.... hunt.”
.. . .
..a‘u.(e....
I ~1hali< it A! .
mw,dmvhmwmmmmmm«

.. ‘1 . ..

‘- ‘4 up . «x.

:99.” 9x... :3

..;;.
xumkuvflumnwzk #12:!“
x. 1.-.. h. 2.2.!
l1‘?’zl“. ICtlI |
£1? .k
a «v.53. 3. I
(I1. “hr u...
fifiaih...’
.5}.-.
....!.

I

5......“101 3-3125...‘

: Ht. ‘.. is! 0
lrhl‘hr o\ Unifi-

33.9...
.v at... ‘60:}.
.- rifllliru...

5v 1% aisle! .

. . x... ,....In..2ow........vzn . num
.é . . .z

.. ‘xe-1.i.i$c..3. bib»:
. ”Thu...“ ui!;i!.3i¢

a? attlynLrtslaonuun.
03.5 a O
:1... a: I! «5‘23,

OI

¢.vllt \c 1,, I 1..., I1:
3 3... t. 3. 3 tint-0.5!! 2.7...-

 

 

 

 

, 2: LIBRARY

Michigan State
University

 

fi—

This is to certify that the
thesis entitled

ENABLING TECHNOLOGY FOR THE USE OF R718 IN A
VAPOR-COMPRESSION REFRIGERATION CYCLE

presented by

BRUCE ERNEST LINDBERG, JR.

has been accepted towards fulfillment
of the requirements for the

MS. degree in Mechanical EngineerinL

 

 

M/éé

Major Professor‘s Signature
7/! /07

Date

 

MSU Is an Afllnmflve Action/Equal Opportunity Enployor

----a-l-I-‘-‘-O-I-l-I-n--o-v-u-

_.-t-.-.-a-.-g-a-u-------<-

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KziPrq/AccsPresIClRCIDateDue.indd

 

 

 

ENABLING TECHNOLOGY FOR THE USE OF R718 IN A VAPOR-
COMPRESSION REFRIGERATION CYCLE

By

Bruce Ernest Lindberg, Jr.

A THESIS

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

MASTER OF SCIENCE
Mechanical Engineering

2009

ABSTRACT

ENABLING TECHNOLOGY FOR THE USE OF R718 IN A VAPOR-
COMPRESSION REFRIGERATION CYCLE

By
Bruce Ernest Lindberg, Jr.

Water as a refrigerant can be an economical and environmentally conscious
alternative to conventional refrigerants. It has a long history of being utilized to cool
various things throughout human history and even beyond. From ocean currents and
floods to the ice in our drinks, water is perhaps the most natural cooling agent on the
planet. In addition to the environmental benefits, the use of water in a vapor compression
refrigeration cycle will also be shown to perform thermodynamically very efficient. This,
in turn, leads to economical benefits in the realm of power consumption as well as
availability.

In fact, water as a refrigerant is already being utilized in commercial applications
in Europe, Israel, and South Africa. The current state-of-the-art technology used in these
countries is discussed. In addition, the challenges that arise from this technology are also
discussed as well as the various characteristics that make water as a refrigerant an
attractive alternative to conventional refrigerants used in the US.

Finally, the enabling, innovative technology that can be shown to address these
challenges is conveyed. The primary component of this technology includes a novel
compressor with a unique, woven impeller design including its novel manufacturing
technique. Like many technologies, the exploitation of this research is not limited to the
immediate motivating concept of water as a refrigerant, but may also be utilized in other

applications such as power generation, propulsion, venting, and more.

To my Loving Grandparents

Jim and Gwen Wright

&

Ernest “Bud” and Marylyn Lindberg

iii

ACKNOWLEDGEMENTS

While conducting this research and compiling my thesis, I have been assisted
greatly by many people in one way or another. The efforts to complete this thesis would
have been impossible without them, and I would like to take this opportunity to thank
them.

First, I want to thank my family and friends for their endless support and patience
that they have given to me all my life. My relationships with you are the most important
part of my life and have allowed me to be the best version of myself that I could be.
Thank you for always being there for me through the joyous times and especially the
difficult times.

I would also like to thank the past and present members of the turbomachinery lab
at Michigan State. You have made the friendly working atmosphere and have shown the
technical skills that has made our lab what it is known for. For their direct involvement
with the goal water as a refrigerant, I would like to first thank Amir Kharazi and his
guidance in introducing me to the lab and the research. Next, I would like to thank
Ludek Pohorelsky and the graduate students, Nathan Holden, Mark Mouland, Kristen
Papuga, Anirban Lahiri, Neelima Reddy, Qubo Li, Jifeng Wang, Mohit Patil, and Blake
Gower for their direct involvement. The research has also benefited from the assistance
of talented undergraduate researchers who should be named. They are Allen Eyler, Eric
Tarkleson, Jakub Mazur, Paul Allen, Darren Fung, Piere Burk, Devesh Chaphalkar, Joel
Darin, Jarreau Jackson, and David Kempf.

I would also thank Michigan State University, the College of Engineering, and the

department of Mechanical Engineering for giving me the opportunity to pursue my

academic career here. In addition, I would like to thank our chief financers, California’s
EISG program and ASHRAE for funding the research. A special thanks to my
committee members Dr. Abraham Engeda and Dr. Alan Haddow for their valuable time
and guidance as well.

Finally, I must thank my advising professor, Dr. Norbert Miiller for his invaluable
guidance and infinite patience while pursuing my education. Dr. Miiller has embraced all
of his students and their diverse spectrum of backgrounds, methods, and opinions. He
has established a working philosophy that encourages creativity, teamwork, and
happiness in an environment where such things come second to results and bottom-line
thinking. I am fortunate to have had the opportunity to work with Dr. Muller, and I will

forever be grateful for what he has taught me about engineering as well as about life.

PREFACE

The purpose of this thesis is to present the broad research conducted at Michigan
State University on the enabling technology for water as a refrigerant (R718). Over
several years the author of this thesis has worked as a part of a collective effort consisting
of numerous members working toward a common goal. It is the objective of the author,
to present his own specific work on the thesis topic in the perspective of a broad
“roadmap” of the technology.

Unlike many theses, this work is an overall proof of concept and not an in-depth
analysis of a specific area of the technology. Therefore, the intimate details of specific
components of this technology are not discussed. Although important to the research,
there are no rigorous optimizations of blade design, surface features, specific
thermodynamic applications, and the like. This thesis does however identify and
discusses several different principles, that when employed together, can be shown to be
feasible in accomplishing the immediate motivation of enabling water as a refrigerant.

Another unique characteristic that arises from a broad approach to a general
research topic is the number of different options to the enabling technology that are
explored. This research does not identify and pursue a single solution to a problem with
a beginning, middle, and end. This thesis does however explore several options and
solutions, all of which have their own inherent advantages and disadvantages.

This thesis is organized in individual chapters that are designed to guide the
reader throughout the enabling technology. Chapter 1 is an introduction that also outlines
the motivation for the technology, followed by Chapter 2 that introduces the refrigeration

cycle employed. Subsequent chapters explain the challenges, advantages, and specifics

vi

R718, plus representative cycle calculations. Chapter 6 summarizes the current state-of-
the-art R718 technology and how it addresses the challenges outlined in Chapter 4.
Going beyond the state-of—the-art, a more advanced solution is presented that may enable
R718 in widespread use. This advanced solution is a compressor design that is described
in Chapters 7 and 8.

Chapter 10 describes the physical realization of the design. To assist with this, a
summary of theoretical and numerical work was added in Chapter 9. This chapter is a
short compilation of the work of Anirban Lahiri. In Chapter 12, recommendations of
further study are given to convey ideas that have not yet been pursued, as well as re-
iterate the importance of an older concept that has not yet been realized - the condensing

wave rotor first investigated by former research group member, Amir Kharazi.

vii

Table of Contents

LIST OF TABLES .............................................................................................................. x
LIST OF FIGURES ........................................................................................................... xi
CHAPTER 1 : INTRODUCTION ..................................................................................... I
CHAPTER 2 : BACKGROUND ....................................................................................... 2
The Vapor-Compression Refrigeration Cycle ................................................................ 2
The Ideal Vapor-Compression Refrigeration Cycle ................................................... 3

The Actual Vapor-Compression Refrigeration Cycle ................................................. 4
Refrigerants ..................................................................................................................... 6
CHAPTER 3 : BENEFITS OF WATER AS A REFRIGERANT ..................................... 8
Environmental Benefits .................................................................................................. 8
Economic Benefits .......................................................................................................... 9
CHAPTER 4 : CHALLENGES OF WATER AS A REFRIGERANT ........................... 10
Working Under Vacuum ............................................................................................... 10
Large Volume Flows .................................................................................................... 11
High Pressure Ratios ..................................................................................................... 1 1
Low Reynolds numbers ................................................................................................ 11
CHAPTER 5 : THE CYCLE ........................................................................................... l3
Mapping COP and pressure ratios ................................................................................ 14
Other Factors that impact COP ..................................................................................... 17
Heat Exchangers ....................................................................................................... l7
Polytropic Efficiency ................................................................................................ 21
CHAPTER 6 : CURRENT STATE-OF-THE-ART ........................................................ 23
CHAPTER 7 : ALTERNATIVE COMPRESSOR CONFIGURATION ........................ 28
Axial Flow Compression .............................................................................................. 28
Multi-Stage Compression ............................................................................................. 30
CHAPTER 8 : NOVEL DESIGN FOR MANUFACTURING ....................................... 33
Composite Materials ..................................................................................................... 33
Filament Winding ..................................................................................................... 34
Fiber Material ............................................................................................................ 34
Matrix Material ......................................................................................................... 36
Winding Patterns ........................................................................................................... 4O
Advatages of Winding Design ...................................................................................... 44
Driving Scheme ............................................................................................................ 46
Brushless Permanent Magnet Motor ......................................................................... 47
Induction Motor ........................................................................................................ 49

viii

CHAPTER 9 : COMPUTATIONAL ANALYSIS OF DESIGN .................................... 51

Pattern Modeling ........................................................................................................... 51
Interfacing for 3-D Modeling .................................................................................... 51
Radius of Curvature around Slots ............................................................................. 51
Fiber Collisions ......................................................................................................... 53
Modeling the Resin ................................................................................................... 54

