁle for Plane 3.4, 4.5 mm inside impeller passage .......... 73
Velocity profile for Plane 3.8, 4.5 mm inside the impeller passage....74
Mass flow rate data from Plane 3 for the 0.4 speed ratio .................... 76
Mass flow rate data from Plane 3 for the 0.8 speed ratio .................... 76
Velocity proﬁle for Plane 4.4, 2.5 mm inside the impeller passage....78
Velocity proﬁle for Plane 4.8, 2.5 mm inside the impeller passage ....79

Mass flow rate data from Plane 4 for the 0.4 speed ratio .................... 83

viii

Figure 43.
Figure 44.
Figure 45.
Figure 46.
Figure 47.
Figure 48.
Figure 49.
Figure 50.
Figure 51.
Figure 52.
Figure 53.

Figure 54.

Mass flow rate data from Plane 4 for the 0.8 speed ratio .................... 83
Velocity proﬁle for Plane 5.4, 1 mm inside the impeller passage ....... 86
Velocity proﬁle for Plane 5.8, 1 mm inside the impeller passage ....... 87
Mass flow rate data from Plane 5 for the 0.4 speed ratio .................... 89
Mass flow rate data from Plane 5 for the 0.8 speed ratio .................... 89
Velocity proﬁle for Plane 6.4, 1 mm outside the impeller passage ..... 92
Contour map of Plane 6.4, 1 mm outside of the impeller passage ...... 93
Velocity proﬁle for Plane 6.8, 1 mm outside of the impeller passage.95
Contour map of Plane 6.8, 1 mm outside of the impeller passage ...... 96
Mass flow rate data from Plane 6 for the 0.4 speed ratio .................... 97
Mass flow rate data from Plane 6 for the 0.4 speed ratio .................... 97

Comparison of all mass ﬂow rate data over all 31 impeller passages.99

ix

CHAPTER 1. INTRODUCTION

1.1 Motivation

Streamwise ﬂows within production automotive torque converters are poorly
understood. They can not be modeled directly using computational means for several
reasons. Boundary conditions depend on converter geometry as well as characteristics of
the working fluid and relative speeds of components within the converter. It is difﬁcult to
determine boundary conditions for use in computational ﬂuid dynamics (CFD) applications
due to the complex geometry and variable speeds of the torque converter components
within a closed loop. The purpose of this study was to aid CFD theorists using Laser
Doppler Velocimetry (LDV) to determine quantitative parameters for use in the design and
modeling of production torque converters. Determination of these parameters includes
mapping the streamwise velocity proﬁles within the impeller and interface region between
turbine and impeller as well as collecting ﬂow rate information for these same regions. This
study would aid in the construction of higher accuracy CFD models of this complex
turbomachinery device. Eventually this information will allow for those familiar with the
design of the torque converter to better understand what effects design changes have on

performance and efﬁciency through modeling techniques.

Torque converters are designed to be a viscous couple between the engine of an
automobile and the transmission. The torque converter design allows it to multiply torque

for start-up and towing and act as a braking device for the automobile during coasting. A

 

   
    

.0 .l. L l. 1.]... l \\

 

 

 

CRANKSHAFI'
PILOT HUB

OIL PUMP
DRIVE HUB

Figure 1. A four element torque converter design

four element torque converter can be seen in Figure 1. The elements are the cover, the

impeller, the reactor or stator, and the turbine.

The cover is designed out of a heavy gauge steel stamping to withstand high
pressures during start-up. It is bolted to the engine via a ﬂex-plate and rotates at engine
speed. The cover is attached to the impeller by a seam weld. The impeller is a collection of
passages for directing ﬂuid ﬂow into the turbine region and is held in place by another steel
stamping that makes up the transmission side of the converter. The impeller blades are held
in place in this stamping using slots machined into the stamping and a shroud ring that holds
the blades in place via folding tabs. The impeller acts as a centrifugal pump, moving
transmission ﬂuid from within its passages as it rotates to the turbine side to do work by
rotating the turbine. The turbine is a slotted steel stamping containing many blades angled
to generate rotation when ﬂuid ﬂows from the impeller side of the converter to the turbine
side. The turbine drives the transmission input shaft. Between the turbine and the impeller

is the reactor (or stator). It is used to redirect ﬂow exiting the turbine back into the impeller

2

TORQUE MULTIPLYING.
REACTOR REVERSING OIL
FLOW FROM TURBINE

    

IMPELLER

Figure 2. Flow path within converter to be studied

passages for the purpose of maximizing the torque added to the turbine. It is piloted by the
transmission case and rotates on a one way clutch. The impeller and turbine are not
connected in any way allowing them to rotate independent of one another. A description of

ﬂow through the components of the torque converter can be seen in Figure 2.

1.2 Design Considerations

The design of a torque converter must allow it to do four basic functions. It must
allow for smooth transitions between gears and maintain efﬁciency and allow for torque
multiplication for start-up and loaded conditions. It must also allow for direct drive
capabilities for near 100% efficiency when cruising speed is reached by locking up the

converter, this forces the impeller and turbine to rotate at the same speed without slippage.

Finally, it must dampen all torsional vibrations from the engine and the transmission due to
gear changes or loading conditions as felt by the operator. There are a number of methods
used to determine whether or not a torque converter meets these criteria. The most useful
tests are performance tests run over numerous load conditions and speed ratios. When
torque converters are compared via these test results, a number of parameters that
determine the capacity and efﬁciency of the device to allow for direct comparison can be
utilized. There are two basic empirically determined equations used to describe the
behavior of a torque converter as a function of torques, speeds and geometric parameters.
The first equation relates the input torque (Ti) to the input speed (Ni) and the diameter of

the torque converter (D).
Ti = C - Ni -D (1)

The factor “C” is a function of geometric design and working ﬂuid properties. This esoteric
variable is represented best by the “K-factor”. K is represented by collecting the diameter

and the C factor into one variable. The K factor is represented by the relation in Equation 2.

K: —i (2)

ff;

This relationship is used over the full range of torque converter operation. As can be seen
in Equation 2, the K-factor negates the need for ﬂuid or geometric parameters and relies
only on performance based information. The K-factor is used to determine the “capacity”
of a particular torque converter design. The efficiency of the design is determined by
multiplying the torque ratio by the speed ratio. In this relationship Tout and Nout represent

the torque output and speed output respectively.

Eff: . - °“‘ (3)

 

 

The relationships in Equation 2 and Equation 3 have been used to benchmark
converter designs for several years. The down-fall of these relationships is that they require
a converter to be designed and built for a performance test in order to determine the
design’s validity compared with other designs. This is a very costly method of breaking
designs. Current CFD models have not matured to the state that the design of a converter
can be based on conclusions from those models alone. An important development would
allow for improvement of the torque converter design without having to build one, such as

improving conﬁdence in CFD models to provide accurate performance statistics.

1.3 Problem Deﬁnition

Two production torque converters, of the same make, used by Ford Motor
Company have been modiﬁed to allow optical access to the impeller and the gap region
between the impeller and turbine regions. Using Laser Doppler Velocimetry, these two
converters are to be used in a study to determine primary ﬂow characteristics of the torque
converter design in question both quantitative and qualitatively. This will allow those
designing the torque converters to better understand the ﬂow phenomena within these
regions and generate primary ﬂow parameters for use in CFD models for improved model
accuracy and consistency. Improvements in the area of design would lead to reduced cost
and time to market of the part as well as improved performance and fuel efﬁciency for
automobiles implementing the improved torque converter. This will eventually allow for

the development of an optimal design, reducing size and weight.

1.4 Literature Review

Flow visualization of turbomachinery is a well established ﬁeld for devices less
complex than automotive torque converters. Many techniques have been used in
visualizing ﬂows within in the torque converter with various levels of success [1-4]. Upton
(1962) developed a large ﬂow table that allowed for visualization of ﬂow external to a
converter setting. This ﬂow table was heavily used for the study of blade cascades in the
1960’s for the development of torque converters [5]. While the level of understanding of
the ﬂows was rudimentary, it was the ﬁrst time ﬂow visualization was useful as a design

tool for torque converters.

An early attempt at viewing ﬂows within a torque converter of any design was
conducted by Fister and Adrian (1983) [6]. The torque converter design studied was a large
industrial model that required two converters be built for the study. One torque converter
used air as the working ﬂuid for the purpose of conducting spark tracer velocity
measurements and was constructed entirely out of clear plastic. The other converter was
designed for optical access through sight holes into the reactor region and was ﬁlled with
water. A dual focal point measuring method (laser-2-focus) was used to determine the
velocities of the ﬂows in that region. It was determined that even though the working ﬂuid
of the two converters were different, the ﬂow proﬁles and performance data acquired from
the test stands were the same. There was not enough information about the ﬂows to make
any ﬁrm conclusions about the natures of the ﬂow within the converter itself under actual

operating conditions.

By and Maloney (1988) described the ﬂows within the torque converter as highly
three dimensional due to complex geometry passages and the different rotational speeds

6

experienced by the components within the device [7]. The complexity of the torque
converters independent rotation of its components was thought to make the understanding
of the general ﬂow characteristics difficult. It was also concluded that internal ﬂow analysis
could be conducted although the techniques developed to do so may prove very difficult to

perfect.

Bahr, Flack and By (1990) constructed a test stand with a automotive design torque
converter constructed completely out of Plexiglas for the purpose of employing LDV
techniques to determine ﬂow distributions within the stator regions [8]. This experiment
resolved instantaneous measurements with turbine and impeller locations and their effect
on the average ﬂows in the passage using encoders for both impeller drive and output
turbine shafts. Flows were measured within the region of interest near the stall condition
and the 0.8 speed ratio condition. It was found that the passing of the blades caused a strong
effect on the average velocity. Although optical access to the ﬂows within the converter
were very good, the geometry required by the use of the relatively weak Plexiglas required
a blade geometry that was non-conventional. The largest constraint on the experiment was
that the Plexiglas converter was not able to operate at actual automotive torque converter

conditions due to the materials used in its construction.

McCarrack (1993) was the ﬁrst to collect data from a production automotive
converter through optical means, testing at operating conditions [9]. He employed the use
of LDV technology in a feasibility of concept study to determine the parameters needed to
study ﬂows in a Ford Motor Company production toque converter. The work details the
construction of the test stand and controller that would operate the torque converter during

the experiment. McCarrack also details the methods used in determining the correct seed

material and a study of the method of distribution for the seed particles within the converter.
Various testing conditions were attempted by varying speed ratio, ﬂow rates to the
converter and temperatures of operation. Results of 2 component LDV measurements were
animated for the first time showing the effect of the impeller and turbine positions during
rotation on the ensemble averaged ﬂow. In this study it was found that the turbine position
had little effect on the ﬂows within the impeller passage but did have a substantial effect
on ﬂows in the gap region between the impeller and turbine. It was recommended that 3
component measurements be made in the passages to allow for a more thorough study of

ﬂow behavior.

By and Lakshminaraya (1995) used eight static pressure taps to determine
characteristics of the ﬂow within an automotive converter [10]. This experiment required
that after a speed ratio was set, conditions inside the converter were allowed to reach steady
state conditions before measurements were made. Results from this experiment showed
that separation occurs along the shroud of the impeller passage at all speed ratios. This
implies that the efﬁciency of the converter is being reduced since the separation may
require more input torque to support the recirculation in these regions. This supports a

conclusion that ﬂows were erratic and highly three dimensional in nature.

Lakshminaraya and von Backstrom (1996) reviewed techniques used in previous
torque converter studies [11]. The techniques evaluated include static pressure
measurements made at surfaces, ﬁve-hole probes, hot wire velocimetry, fast response
sensors and laser velocimetry. The result of this discussion was the determination that none

of the currently used techniques are ideal for the measurement of such complex ﬂows and

that methods still need to be developed to measure both primary and secondary ﬂows

within the converter under actual operating conditions.

Gruver et a1. (1996) studied ﬂows at the entrance, mid-passage and the exit of the
impeller passage using the same experimental setup as the 1990 study conducted by Bahr,
Flack and By [12]. The study employed the use of LDV technology to measure the region
speciﬁed in the impeller passage at a 0.065 and 0.8 speed ratio with a maximum impeller
rotational rate of 1100 RPM. Velocity proﬁles for the regions studied were generated
showing separation and ﬂow reversal in the passage. They were not able to resolve the
position of the data in time for more than a few location relative to the stator and the turbine
[13]. The proﬁles were not in an impeller frame of reference, making comparisons with

other results such as those through CFD means difﬁcult.

Watanabe et a1. (1997) constructed a torque converter to model ﬂows within an
automotive torque converter by extending the torque converter’s radial dimension without
changing the passage geometry [14]. A cavity within the center of the converter was
constructed and optics placed within. This allowed for the direct measurement of ﬂow
within the reactor using laser sheets. The goal of this study was to determine the effect that
varying the stator blade thickness had on the ﬂows local to the blade. The conclusions
drawn from this experiment were that varying the thickness of the stator blades effected the
torque converter’s capacity coefﬁcient more than the efﬁciency or the torque ratio

characteristics.

Dalimonte (1998) conducted an experiment with the same automotive torque
converter used in McCarrack’s feasibility of concept studies [15]. The work done on the

torque converter mapped 2 component ﬂows at the mid-chord of the impeller passage,

9

secondary ﬂows near the exit of the impeller passage and into the gap region of the impeller
and turbine passages. An attempt was also made to map a component of the streamwise
ﬂows in the impeller exit and gap region using two different torque converters. One
converter had optical access to the impeller passage on the face of the converter and the
second converter had optical access along the circumferential of the torus. The
measurements were made at two operating conditions, 0.4 and 0.8 speed ratio at the
operational speeds of 1600 RPM and 633 RPM for the 0.4 conditions and 2000 RPM and
1600 RPM for the 0.8 set based on a realistic, 50 ft-lb constant input torque. Methods of
post-processing data for animation in an automated fashion were also studied. The results
of this study include animations of the measured ﬂow relative to impeller velocity and
position of the turbine blade. Turbine blade passing effects were observed in the gap
regions in the secondary ﬂows but little effect on streamwise ﬂows was observed.
Preliminary ﬂow rate studies were also conducted on measured data. Recommendations
were made as to where further studies should focus attention. The recommended areas of
interest should be at the impeller exit and at the mid chord regions to better understand the

nature of recirculation and ﬂow variations measured at these locations.

Numerous studies have been carried out on the torque converter using a variety of
techniques both computational and experimental [16-22]. The purpose of these
experiments is to determine ﬂow characteristics that aid in the design of a more effective
torque converter. Very few of these studies have been performed on actual automotive
converters and those that have, required significant modiﬁcation be made to the converter.

Fewer yet have been conducted at actual operating conditions. The studies therefore have

10

results that do not necessarily relate accurate information about operational ﬂow

parameters.

11

CHAPTER 2. EXPERIMENTAL SETUP

2.1 Laser Doppler Velocimetry

Velocity measurements ﬁom within the torque converters were made using Laser
Doppler Velocimetry (LDV). Elements making up the LDV system include an Argon-ion
laser, Colorburst (T81) and 4-beam ﬁberoptic probe using a lens with a 350 mm focal
length. A ﬁequency shifter, digital burst correlator (TSI, IFA750), traverse table allowing
for three degrees of translation and one degree of rotation and an IBM compatible PC for
data collection and storage are also required for this experiment. LDV has been used for
non-intrusive ﬂow studies for many years and is one of a very small group of technologies
that can measure ﬂows within closed systems. It requires no calibration under changing
conditions and measures ﬂows, without effecting the ﬂuid ﬂow. It uses only light and a
non-homogeneous scattering particle to obtain velocity measurements. Figure 3 shows the
LDV laser and associated hardware atop the traverse table. The traverse table is used to

position the probe from which the lasers are emitted and velocity information is received.