Structural Analysis ........................................................................................................ 55
Setting up the Simulation .......................................................................................... 55
Simulation Results .................................................................................................... 55
Analysis of Simulation Results ................................................................................. 60

CHAPTER 10 : REALIZATION OF DESIGN ............................................................... 67

Rapid Prototype ............................................................................................................ 67

Hand Winding ............................................................................................................... 67

Automated Winding ...................................................................................................... 71
Modifying the CNC Mill .......................................................................................... 72
Challenge of Fiber Creep .......................................................................................... 75
The Patterns .............................................................................................................. 77

Structural Spin Test ...................................................................................................... 79
High-speed Air-turbine Grinding Tool ..................................................................... 81
Balancing and Vacuum ............................................................................................. 82
RC Racecar Nitro-motor ........................................................................................... 85
High-speed Router Tool ............................................................................................ 86

Integrated Motor ........................................................................................................... 86
Induction Motor ........................................................................................................ 87
Permanent Magnet Motor ......................................................................................... 89

A Brief Cost Analysis ................................................................................................... 90

CHAPTER 11 : CONCLUSION ..................................................................................... 92
CHAPTER 12 : RECOMMENDATIONS FOR FURTHER STUDY ............................ 94
APPENDIX A. EXAMPLE OF CYCLE CALCULATION ............................................ 97
APPENDIX B: EXAMPLE OF VISUAL BASIC CODE USED IN CYCLE ................. 99
APPENDIX C. COMPLETE PATTERN DESIGN TABLE ......................................... 105
REFERENCES ............................................................................................................... I l I

LIST OF TABLES

Table l. Refrigeration comparison chart [7] ..................................................... 12
Table 2. Properties of Kevlar aramid fiber yarns [25] .......................................... 36
Table 3. Abbreviated table of winding patterns (full table in Appendix C) .................. 41
Table 4. Spin test inquiries and quotes .......................................................... 80
Table 5. Air-turbine spin test trials ............................................................... 85

LIST OF FIGURES

Images in this thesis are presented in color.

Figure 1. Schematic example ofvapor-compression cycle ................................................ 3
Figure 2. Schematic T-s diagram for the ideal vapor-compression cycle ........................... 4
Figure 3. Schematic T-s diagram for the actual vapor-compression cycle ......................... 5

Figure 4. Schematic example of actual vapor-compression cycle with system components

............................................................................................................................................. 6
Figure 5. COP iso-Iines for R718 with np=0.7 ................................................................ 15
Figure 6. COP iso-Iine comparison between R718 and R134a with np=0.7 .................... 16
Figure 7. Zoomed-in view of COP iso-Iine comparison between R718 and R134a with

np=0.7 ............................................................................................................................... 17
Figure 8. Effect of heat exchangers on COPS of R7 I 8 and R134a ................................... 19

Figure 9. COP iso-Iine comparison between R718 and R134a with np=0.7 including heat
exchanger effects .............................................................................................................. 20

Figure 10. Zoomed-in view of COP iso-Iine comparison between R718 and R134a with
np=0.7 including heat exchanger effects .......................................................................... 20

Figure l I. Zoomed-in view of data point in a COP iso-Iine comparison between R718
and R134a with np=0.7 including heat exchanger effects ................................................ 21

Figure 12. COP iso-Iine comparison between R718 and R134a with np=0.9 including
heat exchanger effects ....................................................................................................... 22

Figure 13. Effect of polytropic efficiency on COPS of R718 and R134a including heat
exchanger effects .............................................................................................................. 22

Figure 14. Commercialized and installed 100-300 ton R718 chillers developed by ILK
Dresden, Germany ............................................................................................................ 23

Figure 15. Schematic illustration of the current state-of-the-art R7 I 8 refrigeration cycle25

Figure 16. Illustration of current state of the art compressor wheels ................................ 26
Figure 17. Current state of the art units at IDE, Israel ...................................................... 27
Figure 18. Diameter comparison between axial and centrifugal impellers ..................... 29

xi

Figure 19.
Figure 20.

Figure 21.
resin .........

Figure 22.

Compressor stage comparison ......................................................................... 31
Fiber tow being run through a resin bath for wet winding .............................. 38
Automatically woven impeller prototype pattern 6-8 with post-process applied

.......................................................................................................................... 39

Automatically woven impeller prototype pattern 8-C with post-process applied

resin ................................................................................................................................... 39
Figure 23. Schematic illustration of fiber path in pattem"l skip #" ................................. 42
Figure 24. Schematic illustration of fiber path in pattern "1 skip 2" ................................ 43
Figure 25. Illustrative model of woven impeller pattern 8-B ........................................... 45
Figure 26. Commercially available Super Hornet winder from
www.mccleananderson.com ............................................................................................. 46
Figure 27. Schematic, illustrative example of a single-phase permanent magnet motor .47
Figure 28. Schematic, illustrative example of a three-phase permanent magnet motor... 49
Figure 29. Computer model of woven impeller including its support mandrel ................ 52
Figure 30. Magnified detail of a wound impeller model, illustrating the radius of
curvature of the fiber ......................................................................................................... 52
Figure 31. Detail showing collision avoidance of the fiber with segments that have
previously crossed the cylinder ......................................................................................... 53
Figure 32. An Impeller Model in which all fiber collisions occur at the center of the
cylinder ............................................................................................................................. 54
Figure 33. Pattern 7A - Displacement ............................................................................... 56
Figure 34. Pattern 7A - Stress ........................................................................................... 57
Figure 35. Pattern 7B - Displacement ............................................................................... 57
Figure 36. Pattern 7B - Stress ........................................................................................... 58
Figure 37. Pattern 8A - Displacement ............................................................................... 58
Figure 38. Pattern 8A - Stress ........................................................................................... 59
Figure 39. Pattern SB - Displacement ............................................................................... 59
Figure 40. Pattern 88 - Stress ........................................................................................... 60

xii

Figure 41.
Figure 42.
Figure 43.
Figure 44.
Figure 45.
Figure 46.
Figure 47.
Figure 48.
Figure 49.
Figure 50.
Figure 51.
Figure 52.
Figure 53.
Figure 54.
Figure 55.
Figure 56.
Figure 57.
Figure 58.
Figure 59.
Figure 60.
Figure 61.
Figure 62.
Figure 63.
Figure 64.

Figure 65.

Stress vs. skip for patterns with 6-9 slots ........................................................ 61
Stress vs. skip for patterns with 10-14 slots .................................................... 62
Stress vs. skip for patterns with 15-18 slots .................................................... 63
Effect of fiber thickness for 7 slot pattern ....................................................... 64
Effect of fiber thickness for 9 slot pattern ....................................................... 64
Effect of fiber thickness for 16 slot pattern ..................................................... 65
Rapid prototype of woven impeller with electric stator .................................. 67
Dry hand-wound prototype to be used in integrated motor ............................. 68
Hand winding using dry fiber and resin bath .................................................. 69
Hand winding using pre-preg fiber .................................................................. 69
Finished result of hand winding using resin bath ............................................ 70
Finished result of pre-preg hand winding before heat curing .......................... 70
Photo of CNC winding set-up ......................................................................... 72
Portable CNC winding machine ...................................................................... 73
Winding machine with fiber spool and resin bath added ................................ 74
Photo of winding with leader tool ................................................................... 75
Sandpaper applied to outside of mandrel for added friction ........................... 76
Winding shown with actuator to re-position fibers after fiber creep ............... 77
Automated, wet wound prototype of pattern 8-B ........................................... 78
Winding pattern 8C including a shaft .............................................................. 79
Photo of spin test using air-turbine (front view) ............................................. 81
Photo of spin test using air-turbine (rear view) ............................................... 82
Flexible shaft coupled to wheel (left) and static balancer (right) .................... 83
Airflow prevention chamber for spin test ........................................................ 84
Design illustration of vacuum test chamber .................................................... 84

xiii

Figure 66. Test bench for RC racecar nitro motor ............................................................ 86
Figure 67. Prototype woven impeller with integrated motor in an induction machine 88
Figure 68. Woven impeller shown with its corresponding electromagnetic stator ........... 89

Figure 69. Woven impeller with its integrated motor shown still (left) and spinning (right)
........................................................................................................................................... 89

Figure 70. Fume hood for experimentation with magnetic powders (left) and iron-
powder/epoxy mixture experiments (right) ...................................................................... 90

xiv

CHAPTER I : INTRODUCTION

The initial motivation of the research presented in this thesis is for the application
of water as a refrigerant (R718) for conventional, residential use in the United States.
Water may be the most natural cooling agent on the planet. It is a substance that is
essential to preserving life-itself, and as will later be demonstrated, can also be used to
preserve our quality of life as well.

The need for refrigeration is essential for modern civilization. These needs
include, but are not limited to, food and tissue preservation, chemical environments for
use in medical and science industries, and air conditioning. Refrigeration can be defined
as any heat transfer process that reduces the temperature of an object or location below
that of its surroundings [1]. However, the scope of this thesis will be restricted to
refrigeration in a vapor compression cycle.

For a perspective on the demand for residential air conditioning, as well as its
impact on the environment and economy, a look at the state of California makes logical
sense. Perhaps no state in the US. is more recognized as being at the forefront of the
energy battle than California. The rolling black-outs of 2000 and 2001, economically-
crippling energy prices, and increasing demand for air conditioning prompt new
innovation in the area of air conditioning and refrigeration. Already in 1997, 96% of the
households in California used electric air-conditioning and 30% of peak electricity usage

in California is due to air-conditioning [2].

CHAPTER 2 : BACKGROUND

The Vapor-Compression Refrigeration Cycle

Refrigerators, heat pumps, and air conditioning systems operate on cycles called
refrigeration cycles. There are many types of refrigeration cycles. Some examples are
gas refrigeration, cascade refrigeration, absorption refrigeration, and thermoelectric
refrigeration cycles. However, the most frequently used refrigeration cycle is the vapor-
compression refrigeration cycle [3]. In this refrigeration cycle, the working fluid or heat
transferring medium is called the refrigerant. This refrigerant is throttled through four
major components of a vapor compression refrigeration cycle and uses the process of
latent heat rejection and absorption during phase changes. The four components are:

I. Compressor- where the refrigerant in vapor form is compressed and work-energy
is inputted to the cycle

2. Condenser- where heat-energy is rejected to the environment as the refrigerant is
condensed

3. Expansion valve- where the cycle is throttled by expanding the refrigerant

4. Evaporator- where latent heat is absorbed by the refrigerant as it is evaporated

This cycle with its components is shown schematically in Figure l. A common measure
of how a cycle like this performs is a dimensionless quantity known as its Coefficient of
Performance (COP). COP is defined mathematically as

Qin (l)

COP = —
Win

where Qom is the heat absorbed by the evaporator in the cycle and Win is the work input to
the cycle. These two quantities may also be expressed on a rate basis. Essentially, the

COP is the amount of cooling energy per unit of compressor work energy [I].

IQ...

condenser <

 

 

 

A

 

 

 

 

 

 

\

compressor

expansion valve

 

 

> eva porator

lg.

m

 

 

 

 

Figure 1. Schematic example of vapor-compression cycle

The Ideal Vapor-Compression Refrigeration Cycle

For simplicity in introducing the vapor-compression refrigeration cycle, an ideal
cycle is shown in a T-s diagram in Figure 2. The characteristics of an ideal cycle are:
l. Isentropic compression (I-2)
2. Constant pressure heat rejection and complete condensation (2-3)
3. Isentropic expansion through a throttling expansion valve (3-4)

4. Constant pressure heat absorption and complete evaporation (4-1)

TA

 

 

 

 

 

 

 

‘41!

Figure 2. Schematic T-s diagram for the ideal vapor-compression cycle

Although this ideal cycle makes for easy property calculations, it is not practical
for describing the cycles that actually occur in the vapor compression refrigeration
process. For this, irreversibilities such as heat and pressure losses, as well as super-

heating and sub-cooling must be taken into account.

The Actual Vapor-Compression Refrigeration Cycle

The actual cycle differs from the ideal one in many ways. A T-s diagram of this
cycle is shown in Figure 3. Numbered points on this figure are also schematically shown
in a cycle with system components in Figure 4. Most of the irreversibilities that occur in
this cycle can be attributed to pressure losses due to fluid friction as well as heat transfer
between system components. The ideal case described above included isentropic
compression from 1-2. This involves a compression process that is internally reversible

4

and adiabatic. However in reality, compression results in an increase in entropy shown in
process 1-2 in Figure 3. Heat losses during compression can also result in a decrease in
entropy as shown in process 1-2’ in Figure 3. This non-adiabatic compression is
favorable in that it decreases the amount of overall compressor work needed. It is
therefore favorable to cool the compression in an actual cycle.

In an ideal cycle the refrigerant is evaporated and condensed to exactly saturated
vapor and saturated liquid. In reality however, it is difficult to control the state of the
refrigerant so precisely. To ensure complete vaporization before compressing, the
refrigerant is slightly superheated in the evaporator before entering the compressor.

Likewise, to ensure complete condensation, the refrigerant is also slightly subcooled.

I2

TA

 

 

 

V

CI:

Figure 3. Schematic T-s diagram for the actual vapor-compression cycle

 

4 to... 3

1 condenser <

A

 

 

—_

 

 

 

 

 

\——2

compressor

or

expansion valve

 

 

> evaporator

I ,
7 {Q I

l

 

 

 

 

 

 

Figure 4. Schematic example of actual vapor-compression cycle with system components

Refrigerants

A refrigerant, in the case of a vapor compression cycle, can be defined as the
medium that absorbs and releases thermal energy to and from two dissimilar
environments [I]. This refrigerant should possess certain chemical, physical, and
thermodynamic properties that render it safe, effective, and economical to use. More
specifically the refrigerant should be nontoxic and nonflammable in its pure state or when
mixed with air. In addition, the refrigerant should not react unfavorably with system
components such as construction materials, lubricating oil, and moisture in the system.
Lastly, but not in importance, a refrigerant should be economically attractive as well as
environmentally safe, not only in its acquisition, manufacture, and transport, but also its

inevitable disposal.

While many substances possess one or more of these characteristics, several do
not possess all of them, hindering the substance’s applicability. For instance, in the early
days of mechanical refrigeration, ammonia (R717) was widely used. Although ammonia
is a natural substance, has a low boiling point at atmospheric pressure, and has thermally
shown to operate very well in a refrigeration cycle, it is very toxic. This makes this
particular refrigerant extremely dangerous and limits its applicability to highly controlled
commercial applications. In the late 1920’s synthetic refrigerants were developed that
were safe, non-toxic, and non-flammable. These compounds are known as
chlorofluorocarbons (CFC’s). In the 1970’s it was shown that CFC’s depleted the ozone
layer and in 1987 the Montreal Protocol was signed and the manufacture of CFC’s was
phased out by 1996 [1]. Another synthetic refrigerant, 1,l,l,2-tetrafluorethane (R134a),
is one of the leading candidates to replace these CFC refrigerants [I]. R134a is non-
flammable, non-explosive, non-toxic, and has no ozone depletion potential (ODP).
However, it does have some global warming potential (GWP) because of a small green-

house effect.

CHAPTER 3 : BENEFITS OF WATER AS A REFRIGERANT

Often for simplicity’s sake, several textbooks introduce the refrigeration cycle to
novice thermodynamics students by using water as the working refrigerant. However, as
the lesson continues, students are reminded of water’s high saturation temperature at
ambient pressures, as well as its low volumetric heat capacity. The lesson then abruptly

abandons the use of water and subsides to synthetic refrigerants like R12 and R134a.

Environmental Benefits

To offset the continuous threat of global warming and ozone depletion,
regulations and bans of traditional refrigerants have been handed down by governments
and agencies, namely the Montreal Protocol signed by 57 industrialized countries in
1987. Because of these regulations, the investigations of natural refrigerants such as
water are becoming more necessary. Water being completely inert to the environment
has many environmental advantages over traditional refrigerants. It has no global
warming potential (GWP = 0) and no ozone depletion potential (ODP = 0). In addition, it
is non-toxic, and non-flammable. Water can easily be disposed of and needs no
manufacturing or extensive refining. While traditional refrigerants meet today’s
restrictions and standards, it is almost inevitable that these restrictions are bound to
change. Water can be guaranteed not to fall under future restrictions. Also included with
the economic benefits, is the thermodynamic efficiency of a R718 cycle. These
efficiency gains will be discussed in greater detail later, but undoubtedly reduce the

needed energy to operate a chiller using water as a refrigerant.

Economic Benefits

In addition to its many environmental benefits, R718 also includes several
economical advantages as well. The first and probably most obvious advantage is the
availability of R718. Water covers roughly two thirds of the earth’s surface. Special
treatments are not needed. Municipal tap water can be used, as well as filtered river or
stream water. Treated waste water is another possibility. Since the refrigerant is so
readily available, there would be no need to warehouse bulky refrigerant containers. The
gross cost of the refrigerant would be less since water needs no manufacturing. R718
also reduces safety precautions by working with low pressure differences (less than 1
atm). This may cut down on insurance premiums.

When discussing economic benefits, it is most important to include the energy
efficiency of R718 cycles. Therrnodynamically, water can be shown to achieve a high
coefficient of performance (COP). In studies, the COP is shown to be 20-30% higher
than conventional refrigerants [4]. Unlike other refrigerants, R718 can be used in direct
heat exchangers, increasing the efficiency of the cycle even more. This technology will

be discussed later in Chapter 5.