Laser Doppler theory states that the velocity of a particle can be measured by
detecting the Doppler shift frequency of laser light scattered by the particle. The particles
used must be approximately 10 um across. The particles used in the torque converter were
the correct size, 9 pm in diameter, had a high index of refraction and the ability to follow
ﬂow characteristics of the ﬂuid down to 0.1 m/s. The capability of the seed particle to

follow ﬂow characteristics was determined by determining the sedimentation velocity

12

 

Figure 3. Laser Doppler Velocimetry system atop of LDV traverse table

calculating the distance the particle has dropped due to gravitational forces in a ﬂow of a
known velocity over a given distance. If this vertical distance is less than 20% of the ﬁxed
distance through the given distance, the particle shows acceptable characteristics for
following that ﬂow velocity. The particles were plastic core and metallic coated with a
speciﬁc gravity of 2.6.

The scattered light is collected by a photomultiplier. The function of the
photomultiplier is to convert bursts of light scattered in the ﬁ'inge pattern into an analog
signal for interpretation by the data acquisition equipment (IFA-750). The data acquisition
equipment then combines the frequency of the analog signals with a set of encoder values
relating to the position of the impeller and turbine. This related data set is stored in a PC
connected to the data acquisition hardware. A schematic of these events can be seen in

Figure 4. The schematic shows a back-scattering system, where the bursts of light scattered

l3

Receiving Fibers Receiving Lens

Transmitting Fibers Focusing Lens

M \_f_/

Collunators

 

 

    

   

 
  

A—A

   

Fiber Optic

Junction Signal Processor IFA 750

Figure 4. Back Scatter Schematic

from the measurement volume are collected at a physical location in front of the focal point.
In back-scatter mode, the intensity of scattered light is one thousand times weaker than
scattered light intensities using the forward-scattering method. Back-scatter data collection

was used in this experiment due to optical access constraints.

The point of measurement is a volume determined by the half angle of the beams
(K) intersecting at the focal point and the wavelength of light used (3.). This volume is in
the shape of an ellipsoid with the length governed by Equation 4. The height of the ellipsoid

is determined by Equation 5.

4f)»
1 = —— 4
m nDe-2sinK ()

4H.
d =——— s
m nDe_2cosr< (‘)

The measurement volume is thus deﬁned by Equation 6

3
v ___ (40¢ . 1 (6)

D3432 6((cosrc)2 - sink)
e

Where r is the focal length, it represents the wavelength of the light used and 1),;2
represents the diameter of the unfocused Gaussian laser beam measured at Me2 of the
center line intensity. The frequency of the signal converted by the photodetector increases
proportionally to the half angle value. Therefore, the half angle can limit the range of
measurable velocities. Velocity measurements have been made from 1 rim/sec to 1000 m/
sec [23]. This wide range of measurements is made possrble, in part, by changing the half
angle to allow for measurement of the velocity range. The LDV setup used in this
experiment used a half angle of 4.0°. The half angle generates a fringe pattern when two
beams of polarized laser light interfere with each other at the focal point. Figure 5 depicts

the measurement volume at the focal point and the generated fringe pattern.

As the beams of the unshifted and that of a shifted beam cross, a fringe pattern
moving in a known direction, is formed. Shifting the frequency of a laser beam refers to
changing the frequency of the laser light by a known quantity either up or down. Frequency
shifting is commonly used in LDV measurements. The purpose for this is to measure ﬂow
reversal, to measure small component velocities in the presence of large component
velocities and to increase the range of velocities that can be measured by the signal
processor. The frequency of the signal collected by the signal processor is ﬁxed at 40MHz

by a Bragg cell. This ﬁ'equency is then down mixed to a value greater than twice the

15

   
   
 
 

 

The measurement
volume is located
at the focal point

The component of velocity measured
is that which lies in plane with the
two laser beams, perpendicular to
the centerline between the beams.

Figure 5. Measurement volume and fringe pattern generated at focal point

Doppler frequency for the corresponding negative velocity. The direction of the ﬁinge
motion is normally oriented in a direction opposing the mean positive ﬂow. Thus any
velocity measured that has a frequency less than the shift value is negative and those with
greater frequencies are positive. The formula used to resolve positive and negative velocity

measurements is as follows.

Frequency = ShiftFrequency j: IVelocityl /(FringeSpacing) (7)

The velocity value used in Equation 7 is determined by the result of Equation 8.
Velocity = DopplerFrequency - FringeSpacing (8)

l6

The Doppler frequency is simply the frequency at which the signal processor receives light

burst signals from the incidence of seed particles passing through the focal point.

The data acquisition equipment has been pre-programed with the characteristics of
the laser beam and its shiﬁ parameters. The data acquisition equipment calculates the ﬁinge
spacing using the wavelength of the laser light and the half angle between the beams as seen

in Equation 9.

 

A
d = 9
f 2sinrc ( )

The velocity of the particle passing through the fringe pattern is then calculated by
multiplying the fringe spacing by the frequency of the Doppler shifted light scattered by the
particle. The frequency of Doppler shifted light is equal to the number of fringe patterns
scattered in a ﬁxed amount of time over a known distance and is measured in Hertz (Hz).
This information can then be used to infer direction and velocity directly by noting the
orientation of the shifted beam to the unshiﬁed beam and their orientation in Cartesian
space. These calculations are carried out in the IFA 750 signal processor which collects
data from all the sensing equipment used. The information regarding the adjusted
frequency, representing the velocity of the particles through the measurement volume, is
then stored in a PC running TSI PHASE software. The TSI PHASE software is used to

control and program the data acquisition equipment.

The laser used was a multi-frequency Argon-ion laser capable of producing three
wave lengths of light. The wave lengths of light are green at 514.5 nm, blue at 488 nm and
violet at 476.5 nm. This experiment used the green wavelength. These wavelengths are

separated by a device called the ColorBurst, manufactured by TSI. It is composed of a set

17

of optics used as beam splitters for certain wavelengths of light. Once separated, each
distinct wavelength of laser light can be routed via ﬁber optics to the LDV probe face where
their orientations, on the face of the probe, determine which velocity components they will

measure.

2.2 Test Rig, Drive motor, Dynamometer motor and Controller

The torque converter ﬁxture is a modiﬁed performance test rig designed to allow
optical access into the torque converter via the available window conﬁgurations. The test
rig consists of a large bedplate on which the torque converter stand is mounted. Two motors
are connected to the impeller and the turbine using belts. Driving the impeller section of the
torque converter is a 50 Hp DC motor. A 50 Hp Universal Dynamometer is connected to
the output shaft on the turbine assembly. Figure 6 shows the test stand with the green
Universal Dynamometer in the foreground, the blue drive motor in the aft below the
bedplate and the test ﬁxture on the top. The drive motor is used to simulate the engine
output to the impeller, while the dynamometer simulates the load conditions placed on the
torque converter. The operational limits of the drive motor are 2500 RPM which produce a
5000 RPM rotation rate for the impeller and shell of the torque converter due to the 2:1
gearing ratio. The Universal Dynamometer is set at a 1:1 gearing ratio and has an upper

limit of 5000 RPM on its operation.

The bearing assembly connecting the torque converter to the motors has several
features. This assembly allows oil to ﬂow in and out of the torque converter, as well as
access to the bearings for oil lubrication. The design of the bearing assemblies includes

strain gauges attached to the shafts of the stator, the turbine and the impeller. By measuring

18

 

Figure 6. Test stand with torque converter drive and dynamometer motors

the torsional deﬂection of the shaft and the knowing the shafts’ material and geometric
properties it is possible to compute the torque placed on the shafts by the stator, turbine and
the impeller. This allows for input and output torques to be recorded in real time. At either
end of the bearing assemblies, attached to the shafts, are a set of encoders used to
determining angular location during rotation. There are encoders for both the impeller and
the turbine. Each encoder generates one square pulse per revolution (CPR) and two sets of
1024 square pulses per revolution. This gives the encoders a resolution of 009°. They are
connected to the Powertek controller. The Powertek controller records the encoder OPR
signal for cycle information and transmits the encoder pulses to a set of rotating machinery
resolvers (RMR’s). The RMR’s are used to correlate encoder pulses ﬁ'om both encoders

and pass the information on to the IF A 750. The [PA 750 processes LDV measurement

i

_ .. _._. . __“md

.ml no." 3.1m . u 1' -' w lLu‘ -w»l x.“

1mm Minna-t
'1'. w 1 m 11
mm! m in» ‘; " " ‘ 1 .un rtvrw
”urn n “M ‘1' H I .— 7! went: 31w.
-.‘ i‘iil'i :r > I -' ‘r'r Hm} H r.

K 1 I1}; 1’) - r "19.» I
r. _— - .-. , " ."~ » .. .‘ ' r ‘ z.

121mm 5 at u PM
5m“:
' the rar- -

M We

“not N.- [-

',-.

 

Figure 7. LabVIEW VI used by Powertek controller

information in conjunction with the angular location data delivered via the RMR’s. There

is a RMR device for both the impeller and the turbine encoders.

The Powertek controller serves several functions; it monitors and sets conditions to
be studied in the torque converter, it contains the majority of the fail-safe checks on the test
rig and operates independent of the LDV data acquisition equipment. The controller
calculates input and feedback torque from the impeller, stator and the turbine shaﬁs. It also
monitors charge, discharge and bearing oil pressures, temperatures and ﬂow rates. It
controls the drive motor RPM and the dynamometer RPM internally through independent
controllers. The Powertek controller is a complex data acquisition system. The user

operates the controller with a LabVIEW software interface on an IBM PC compatible

20

computer. This interface, known as a VI or virtual instrument, relays information being

monitored back to the operator. The V1 interface can be seen in Figure 7.

2.3 Transmission Oil Circulation and Temperature Control System

The combination of the controller and oil supply system allows for steady state
torque converter operations in a repeatable manner. The oil supply cart, seen in Figure 8,
controls the ﬂow of oil to and from the test stand for the bearing assembly and the torque
converter as well as controlling temperatures and pressures of the oil. Regulators are used
to control pressures and ﬂow rates of the oil circulation system for both the bearing
assembly and the torque converter. The temperature of the bearing and the transmission oil
is controlled via a water cooled heat exchanger. This allows for control of the temperature
conditions in the torque converter, limiting the chance of over heating either the converter
or the bearings. The oil supply system also aides in developing steady state conditions of
the pressures and temperatures in the torque converter. The systems involved in circulating
the oil within the torque converter and the bearing assembly are independent of each other.
Oil from the bearing system is sent through a ﬁlter and is circulated by a vacuum pump.
The transmission oil is not ﬁltered, but sent through a reservoir where seed particles used

in determining the ﬂow velocity can be added and transmission oil changes can be made.

2.4 Torque Converter Modiﬁcations

The automotive torque converters used in this experiment are Ford Motor Company
production design torque converters. They have been modiﬁed to allow for access to the
inside of the converter as well as the addition of plexiglas windows. These windows

provide optical access to the flows within the impeller passage, exit and the gap between

21

 

Figure 8. Oil cart used to supply bearing and charge oil to test stand

the impeller and turbine assemblies. Initial modiﬁcations to the converter included welding
a ﬂange along the circumferential rim of both the impeller side and engine side of the
housing. The converter can then be bolted together with cap screws and this allows access
to the interior of the torque converter. Next, access windows were cut through the shell of
the impeller housing. Window mounts were then welded to the shell allowing for the
installation of acrylic windows fashioned to ﬁt into the geometry of the passage on the
interior side.

Two converter window conﬁgurations resulted The front window is used for
measurements within the impeller passage and secondary ﬂow measurements in the gap
region between impeller and turbine assemblies. The side window, allows measurement of
ﬂows traveling from the impeller exit to the turbine region along the axial component. The
window design required that the material used for the window have similar index of
refraction as the transmission oil. The material must be shaped to follow the contour of the

22

internal geometry of the torque converter so the installation would not disturb the ﬂow. The

speciﬁcation of the torque converter design used can be seen in Table 1.

Table 1. Speciﬁcations of torque converter used
Converter Speciﬁcations

 

 

 

Diameter 291.4 mm
Axial Length 95.5 mm
# Impeller Blades 31
# Turbine Blades 27

 

 

 

 

 

 

 

The front window converter has an acrylic window with an index of refraction of
1.48 ~ 1.5. This value corresponds to the Shell 212 transmission oil’s index of reﬂection of
1.49, used in the experiment. These nearly equal indexes of refraction assure that the beams
from the laser and back scattered light will only reﬂect at the interface between air and the
window itself. The interface between the oil and the window media will have a negligible

amount of refraction. The angle of refraction is governed by Snells’s Law.

NI sine = stintb (10)
Where N1 represents the reﬁactive index of air and N2 represent the index of refraction for

Plexiglas window. The angles 0 and 9 represent the incident angle of the laser from the
normal of the window surface and the angle from the same normal that retracts
respectively.

The location of the front window is on the transmission side of the converter. Direct
optical access can be gained into an impeller passage from just below the halfway point up
the passage to the impeller exit. The optical access through this window allows for

measurements of the ﬂow passing through the impeller passage, impeller exit and gap

region to be measured depending on the type of ﬂow requiring measurement. The front

23

 

Figure 9. Front window torque converter assembly

window assembly allows for streamwise ﬂow measurement along the impeller passage
from the mid-chord to near the exit of the ﬂow from the impeller section of the converter.
Secondary ﬂows can be measured off axis starting from bottom of the window to the top.

The torque converter assembly with the ﬁ'ont window modiﬁcation can be seen in Figure 9.

The implementation of the window requires it to be secured to the window mount
via cap screws. Considerations regarding the temperatures and stresses the window
experiences under operation must be made since operating at high temperatures can weaken
the window and cause deformation or mechanical failure of the window. Stressing
Plexiglas can cause mechanical birefringent photoelastic properties of the material to effect
the laser light used. Under unstressed conditions, mechanical birefringence does not occur,

so it does not effect the polarization of the light passing through the window. However,

24

 

Figure 10. Plexiglas window used in front window assembly

under stressed conditions, the polarization will be shifted causing the fringe patterns that
normally occur in the focal region to diminish or not occur at all. The front window itself

can be seen in Figure 10.

Side window modiﬁcations were made using the same procedure and materials.
The Plexiglas window was placed at the top of the impeller passage providing for optical
access into the impeller exit and the gap region between the impeller and the turbine shell.
This window allows for two component measurements of the primary and tangential ﬂows

through these regions or single component measurements of either ﬂow direction.

The complex contour of the inner surface of the window varies greatly in all three
dimensions. This can cause numerous difﬁculties at the interface between the oil and the
window media. In theory, the divergence of the beams in plane or deﬂection of a single
beam above or below the location of the focal point, could occur eliminating the focal point
entirely or reducing the size of the measurement volume and thus effecting the data rate.

Since the indexes of refraction for both types of media present are approximately the same,

25

 

Figure 1 1. Side window impeller assembly and the window used

the effect of the angle of incidence is greatly reduced. The amount of refraction that occurs
at this interface has a negligible effect on the location or focal position. The impeller

assembly with side window modiﬁcations can be seen in Figure 11.