CHAPTER 4 : CHALLENGES OF WATER AS A REFRIGERANT

Just as R718 can be shown to be beneficial as a refrigerant, there are also several
key characteristics that complicate its immediate application in refrigeration cycle, and

are outlined as follows:

Working Under Vacuum

As explained in Chapter 2 of this thesis, the vapor compression cycle works by
absorbing heat from a cold space during a phase change. That is to say, as heat is
absorbed from the cold space to the evaporator, the refrigerant then evaporates or boils.
A novice student of elementary science can tell you that at standard pressure, water
evaporates at 100°C. However, as pressure decreases, water evaporates at lower
temperatures. Since most applications require heat removal at temperatures well below
ambient temperatures, cycles using R718 must he ran at low pressures under a coarse
vacuum.

Working under vacuum comes with its own inherent challenges. A pump must
evacuate piping in the cycle and sealing must be employed to maintain the low pressure.
Challenges in sealing may occur at joints and couplings within the cycle. Many systems
are pressurized and the challenge is to contain the contents from escaping into the
ambient environment. However under vacuum, it is a task to keep the ambient
environment from contaminating the system. In addition, most mechanical and electrical
components are designed for operation under ambient pressure. Their immediate

application may need to be modified for use under ultra-low pressure.

10

Large Volume Flows

Since the cycle works under vacuum pressure, the volumetric cooling capacity of
water vapor is very low. Hence a cycle using R718 needs a larger volume flow of about
200 times higher than traditional refrigerants. That being said, the majority of turbo
compressors for refrigeration are centrifugal or mixed-flow, and although capable of
achieving greater pressure ratios per stage, these types of compressors are not suited for

high volume flows.

High Pressure Ratios

Due to the thermodynamic properties of water vapor, its application in a
refrigeration cycle requires a high pressure ratio to be achieved. This pressure ratio is
about double, or triple that of a cycle using traditional refrigerants like R134a or R12. A
comparison of properties like this can be shown in Table 1 below. It should be noted that
calculations in this table were done using an evaporator temperature of 10°C and a

condenser temperature of 50°C.

Low Reynolds numbers

Due to the high pressure ratio required, turbo impellers need to have
approximately two to four times higher circumferential speed, depending on the impeller
design. The speed of sound is approximately 2.5 times higher, so material limitations are
of more concern than Mach number restrictions. Reynolds numbers are like 300 times

lower and the specific work transmission per unit mass has to be around 15 times higher.

[5], [6], [7]-

ll

Table 1. Refrigeration comparison chart [7]

 

 

 

 

 

 

 

 

 

Capacity Pressure Compressor
Refrigerant ODP GWP
[kJ/kg] Ratio Temp out |°C]

R718 0 0 2309.0 10.0 223

R717 0 0 102.9 3.03 99

R12 1 8500 106.5 2.88 55

R22 0.034 1900 145.6 2.85 67

R290 0 20 250.1 2.66 54

R134a 0 1600 131.9 3.18 62

R152a 0 190 228.0 3.15 52

 

 

 

 

 

 

12

 

CHAPTER 5 : THE CYCLE

To describe the characteristics of water used in a vapor compression cycle, as well
as to demonstrate the performance of R718, a cycle analysis was conducted. When
describing any vapor compression cycle calculation, it is important to first disclose the
assumptions made. To make the calculations as straightforward as possible, several
simplifying assumptions were made, and are as follows:

1. Compression is done in a single stage.

2. There is no superheating or sub-cooling.

3. There are no pressure losses in the piping and no energy transfer to the
environment.

The cycle was calculated with Microsofi Excel using an assumed polytropic
efficiency. Properties such as temperature, pressure, enthalpy, and entropy, were
calculated at each of four different stages in the refrigeration cycle. These stages were
the evaporator, compressor, condenser, and expansion valve. Using the pressure ratio
obtained, from these calculations, an isentropic efficiency was calculated based on the

formula

 

Ila - 1
”isentropic = ( a ) (2)
n Tlpolytropic ... 1

Here npolympic is the assumed polytropic efficiency, [1 is the pressure ratio, and a is the

isentropic exponent or

13

 

a = (3)
K
where K is the ratio of specific heats.
CP
K = — 4
Cu ( )

With the new calculated isentropic efficiency, properties were iteratively re-calculated to
reflect better accuracy. Using these properties, the cooling load and compressor work

were calculated and thus the COP could be calculated by equation (1).

Mapping COP and pressure ratios

As with calculating the COP of any refrigeration cycle, the COP varies with
properties and cycle characteristics. Two important properties that affect COP are the
evaporator temperature as well as the condenser temperature. This is intuitive since one
would expect a cycle to be more efficient the less cold the “cold space” needed to be. By
the same logic, one would also expect better efficiency the colder the ambient space the
heat was rejected to was. Since the evaporator and condenser temperatures are perhaps
the first criteria when designing a refrigeration cycle, COP values were plotted with
respect to these criteria.

Plotting COP versus evaporator temperature alone fails to convey the effects of
condenser temperature on COP. Likewise, plotting COP versus condenser temperature
alone doesn’t describe the effects of the evaporator temperature on COP. To essentially

describe three parameters simultaneously, COP iso-lines were plotted for evaporator

14

temperature versus temperature lift from condenser to evaporator. This was done by
using a Visual Basic code to locate combinations of evaporator temperature and
temperature lift that yield a particular COP value. Integer COP values from 2 to 10 were
used and iso-lines were mapped. A base line plot of this using R718 as the refrigerant is
shown in Figure 5. Here one can see the large operating range of R718 as well as the
general trend for decreasing COPs with increasing temperature lift. The vertical yellow
line represents the optimum operating temperatures to maximize COP. These optimal

temperatures indicate high efficiencies for R718 in the case of using a heat pump.

 

F . . ,, Ls --- -._. - --- - -I,-_-L-- ~-— -~ ___- - 110
~ 100
r 90
- 80
i 70

 

Tllft

‘50

 

‘40

x7
‘8
‘9 . 30
W 10

 

 

 

 

 

 

 

 

 

 

i

i " 20

I i 10

L 0

-20 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300
To

 

 

 

Figure 5. COP iso-lines for R718 with np=0.7

To compare COP values of R718 to other conventional refrigerants, COP iso-lines can be

mapped on the same plot such as in Figure 6. In this case, R134a is being compared. It

15

is worth noting the operating temperatures most favorable to R134a as well. The
intersection of each refrigerant’s iso-lines is marked with a green dashed line. From this,
it is shown that for operating temperatures on this line, R134a and R718 have the same
COP. Furthermore, one can make the valid assumption that R718 has a higher COP
value for operating temperatures to the right of this intersection line. To accentuate this
range of operating temperatures favorable to R718, Figure 7 shows a plot zoomed in on

this range.

 

 

 

110

R7—8
‘ 90
2
‘ ‘ 80
1
l
1 - vo
‘ 3
‘ 60
I

A ‘ _ 20
l:—
\ - 1o

.
T lift

it

I

1
1
1
1
1
l
1
1

ll
1:11

1 o

l L AL I I I I I I I I I I I I I I I I o
-90 -7o -50 -30 -10 10 30 50 70 90 110 130150170190 210 230 250 270
Te

 

 

 

 

Figure 6. COP iso-line comparison between R718 and R134a with np=0.7

 

 

 

 

 

7 7 I '7 ’ ’ " 3°
2
‘ 70
‘ 60
3
\\
\ ‘ 50 a
\ 4 ':
\ ’-
\ 40
\ L‘ 5
M \ ‘ 7
. “J ‘ 3
”4
I I I I I I I I I I I I I I ‘ 1o
-5 0 5 10 15 20 25 30 35 40 45 50 55 60 55 70 75 80
Te

 

 

 

Figure 7. Zoomed-in view of COP iso-line comparison between R718 and R134a with
np=0.7

Other Factors that impact COP

Heat Exchangers

By Figure 7, it can be argued that the operating temperatures where R718 has a
higher COP than R134a is quite limited. However, an important characteristic of the two
cycles was neglected. One of the biggest advantages of using R718 is that it may be used
in direct heat exchangers. This means that there doesn’t need to be such a substantial
temperature difference to drive the heat transfer in the evaporator and condenser. The
resulting effects are a larger temperature lift needed for the same cooling load. This, in

turn, decreases the COP. In the case of R134a, this decrease in COP is much more than

for the case using R718. A conservative assumption of the temperature difference needed
in R134a heat exchangers would be 5°C. Conversely, R718 would only need a 1°C
temperature difference. A demonstration of the effect the different heat exchangers have
on their respective COPS can be Shown in Figure 8 Here a COP of 5 is used, and one can
see the intersection point go from an evaporator temperature of about 35°C to a much
lower temperature below R718’s operating range. Figure 8 also shows the COP 5 line of
R134a dropping well below the R718 line indicating a better COP for R718 at these
conditions. A new, more accurate plot of all the COP iso-lines can now be observed in
Figure 9. This Shows a much larger range of operating temperatures where R718 has a
higher COP when compared to the graph in Figure 6. Like Figure 7, Figure 10 shows a
zoomed view of these optimal temperatures, but with the heat exchanger effects taken
into account. From this, one can assess the fact that with these conditions, R718 has a
higher COP than R134a for evaporator temperatures greater than -1°C and for any
temperature lift. Another interesting observation is that from this plot it can be deduced
that for the particular case of a 12°C evaporator temperature and a 20°C temperature lift,
the COP of R718 is 10 and the COP of R134a is 7. This point is demonstrated by the

shaded oval in Figure I I.