CHAPTER 3. EXPERIMENTAL PROCEDURE

3.1 Performance Testing

A constant input performance test was run on both front and the side window torque
converter conﬁgurations. A constant input performance test requires applying a constant
torque to the impeller as the turbine RPM is increased by increments of 200 RPM to lock-
up conditions. The performance test was run at a set of operating conditions speciﬁed by
Ford Motor Company based on a performance test run previously. The Ford data was ﬁ'om
the same model torque converter without window or accessibility modiﬁcations. This was
done to compare performance data with the previous Ford performance test results. By
matching the Ford performance test, it is likely that the modiﬁed converters used in the ﬂow
studies would have ﬂow conditions behaving in a similar fashion to those in a non-modiﬁed
converter. The conditions of the performance test ranged from turbine speeds of 400 RPM
to 2000 RPM with the impeller torque feedback at 67 N-m of torque and the RPM necessary
to maintain the constant torque constraint. The pressures, temperatures and ﬂow rates of the
charge, discharge and drain oil as well as torque measurements of the turbine, stator and the
impeller were recorded on the Powertek controller. The data was later used for analysis
purposes when comparing the computed K-factor and the efﬁciency to the values recorded

by the Ford performance test.

27

3.2 LDV Torque Converter Test Conditions

The point of using a production torque converter to take measurements is to
measure actual conditions occurring during operation in a vehicle. The design of the data
acquisition and post-processing system allows it to ensemble average measured velocities
using encoder values of 1 025° about 05° increments across the impeller and turbine
passage. Since an averaging method is used, conditions contributing to the ﬂow behavior,
such as temperatures and pressures of charge and discharge oil from the converter, should
not be changed signiﬁcantly. This is done in order for the average to be representative of
the actual ﬂow behavior. Although small pertubations are present in the values measured,
the average of the ﬂuctuating values is set to correspond to the desired operating conditions.
Therefore, slight variations about a constant value allow for a good representation of the
desired steady state conditions. This requires running the torque converter for several
minutes prior to taking any measmements in order for the temperature and the pressures to
attain a steady state value at the measurement speed ratios. The conditions held at steady

state for the measurements taken can be seen in Table 2

Table 2. Conditions used for the ﬂow measurements
0.4 Speed Ratio (1600/633 RPM) 0.8 5110011 Ratio (2000/1600 RPM)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Oil Temp (°C) Pressure (PsiG) Oil Temp (°C) Pressure (PsiG)
Vent 40 75.62 Vent [30 76.16
Charge 40 59.83 Charge 30 65.16
Discharge 65 33.86 Discharge 50 36.79
3.3 Coordinate Systems

The LDV traverse table is designed for translation in 3 dimensions. It is equipped

with stepper motors for automated translation with a maximum resolution of 1 pm. There

28

are two coordinate systems used during the collection and post-processing of the measured
velocities; a Global coordinate system containing an absolute reference from which all
locations in space can be determined and a local coordinate system, positioned arbitrarily
in the Global coordinate system’s space for the purpose of simplifying spatial referencing.
The local coordinate system is set by the LDV traverse table and requires a reference point
for all of its translational movement to be relative to. A location different than that used by
the torque converter blueprints was chosen as the local coordinate system reference. Setting
the local coordinate system is done by positioning the LDV laser focal point on the test
stand or converter itself and referencing the local coordinate system from that location. The
Global origin, as set by the torque converter blueprints, are located at a position within the
converter that is impossible to reference by using this method. In addition to not being able
to reference the same location as used in the Global system, the X and Y traverse table
translation axis system is swapped in relation to the Global system. Therefore, for the ﬁ'ont
window setup, a point on the test stand was chosen that would simplify the translation from
the local coordinate system to the Global system. The side window conﬁguration also used
the same local coordinate system as the front window setup, but with a different reference
location. A comparison of the Ford coordinate system orientation and the local coordinate
system orientation can be seen in Figure 12. With respect to the turbine and impeller
sections of the torque converter, which the torus in the ﬁgure is meant to represent, the
perspective is that of viewing the ﬁgure ﬁ'om the turbine side of the converter. The Z axis
in the Global coordinate system refers to the vertical direction in space and the Y axis refers
to the axis of rotation and follows the axial center line of the assembly. By referencing a

point common to both Global and local systems, positions in both coordinate systems can

29

Z axis represents
vertical axis

  
 
 
     
 
 
   
   
 

Direction

1 of Rotation

Y 05518
represe ;- _

   
  

Local system
(LD V Coordinate
System)

‘I

   

Global Coordinate System“?

Figure 12. Coordinate systems used as viewed from the turbine side

be interchanged. Measurements made in the side window conﬁguration are parallel to the

Global X-Y plane. Front window conﬁgurations are made on planes that require a

deﬁnition in 3 dimensions.

3.4 Front Window Measurements

After determining that all sensors and ﬂow rate meters were functional and the

optical qualities of the oil had not suffered due to oxidation, a number of steps were

followed to setup the LDV system hardware and the references for the coordinate systems.

The ﬁrst step was to set the local coordinate system so that the location of the focal point

could be determined in space relative to a ﬁxed point on the LDV test rig. The point chosen

30

 

 

Traverse Table / b X
Reference Point /
Front window

Torque Converter
Test Fixture .

 

 

 

Figure 13. Traverse table zero reference point for front window

to reference the local coordinate system was along the vertical axis at a point near the top

of the converter ﬁxture as seen in Figure 13.

This is done by positioning the laser focal point at the location chosen on the test
ﬁxture and setting the traverse table coordinate measuring system to zero values for all
three coordinates. The position of the focal point is then known to within 1 0.25 mm along
each Global coordinate axis. This results in a position uncertainty of the plane of data

acquired to be within : 0.25 mm of its recorded position along the Global coordinate axis.

The angular location of the impeller portion of the torque converter is then
referenced to the local and Global coordinate systems. This is done by choosing a point on
the shell of the converter that represents a known location in the Global coordinate system
and setting the encoder zero degree location at a position that allows for easy relation to the
local coordinate system. Knowing where a position in the Global coordinate system lies in
relation to the local coordinate system, allows all positions recorded in the local coordinate
system to be transferred to the Global coordinate system. Setting the encoder required that
the laser be focused on a known location on the shell of the converter and then using a

Tektronix 2445 oscilloscope to set the encoder to zero degrees at that position. The

31

oscilloscope is used to view the trigger signal which represents a full rotation of the encoder

ﬁ'om the zero position.

The turbine encoder setup requires a similar procedure where the focal point of the
laser is located at a location representing a known position in the Global coordinate system.
This must be done while viewing the turbine though the window conﬁguration used. The
location of the zero points chosen for use with the ﬂoat and side window conﬁgurations
was selected to allow for maximum optical access to the ﬂow within the converter. This
required the rotation of the ﬁont window converter to 27.6° oﬁ‘ the vertical axis,
maximizing the width of the window through which data could be acquired. That angle
matches the angle of the vertical edges of the window following the curvature of the blades,
making the window edge a line parallel to the vertical axis when rotated to this angle. Since
the beams from the LDV probe were oriented to view streamwise ﬂow through the passage
in this window conﬁguration, they lie in the vertical plane. The orientation of the edge of
the window parallel to this plane causes both beams to be broken at once, thus allowing for
the complete window width to be measured. The component of velocity measured was

therefore 27.6° off axis, parallel to the blades.

Uncertainties of the encoder values after zeroing are 1 2° about the zeroing point
centered along the Global Y axis in the Global X-Z plane for both the turbine and impeller
encoders. This results in a uncertainty of angular position of the entire plane of data

collected to be at maximum 1 2°.

All measurements required the system to reach steady state pressures and
temperatures. Therefore the test stand was allowed to run for several minutes at the

measurement speed ratio before data acquisition was attempted.

32

Using this window conﬁguration, two planes were measured at speeds of 1600 and
2000 RPM, 1 3 RPM, with the respective turbine speeds set at 633 and 1600 RPM, : 3
RPM, representing two speed ratios, 0.4 and 0.8. The orientation of the planes in 3-
dimensional space were such that they were orthogonal to the streamwise ﬂow traveling
through the plane. The positioning of the plane requires the calculation of the focal point
location as it travels from the LDV probe face through the air and reﬁacts at the window
surface to the focal location in the oil. By knowing the orientation and location of the focal
point in 3—dimensional space it was then possible to calculate the slope of the plane to be
measured by using the geometry of the passage to determine the average slope of the shell
side and the shroud side of the passage and orient the probe to measure ﬂow velocities
orthogonal to that slope. The position of the ﬁrst plane was chosen so that the streamwise
ﬂow would be traveling vertically requiring a plane with an inclination of 0° degrees from
the Global Y axis in the Y-Z plane within the passage. This measurement occurred 91.45
mm from the axis of rotation along the Global 2 axis and can be seen in spatial relation to

the passage itself in Figure 14.

The second plane measured through the front window was oriented 7.96° ﬁom
horizontal in the passage. It was located 93.74 mm along the Global Z axis from the axis of
rotation on the shroud surface within the passage. The location of the plane, in relation to
the passage, can be seen in Figure 14 as well. The inclination of the plane required an
inclination of the probe through 12° ﬁ'om the Global Y axis in the Y-Z plane due to
reﬂection through the window. Each plane was generated by measuring velocities along
lines following the path of rotation. Each line was separated by a distance of 0.5 mm in the

oil, requiring a 0.335 mm increment along the Global Y axis direction for Plane 1 and 0.32

33

   
 
   
  

Plane 2. Oriented 7. 96° Direction of
from the Y axis in the Rotation

Y-Z plane
Plane 1. Oriented R
' 0° from the

Y axis in the
Y-Z plane.

     
  
   
 

Perspective through
Approximate ﬂow the shell into the
direction through impeller passage

the passage ‘5

Figure 14. Orientations of Planes l and 2 with respect to the impeller passage

mm distance increment of the probe along the hypotenuse of the triangle formed by angling
the plane at 796°. The difference in incrementation used for the 0° inclined plane is due to
refraction of the beams through the window, both increment values result in a 0.5 mm

increment in oil.
3.5 Side Window Measurements

The side window converter was set up for the purpose of measuring four planes near
the exit and outside the exit of the impeller passage. The local coordinate system was set
using a procedure similar to that used to set the front window conﬁguration encoders. One
difference in the side window as compared to the front window setup was the use of a

location on the converter itself as the zero for the local coordinate system. The zero was set

34

by positioning the focal point of the laser on the seam that was formed by fastening the
transmission or impeller side and the engine side of the torque converter together even with
the axis of rotation. The laser focal point was set at a point on the impeller housing that was
known under the Global coordinate system for the purpose of setting the impeller encoder.
The turbine encoder was then set by locating a position on the turbine shroud known in the
Global coordinate system and setting the encoders based on its location through the
window even with the center of rotation. The four planes measured through this window
were all measured with the converter window rotated 17° from the horizontal. The probe
was rotated 12.6° clockwise from horizontal to align the plane the beams were in with the
edge of the window that would break the beams during measurement. This maximizes the

width of the window through which data can be acquired.

In order to ensure steady state temperatures and pressures in the system, it was run
several minutes at the measurement speed ratio before data acquisition was attempted. The
planes measured were 4.5 mm, 2.5 mm and 1 mm from the exit of the impeller passage,
with one plane 1 mm outside the impeller passage in the gap region. The planes and their

orientation in 3-dimensional space to the impeller passage can be seen in Figure 15.

Each plane measured streamwise velocities, indexed by the impeller and turbine
encoder values, along the path of rotation. The cartesian location of the probe orientation
in the LDV coordinate system was recorded by hand for every line measured. The
corresponding velocity values and encoder values were combined in the [PA 750 and
recorded in the IBM PC controlling the LDV system. The distance into the converter itself

between lines of measurement was 0.5 mm, requiring a 0.335 mm increment of the beams

35

Plane 5
Plane 4 . .
ll Direction of
Rotation

v-\

    
     
 

W

Perspective through

     
  

Approximate the shell into the
direction of ﬂow impeller passage
through the passage

Figure 15. Location of Planes 3 through 6 in relation to the impeller passage

in air into the converter through the window. The differences of incrementation in oil and

air are caused by the refraction of the beams as they pass ﬁom air into the window.

3.6 Post Processing of Data

After the data set for an entire plane has been stored on the controlling PC running
the TSI PHASE soﬁware, the data ﬁles are run through a number of post processors for
conversion of ﬁle formats and calculations. A ﬂow chart of the post processing software
the data ﬁles are run through can be seen in Figure 16. Initially, the data is in a proprietary

binary format that is not immediately useful for ﬂow visualization or calculations. The

36

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

* M th AD
[Raw Data -> Velocity Files SGI Animation Az'mfﬁon
* I f g A 4
PH TQ VEL Output File FRDARRA Y
Statistical Files SW?" , (””15") S" g"
V. V iles ‘ . MASSF LOW
f Soﬂw are Software

 

 

 

 

 

 

Figure 16. Post processing flowchart

binary ﬁle contains all information pertaining to the shiﬁ frequency, shift direction, ﬁ'inge
spacing, half angle, as well as Doppler shift frequency data and the encoder values these
frequencies correspond to. Conversion of this ﬁle to an ASCII, or text ﬁle, is completed by
software provided by TSI. It uses the fringe spacing, shift information and frequency
information to determine the actual velocity (in the frequency domain) and transfers all
encoder and cycle information to ASCII format. With the data in ASCII format, it is then
processed by a FORTRAN program named PHTQVEL to change velocity information

from the frequency domain to the time domain (meters per second).

The PHTQVEL software produces an output that has all velocity information in
terms of meters per second as well as encoder values in ASCII format. The PHTQVEL
software has been modiﬁed to handle data ﬁles in batch mode, requiring only information
about a sequence of ﬁles to be processed be passed on to the soﬁware by the user. This
allows for the processing of an entire plane of data at a time and reduces the amount of user
intervention needed to complete the process. The output of this soﬁware is a ﬁle with the
distinction of the “v.v” in its ﬁle name, denoting the conversion from frequency to time

domain on the ﬁle.

37

The v.v ﬁle contains a portion of the information needed to process the data. In its
form in the v.v ﬁle, all the velocity data and encoder information are present but there is no
information regarding its location in 3-dimensional space in either the local or Global
coordinate systems. A primary goal of the post processing is to generate an output that
allows those analyzing the data to do so while understanding its relationship to the spatial
relationship to the impeller blades, shroud and shell surfaces. There is also a need for
relating the data over a period of time. This period is deﬁned as the time it takes for a single
passage within the turbine to pass ﬁ'om its initial location (given by blueprint information)
across the impeller passage. Thus, there is a need to process the velocity and angular
location information and add information regarding where in space the measurements were
made in reference to either the Global or local coordinate system. In order to maintain
consistency with a standardized coordinate system, the coordinate system of choice was the
Ford coordinate system. This requires a transfer of coordinate system information from the
local coordinate system, where all measurements were made and reference zeros were set,
to the Global coordinate system. The format for the output of the next post processing
program is therefore cartesian Global coordinates with velocity information in sets that can
be selected by turbine location. The animation software requires X, Y and Z coordinate
information, animation We value, and U, V and W values for velocity vectors. The
Global coordinate system can be seen in Figure 17. The LDV coordinate system not only

has a different origin, but the Y and X axis are reversed.

The TQANLIZR FORTRAN code was written to convert the “v.v” ﬁles of an entire
plane into an animation output ﬁle for use on the SGI Explorer animation soﬁware. The

TQANLIZR code requires that information regarding the sequence of the “v.v” ﬁle set,

38

in"

 

Figure 17. Location of impeller passages relative to Global coordinates

coordinate information in local coordinates, shift direction and hardware speed be speciﬁed
by the user. This information is then used to generate a batch ﬁle that allows for the
processing of each data ﬁle sequentially. Aﬁer generating a batch ﬁle, the software will
never need to generate another batch ﬁle for the data set. It will read the batch ﬁle
information, requiring only the name of the batch ﬁle be entered by the user. The batch ﬁles
are read into the program and each data ﬁle is processed in order. Before each data ﬁle is
read, information regarding the focal point location is analyzed and constraints are
calculated based on the radial location of the measurements within the passage. Data
indexed by encoder values must have a maximum and minimum value. The maximum and
minimum angles correspond to the blade locations on the suction and pressure side of the

passage at the radial location of measurement. Determining the maximum and minimum

39

angles within the passage is done by determining the radial focal position through one of
the three optical routines designed to compensate for refraction in the windows. A cubic
spline routine is used to determine at what location in 2-d space the radial measurements

would strike the blade surfaces when following the path of rotation.