18

 

 

 

I — Heat exchangers neglected

— Heat exchangers included

 

/

 

 

\

R718 COP drop

R134a COP drop

 

 

5 15 25 35
Te

45 55

65

75

80

70

60

50

30

20

T lift

 

Figure 8. Effect of heat exchangers on COPs of R718 and R134a

 

 

 

*‘ * ' ‘ r r , ' , ' r r' "‘ 110
a 100
‘ 90
- 80
‘ 70

‘60

T lift

<50
-40
‘30
[20
41o

 

 

l I l I l I I r I I 1 I L 1 EL J I o

400 -80 -60 40 -20 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300

Te

 

 

Figure 9. COP lso-line comparison between R718 and R134a with np=0.7 including heat

 

exchanger effects
, , , , a — so
70
3 so

\\ .
____- 6

Tlift

 

 

 

 

 

/ ’—
q 30
———’ “‘.~ 3
’J—L.‘ .
— "
*— 7 “ _
,!————.-_’—— 1o 20
_._.—————— \ \
I I I 1 1o
6 1 5 35 55 75
Te

 

 

 

Figure 10. Zoomed-ln View of COP iso-line comparison between R718 and R134a with
np=0.7 including heat exchanger effects

20

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l
I
1 - 24
i 7 #»
I
l ’ ‘
i 8., \j '5‘
I i l-
g__ 16
10
‘ 8
I L J 0
-5 0 5 10 Te 15 20 25 30 35

 

 

 

Figure 11. Zoomed-in view of data point in a COP iso-line comparison between R718 and
R1343 with np=0.7 including heat exchanger effects

Polytropic Efficiency

As stated before, the cycle was calculated using an assumed polytropic efficiency.
This assumption in itself can have a significant effect on the COP iso-lines. To
demonstrate this, COP iso-lines were mapped on temperature ranges as before, except
now with an assumed polytropic efficiency of 0.9. This plot can be shown in Figure 12.
Again, this plot takes into account the heat exchangers as before. It is evident by Figure
12 that the COP iso-lines for R718 and R134a have both increased. However, as shown

in Figure 13, R718 iso-lines have comparatively increased more.

21

 

T lift

3¢¢~I a v.
I
b
O

 

 

i
I
|
l
l
1
1
1
l

 

I L 1 I_ g—I I r I I o

-100 -80 -60 40 -20 0 20 40 60 80 100 120 140 160 180 200 220 240 260

To

 

 

 

Figure 12. COP iso-line comparison between R718 and R134a with np=0.9 including heat
exchanger effects

 

I 5 L L. 7 ,7 I -
I ‘ —Po|ytropic Efficiency of 0.9 i
i
1

  
  

. 60
— Poltytropic Efficiency of 0.7 ,

'50

 

-40

Tllft

'30

20

 

 

. y , r , . 0
-50 -30 -10 10 30 50 70 90 110 130 150 170 190 210 230 250
To

 

 

 

Figure 13. Effect of polytropic efficiency on COP: of R718 and R134a including heat
exchanger effects

22

CHAPTER 6 : CURRENT STATE-OF-THE—ART

Although the above calculations in the previous chapter convey R718 as a viable
alternative to traditional refrigerants, the challenges outlined in Chapter 2 must still be
addressed. To address the challenges of using water in a compression refrigeration cycle,
innovative technology is necessary. The simple application of conventional refrigeration
technology and compressors has repeatedly been found to be neither economic nor
competitive in the US market [5], [6]. However, R718 units have been installed outside
of the US. in countries like Germany, Israel, and South Africa as Shown in Figure 14.
Still, typically the compressors and complete units are deemed to be too expensive, too

big, and unable to scale down to smaller capacities below 100 ton.

CargoLux

, ' U
N .‘u‘fwr

. 717
Figure 14. Commercialized and installed 100-300 ton R718 chillers developed by ILK
Dresden, Germany

 

 

23

The most important component of 3 R718 chiller is the compressor. There are
world-wide only two companies that have developed and commercialized R718
compressors. The first one is IDE Technologies Ltd., Israel [8] and the second. one is at
Institut fiir Luft- und Kéiltetechnik (ILK) Dresden gGmbH (Institute of Air and
Refrigeration), Germany where the Associate Professor Dr. Norbert Miiller has been a
key engineer in the development effort [9], [10], [11]. Derived from desalination, the
larger IDE compressors have also been used for designs by Danish Technological
Institute (DTI) and the Sabroe Refrigeration Company in Denmark [12] and by
INTEGRAL, Germany [13]. The IDE centrifugal compressors have been developed into
two products called ECOVIM (Vacuum Ice Machine) [14] and ECO-CHILLER [15].
Both the IDE and the ILK compressors are of centrifugal diagonal-flow (axial inlet and
radial outlet) type with stationary inlet guide vanes, and typically installed in a two stage
configuration as shown schematically in Figure 15. The compressor wheels are
approximately l-2 meters in diameter. With a radial diffuser, the compressor diameter
reaches approximately 2-4 meters. The lower values are for the ILK compressors and the
larger values for the IDE compressors. Capacities of these chillers are 100 ton and 300
ton respectively. It is interesting to note that both the lower and higher capacity chillers
have almost the same in technology and price, indicating that these chillers’ size and cost

do not scale with capacity [16].

24

 

 
   
  

throttling

      
   

     

valve .
. I I 1‘ a
. . I ® intercop‘l’r
I I I I
I I I I
O I m:- [ZA 0 0 '0'“

 
 

   

condenser

a e\
‘\.

Cooling tower L t n; =

throttling valve

    
  

ressor
stage 2.

   
 
 
   

 

 

 

Figure 15. Schematic illustration of the current state-of-the-art R718 refrigeration cycle

The compressor wheels are lightweight constructions with extremely thin blades
made of titanium or composite material sheets as illustrated in Figure 16. While energy
savings of 20% have been Shown with this technology [17], the fluid mechanics design is
still much compromised for stability. This indicates the potential for an even more
efficient design. The current state of the art impellers are very unique in that they are not
milled like usual high-performance impellers. Some of the notable features of these
impellers can be shown in Figure 16 and are outlined as follows:

0 The blades of these wheels are flat and straight, not aerodynamically curved.

0 Same is true for the hub contour that even shows cut backs between the blades.

0 Technology does not allow for an outer shroud introducing a tip clearance with
leakage flow.

0 These state-of-the-art compressors work with stationary guide vanes (IGV) and
guide elements in the diffuser that both limit the operating range and introduce

friction surfaces that do not transmit compression work.

25

The features listed above have limited the achieved compression efficiency to
about 60 to 70 percent. The wheel assemblies are built from multiple parts in a laborious
process [18], [19], [20]. The design is a major reason for the high cost that does not scale
with size, making only larger units greater than 100 ton marketable, and this often only in

protected markets.

   

mechanically

critical region

Figure 16. Illustration of current state of the art compressor wheels

Both the IDE and ILK compressors work with a direct electric motor drive.
However, two different approaches have been followed. IDE has placed the motor
outside of the unit under atmospheric conditions as shown in Figure 17. This addresses
the challenge of sealing the shaft. ILK has placed the electric motor drive inside the unit
under water vapor atmosphere and vacuum, having resolved the greater challenges of

electric insulation, heat removal, and bearings.

26

 

Figure 17. Current state of the art units at IDE, Israel

There has also been the concept of using multistage axial compressors in the
HydroFrioTM chiller introduced by INTEGRAL [13], [21]. These compressors have been
derived from gas turbine compressors and apparently resulted in too high costs for the

HVAC&R market and have not been commercialized.

27

CHAPTER 7 : ALTERNATIVE COMPRESSOR CONFIGURATION

Axial Flow Compression

Turbo compressors are classified as either axial-flow or radial (centrifugal flow)
compressors. Traditionally, the centrifugal compressor has been the more rugged and
lower cost-type, while the axial-flow compressor has offered better efficiency. These
differences, however, have become much less Significant in recent years due to advances
in technology, particularly with regard to efficiency [22]. Flow streamlines through
axial impellers have a radius that is constant or nearly constant, while in centrifugal
impellers, the streamlines undergo a substantial increase in radius. For this reason,
centrifugal compressors are able to achieve a higher pressure ratio in a single stage. This
can be described mathematically with Euler’s equation for turbomachinery shown in

equation (5).

e = uzcuz — ulcuz . (5)

Here 6 is the specific (per unit mass) shaft work done on the fluid, uz and u. are
the tangential velocity of the impeller at outlet and inlet respectively, and cu; and out are
the tangential component of the fluid velocity at outlet and inlet. From this equation, it is
obvious that a change in radius leads to a change in tangential velocities and in turn, will
lead to a larger work transmission.

Although centrifugal compressors are traditionally capable of achieving higher
pressure ratios in a single stage, axial compressors can achieve significantly greater mass

flow rate per unit frontal area [22]. Also Since the radius doesn’t increase substantially,

28

axial impellers are much smaller in diameter than centrifugal compressors. An example
of this size difference is Shown in Figure 18. Here it is Shown that the diameter of an
axial compressor is one fourth the diameter of a centrifugal compressor with stationary
guide vanes. Since area increases with the square of diameter, centrifugal compressors
can take up to 16 times the area than axial compressors. Current state of the art industrial
water vapor chillers in use in Europe employ centrifugal compressors with typical
diameters of around 1 m. By comparison a chiller using an axial flow compressor
yielding a similar cooling capacity would have a diameter of around 0.5 m. In a general
sense, unit costs reduce with size and become more applicable in cases with tight space

constraints.

 

 

.1

 
   
    

Counter
rotating
Axial

 

Radial with
guide vanes
(blue)

 

 

 

 

4x diameter of axial with same capacity

 

C1 1

Figure 18. Diameter comparison between axial and centrifugal impellers

 

 

 

 

29

Multi-Stage Compression

To achieve the necessary pressure ratios, efficiencies, and Mach number
constraints, water vapor must be compressed in multiple stages [23]. In simulations,
when the desired pressure ratios are put in place with a single stage compressor, the
results are unreasonable. For example, the tip speeds are much too large for most modem
equipment. Velocities over 500 m/s are not achievable by most of the widely available
materials. The forces placed on the compressor itself, at this velocity, would cause high
stresses, this causes the components within the compressor to fatigue quickly or yield.
Several configurations are possible to achieve the needed work transmission. These may
or may not include stationary guide vanes. An illustration of some possible
configurations can be shown in Figure 19. When using water vapor, the compressor exit
temperatures tend to be higher than more commonly used refrigerants, and with multiple
stages, this temperature change becomes amplified.

Multi-stage compression also allows for inter-cooling. Inter-cooling between
stages reduces the required work input to the compressor [3], [23]. This, in turn, increases
the overall efficiency of the compressor. It also allows for the temperature of the water
vapor to be decreased in between stages; decreasing the overall compression temperature
changes. Flash inter-cooling is one of the ways to effectively achieve compression
temperature adjustment. With this type of cooling, liquid water is injected into the fluid
exiting the compressor. The water is to then to be evaporated by the superheated vapor,
bringing the fluid to the saturation temperature. The negative effects are that the mass

flow increases between each of the stages [6].

30

 

1 x work transmission 2 x work transmission 3 x work transmission

 

 

 

i>

(b)

i

(c)

\\\\\\\\\ «-
1 gm ~

 

 

 

 

 

///////// i -'

 

 

/////////. -
my -

//////// -

 

 

 

 

 

 

 

 

 

(a)

 

 

 

 

 

Figure 19. Compressor stage comparison

As explained in Euler’s Equation (5), the work transfer comes from the change in
the tangential component of the fluid velocity. Because of this, axial compressors often
employ the use of stationary guide vanes to add or remove a pre-swirl from the rotating
impellers. An example of this configuration can be shown in Figure 19(a). Here the
guide vanes in blue introduce a pre-swirl before the impeller, and then remove the swirl
after the impeller. There are several reasons for this, an important one is symmetry. If
the dynamics of the fluid remain the same throughout each stage of compression, the
hardware for each of the compressor stages can be designed the same. The impellers
being of the same material, rotating in the same direction, and at the same speed also
allows all them to be fixed to the same shaft. This in turn increases the manufacturability
of the entire compressor.

Regrettably, with the ease of manufacturability also come difficulties in size
constraints. The additional guide vanes between stages require the overall compressor to
be much larger to accommodate the additional hardware. However by counter rotating
the impellers, these guide vanes can be eliminated. Eliminating these guide vanes makes
the compressor smaller and more adaptable to various applications. A direct correlation
can also be made between a decrease in price and size reduction. In addition to reduced

prices, the hydraulic losses caused by the guide vanes are also eliminated. Removing the

31

 

need for guide vanes is done by counter-rotating the impellers. This process utilizes the
swirl after an impeller stage to gain additional work transmission in the next stage. In
equation (5) the work transmission comes from the difference of the tangential velocities
of the fluid at inlet and outlet. 1f the velocities at inlet and outlet are in opposite
directions, twice as much work can be transferred to the fluid in a single stage, and thus
more compression of the vapor. This is done by rotating the impeller in the opposite
direction of the preceding impeller. A schematic example of counter rotation can be
shown in (b) and (0). Here one can observe the additional compression for the same size
compared with the configuration in (a). This is also known as an increase in power

density (work transferred per unit volume).

32

CHAPTER 8 : NOVEL DESIGN FOR MANUFACTURING

Because of the low density of gases, the majority of the forces seen by the
impeller are not from the gas passing through the blades, but forces in its radial direction
due to its own inherent mass rotating at high speeds. Because of this, the impeller must
be of lightweight and strong material. The impeller being lightweight reduces safety
issues that arise from using heavy materials as well as reduces the forces inflicted on the
bearings. Lightweight materials also reduce the need for extensive balancing. As stated
in Chapter 6, conventional water vapor compressors, have impellers made from titanium
or formed composite Sheets. Although strong and lightweight, they are made component
by component by hand and are time consuming and extremely expensive to manufacture
or prototype. To address this issue, new technology being investigated at Michigan State

University involves the use of a wound or woven composite impeller.

Composite Materials

In short, a composite material is essentially any material that is made up of high
strength fibers arranged in a matrix. The benefits of composite materials are widely
known and vast. Most notably are their high strength and low weight which, as noted
above, make composite materials particularly attractive for impellers for vapor
compression. Composite materials are also anisotropic which allows for the fibers to be
arranged in the direction of the highest force. Interestingly, composite materials can be
both man-made substances as well and natural materials. Examples of naturally
occurring composite materials are wood or bone, whereas examples of man-made

composites can be concrete or fiberglass moldings for boats.

33

Filament Winding

Filament winding techniques has been well investigated over the last thirty years
[24], [25]. In a simple description of the process, reinforcement fiber tows in bundle
form are fed trough a wet bath wherein they are impregnated with resin, and are then
uniformly and regularly wound onto a rotating mandrel. Once the desired architecture
and lay up thickness are achieved, the wheel can be cured and the mandrel removed.
This technology is typically used for the fabrication of parts that are axial-symmetric,
such as pipes, containers, pressure vessels, rocket motor cases, and other tubular
structures [24], [25]. Depending on the desired properties of the product, winding
patterns such as hoop, helical, and polar can be developed. Various curing system such
as drying in the oven, through hot oil, heat by lamps, steam, autoclave and microwave are
possible for different applications.

The principal advantages of filament winding are its low labor cost and its
reproducibility. This is due to robotic motion via computer coded design. Thus, one can
apply this filament winding technology to manufacture composite impellers at a low
labor cost. In addition, various weaving patterns can also be achieved. These patterns
will be discussed in the following section. Another advantage of filament winding is the
ability to interweave motor components. This could be induction wires or magnetic
materials dispersed within the matrix. Motor schemes like this will also be discussed in

greater detail later.

Fiber Material

In recent years, the fiber material that has perhaps been given the most attention is

carbon fiber. This is undoubtedly because of its low density (about 2.268 g/cm3) and

34

high strength (up to 1000 GPa for single crystal graphite [25]. Certainly properties like
these warrant its widespread use in applications such as aerospace, sporting equipment,
and auto racing. However, several other fibers exist as well, each with their own intrinsic
properties associated with them. Some examples include glass, boron, and organic fibers.
The broad vision of this research undoubtedly includes the use of carbon fiber for the
characteristics mentioned above. Initial experience has lead to the experimentation with
the aramid fiber, Kevlar®.

Kevlar®, based on the chemical poly(p-phenylene terephthalamide) was created
in 1965 by DuPontTM originally for the replacement in steel in radial tires, but has been
made most famous by its use in bullet-proof fabrics. Although its modulus is less than
that of carbon fiber, Kevlar® has a lower density and has proven to be less brittle, which
make it useful in a variety of applications. In the case of winding turbo impellers, these
properties are favorable since impellers need to be light-weight, and also since patterns
sometimes have sharp turns where fibers could break or fray. In addition to these
properties, Kevlar® also has a small negative thermal coefficient of expansion that may
counter dynamic deflections with thermal contractions at high compressor temperatures.
Kevlar® also has good vibration damping characteristics. Composites of Kevlar®/epoxy
show about 5 times the vibration dissipation than glass/epoxy composites [25]. There are
several types of Kevlar® yarn that have a variety of properties and can be noted in Table
2.

For convenience and availability, dry, spun Kevlar® yarn or thread was obtained
for initial prototyping. However, tows made of clusters of entire fibers are also available.

The Kevlar® thread was purchased from Atlantic Thread and Supply CompanyTM out of

35

Baltimore, MD. Specifically, the thread that purchased and used for this research is

CRAQ-SPUN® Tex-35 Glazed which comes in black or natural colors.

sizes are available.