An adjustment must be made to all the encoder values to transfer them from the
local coordinate system to the Global coordinate system. This adjustment requires that a
change be made to the software every time the impeller and turbine encoders are set. The
location of the impeller and turbine encoder zeros relate to positions in the Global
coordinate system along the impeller and turbine blades. By using the location of the
encoder zeros in the Global coordinate system, a constant angle of adjustment (the angle of
the zero point in the X-Z plane from the Z axis) can be subtracted oﬂ‘ of the encoder values
to adjust them from the local coordinate system in which they were measured to the Global

coordinate system.

With the adjustment made, the encoder values for the impeller and turbine are run
through an algorithm to determine where in the passage the point corresponding to the
encoder values should lie. The impeller encoder values are straight forward in their
calculation for location in space. This is due to their correspondence with the area available
within the window that can be measured, plus one passage width. The presence of the extra
passage width is due to the encoder zeroing technique that chooses a point in the previous
passage as the zero location. By subtracting the extra passage width off the encoder value
the remaining angle is always less than 11.6°, or the width of the impeller passage. Using
the radius of measurement and this angle, the cartesian location in Global space is easily

determined.

40

The turbine angle calculations are more complex. The turbine angle represents a
unit of time in the animations and therefore doesn’t imply position in space for the data
collected. It does determine at which time the data was collected. The turbine angles are
over a full 360° range and a modulus ﬁinction is run on the encoder values to reduce it to
the angle within one turbine passage width on either side of the passage of interest. This
wide range of angles is the result of the adjustment made to the encoder values to transfer
the angular location from local space to Global space. Relational operations are used to
reduce the angle further to ﬁt correctly within the range of 0° and 133° after the adjustment
angle has been added back on. This is the width of the turbine passage. After determining
the correct location of the angles in space and time, they are collected into groups of : 0.25°
degrees around each O.5° increment across the minimum and maximum angles at that

radius of measurement.

Since the velocities are collected as one component off axis in both front and side
window conﬁgurations, u and w components (tangential and vertical) are required to
represent the velocity in the Global coordinate system for the front window data set and v
and u (axial. and tangential) are required for the side window data sets. The velocities along
the tangential components must have the tangential velocity of hardware speed subtracted
off their value in order to transfer the data into the impeller’s frame of reference. The
velocities of each group of impeller and turbine angles are ensemble averaged and stored
in an array with a corresponding X, Y, Z and turbine angle scaled from 1 to 27. The scaling
of the turbine angle across this range is done to account for animation frame requirements
of the SGI soﬁware. Every data ﬁle is sent through this routine and the array values for the

output ﬁle are written, resulting in a data ﬁle with 27 frames of animation data.

41

The output ﬁle of the TQANLIZR code can be sent to the SGI software for
animation and detailed analysis of flow trends in virtual 3-d space. The data ﬁle sent to SGI
animation software is mapped into position in 3-d space within a wire-frame of the impeller
passage complete with a turbine blade rotating to the proper position per frame of
animation.

The output ﬁle can be used in two other codes written for analysis purposes. The
MASSF LOW FORTAN code has 3 functions. It will read the “frd” ﬁle and generate mass
ﬂow rate information based on turbine position, as well as per vector location. It will also
average the 27 frames of velocity data into a single frame. The flow rate per turbine angle
is most useful for measuring the variation of the mass ﬂow rate over the 27 turbine
positions, and is written to a ﬁle called F LOWPTURB.frd. The mass ﬂow rate per vector
is called F LOWPCELL.frd. The averaged ﬂame of animation has the same name as the
“f'rd” ﬁle read into the program, with the exception that the extension has been changed to
“.ave”. The program recognizes the orientation of the plane in space by either preset
conditionals to recognize ﬁle names or an algorithm that determines which coordinate set
(X,Y or Z) doesn’t vary, thus determining which plane the measurement was made in. The
MASSFLOW’s most sensitive calculations are those that pertain to determining the area
measured. The area of a planes is calculated as a grid of cells with a vector centered in each
cell. Depending or the change in the radius between each vectors location in the axial or Y
direction, the grid is either rectangular or varies based on the changing radius. Figure 18

depicts the grid shapes area is calculated for.

The area of the cells are summed for each turbine angle in the FLOWPTURB

routine and then multiplied by the density of the oil and resultant velocity of the vector

42

 

Figure 18. The three type of cell geometries used in calculating area

centered in the cell. The FLOWPCELL ﬁle has the mass ﬂow rate calculation result for
every cell in the same format as the input “frd” ﬁle. This would allow for an animation of

the mass ﬂow rates though the same routines used for the velocities.

The mass ﬂow rate calculations completed in the MASSFLOW program only
represent mass ﬂow rate through the area measured. The orientation of the measured plane
in the passage in Planes 3 through 6 requires only extrapolation of the area measured over
the total area for comparison of mass ﬂow rates from one plane to the next. Planes l and 2
are oriented such that the area measured is actually greater than that seen by the streamwise
ﬂow. This is due to the fashion in which data must be collected, along the path of rotation.
The actual area the streamwise ﬂow is exposed to is the area normal to the surfaces of the

pressure and the suction blades above and below the plane measured. These areas are seen

43

' Smairiwlseﬂows
experience this
area instead of
the area
measured.

 

Figure 19. Orientation of areas used for mass flow rate calculations

in Figure 19. The probe was oriented to nm parallel to the window edge which allows it to
approximate the angle running parallel to the blade curvature. Positioning the plane to run
orthogonal to the streamwise flow through the center of the passage makes the area
measured an accurate method of measuring the mass ﬂow in the Global Y-Z plane.
Compensation for mapping an area larger than the stream wise ﬂow actually experiences
in the X-Z plane is made by taking the cosine of 27.6° of the horizontal distance between

vectors and then extrapolating over the actual area experienced by the ﬂow.

The uncertainty of the mass ﬂow rate calculations is less than 2% of the calculated
value. The positional uncertainty of i 0.25 mm on the radial distance from the axial
centerline of the torque converter generates a 1% difference in grid area with an additional
uncertainty of z 1% for computational truncation errors. This results in an overall

uncertainty of the mass ﬂow rate calculations to be within 2% of the reported values.

The FRDARRAY software is used in conjunction with the commercial MathCAD
software. It reads the “ d” ﬁles and sorts the data points into an array representing 2-d
spatial coordinates, velocity resultants and animation frames. The software prepares an
output ﬁle that arranges the vector information of the “.frd” ﬁle into a format for parsing
by MathCAD. The FRDARRAY software also has a routine that allows for the holes in the
data set to be ﬁlled with the average value of the eight nodes surrounding the hole. The
values of nodes outside the array limits, or that are holes themselves, are not used in
determining the average. The generation of the array and the ﬁlling of the holes within the
array results in two ﬁles for output, a “mm” for the array without holes ﬁlled and a
“s.anm” for the array that has had holes ﬁlled in with an average value of the surrounding

nodes.

45

CHAPTER 4. RESULTS AND DISCUSSION

4.] Performance Test

The constant input performance test was run with the input torque set at 67.8 N-m.
The turbine feed back torque was recorded as the RPM of the turbine was varied item 400
to 2000 RPM. The purpose of the performance test is to compare the result to known

performance results from a non-modiﬁed, same make converter.

Performance tests were nm on both the front and side window torque converters
through the same set of speed settings. Calculations for efﬁciency and the K-factor were
made for both converter designs using the data collected in the Powertek controller.The
data was compared with the Ford performance data. The results of the comparison of input
RPM to output RPM are seen in Figure 20. This graphic compares the speed settings used

with data generated on a Ford test stand.

The results of the front and side window data follow the curve generated by the Ford
data. The difference between the front window data and the Ford performance curve is an
average of 3.2%, with a maximum deviation from the of 6% along the dependent axis. The
side window data correlates to the Ford test data within 2%, with a maximum deviation of

6.2% over the range of turbine speeds tested along the dependent axis.

The calculation of the torque converter efﬁciency and its percentage difference is a
scalar of the input vs. the output speeds and has the same percentage difference. The 1(-

factor calculation results in a different percent difference for both side and front windows.

46

Comparison of Given Performance test data and test rig
perfomance data (constant input, 50ft-lb, Side and Front
window converter)

 

 

 

 

 

 

 

 

 

 

 

2500 _ ._ - _ . . -. - _ a a- ﬁn... w _ m- --_, _.-__, .s.-- ___a..- ,, __ _
2000
5.7
a
a.
g 1500
h
.2
_E_ 1000
2
G.
K
500 +Side \Mndow Test __—
+F0fd TOSt '
+Front \Mndow bet
0 l l 1 l 1
0 500 1000 1500 2000 2500
RPM Turbine (output)

Figure 20. Comparison of Ford data vs. the front and side window data

A plot of the MSU data and the Ford test data can be seen in Figure 21. The plots shows the
data item the modiﬁed converters following the curve of the Ford test data with a small
margin of difference. The results from the front window converter average 1.6%
differences from the Ford data with a of maximum 4.6% along the dependent axis. The side
window converter averaged 1.7% difference in value with a maximum 4.7% along the
dependent axis. Since the data collected ﬁom both the ﬂoat and side window torque
converter conﬁgurations compare favorably with the Ford test data, a high degree of

repeatability of performance test data from the MSU torque converter test stand is apparent.

47

K Factor Data from Front, Side and Ford Performance Test
(constant input, 50ft-lb, Side and Front window converter)

 

 

 

 

 

360. __ _ _.- ._ .- - , _ -- _- - _-____ -_ _..--_ _- _- --,_---,-.__F_..
_._SideWindowTest
340._ +FOMT$t
+FiontWindowtest [X
320

 

.. //
.. /
.. .%

240

K Factor

 

 

 

 

 

a 0:1 62 0:3 0:4 0:5 0; 0:7 0:3 019 i
Speed Ratio (Turbine speed I lnpeller speed)

Figure 21. K-f'actor data from Ford tests and modiﬁed converters

The small differences in data from the test rig and the modiﬁed converters suggests
that the converters perform very similarly to Ford’s non-modiﬁed converter tests. Due to
the temperature limitations of the test rig used at the Michigan State University Engine
Research Laboratory (MSUERL) the values may explain the slight variations from the Ford
test data. The deviations could result because the maximum temperature the MSUERL test
rig may safely operate at is 70° C, there is no upper limit to the temperature used in the Ford
test facility which recorded a maximum of 123° C. The lower temperature of the MSUERL
test rig operation results in a higher viscosity working ﬂuid compared to the viscosity of the

Ford test rig ﬂuids. This may result in slight variations of the values recorded versus the

48

Ford test data. Another possible cause of the small deviation in values received from the
torque converter test stand is the averaging that must be used to gather data over a range of
input torques 1: 6.8 N-m of the target constant torque input. The system torque input varies
slightly over small time periods and the average was 67.8 N-m for the data collected. The
accuracy of this average is related to the amount of time over which sampling occurred. The
experiment consisted of 30 seconds of sampling per turbine speed after steady state
temperatures and pressures had been achieved. The exception to this is the low speed ratio
data points that were only allowed to reach a critical temperature of between 65°C and 70°C
before conditions were changed and the converter was allowed to cool. If the data sampling
period, per condition, was increased, the data represented in the average would be more

accurate as long as steady state was maintained.

4.2 Front Window Velocity Measurements (Planes 1 and 2)

The data presented in the following sections require an overview of the coordinate
system and how the ﬁgures will be presented. Due to the 3-dimensional nature of the
passage where measurements were collected, spatial references are required to maintain
perspective and derive information about the velocity proﬁles displayed. The data will be
presented using the Global coordinate system. This coordinate system seen in Figure 12 on
page 30 has the axis of rotation of the converter as the Y axis, the radial axis of the
converter assembly is represented by the Z axis in the vertical and the X axis in the
horizontal. Figure 22 depicts the solid model reference within the Global coordinate system
and terminology used in the following result ﬁgures. Data will be displayed in two formats
using a surface or contour plot of the velocity proﬁle and a SGI vector ﬁeld. The SGI vector
ﬁeld animation displays the data in 3-dimensional wireframes of both the impeller and the

49

. . Direction of
Suction Side of plane Rotation

 
 
   
  
   
 
  

 

Shroud Side
of plane _ 1,,
£2310 Suction , / Pressure
plane Side of ' Side of
plane plane
Pressure Side
of plane x
View through shell into
View from above the passage the i ller passage

Figure 22. Impeller passage and related terminology

turbine passages. The velocity proﬁle graphics represent a single velocity proﬁle
containing the average velocity values of all 27 sequential frames for each ensemble
averaged velocity measurement. Both perspectives will be referenced by the location of the
plane where measurement occurred using a 3-dimensional solid model representation of the

passages in the Global coordinate system.

The data will be presented by plane and then by speed ratio. The speed ratio will
proceed from the 0.4 speed ratio at 1600 RPM for the impeller and 633 RPM for the turbine
to the 0.8 speed ratio at 2000 RPM and 1600 RPM for the impeller and turbine respectively.
The planes and corresponding speed ratios will be labeled using the convention “Plane
number.speed ratio” (e.g. Plane 1.4 represents Plane 1 at the 0.4 speed ratio). Comparisons

of ﬂow rate data and trends will be discussed at the end of each section.

50

4.2.1 Plane 1

Plane 1 was measured from the surface of the window in the interior of the passage
to a point near the surface of the shroud where streamwise velocity data could no longer be
collected. The results from Plane 1 can be seen in Figure 23 through Figure 28. The plane
was constructed of lines from data collected individually from the LDV system along the
Y axis, the axis of rotation. This plane was not inclined from the horizontal and therefore
lies in the X—Y plane. The Global coordinates over which the plane was measured can be

seen in Table 3.

Table 3. Location of Plane 1 in Global cartesian space

 

 

 

 

 

 

 

Global Coordinates of Plane 1 (mm)
X -14.78 —> -l.28
Y -27.26 —) -12.26
Z (radial) 91.47

 

 

These coordinates have been translated through the 27.6° off axis angle to orient it
to a location on axis. Data is represented in a standard fashion bounded geometrically by a
single set of impeller and turbine passages. This is done to allow for analysis of the data to
be conducted using spatial references and to allow for more insight into what trends are

observed.

The data collected at the 0.4 speed ratio condition, shown in Figure 23, shows a
fairly uniform distribution of the flow over the width and depth of the area measured The
highest velocity ﬂows appear to reside along the suction side of the passage. The magnitude
of the velocities in this region are 2 to 3 m/s greater than those present in the remainder of

the area measured. The overall uniformity and distribution of the higher velocity flows

51

along the suction side of the passage suggests that the momentum of the ﬂuid on the
impeller pressure side has not over come the momentum the ﬂuid had when it was initially

directed into the impeller passage by the stator blades.

Additional data sets were collected for both speed ratio conditions of Planes 1 and
2, they will be denoted as plane Xa and plane Xb for the ﬁrst and second sets of data
collected respectively (e.g. plane la and 1b). The collection of the same plane, under the
same conditions, for a second time had two purposes; it allowed for repeatability
comparisons and decreased the likelihood that a sparse matrix of data would be collected
for a single operating condition. Plane 1a.4 was re-measured to obtain plane lb.4 and can
be seen in Figure 24. The results for plane lb.4 are very similar as compared to plane 1a.4.
The data collection for plane lb.4 was superior to that of plane 1a.4 in the quantity of
velocities measured. Twice as many points were collected in this data set as the ﬁrst. Plane
lb.4 has very similar trends, at the same magnitudes, as plane 13.4. There is a high velocity
ﬂow distribution along the suction side of the passage within an otherwise uniform ﬂow
distribution. The repeatability of this data supports the likelihood that the data is an accurate

representation of the ﬂow through the passage at this location.

Plane 1a.4 and plane lb.4 are unique in that they are not exact copies of each other
as far as the experimental setup for Planes l and 2 are concerned. Plane 1a.4 had an
increment of 0.75 mm between each line of data while plane lb.4 has only 0.5 mm
difference between each line. Since the half angle of the LDV system is changed due to
refraction through the window into the converter, the distance that the probe emitting the
laser travels in air on the traverse table is not the same distance to focal point travels in the

oil. Theoretical calculations and experimental results showed that the distance the focal

52

 

Plane 1, 0.4 speed ratio (1600/633)

Pressure

 

       

Shroud Side

Shell Side

NIMl

 

Figure 23. Velocity proﬁle for plane 1a.4, lower mid-chord measurement

53

Plane 1. 0.4 speed ratio (1600/633) v2

Pressure

 

ANIMl

 

Figure 24. Velocity profile for Plane 1b.4, lower mid-chord measurement

54

point traveled in oil was as 1.5 times greater than the distance the traverse table traveled
when moving along the normal from the surface of the window. At this point a software
aide was developed to predict locations in the converter based on refraction and the local

and global coordinate systems resulting in 0.5 mm increments in all following planes.

Plane 1a.8, seen in Figure 25, has a signiﬁcant area exhibiting ﬂow reversal of l to
2 m/s in the suction/shroud corner. This reversal shows that not all the ﬂuid in the passage
is being pumped by the impeller and that some is actually moving down the impeller
passage. The net effect of this would be a loss of momentum transferred to the turbine
section, since not all of the area available is used for positive mass ﬂow. The high velocity
regions in this plane extend from the suction/shell side along the shell and down the
pressure side of the plane to the shroud in order of decreasing velocity respectively. It is
unlikely that the impeller is imparting much momentum to the ﬂow at this speed ratio since
the impeller and turbine assemblies are rotating at near the same speeds. Therefore, it is
suspected that this ﬂow develops at the entrance of the impeller passage since the reduced
amount of mixing of the ﬂuid ﬂows due to the near sychronization of the impeller and

turbine assemblies.

Plane 1b.8, seen in Figure 26, exhibits a very similar proﬁle with the same
magnitude high velocities tending around the suction/shell to the pressure/shroud. It also
has the same magnitude ﬂow reversal in the same position as the plane 1a.8. The repeated
trends in the ﬂow distribution add conﬁdence in the probability that results are accurate.
Both planes 1a.8 and 1b.8 have results that are supported by the work of Gruever with a

ﬂow reversal near the shroud on the suction/shroud corner [12].

55

 

Plane 1, 0.8 speed ratio (2000/1600)

Pressure
Side

Scale_mps

     

ANTMI

 

Figure 25. Velocity profile for plane 1a.8, lower mid-chord measurement

56

 

Plane 1, 0.8 speed ratio (2000/1600) v2

Pressure

Scale_mps

ANIMI

 

Figure 26. Velocity profile for plane 1b.8, lower mid-chord measurement

57

The mass ﬂow data for plane la and 1b were made from the measured area without
any ﬁlling of holes in the data set by averaging. The values are the results of the measured
data only. An extrapolation based on the entire passage area normal to the streamwise ﬂow
in 3 dimensions was performed. The calculation for this area were made by hand and are
accurate to i 7% based on blueprint tolerances and were checked against CAD drawing
area calculations with a difference of only 2%. The value of the total area was calculated to

be 287.1 mmz. The measured area for the four planes constituting Plane 1 can be seen in

 

 

 

 

 

 

Table 4.
Table 4. Measurement area calculated from probe locations in planes 1a and lb.
Plane and Speed Ratio Area (mmz) Percent coverage
Plane 1a.4 151 53%
Plane lb.4 190 66%
Plane 1a.8 196 68%
Plane 1b.8 190 66%

 

 

 

 

 

The mass ﬂow rate results from planes 1a.4 and lb.4 can be seen in Figure 27. This
plot represents the extrapolated values of the mass ﬂow rate over the entire area of the
passage orthogonal to the streamwise ﬂow. A single period sinusoid can be seen over the
27 turbine positions for each plane. The amplitude of both sinusoids amounts to a 7%
variation of the ﬂow rate ﬁ'om the average ﬂow for the 27 turbine positions. The lowest
ﬂow rate occurs when 33% of the turbine passage width has rotated across the exit of the
impeller passage. The highest ﬂow rate occurs at 66% of this same distance. The frequency
of the sinusoid is 435 Hz. This ﬁequency corresponds to the frequency of the passing of the

turbine blades in front of the impeller exit at every 0.062 second intervals.

58

Flowrate vs. Turbine Position

 

0.4 Speed Ratio (1600I633 RPM)
1.5 . Radiuseiﬂeﬁyrygntﬁl-Sé "110.999 [hamster __ ._ _
1.5 \
+Plane 1a|
1.45 . ___—

 

1.41 A

\ /
W

1 3 5 7 9 11 13 15 17 19 21 23 25 27
Turbine Position

 

 

 

no. Flow m (kglaoc)

ii

 

1.3

 

Figure 27. Mass ﬂow rate data from planes 1a and lb for the 0.4 speed ratio

Plane 1a.8 and plane 1b.8 data also exhibits a similar sinusoid in the mass ﬂow rate
calculations, seen in Figure 28. The variation in mass ﬂow rate due to the amplitude
compared to the average mass ﬂow rate of the sinusoid is 4% for plane 1a.8 and 5% for
plane 1b.8 at this speed ratio. There is less ﬂuctuation in the ﬂow rate for the 0.8 speed ratio
data than the 0.4. This could be due to the reduction in ﬂow rate or the reduction of rigorous
mixing. These effects would be caused by the low relative speed of the turbine to the
impeller blade. The frequency of the sinusoids is the same at 180 Hz. This frequency
coincides with the rate of turbine blades crossing the exit of the impeller passage. The
maximum ﬂow rates occur with the turbine rotated 50% of its passage width across the
impeller passage. Minimum ﬂow rates occur when the turbine blades lie on either side of

the impeller passage.

59

Flowrate vs. Turbine Position
0.8 Speed Ratio (200011600 RPM)
Radius of Meaeurment I 91.45 mm, 0 deg inclination

 

 

 

 

 

 

 

 

 

 

 

 

 

 

+Plane1a
1.16 +Ptano1b—
- 1.15
g 1.14 [fr/éAm \
3 1.13 J
l 1.12. VA\\
1.11
fi‘f v
1.1 . . a . . . a a . . - . .
1 3 5 7 9 11 13 16 17 19 21 23 25 27
Turbine Posnton

Figure 28. Mass flow rate data from planes la and 1b for the 0.8 speed ratio

4.2.2 Plane 2

Plane 2 was measured with an inclined orientation at 7.96° off the Global X-Y axis.
The results of measurements collected from Plane 2 can be seen in Figure 29 through
Figure 34. Since velocity measurements are made along the path rotation at the radius of
interest, the radius curvature of the plane generated by the measurements decreases as the
distance from the shroud surface and radius of interest decreases. The range of radial Z, X
and Y values used to deﬁne the planes measured can be seen in Table 5 in Global

coordinates.

As was the case in Plane 1, there were two sets of data for each speed ratio condition

collected all at the same location. They will be referred to as plane 2a and plane 2b.

60

Table 5. Location of Plane 2 in Global cartesian space

 

 

 

 

 

Global Coordinates of Plane 2 (mm)
X ~12.74—)-1.34
Y -28.98 —-) 12.66

Z (radial) 96.12 -—) 93.77

 

 

 

 

It can be seen in Figure 29 and Figure 30 that the distribution of the 0.4 speed ratio
velocity proﬁle is still fairly uniform as compared to the plane 1a.4 and lb.4 velocity
proﬁles. The difference in both planes 2a.4 and 2b.4 from planes 1a.4 and lb.4 is the steep
decline in the velocity near the shell and the slight rise and abrupt decrease of the velocity
at the suction/shell side corner. The largest velocity region still lies along the suction side
of the blade but now it occupies primarily the suction/shroud corner. There is a uneven
distribution on the pressure side about half the distance to the shroud that has slightly higher
velocities than the surrounding points. The average velocity in this plane is about 6.0 to 6.5
m/s with maximum velocities of 8.0 m/s. The velocity proﬁles of plane 2a.4 and 2b.4 have
similar features as the 0.4 speed ratio work conducted by Dalimonte at the mid-chord of the

same converter [15].

Planes 2a.8 and 2b.8, seen in Figure 32, represent the results from the second set of
data taken at this location at the 0.8 operating condition. This data set is not as full as plane
2a.8 due to some difficulty in collecting data near the shroud of the converter. It does
exhibit the same trends as plane 2a.8 with the declining velocities on the shell surface and
a slight rise and abrupt decline of velocities in the shell/suction side corner. The highest
velocity region is along the suction/shroud side at a velocity of 7.5 m/s to 8.0 m/s during
various turbine positions. The average velocity is 6.5 m/s with a generally uniform ﬂow

distribution. It appears that the pressure side of the impeller does not begin to deliver the

61

Plane 2, 0.4 speed ratio (1600/633)

    

Pressure

 

‘V‘V.'. - . -

A?

   

V

w!

Scale_mps

Sucﬁon
Side

ANIMl

 

Figure 29. Velocity proﬁle for plane 2a.4, upper mid-chord measurement

62

         

Plane 2, 0.4 speed ratio (1600/633) v2

Scale_mps

Shell Side

Pressure

1 I it‘O‘.‘ .

.<.<.t... ‘O‘.‘. . . . I.

63"...” NOOwOWOHO. .
0

Side

Shroud Side

ANIMI

 

Figure 30. Velocity proﬁle for plane 2b.4, upper mid-chord measurement

63

majority of the momentum at this height in the passage. The high velocity values along the
suction side of the passage suggest that momentum imparted to the ﬂuid at an earlier stage

in the impeller passage is still dominating the velocity proﬁle.

Planes 2a.8 and 2b.8 can be seen in Figure 31 and Figure 32. They have very similar
velocity proﬁles. There is no recirculation in the 0.8 speed ratio data, however, there is a
depression in the velocity proﬁle in the same area where ﬂow reversal was occurring in the
Plane 1, 0.8 speed ratio data sets. The velocity in the depression is a minimum 1 m/s while
the largest velocities, near the shell/pressure side reach 5.5 m/s. The velocity distribution
in the 0.8 speed ratio data is primarily along the pressure side of the passage, with highest
velocities on the shell half way between the pressure and the suction sides of the passage.
These tends are supported by the Dalimonte whose work at the mid-chord of the same
converter shared similar results [15]. A similar feature to the 0.4 speed ratio data is the
small rise in velocity in the shell/suction side. Plane 2a.8 as seen in Figure 31 on page 65
has a few discontinuities in the velocity proﬁle near the suction and pressure side leading
edges. This is due to a lower number of velocity data values available for ensemble
averaging at those points in the proﬁle. This decreases the accuracy of the averages
representation of the ﬂow velocity and allows a small amount of data that doesn’t

accurately represent the ﬂow to dominate the average value.

The mass ﬂow rates for Plane 2 were calculated without ﬁlling holes in the data set
to adjust for unmeasured data. Each data set of Plane 2 has a slightly different percent
coverage of the total area. This information can be seen in Table 6. The area the streamwise
flow experiences when passing through the passage at Plane 2’s location was calculated by

hand and determined accurate by j; 7%. The accuracy was determined using the

64

 

Scale_mps

Pressure

Plane 2, 0.8 speed ratio (2000/1600)

Shroud side

ANIMI

 

Figure 31. Velocity proﬁle for plane 2a.8, upper mid-chord measurement

65

 

Plane 2, 0.8 speed ratio (2000/ I600)

Pressure

Scale_mps

 

ANIMI

 

Figure 32. Velocity proﬁle for plane 2b.8, upper mid-chord measurement

66

Table 6. Measurement area calculated from probe locations in planes 2a and 2b.

 

 

 

 

 

 

Plane and Speed Ratio Area (mmz) Percent covered
Plane 2a.4 196 67%
Plane 2b.4 198 68%
Plane 2a.8 167 57%
Plane 2b.8 189 65%

 

 

 

 

 

approximations made during the calculations and the tolerances of the blueprints. A CAD
model was used to verify the approximations made during the hand calculations and a

difference of 4% resulted. The calculations resulted in an area of 291 .5 m2.

The variation of the mass ﬂow rate determined by the variable turbine position for
plane 2a.4 and 2b.4 data can be seen in Figure 33. A clear sinusoid is only observable in the
plane 2a.4 data. The variations of some points between large values limits the quality of
plane 2b.4’s mass ﬂow rate data. Plane 2a.4 does contain a sinusoid of an amplitude that is
6% of the average ﬂow rate of 1.54 kg/sec over all 27 tribune positions. Plane 2b has an
average mass ﬂow rate of 1.47 kg/sec over the 27 turbine positions. The ﬁequency of this
turbine in plane 1a is 435 Hz corresponding to the difference in the impeller and turbine
speeds. The maximum ﬂow rate occurs when the turbine passage has moved 33% of its

width across the impeller passage and the minimum occurs at 66% of that distance.

The mass ﬂow rate data for planes 2a.8 and 2b.8 can be seen in Figure 34. The 0.8
speed ratio data sets for Plane 2 do not appear to exhibit a sinusoidal pattern. The highest
velocities are when the turbine blades are positioned on either side of ﬁre impeller passage
and the minimum occurs as ﬁre turbine blade is 50% of its passage widﬁr across the impeller
passage. The ﬂuctuations in the mass ﬂow rate do not reﬂect on ﬁre quality-of ﬁre data. The

maximum and ﬁre minimum values for this data set represent only 2% of the total mass ﬂow

67

Flowrate vs. Turbine Position
0.4 Speed Ratio (16007633 RPM)

Radius of Measurment a 93.74 mm, 7.96 deg inclination
1.6 . - . - - - _ h -. --

1.58 q — ’ +P—lane—2a-
156‘ +Plane 2b F—
... \ ;
1.52
1.5
1.48 m
1.46 R

..... \/ V

1 .42

1-4 . s . . . . ﬂ . . r . . 1'
1 3 5 7 9 11 13 15 17 19 21 23 25 27

Turbine Position

 

 

 

 

 

 

 

 

 

 

 

Mass Flow Rate (kgIsec)

 

 

 

Figure 33. Mass flow rate data from planes 2a and 2b for the 0.4 speed ratio

rate. This low value accounts for the large ﬂuctuations due to the scale that must be used to
observe any changes in ﬁre mass ﬂow rate value over ﬁre 27 turbine angles. The ﬂuctuations
from ﬁris broad trend do not allow for the observation of a clean sinusoid, although one may
be present if more data were collected and averaged against ﬁris test. The fact that the
minimum appears in the middle of two equal maximums equidistant from its position
supports ﬁris hypothesis. However, wiﬁr such a low percentage ﬂuctuation in ﬁre average
ﬂow rate, it would most likely require a vast amount of data to average out the ﬂuctuations
with the current scale. The average mass ﬂow rate for this data set is 1.07 kg/sec for boﬁr

data sets.