Table 2. PrOperties of Kevlar aramid fiber yarns [25]

However, other

 

 

 

 

 

 

 

 

 

Property K 29 K 49 K 68 K 119 K 129 K 149
Density (g cm-3) 1.44 1.45 1.44 1.44 1.45 1.47
Diameter (pm) 12 12 12 12 12 12
Tensile strength (GPa) 2.8 2.8 2.8 3.0 3.4 2.4
Tensile strain to fracture (%) 3-5'4-0 2-8 3-0 4-4 3.3 1.5-1.9
Tensile modulus (GPa) 65 125 101 55 100 I47
Moisture regain (%) at 25°C, 65% RH 6 4-3 4-3 ' - 1.5

-4.0 -4.9 - - - -

Coefficient of expansion (10" K")

 

 

 

 

 

 

 

Matrix Material

While most of the attention in a composite material is given to the fiber material,

the matrix material is also of importance. While the fiber material is responsible for

“carrying the load” associated with composite applications, the matrix material is

responsible for many material tasks in its own right. These include positioning fibers as

well protecting the fibers from transverse loads buckling, and unfavorable environments.

Matrix materials can be made of ceramic and metal material, however polymer

matrix materials are the focus of this research. Although polymers are structurally more

complex than metal or ceramic material, they are inexpensive and easily processed,

36

 

which make them ideal for woven composite impellers. More specifically, the matrix
material used in this research is a two component resin and hardener epoxy. This gives to
a variety of curing methods including self—hardening, heat-curing, as well as UV-curing.

There are three ways in which matrix material can be applied. The first is a post-
process application where large amounts of fibers are initially oriented, and then the
matrix material is applied after. This may be favorable for simple composite geometries
in that it reduces waste and messes. However, in a post-process matrix application, it
may be difficult to distribute matrix material evenly throughout as well as control fiber to
matrix ratios. Examples of post-process matrix application can be shown in the impellers
in Figure 21 and Figure 22.

Another convenient method of applying matrix material is through the use of pre-
impregnated (pre-preg) tows. Pre-preg tows are a bundle of fibers that already contain
the epoxy material which is solidified after controlled curing. The hand wound prototype
shown in Figure 50 and Figure 52 were made of this pre-preg material. This method is
by far the most convenient in terms of eliminating messy resin baths, however it is only
available in certain fiber tows. This limits its applicability where specialized fibers and
various tow thicknesses are desired. Using a pre-preg tow also limits the custom control
of the matrix distribution as well because quantities are pre-deterrnined.

The third and most common method of matrix application is the resin bath or wet
winding. This method is practiced by running fiber tows first through a resin bath and
then on the mandrel or winding surface. Photos of this method are Shown in Figure 20
and Figure 49. Although cumbersome and messy, this matrix application allows for the

outright control of distribution on the fiber and can be applied to any geometry.

37

 

Figure 20. Fiber tow being run through a resin bath for wet winding

38

 

Figure 21. Automatically woven impeller prototype pattern S-B with post-process applied
resin

 

Figure 22. Automatically woven impeller prototype pattern 8-C with post-process applied
resin

39

Winding Patterns

AS mentioned above, several winding shapes can be created. These different
shapes have various advantages and characteristics unique to their geometry. Table 3
shows the classification and various characteristics of several winding patterns. Patterns
are first named by the number of blades, or in the case of the winding mandrel, the
number of slots. The patterns are then assigned letters corresponding to the relative size
of the inner diameter. Patterns of a given number of blades are assigned with an “A” for
the largest inner diameter and “B” for the next size smaller inner diameter, etc. Some
patterns have the unique characteristic of having a zero inner diameter. Examples of

these patterns include 6-B, 8-C, and lO-D.

40

Table 3. Abbreviated table of winding patterns (full table in Appendix C)

 

Number

of Points Pattern
5 1. skip 1
5 1. skip 2
5 2. skip 1
5 2. skip 2
5 3. skip 1
5 3. skip 2
5 4. skip 1
5 4. skip 2

6 1. skip 1
6 1. skip 2
6 1. skip 3
6 2. skip 1
6 2. skip 2
6 2. skip 3
6 3. skip1
6 3. skip 2
6 3. skip 3
6 4. skip 1
6 4. skip 2
6 4. skip 3
6 5. skip 1

6 5. skip 2
6 5. skip 3

71.sk1p1
71.skip2
7 1. skip3
7 1, skip4
7 2. skip1
7 2. skip2
7 2. skip3
7 2. skip4
7 3. skip1

 

Not
Possible Shape N
5—A

5-A
5-A

5-A
5-A
5-A

6-B

7-A

dildo

0.5

0 Stack-up in the center

0.5

0.5

O Stack-up in the center

0.5

0.62349
0.22252
0.22252
0.62349
0.62349
0.22252
0.22252

0
0.62349

Notes

Sharp turns

Sharp turns

Sharp turns

Sharp turns

Sharp tums

Solid outside for 3 separate
slxths of the circuit, normal
0 inside. stack-up in the center

Sharp turns

Shapes:

5—A

 

7-A

 

The second column of Table 3 called “pattern” refers to the repeated winding fiber path.

Patterns are in the form, “# skip #”. Starting at an arbitrary slot in a winding mandrel, the

first number refers to how many slots the fiber must wrap around the outside before

entering the slot. For example, if the pattern is “1 skip #”, the fiber must start at an

arbitrary Slot, wrap around the mandrel, and enter the next slot. This is illustrated in

Figure 23.

41

/.\

\%

Figure 23. Schematic illustration of fiber path in pattem"1 skip #"

The second number is the pattern refers to how many slots the fiber must skip
before exiting through the next slot. For example, if the pattern is “I skip 2”, the fiber
must enter the first Slot after wrapping around the outside of the mandrel. Then, the fiber
must skip 2 slots in the middle of the mandrel before exiting out the next slot to repeat the

pattern again. This is illustrated in Figure 24.

42

 

Figure 24. Schematic illustration of fiber path in pattern "1 skip 2"

The ninth column labeled “di/do” is the ratio of the inner diameter to the outer
diameter. This is elegantly calculated by Allen Eyler for each shape by the general

equation

 

EL =lj(1 — cosEn—N)2 + (Sinai/Y)2 (6)

where S is the number of slots in the winding mandrel or blades, and N is a pattern factor
used in the calculation. This pattern factor starts with zero for patterns that have no inner
diameter like 6-B or 8-C. When the pattern has an even number of slots (i.e. when S is an
even number) this factor is a whole integer (0, 1, 2, 3,. . .). When S is an odd number, N is

assigned numbers that start at 0.5 and increase by l (i.e. 0.5, 1.5, 2.5,...). N factors start

43

at the smallest inner diameter and increase with increasing inner diameter for patterns
with the same number of slots. For example, for patterns with 8 slots (i.e. 8-A, 8-B, and
8-C) the N factor is 0 for 8—C because it has a zero inner diameter. The N factor is 1 for
8-B because it has the next smallest inner diameter, and 2 for 8-A because it has the
largest inner diameter for patterns with 8 slots. An example of the N factors for odd
slotted patterns can be the 5:7 patterns (i.e. 7-A and 7-B). Here the N factor would be
0.5 for 7-‘B because it has the smallest inner diameter of the 7 Slotted patterns. The N
factor would be 1.5 for 7-A because it has the next largest inner diameter. These are the
only patterns for 5:7. It can be Shown that patterns with zero inner diameters are only
possible with patterns that have an even number of slots. Therefore, these are the only
patterns that may have an N factor of zero. This di/do ratio is very useful for selecting
specific patterns for size constrained applications like hub diameter. It is also useful
when selecting various bearing and shaft/axel configurations.

In addition to different Size characteristics, the various winding patterns have
other characteristics that are worth mentioning. These include fiber paths with sharp
turns that may be less favorable for brittle fiber material. Also, some patterns have areas
where fibers “stack-up” relative to other areas of the impeller because of frequent
overlapping. Characteristics like these are important to identify because they can be

advantageous or undesired depending on application.

Advatages of Winding Design

A schematic illustration of a woven impeller can be shown in Figure 25. Here a
star shaped configuration is depicted, however virtually any shape can be made. These

may include curved blades that increase efficiency. There are several advantages to this

44

design. One is that the woven composite impeller can be mass-produced and rapidly
prototyped using a readily available multi-axis winding machine greatly reducing costs.
An example such a machine is the McClean AndersonTM Super Homet® SX9000 shown

in Figure 26.

 

 

 

 

 

Figure 25. Illustrative model of woven impeller pattern 8-B

By using commercially available software, a design code can be electronically
written and carried out by the fiilly automated winding machine. , Utilizing this
technology also reduces costs by eliminating the need for additional assembly of
components. Various geometries can by constructed by the winding machine including
radial-flow and mixed-flow impellers. In addition, the computer design also allows for
an outer shroud. An outer shroud adds additional sealing and strength in the tangential
direction, thus allowing for an integrated motor at the outer diameter. The star shape
depicted in Figure 25 also employs its geometry to segregate the turbulent, separated flow

around the hub or inner diameter from the uniform flow along the blades. This flow

45

segregation allows for better efficiency. Structural advantages are discussed in greater

detail later in Chapter 9.

 

 

 

 

 

Figure 26. Commercially available Super Hornet winder from www mm

Driving Scheme

In addition to the geometrical freedoms this impeller offers, the woven or wound
impeller also allows for various elegant methods of driving the impeller. Conventional
turbo-compressor design employs a separate motor to drive an impeller via a shaft.
Although the woven impeller technology can most certainly be used for this driving
design, the scope of this thesis includes a more elegant and compact design involving the
integration of motor components. By doing this, the compressor impeller in-itself
becomes an integral component in the electric motor. Electric motors involve two major
components: the stator and rotor. By integrating electrical or magnetizable components
in the impeller, the woven impeller can essentially become the rotor portion of the

electric motor. The two motor schemes explored in this research are the brushless

46

permanent magnet motor approach as well as configuration known as an induction

machine or induction motor.