68

Flowrate vs. Turbine Position
0.8 Speed Ratio (2000/1600 RPM)
Radius of Measurment = 93.74 mm, 7.96 deg inclination

+
+ Plane 2b

1.095

1.09

1.085

2‘37

1 .075

lbs. Flow Rate (kghoc)

§§§i§

 

Turbine Position

Figure 34. Mass ﬂow rate data from planes 2a and 2b for the 0.8 speed ratio
4.2.3 Summary of Mass Flow rate results for Planes 1 and 2

There are 31 passages present in the torque converter studied. Using the information
collected from a speciﬁc data set, it is possible to extrapolate ﬁre data over every passage
in ﬁre torque converter and compare results of all the data sets at once for conservation of
mass purposes. The values displayed in Figure 35 show ﬁre mass ﬂow rates in the impeller
for each of ﬁre respective data sets. The values of the Plane 1.4 data set varies 7.6% from
the average value of 44 kg/sec. Plane 1.8 data set has a 1% difference between plane 1a.4
and lb.4 and its average mass ﬂow rate. The average mass ﬂow rate for this data set is 35
kg/sec. The Plane 2.4 data sets had a difference of 5% with an average value of 47 kg/sec.
The variation of the 0.8 speed ratio data sets for Plane 2 is less than 1% of ﬁre average of

33 kg/sec. The percent difference in 0.4 speed ratio data sets of Planes 1 and 2 show a 3%

69

Total Mass Flow From All 31 Passages In lm peller Assembly

 

 

 

has Flowm (kghec)

 

 

Benet 0.4 Pierrot 0.8 Plene2 0.4 Flene2 0.6
Measurement Plane Type

Figure 35. Comparison of all mass flow rate data over all 31 impeller passages

difference from the average of the two planes. The 0.8 speed ratio data has a 2.5%

difference in the values between the average of the two planes.

4.3 Side Widow Velocity Measurements: Plane 3

Plane 3 represents ﬁre ﬁrst plane measured using ﬁre side window torque converter
conﬁguration. Plane 3 lies in the X-Z plane and is located inside the impeller passage, 4.5
mm from the impeller passage exit. One set of data was collected for each speed ratio.
Results collected from Plane 3 can be seen in Figure 36 ﬁrrough Figure 39. Advancements
in analysis techniques allowed for problematic data sets to be identiﬁed and remeasured
immediately. The side window involves challenges not present in ﬁre ﬁont window data

acquisition method. The full width of ﬁre impeller blade passes directly under the lasers and

70

scatters light which is collected and sent to the signal processor which in turn interprets the
signal as noise and disregards it. In a backseatter data acquisition mode ﬁre noise
accompanying a good signal can make velocity measurements difﬁcult. This reduces the
amount of data collected and reduces the accuracy of the averaging. The shroud also
reﬂects light which is interpreted by the IF A 750 as noise. Often, when measuring near the
shroud, ﬁre laser intensity must be lowered and this makes data acquisition difficult due not
ﬁom reﬂection, but from to ﬁre lack of light to scatter back to ﬁre collector. Therefore, data
was inﬁequently collected near the shroud surface. The range of coordinates relating the

position of Plane 3 in the Global coordinate system are available in Table 7.

Table 7. Location of Plane 3 in Global cartesian space

 

Global Coordinates of Plane 3 (mm)

 

 

 

 

 

 

 

x 0.51 —> 13.69
Y -6.79
z (radial) 116.49 —) 106.71

 

The ﬂow ﬁrrough the planes is exiting ﬁre impeller at -12.6° ﬁ'om ﬁre Y axis in ﬁre
Global X-Y plane. This angle is parallel to ﬁre angle of ﬁre blades in ﬁre Global X-Y plane
at the exit and corresponds to the angle ﬁre access window in the Global X-Y plane. The
probe was rotated along its axis to match ﬁris angle to maximize the measurable area and
to align the beams with the streamwise ﬂow exiting the impeller. Figure 36 shows the
velocity proﬁle of the ﬂow passing ﬁrrough Plane 3.4. Holes in ﬁre velocity proﬁle were
evident for this data set. They appeared in random location through the frames of
animation. This occurs when data is not collected at locations due to low numbers of
velocity measurements taken at ﬁre location or ﬁre LDV systems inability to measure at the

location. These holes in the data set were ﬁlled using the FRDARRAY code. Only 1% of

71

the data array in the surface plot had holes. This small percentage was averaged with the

surrounding data and ﬁlled with ﬁrat average value.

The ﬂow no longer has the highest velocity region along the suction side. The ﬂow
has distributed itself along ﬁre shell and near ﬁre pressure/shroud side. The highest velocity
region is at ﬁre pressure/shroud side at 4.2 m/s. The lowest value is along ﬁre mid-portion
of the suction side at 2 m/s. A low velocity region near the suction side dissipates and the
velocities become more uniform as the turbine crosses the impeller exit. During this time
the velocity in ﬁre low velocity region reaches 3 m/s, comparable to ﬁre values bordering
this low velocity area. The depression occurs as the turbine passage approaches a position
bracketing the impeller passage. There is a decrease in the velocities as the shell surface
and suction side is approached. This may be due in part to the curvature of the housing
which is less ﬁran 2 mm along the Global Y axis from dramatically increasing its radial
distance from the axial center line in ﬁre gap region to extend above ﬁre turbine assembly

and connect wiﬁr the engine side of the housing.

The 0.8 speed ratio data set, seen in Figure 37, has a very smooth distribution,
increasing from ﬁre suction side towards the pressure side, although there are negative
velocities from half way across ﬁre measured area to ﬁre edge of the suction side. The
velocity varies linearly across ﬁre measured region from -2.0 m/s to 2.2 m/s and this proﬁle

remains unchanged during animation.

The mass ﬂow rate calculations were based on ﬁre area measured and did not
include any method of ﬁlling in holes to alter the mass ﬂow rate value. Plane 3, 0.4 and 0.8
speed ratio data sets cover ﬁre same percentage of the total area available. The total area

was calculated to be 313 mm2 with an accuracy of _+_ 6% based on the tolerances of the

72

 

 

Plane 3, 0.4 speed ratio (1600/633)

Shell Side

     

Pressure
Side

Shroud side

ANIMI

 

Figure 36. Velocity proﬁle for Plane 3.4, 4.5 mm inside impeller passage

73

 

Plane 3, 0.8 speed ratio (2000/I600)

Pressure
Side

Shroud side

ps

Scale in

   

Shell Side

  

 

Figure 37. Velocity proﬁle for Plane 3.8, 4.5 mm inside the impeller passage

 

ANIMI

74

blueprints used in this calculation. The percentage that each plane covered of the total area

can be seen in Table 8.

Table 8. Measurement area calculated from probe locations in Plane 3

 

Plane and Speed Ratio Area (mmz) Percent covered
Plane 3.4 127 40%
Plane 3.8 124 40%

 

 

 

 

 

 

 

 

As seen in Figure 38, it is difﬁcult to distinguish any trend in the mass ﬂow rate data
for the 0.4 speed ratio data set. It has a maximum ﬂow rate of 0.280 kg/sec and a minimum
of 0.258 kg/sec. The percentage difference in ﬂow rate from the average ﬂow rate was 4%
over ﬁre 27 turbine positions. It is possible that a sinusoid is present, but a signiﬁcantly

greater amount of data would be needed to average out the ﬂuctuations.

A sinusoid is present in Plane 3.8, ﬁre data supporting ﬁris can be seen in Figure 39.
Its variation is 97% of ﬁre total mass ﬂow rate with an average value of 0.030 kg/sec
passing ﬁrrough the area measured. The maximum mass ﬂow rate was 0.047 kg/sec and ﬁre
minimum 0.018 kg/sec.The maximum value occurs with 25% of the turbines passage width
across ﬁre impeller passage. The minimum occurs near 75% of the turbines passage widﬁr
as it passes ﬁre impeller section. The frequency of the sinusoid is 180 Hz or the same

frequency as the rate at which the turbine blades cross in ﬁont of ﬁre impeller passage.

4.3.1 Plane 4

Plane 4 is located 2 mm from Plane 3 in the direction of the impeller passage exit,
ﬁre results ﬁ'om which can be seen in Figure 40 through Figure 43. The LDV probe was
oriented to -12.6° ﬁ'om the Y axis as was ﬁre case in Plane 3. The was done to maximize

the area that could be measured. This allows the streamwise ﬂow velocity to be accurately

75

0.285 .

0.28
0275
0.27
i ....
0.26
0.255

0.25

0.01

Flowrate vs. Turbine Position
0.4 Speed Ratio (1600/633 RPM)
Axial location I - 6.79 mm, dde window conﬁguration

__ ,.._ ___.. -_ _. - F. _..._.=...

1

 

7"er

t

w @—
/ \N

f \Y.
a/ \ ..

 

 

W V“

 

12 3 4 5 6 7 8 9101112131415161718192021222324252627
TurbinePosltlon

Figure 38. Mass ﬂow rate data from Plane 3 for the 0.4 speed ratio

Flowrate vs. Turbine Position
0.8 Speed Ratio (2000/1600 RPM)
Axial location 8 - 6.79 mm, dds window configuration

A“
\\
\ .
V \\ /

W

 

 

 

 

 

 

 

12 3 4 5 6 7 8 9101112131415161718192021222324252627
TurbinePositlon

Figure 39. Mass ﬂow rate data from Plane 3 for the 0.8 speed ratio

76

determined by making velocity measurements running parallel to the blade. Plane 4 was
measured in the X-Z plane at a position along ﬁre Y axis. The range of Global coordinates

that can be used to determine ﬁre area measured are displayed in Table 9.

Table 9. Location of Plane 4 in Global cartesian space

 

 

 

 

 

 

 

Global Coordinates of Plane 4 (mm)
X 1.52 —) 14.68
Y - 4.79
Z (radial) 116.24 —> 104.33

 

 

The velocity proﬁle of Plane 4.4 has no ﬂow reversal. Data near the shell was
difﬁcult to acquire and therefore there is a small region near the shell not measured. It is
believed ﬁrat if more velocity data was measured in ﬁris region, a ﬂow reversal would have
been observed. This hypoﬁresis is derived from ﬁre strong reversal measured very near the
shell surface in Plane 3. Compared to previous data sets, this set is different in that the high
velocity regions sit squarely in ﬁre shroud/pressure side of the area measured. All velocities
decrease in magnitude as ﬁre distance from this area increases. The maximum velocities
measured in Plane 4.4 are 6 m/sec, with a minimum velocity of 1 m/sec about half way
between the shell and shroud on ﬁre suction side of the area measured. The velocity proﬁle
is erratic near the shell. There are signiﬁcant variations in the magnitude of the velocity for
almost every point in this region. This was caused by the low data rate near ﬁris surface.
The number of data points recorded near the shell were at ﬁre minimum of 10
measurements per 0.5 degrees, reducing the accuracy of the averaging meﬁrod used. The

results for Plane 4.4 can be seen in Figure 40.

The Plane 4.8, seen in Figure 41, contains a phenomenon seen in Plane4.4 speed

ratio data set. There appears to be a linear variation of ﬁre velocities ﬁom ﬁre pressure to

77

 

Pressure

Plane 4, 0.4 speed ratio (1600/633)

Shroud side

 

Figure 40. Velocity proﬁle for Plane 4.4, 2.5 mm inside the impeller passage

78

 

Plane 4, 0.8 speed ratio (2000/1600)

Side

Shroud side

   

Shell Side

 

Figure 41. Velocity proﬁle for Plane 4.8, 2.5 mm inside the impeller passage

79

the suction side of the impeller passage. Flow reversal occurs 2/3 of the distance across the
area measured from the pressure to the suction side. The linear variation of velocities from
pressure to suction side suggests that the area beyond ﬁrat measured may also exhibit a
similar trend. There is some deviation from ﬁrat trend near the shell surface. There appears
to be a small rise in the velocity on the suction side when compared to ﬁre velocities closer
to ﬁre shroud on the suction side. The maximum velocity is found along the pressure side
at 3.75 m/sec and a minimum velocity of - 2.1 m/sec is located mid way along ﬁre suction

side of the data measured between the shroud and shell surfaces was observed.

The mass ﬂow rate calculations carried out for ﬁris data set did not use any method
of averaging for ﬁlling in holes in the area measured. The amount of area measured as a
percentage of the total area varies for each speed ratio. The 0.8 speed ratio measurements
are typically twice as difﬁcult to acquire data and ﬁrerefore only cover a fraction of the area
measured in the 0.4 data sets. Data was difﬁth to obtain in Plane 4.8. As seen in Table 10,
less ﬁran 50% of the passage was measured in the Plane 4.4 but almost half of that area was

not measured in Plane 4.8 resulting in a very low percentage of the total area covered. The

Table 10. Measurement area calculated from probe locations in Plane 4

1

Plane and Speed Ratio Area (mmz) Percent covered
Plane 4.4 134 44%
Plane 4.8 76 25%

 

 

 

 

 

 

 

 

 

 

reason for ﬁre difﬁculty in measuring the higher speed ratio revolves around the two main
issues in LDV measurements, optical access quality and seed particle distribution. The
scam in ﬁre side window converter leaks a very small quantity of oil. Over time, ﬁre oil,

spun outward ﬁ'om ﬁre rotating torque converter, lands back on ﬁre window due to the

80

enclosure used to house the torque converter. A very thin layer of oil on the window has a
signiﬁcant effect on the data rate achievable and therefore the quality of data collected. It
was during the of data acquisition of Plane 4 ﬁrat it was determined ﬁrat a layer of oil ﬁrat
was undetectable to the eye was present on the window after running the test setup for long

durations. With frequent cleaning, ﬁris problem can be avoided completely.

The seed distribution poses a unique challenge under the 0.8 speed ratio conditions.
The ﬂow may not be sufﬁcient to carry the seed particles causing ﬁre seed to be
redistributed by a lack of mixing in ﬁre ﬂows, reducing the homogeneity of ﬁre distribution.
The inertial effect on the distribution is unknown but may play a role if the velocity in a
sufﬁcient percentage of ﬁre passage is near zero. This would allow the seed particles in that
region to break free of the ﬂow under inertial inﬂuence and redistribute in other areas of
the velocity ﬁeld. The result is a low concentration of seed in various areas in ﬁre ﬂow,
making it difficult to measure there. Additional work must be done to ﬁnd a seed material
that meets ﬁre requirements of ﬁre experiment and has a density closer to that of the working
ﬂuid. This would reduce ﬁre inertial effects that high speed ratio conditions may have on

ﬁre seed distribution.

The mass ﬂow rate results of Plane 4 for ﬁre 0.4 and 0.8 speed ratio both exhibit
sinusoids of ﬁre same nature as observed in earlier planes. The average ﬂow rate for the 0.4
speed ratio case is 0.45 kg/sec, with maximum and minimum values of 0.47 kg/sec and 0.41
kg/sec. The frequency of the sinusoid is 435 Hz, which corresponds to ﬁre rate of the turbine
blades passing ﬁre impeller passage exit. The minimum value doesn’t appear to correspond
to the trend apparent in ﬁre remaining data points. It is a signiﬁcantly smaller mass ﬂow

rate than any other recorded for this speed ratio with a discontinuity between itself and ﬁre

81

next point. The minimum for value on the sinusoidal curve is 0.43 kg/sec. This amplitude
in mass ﬂow rate variation is 11% of the average ﬂow rate. This is signiﬁcantly less ﬁran
ﬁre variations seen in Plane 3 for the same speed ratio condition. Due to the area measured
in ﬁre passage, it is likely ﬁrat the change in ﬂow distribution in Planes 3 and 4 are not
captured due to the limited coverage. That is, as ﬁre velocity proﬁle changes in spatial
perspective, ﬁre area measured may not account for ﬁre changes made to ﬁre velocity
proﬁle. If the areas for Planes 3 and 4 measured the entire passage, ﬁre ﬂuctuations in ﬂow
rate may be much closer in comparison. Figure 42 shows the results of the mass ﬂow rate

calculations.

Figure 43 depicts ﬁre plotted results of the mass ﬂow calculations over ﬁre 27
turbine positions resolved for the 0.8 speed ratio case of ﬁris plane. Plane 4.8 resulted in a
fairly smooﬁr sinusoid with an average value of 0.085 kg/sec. The maximum and minimum
values were 0.100 kg/sec and 0.070 kg/sec respectively. The maximum value obtained is
near 33% of the turbine passages width as it passes ﬁre impeller section. The minimum
occurs near 66% of the same distance. The variation about the average ﬂow rate due to the
periodic nature of ﬁre data is 35%. The sinusoid has a frequency of 180 Hz which is the
same frequency determined for all of the sinusoids observed at the 0.8 speed ratio condition
up to ﬁris point. This value corresponds with ﬁre rate of the turbine blade passing the face
of impeller passage exit. The ﬂow rate in Plane 4.8 is signiﬁcantly less ﬁran Plane 4.4 due

to the limited area measured.

82

lbss FIoszteargisec)

m Flow Rate (W)

Flowrate vs. Turbine Position
0.4 Speed Ratio (1600/633 RPM)
Axial location = - 4.79 mm, side window conﬁguration

 

 

 

0.49. __ _ - .
0'48 /A\/‘—V\ i:i—_'Ham4'
0.47 v '
0.46 g 4?
0.45 F\_‘\

 

MN

 

X

\ A .. /‘
0.3/ \V/\_/W

0.42

 

 

0.41

 

0.4

 

 

12 3 4 5 6 7 8 9101112131415161716192021222324252627
TurbinePosition

Figure 42. Mass flow rate data from Plane 4 for the 0.4 speed ratio

Flowrate vs. Turbine Position
0.8 Speed Ratio (2000I1600 RPM)
Axial location I - 4.79 mm, dde window conﬁguration

0.1 i - a M W
0.09 //~.// \v\
0:07 ‘/ m M

' V

0.11

 

 

-—. - _ ___—___...h . __-_.__.-.

0w

 

 

0.05

 

12 3 4 5 6 7 8 9101112131415161718192021222324252627
TurbinePosltion

Figure 43. Mass flow rate data from Plane 4 for the 0.8 speed ratio

83

4.3.2 Plane 5

Plane 5, the results of which can be seen in Figure 44 through Figure 47, is oriented
in the X-Y plane and is similar to Planes 3 and 4. It is located -l.0 mm along ﬁre Y axis of
ﬁre Global coordinate system from Plane 3 approaching ﬁre impeller exit. This location
corresponds to a plane located 0.71 mm inside of ﬁre impeller passage from ﬁre exit.
Coordinates used to deﬁne ﬁre area measured are available in Table 11. This plane
measured velocities along a 12.6° angle ﬁ'om the Y axis, in ﬁre X-Y plane. This angle is
parallel to the blades at ﬁris location ﬁrus allowing for proper measurement of stream wise

ﬂow through ﬁre plane and maximizes ﬁre widﬁr of ﬁre passage available for measurement.

Table 11. Location of Plane 5 in Global cartesian space

 

 

 

 

 

 

 

Global Coordinates of Plane 5 (mm)
X 1.51 —> 14.58
Y - 3.0
Z (radial) 116.47 —> 104.92

 

 

The velocity proﬁle for the 0.4 speed ratio case of the shows very similar trends to
those seen in Plane 4 at the same operating conditions. Plane 5 was not measured near the
shell surface, but ﬂow reversal is evident in the area measured near that region. The reversal
occurs as the radial distance of ﬁre shell increases at this location along the Y axis as the
shell increases its radial distance in order to account for ﬁre larger turbine assembly.
Maximum velocities observable in the proﬁle are 5.25 m/sec occurring in ﬁre shroud/
pressure comer of the area measured. The minimum value of 1.50 m/sec occurs mid way
between the shell and ﬁre shroud on ﬁre suction side of ﬁre area measured. There is a slight
rise in the velocity as it approaches the shell before it becomes negative, ﬁris trend was
evident in both Plane 3.4 and 4.4. Plane 3.4 appears to have a much more pronounced

84

increase in velocity than Plane 4.4 in this same region. Planes 4.4 and 5.4 have only a slight
increase of 1 nr/sec. There is a dramatic increase in the velocity in a small region near ﬁre
shroud, half way between the pressure and suction sides. It is present in every ﬂame of data
over the 27 turbine positions. Its magnitude is 5 m/sec and it is fairly constant ﬁrroughout
ﬁre progression of turbine positions, with only a slight decline in the mid—passage position.

The data for ﬁris operating condition can be seen in Figure 44.

Plane 5.8 continues ﬁre trend seen in Planes 3 and 4 during ﬁris operating condition.
Figure 45 depicts ﬁre results from Plane 5.8. There is a variation in ﬁre velocity proﬁle that
is nearly linear, from pressure to suction side with ﬁre highest velocities along ﬁre pressure
side of ﬁre passage. Flow reversal occurs at about 2/3 ﬁre distance across ﬁre area measured
ﬁom ﬁre pressure to suction side. The minimum value present is -3.0 m/sec along the
suction side, wiﬁr the maximum value of 4.5 m/sec residing along the pressure side of the
measured area. There is a rise and fall of the velocity on the suction side when approaching
the shell. This phenomenon was present in Plane 4.8. This ﬂuctuation changes the velocity
Item -2.0 m/sec to near zero m/sec. The event in Plane 4.8 was much smaller in magnitude
difference from ﬁre surrounding values by l m/sec. There are some larger velocity data
values along the shroud side, repeating an event seen in the Plane 4.8. There is no ﬂow
reversal apparent along ﬁre shell other ﬁran the general decline of the velocities ﬁom
pressure to suction side. Velocities near the shell sm'face of Plane 5.8 were not able to be
measured ﬁrerefore the lack of ﬂow reversal in ﬁris region is still consistent with the

previous data.

There was an improvement of the percent area covered in Plane 5.8, as it was

recognized that the surface of the window must be cleaned very frequently in order to

85

ps

Scale m

 

Shell Side

 

Pressure

Plane 5, 0.4 speed ratio (1600/633)

Shroud side

 

ANIMl

 

Figure 44. Velocity proﬁle for Plane 5.4, 1 mm inside the impeller passage

86

 

Plane 5, 0.8 speed ratio (2000/1600)

Pressure
Side

Shroud side

    

Shell Side

 

ANIMI

 

 

Figure 45. Velocity proﬁle for Plane 5.8, 1 mm inside the impeller passage

87

measure velocities at lower radial locations in the passage. The amount of area measured
in the 0.4 case and 0.8 case is nearly the same. This leaves particle distribution as the next
difficulty to over come. The total area in the passage at ﬁris point was calculated to be 307

mmz. Table 12 shows the percent of the total area cover for each speed ratio condition.

Table 12. Measurement area calculated from probe locations in Plane 5

 

Plane and Speed Ratio Area (mm) Percent covered
Plane 5.4 122 40%
Plane 5.8 109 36%

 

 

 

 

 

 

 

 

The mass ﬂow rate data for Plane 5 at the 0.4 and 0.8 speed ratio conditions,
displays a sinusoidal trend. Plane 5.4 has a maximum value of 0.440 kg/sec and a minimum
of 0.350 kg/sec. The minimum value appears as a discontinuity in ﬁre data set similar to that
observed in Plane 4.4 mass ﬂow rate data. The minimum value observed on the sinusoidal
trend in this data set is 0.380 kg/sec. The percent variation about the average mass ﬂow rate
of 0.405 kg/sec is 15%. The percent variation of ﬁre mass ﬂow rate about ﬁre average ﬂow
rate value compares more favorably with Plane 4.4 which varied 11% ﬁ'om the average
mass ﬂow rate value as compared to ﬁre 97% variation determined in Plane 3.8. The
maximum value occurs when 33% of ﬁre turbine passage width has crossed in front of the
impeller passage exit. The minimum occurs at 66% of ﬁris same distance. The ﬁequency of
ﬁris oscillation in mass ﬂow rate is 435 Hz. The frequency of variation of ﬁre ﬂow coincides
wiﬁr the frequency of the turbine blades rotation past the impeller passage exit. Figure 46

depicts the results of ﬁre mass ﬂow rates for Plane 5.4.

Plane 5.8 mass ﬂow rate data, seen in Figure 47, has a minimum value of 0.12 kg/

sec and a maximum of 0.17 kg/sec. The maximum and minimum values occur at 33% and

88

his. Flow Rate (kolsec)

lbs. ”091th (W)

Flowrate vs. Turbine Position
0.4 Speed Ratio (1600/633 RPM)

Axial location I - 3.0 mm, dde window conﬁguration
0.45 . - - _- -

°-“ /A\ Finis]—

0.43

0.42 f \\ -

f V

0.4—I a z

0.39 f \e\. A /\ a

0.30 V V W

0.37

0.30

oss‘f
12 3 4 5 e 7 a 9101112131415161718192021222324252627

TurbinePositlon

 

 

 

 

 

 

 

 

Figure 46. Mass flow rate data from Plane 5 for the 0.4 speed ratio

Flowrate vs. Turbine Position
0.6 Speed Ratio (2000I1600 RPM)

Axial location I - 3.0 mm, dde window configuration
0.18. . _ _---_- ___--- - .

 

 

—.—.——.—-—. _ ..._ _ ___—__—h-—. ._ --—— -—-... — ---

0.17 XV...

0.16 wv/ \‘x
0.15 A A
0.14 W \ .

/ V

 

 

0.13

0.12

 

0.11

 

 

0.1

 

12 3 4 5 6 7 8 9101112131415161718192021222324252627
TurbinePositlon

Figure 47. Mass flow rate data from Plane 5 for the 0.8 speed ratio

89

66% respectively of the turbine passage’s width as it crosses the face of the impeller
passage exit. The average mass ﬂow rate for the 27 turbine angles resolved is 0.145 kg/sec.
The amplitude of the sinusoid causes a 34% ﬂuctuation in mass ﬂow rate about ﬁre average
mass ﬂow rate. The frequency of the sinusoid is 180 Hz, matching the frequency at which

the turbine blades cross the face of the impeller passage.

4.3.3 Plane 6

Plane 6 is oriented along the Y axis as a plane parallel to ﬁre X-Z axes. Results ﬁom
this plane can be seen in Figure 48 through Figure 53. It is located -1.29 mm from ﬁre origin
of the Global coordinate system along ﬁre Y axis. This location corresponds to a plane
parallel to the face of the impeller exit in the X-Z plane, 1 mm outside of ﬁre passage in the
region between ﬁre impeller exit and the turbine entrance. The probe was oriented at a 12.6°
angle ﬁ'om ﬁre Y axis in ﬁre X-Y plane. This angle is parallel to ﬁre blades at ﬁre exit of the
impeller section. Assuming the ﬂow continues from the exit at ﬁris angle, ﬁris positioning
will allow for accurate measurements of streamwise ﬂow through ﬁre plane and maximizes
the width of the passage available for measurement. There are no blades in this region to
obstruct the laser beams, it is possible to measure regions behind ﬁre pressure blade in ﬁre
extreme shroud/ suction side corner of ﬁre next passage.Table 13 contains the coordinates

that deﬁne ﬁre area measured.

Table 13. Location of Plane 6 in Global cartesian space

 

Global Coordinates of Plane 6 (mm)
X 2.61 —) 16.11
Y -1.29
Z (radial) 121.08 —) 101.41

 

 

 

 

 

 

 

 

90

The velocity proﬁle for the 0.4 speed ratio case measured velocities very near the
surface of ﬁre shell. There is a region of highly negative ﬂows above the impeller passage
as the shell proﬁle extends above ﬁre turbine shroud. The ﬂow in ﬁris region has negative
velocities of up to -5 m/sec in magnitude. This is the largest magnitude ﬂow reversal
observed in all ﬁre planes. It is hypothesized that ﬂow recirculates in the high static pressure
region, above ﬁre impeller and turbine passages, into ﬁre low static pressure region in where
the ﬂow is exiting from the impeller passage. There is a shearing layer occurring at ﬁre top
of the impeller passage ﬁrat requires 3 mm to go from positive to highly negative in ﬁre
region above the passage. The remainder of ﬁre velocity proﬁle has ﬁre highest velocities
in the pressure/shroud side wiﬁr decreasing velocities as the distance from ﬁris corner
increases. The maximum velocities present in ﬁre pressure/shroud comer are 8 m/sec.
Velocities along the suction side of the passage are between 4 m/sec and 2 m/sec decreasing
ﬁ'om shroud to shell. Figure 48 depicts the plots of the velocity proﬁle for this speed ratio.
The extreme shroud/suction corner of ﬁre next passage can be observed in ﬁris plane. It can
be seen in Figure 49 in the upper right corner of the contour plot. The velocities of ﬁrat

region can be seen in the contour plot as well.

The 0.8 speed ratio also shares a highly negative ﬂow region above the impeller
passage. The negative velocities in this region are the same as ﬁrose observed in the 0.4
speed ratio case at 5 m/sec. The shearing layer is not as pronounced, only requiring 2 mm
to go ﬁom the positive or slightly negative velocities in radial distance of ﬁre impeller
passage to high negative velocities above ﬁre impeller passage. The maximum velocities
reside along the pressure side of ﬁre area measured wiﬁr a magnitude of 5 m/s. The velocity

proﬁle still varies at a nearly linear rate from pressure to suction side reducing to a

91

Plane 6, 0.4 speed ratio (1600/633)

 

Shroud Side

Pressure
Side

Scale_mps

     

Shell Side

 

Figure 48. Velocity profile for Plane 6.4, 1 mm outside the impeller passage

92

Shroud Side

6. 4} 7.073 7805

Suction
Side

 

Scale_mps

Pressure
Side

 

l 2 3 4 5 6 7 8 9 I0 11 12 13 14 15 16
ANIMl ShellSide

Figure 49. Contour map of Plane 6.4, 1 mm outside of the impeller passage

93

minimum value along the suction side of -3 m/sec. Figure 50 contains the 0.8 speed ratio
velocity proﬁle. Figure 51 has a contour plot depicting ﬁre highly negative ﬂows in ﬁre
shroud/suction side corner of the next passage. This can be seen in ﬁre upper right hand

comer of the contour plot.

The amount of area covered by the measurements made for ﬁre 0.4 and 0.8 speed
ratio can not be directly compared to a total area since it is not bound by any geometric
constraints. Therefore a percentage of ﬁre total area covered can not be directly calculated.
2

The areas that were measured in the 0.4 and 0.8 cases was 259 mm2 and 243 mm

respectively. This is a difference of only 6% from ﬁre larger of the two areas.

The mass ﬂow rate data for Plane 6.4 does not reveal any solid trends. This can be
seen in Figure 52. The maximum mass ﬂow rate for this speed ratio data set was 0.74 kg/
sec and ﬁre minimum 0.69 kg/sec. This is a 7% variation ﬁ‘om the average ﬂow rate of 0.71
kg/sec. The mass ﬂow rate does increase from the minimum value to the maximum and
then back down to the minimum value again. However, ﬂuctuations in the data make a clear
sinusoidal trend unclear. Additional measurements need to be made under ﬁre same
conditions to average out the ﬂuctuations. The rise and fall of the ﬂow rates in a broad
sense, indicates a dependency based on ﬁre passage of the turbine blade in front of the

impeller passage exit but a sinusoidal trend is not evident ﬁ'om this data.

The 0.8 speed ratio data has a clean sinusoidal trend present. This can be seen in
Figure 53. The maximum and minimum values present in ﬁris data set are 0.22 kg/sec and
0.1 kg/sec respectively. The average mass ﬂow rate was found to be 0.15 kg/sec. This
results in a 80% variation in the mass ﬂow rate compared to the average mass ﬂow rate.

The maximum ﬂow rate occurs after 33% of the turbine passage has passed in front of the

94

 

 

 

 

Shell Side

       

Pressure

Plane 6, 0.8 speed ratio (2000/1600)

Shroud Side
Sucﬁon

.I -—__.3

ANIMl

 

 

Figure 50. Velocity proﬁle for Plane 6.8, 1 mm outside of the impeller passage

95

Shroud Side

Suction
Side

 

 

Scale_mps

Pressure
Side

 

7 8 9 10 17
ANIMl Shell Side

Figure 51. Contour map of Plane 6.8, 1 mm outside of the impeller passage

96

Flowrate vs. Turbine Position
0.4 Speed Ratio (16001633 RPM)

Axial location I - 1.29 mm, side window configuration
0.75. - _ __ _ _ _ ___ -_ _

0.74 4:” Ham 6
0.73 m
0.72 f

A / \/\ /\-

g 0.7 N w V/‘M
é .

i

 

 

 

 

 

 

 

 

 

12 3 4 5 6 7 8 9101112131415161716192021222324252627
TurbinePosltion

Figure 52. Mass ﬂow rate data from Plane 6 for the 0.4 speed ratio

Flowrab vs. Turbine Position
0.6 Speed Ratio (2000I1600 RPM)
Axial location I - 1.29 mm, dds window configuration

 

 

 

 

 

0.25 ., . - __s____._ _
- M
02
g 0.15 //
s 1/ \d
3 0.1 V
ll.
1....
0 . .

 

 

12 3 4 5 6 7 8 9101112131415161718192021222324252627
TurbinePosition

Figure 53. Mass flow rate data from Plane 6 for the 0.4 speed ratio

97

impeller passage exit. The minimum occurs just beyond 66% of the same distance across

the face of the impeller passage exit. The frequency of this sinusoid is 180 Hz.

4.3.4 Summary of Mass Flow rate results for Planes 3 through 6.

A periodic variation of the mass ﬂow rate is present in many of the measurements
made. The frequency of ﬁris variation is the same as the ﬁequency of the turbine blades
passing the impeller passage exit. This indicates a relationship in ﬁre ﬂow wiﬁr ﬁre blade
passing event. The phasing of ﬁre sinusoid along ﬁre passage could not be studied in this
experiment due to the sparse number of volumeteric measurements made between Planes
2 and 3. Between Planes 3 and 5 ﬁrere is no difference in phasing between the 0.4 and 0.8
data sets. The peaks of the sinusoid fall in the same location for each of ﬁre respective speed
ratios, independent of the plane for Planes 3 through 5. Plane 6 has sinusoidal ﬂuctuations
that have a slight phase shift with respect to the oﬁrer side window data. This may be due
to ﬁre deceleration of the ﬂow after exiting the impeller passage or it proximity to ﬁre

turbine blades ﬁremselves.

It is diﬁicult to compare the results of Planes 3, 4, 5 and 6 due to ﬁre linrited amount
of area each measured. The percentage of the total area measured required for a good
comparison is near 70% of ﬁre total area. Since only extrapolation of the data is used to
determine total ﬂow rates over the entire passage area a signiﬁcant percentage of ﬁre area
must be measured in order to represent the remaining unmeasured area with a reasonable
amount of accuracy. Less ﬁran 50% of the total area was measured for each of ﬁre side
window data sets. The amount of area measured in the side window conﬁguration cases is

insufﬁcient to extrapolate a good representation of the ﬂows in the remaining 50% to 60%

98

of the passage. Comparing the extrapolated data over the total passage area conﬁrms this,
as seen in Figure 54. The data shows that conservation of mass can not be determined using
extrapolated data. The change in mass ﬂow rate from one plane to the next is appreciable
showing ﬁre error incurred during extrapolation is signiﬁcant. Planes 4.4 and 5.4 measured
similar areas and had similar velocity proﬁles therefore they have similar mass ﬂow rate
results. This is not useful for drawing conclusions about total mass ﬂow rates ﬁrrough the
plane at that location since insufﬁcient area was measured to extrapolate wiﬁr. The 0.8
speed ratio data shows no comparable mass ﬂow rate values at all since the areas between

each data set varied and ﬁre velocity proﬁles changes signiﬁcantly from Plane 3 to Plane 6.

Comparison of Mass Flow Rates Across Planes 3 thru 6

8 81

 

N
0|

 

 

8

 

_a
U!

 

Mess Flosztetlthsec)

 

 

0.4 SR 0.8 SR
Plane Type

Figure 54. Comparison of all mass ﬂow rate data over all 31 impeller passages

99

CHAPTER 5. CONCLUSIONS

5.1 Summary and Conclusions

The goal of this research was to study the streamwise characteristics of fluid flow
within an automotive torque converter. This study was conducted to gain insight into trends
that maybe useful in the development of improved torque converters. Transient mass ﬂow
rate values for six planes under two operating conditions were determined and the velocity
proﬁles of the streamwise flow were quantiﬁed and displayed relative to spatial and
temporal references of the impeller and turbine assembly positions. Two planes were
measured within the impeller passage at approximately mid-chord between the impeller
exit and impeller entrance orthogonal to the streamwise flow. Three planes were measured
approaching the impeller exit and one was made in the gap region between the impeller and

turbine assemblies.

The following conclusions can be drawn based on the results of this research.

1. One component LDV measurements are a feasible method for measuring streamwise

flow within the torque converter impeller passage and gap region.

2. Repeatability of performance and ﬂow measurements was established by running
performance tests on both torque converter window conﬁgurations as well as re-

measuring several planes of velocity data for comparison purposes.

3. Direct comparison of the mass ﬂow rate data can be made with CFD results over the

measured area for any plane using the streamwise measurements.

100

. Planes l and 2 cover approximately 70% of the passage and allow for veriﬁcation of

CFD total flow rate results over the entire area at that location in the impeller passage.

. High velocity regions progressively move from shell, suction side to the pressure blade,

near the surface of the shroud in Planes 1 through 4 with the 0.8 speed ratio conditions.

. About 20% of the ﬂow area measured in Plane 1 at the 0.8 speed ratio conditions near
the suction, shroud comer exhibited flow reversal that results in flow recirculation in

this and the surrounding regions.

. Separation of the streamwise ﬂow along the shroud surface was not observed in the
mid-chord measurements as reported by Lakshminaraya et al., Dalimonte and Gruever

et al. in this same region [1 l], [13], [15].

. Momentum from the impeller pressure side overcomes the fluid momentum at a
location closer to the impeller exit in the 0.4 speed ratio operating condition as
compared to the 0.8 speed ratio operating conditions. This can be seen as the high
velocity regions of the 0.4 speed ratio condition transition from the suction side of the
passage to the pressure side of the passage where momentum is being delivered to the

ﬂuid.

. The large secondary flow present in the work of Dalimonte at the mid-chord supports a
conclusion that there is a large amount momentum imparted by the impeller lost to the
secondary flows when compared to magnitude of streamwise flows [15]. This results in

a overall loss of efﬁciency with respect to converter performance.

101

10. The highest velocity regions of the planes measured indicate a transition of this flow
field region from locations near the suction and shell surfaces of the impeller to the
pressure blade, shroud corner as flow progresses through the impeller passage and out

into the gap region, regardless of the speed ratio.

11. Planes 3 through 6 show flow reversal occurring in the area near the shell surface and
progress far into the impeller passage from the impeller passage exit, near shell surface.
This recirculation is thought to occur as the high static pressure region above the gap
and impeller exit region transitions into the low static pressure region in the impeller

passage near the exit.

12. Planes 3 through 6 exhibit reversed ﬂow regions that appear to dominate the suction
side of the exit planes. These regions are thought to continue to be exhibit ﬂow reversal
through the remainder of the suction side of the passage contributing to signiﬁcant flow

recirculation in this area

13. Measurements from Plane 6, at both the 0.8 and 0.4 speed ratio conditions, contain data
about the velocity of the streamwise flow near the suction side, shroud corner of ﬁre
impeller passage exit. This highly negative ﬂow observed in this region supports the
conclusion that the suction side of the flow is dominated by negative values by the high

negative magnitude it exhibits.

102

14. Temporal mass flow rate variations result in a single period sinusoid visible in most
planes with a dependency on the turbine blade position. The sinusoids have a
periodicity that is the same as the period of the turbine blade crossing the impeller
passage exit with a frequency of 435 Hz for the 0.4, 1600/633 RPM, speed ratio and

180 Hz for the 0.8, 2000/1600 RPM, speed ratio.

103

CHAPTER 6. RECOMENDATIONS

6.1 Recommendations

The study of streamwise ﬂows within an operating automotive torque converter can
be facilitated by a number of recommendations. The design of the torque converter makes
any measurement of ﬂuid ﬂow within in inherently difficult. This difﬁculty is enhanced by
the need to employ several complicated systems in order to acquire information about the
ﬂows from this device. Improvements to the techniques used in this study as suggestions
for areas of further study or derived from experience are made based on events and results

resulting directly from this experiment.

The following are recommendations made based on results and experiences gathered from

this study.

1. The goal of any new torque converter window design should be to allow for total
coverage of the passage, taking into account the signal to noise interference generated

by the scattering of light from the hardware inside the torque converter.

2. The window designs for both the front and side window torque converter
conﬁgurations need to be redesigned to span more than one passage. This will allow for

measurement of the total passage.

104

. The surface of the shroud and the blades within optical access of the window need to be
made less reﬂective. A reduction of the signal to noise ratio when measuring near these

surfaces is necessary to achieve satisfactory data acquisition near those surfaces.

. The oil leakage about the circumferential seam of torque converter housing should be
eliminated. This will improve the overall ease of data acquisition and reduce the need

for frequent shut downs of the system for cleaning.

. Several planes of streamwise ﬂow measurements at the impeller exit and entrance
covering the total area in the passage need to be made to aid the understanding of mass

ﬂow variations through the impeller passage.

. Reduction of the size of areas exhibiting ﬂow reversal within the converter should be a

prime objective for torque converter product engineers.

. The implementation of CFD modeling techniques should be employed in conjunction
to experimental measurements of ﬂows within the torque converter to accelerate the
process of improving and implementing changes to the theoretical model. This would
also add insight as to how these changes can be directly applied to the design of torque

converters.

105

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[91

LIST OF REFERENCES

Lee, K., Novak, M. and Schock, H., "3-D LDV Measurement of In-Cylinder Air
Flow in a 3.5L Four-Valve SI Engine," SAE Paper No. 950648, 1995

Hwang, K., Lee, K., Mueller, J., Stuecken, T. and Schock, H., "Dynamic Flow
Study in a Catalytic Converter Using Laser Doppler Velocimetry and High Speed
Flow Visualization," SAE Paper No. 950786, 1995.

Hascher, H., Kasser, J., Novak, M., Lee, K. and Schock, H., "An Evaluation of
Turbulent Kinetic Energy for the In-Cylinder Flow of a Four-Valve 3.5L SI Engine
Using 3-D LDV Measurements," SAE Paper No. 970793, 1997.

Kasser, J., Hascher, H., Novak, M., Lee, K. and Schock, H., "Tumble and Swirl
Quantiﬁcation within a Motored Four-Valve SI Engine Cylinder Based on 3-D LDV
Measurements," SAE Paper No. 970792, 1997.

Upton, E., "Applications of Hydrodynamic Drive Units to Passaenger Car

Automatic Transmissions," SAE Design Practices - Passenger Car Automatic
Transmissions, Vol 1.

Fister, W. and Adrian, F. W., "Experimental Researches of Flow in Hydrodynamic

Torque Converters," Proceeding of the 7th Conference of Fluid Machinery in
Budapest, Hunga_ry, Vol. l,pp. 210-224, 1983.

By, RR. and Maloney, J .E., "Technology Needs for the Automotive Torque
Converter-Part1: Internal Flow, Blade Design, and performance," SAE Paper No.
880482, 1988.

Bahr, H.M., Flack, R.D., By, RR, and Zhang, J.J., "Laser Velocimeter
Measurements in the Stator of a Torque Converter," SAE Paper No. 901769, 1990.

McCarrick, D.W., "Experimental Investigation of the Flow Field in an Automotive
Torque Converter", A Thesis, Michigan State University, E. Lansing, Michigan,
1993.

106

[10]

[11]

[12]

[13]

[14]

[15]

[161

[17]

[181

[19]

[20]

By, RR. and Lakshminarayna, B., "Measurement and Analysis of Static Pressure
Field in a Torque Converter Pump," Jgumal of Fluids Engineering, Vol. 117, pp.
109-115, March 1995.

Lakshminarayna, B. and von Backstrom, T.W., "Perspective: Fluid Dynamics and
Performance of Automotive Torque Converters: An Assessment," Journal of Fluids
Engineering, Vol. 118, pp. 665-676, December 1996.

Gruver, J.K., Flack, RD. and Brun, K., "Laser Velocimeter Measurements on the
Pump of an Automotive Torque Converter: Part 1- Average Measurements,"
Transactions of the ASME, Vol. 118, pp. 562-569, July 1996.

Gruver, J .K., Flack, RD. and Brun, K., "Laser Velocimeter Measurements on the
Pump of an Automotive Torque Converter: Part II- Unsteady Measurements,"
Transactions of the ASME, Vol. 118, pp. 570-577, July 1996.

Watanabe, H., Kurahashi, T. and Kojima, M., "Flow Visualization and
Measurements of Torque Converter Stator Blades Using a Laser Sheet Lighting
Method and a Laser Doppler Velocimeter," SAE Paper No. 970680, 1997.

Dalimonte, L.J., "Measurements of Flows in the Impeler and Turbine Side of an
Automotive Torque Converter at 0.4 and 0.8 Speed Ratio", A Thesis, Michigan
State University, E. Lansing, Michigan, 1998.

Private communication, G.R. Becraft, Ford Motor Company Torque Converter
Laboratory, Livonia, Michigan, 1992.

Kubo, M. and Ejiri, E., "A Loss Analysis Design Approach to Improving Torque
Converter Performance," SAE Paper No. 981100, 1998.

Abe, K., Kondoh, T., Fukumura, K. and Kojima, M., "Three-Dimensional
Simulation of the Flow in a Torque Converter," SAE Paper No. 910800, 1991.

Ma, W., Luo, B., and Wu, 8., "The Research on Quasi-Three Dimensional Flow
Design of Hydrodynamic Torque Converter Blades," SAE Paper No. 912201, 1991.

Ejiri, E., "A New Approach to Developing a More Efficient Torque Converter
Stator," SAE Paper No. 901765, 1990.

107

[21]

[22]

[23]

Schultz, H., Greim, R. and Volgmann, W., "Calculation of Three-Dimensional
Viscous Flow in Hydronamic Torque Converters," Transactions of the ASME, Vol.
118, pp. 578-579, July 1996.

Khalid, A., Legrand, J. and Robert, J .M., "Turbulent Flow Induced by an Impeller in
a Closed Toroidal Loop," Journal of Fluids Engineering, Vol. 118, pp. 677-684,
December 1996.

"Advances in Fluid Flow Diagnostics," TSI Fluid Mechanics Instrument Division,
pp. 9-43, June 1998.

108

"1111111111111111“