Brushless Permanent Magnet Motor

For simplicity, a permanent magnet motor can be introduced as two
electromagnets on a stator surrounding a permanent magnet in its center as the rotor. A
schematic illustration of this is shown in Figure 27. Each pair of stator windings make up
a motor phase, thus Figure 27 is a Single phase motor. A current running through the
windings of the stator induces an electromagnetic field that interacts with the permanent
magnet rotor. When the Similar poles of the permanent magnet rotor, and the
electromagnetic stator repel each other, the rotor moves to the opposite poles of the
stator. Reversing the current through the stator reverses the poles, and the rotor is
repelled again. Repeating this creates a motor torque.

i

._k__%\\

 

Figure 27. Schematic, illustrative example of a single-phase permanent magnet motor

47

Although this simple single-phase example is useful for describing the general
principle of the permanent magnet motor, it is more common to have several pairs of
stator windings. An example of a 3-phase motor is shown in Figure 28. AS the number
of phases and poles increases, the torque characteristic on the rotor changes as well. An
important distinction to make is the mechanical position of the rotor, and the electrical
position. The mechanical position 0”, of a rotor is defined by the number of poles facing

the air gap M... and the electrical position 6..

a, = —— 9m (7)

So, if a motor with a 4-magnet pole (2-magnet pairs) rotor does one entire
electrical rotation, it has only done 180 degrees of mechanical rotation. The electrical
frequencyfe and mechanical frequencyfl" are related to each other through the number of

poles.

fe = Nme (8)

Thus, for a certain mechanical rotational speed, the electrical frequency required
increases with the number of magnet poles. This is where the motor design is often
constrained by the limitations of the input device (frequency converter) that is used to
drive the motor. Therefore, fewer magnet poles are used for the design of high-speed
motors to keep the electrical frequency down. The drawback for using fewer magnet
poles, however, is that the torque production is much lower. The choice of number of

magnet poles also ties into the efficiency and overall losses of the design. Since driving

48

the compressor impeller requires high speed and low torque, a motor design with fewer

phases is desirable.

 

Figure 28. Schematic, illustrative example of a three-phase permanent magnet motor

This type of electric motor concept can be integrated into the woven composite
impeller by introducing magnetic material to the wheel. This may be done by attaching
magnets to the outer shroud of the impeller or by magnetizing the impeller material itself.
One interesting method of magnetizing the woven impeller is by introducing a material
into the resin during winding that is either magnetic itself or has the ability to be
magnetized. This includes the use of various iron compounds and mixing them within

the resin.

Induction Motor

Another option for an integrated motor is the use of an induction machine. This
motor configuration is an attractive option for initial experimentation because of its ease

of manufacture. In an induction machine, AC power is used to “induce” electromagnetic

49

fields that are out of phase with one another and essentially “magnetize” a conductive
rotor. Because of this, the rotor need not have permanent magnets, but only be made of
conductive material. Since the rotor only needs to be constructed of conductive material,
the integrated woven impeller does not need to be magnetized, nor have magnets placed
within the impeller. This is advantageous because integrating magnets and magnetizing
the impeller can be a difficult task, especially in the initial demonstrations of feasibility.

The easiest design for an integrated induction machine is simply to wind the
impeller on a mandrel that is made of cOnductive material and not remove it. The
impeller and mandrel can then be placed in an electric stator designed for induction
machines. A picture of this configuration can be shown in both Figure 67 and Figure 68.
Different bearing configurations are also possible and will later be discussed in Chapter
10.

Although the ease of manufacture makes this integrated motor option very
attractive, it is not without its own inherent challenges. One of these challenges is that it
requires a 3-phase power supply, which although available in many laboratories and
industrial settings, is not readily available in a residential setting. Induction machines
also require small air-gaps in between stator and rotor which makes for tight machining

tolerances.

50

CHAPTER 9 : COMPUTATIONAL ANALYSIS OF DESIGN

Pattern Modeling

To generate accurate computer models of the winding patterns classified in
Chapter 8, a computer code was written in C++ by lab member, Allen Eyler [26]. This
code could generate several outputs automatically that could be directly used in
applications such as: 3-D modeling software like UnigraphicsTM, matrix simulation,
winding mandrel design, as well as the accommodation for material variation. This code
was also employed for physical rapid prototype generation as well as computational

structural analysis and preliminary simulations in computational fluid dynamics.

Interfacing for 3-D Modeling

In addition to a series of points Eyler’s code generated, it also outputted GRIP
(Graphics Interactive Programming) commands. GRIP commands are an interface
language supplied by UnigraphicsTM to generate 3-D models. They specify the arcs,
lines, surfaces, splines, and support mandrel of any winding pattern. Manual inputs and
code execution took less than 2 minutes to complete the 3-D models like the one shown

in Figure 29. Here we see a 3-D model of pattern 8—2-2 including its support mandrel.

Radius of Curvature around Slots

The program is sophisticated in that it not only models the winding pattern
shapes, but also models the actual fiber paths and behaviors as they are constructed. The
computer code calculates and specifies the lines, arcs, and even the radius of curvature as
the fiber wraps around and through the mandrel slots. Figure 30 illustrates this specified

radius of curvature.

51

 

Figure 29. Computer model of woven impeller including its support mandrel

   

\

Figure 30. Magnified detail of a wound impeller model, illustrating the radius of curvature
of the fiber

52

Fiber Collisions

In addition to the radius of curvature, the model also elegantly takes into account
fiber collisions. In reality, while winding 3 fiber across the middle of the mandrel, a fiber
would come up and over the previously placed fiber below it. However, computer
modeling avoids these collisions by defining separate line segments that fibers follow to
overlap nicely. An illustration of these defined lines segments and fiber overlapping can
be shown in Figure 31. Although these fiber collisions are not much of an effect on some
patterns, other patterns such as 6-2-2-B shown in Figure 32 have fibers that all cross the
center of the impeller. This makes for a large accumulation of fibers stacking up in the

middle that the model needs to account for.

 

 

 

Figure 31. Detail showing collision avoidance of the fiber with segments that have
previously crossed the cylinder

 

Figure 32. An Impeller Model in which all fiber collisions occur at the center of the cylinder

Modeling the Resin

When considering a composite material, modeling the fiber material is only half
the task. The matrix or resin must also be taken into account for accurate modeling. This
is done by connecting points on the fibers and creating a surface. This can be tedious
since there can be tens of thousands of points to consider. For this, the original code
must generate defined surfaces in the GRIP outputs. This is done by locating endpoints
of line segments and matching them to their specified layer. Once enough points and
layers are determined, a specified surface can be defined. Repeating this for the entire
impeller defines all surfaces on the woven wheel. These surfaces serve to model the resin

applied during the actual wheel manufacturing.

54

Structural Analysis

Recall from Chapter 4 that to achieve the necessary pressure ratios to employ the
use of water as a refrigerant, very high tip speeds must be achieved. Also recall that the
majority of the forces on the impeller are from its own inherent mass rotating at high
speeds. Because of this, computational simulations must be analyzed to assure the woven
impellers withstand these high speeds. These simulations were completed and analyzed
by lab member, Anirban Lahiri [27] using FEM (Finite Element Method) along with

ANSYSTM software.

Setting up the Simulation

The material properties used in the simulation were from the Kevlar® 49
properties tabulated in Table 2. Rotational speeds were ramped up to 120,000 RPM to
simulate motor startup and operation. To set-up the meshing of the geometry, Eyler’s
code [26] was again employed to be used in a MATLAB® code to then generate an
ANSYSTM log file. The geometry was then specified and a mesh could be applied. The
meshing used in this analysis was Quad 8 node 82 element. This is a quadratic element
and provides a better accuracy than its corresponding linear element. It is also an
appropriate element for a plane stress analysis. The geometry can be map meshed too,
however a free meshing provides pretty good elements with almost zero distorted
elements. Stresses and displacements were analyzed while varying parameters such as

skip number and fiber thickness.

Simulation Results
The displacements and stresses were determined and plotted. A small sample of
these plots is shown in Figure 33 through Figure 40 below. The following figures

55

determine a general nature or trend of various winding patterns. It is very important to
note that the displacements conveyed in the figures were greatly exaggerated for
detection purposes within each individual pattern, and that the actual displacements
are much smaller in reality. It should also be noted that the scales of each plot are greatly
different as well. The plots are to observe the relative displacements of locations within

each pattern themselves.

 

NDDAL SOLUTION AN
STEP-I OCT 13 2008

son -1 16:57:16

DHX -.IODE-03
SI‘IX =.100E-03

 

0 .22313-134 .44oE—04 .668E-04 .891E-04
.lIIE-04 .3345-04 .557E-04 .7305-04 .1005-03

 

 

 

Figure 33. Pattern 7A - Displacement

56

 

 

AN

1
NODAL SOLUTI OK

 

 

 

STEP-J. OCT 13 2008
SUB _1 16:56:01
TIM-I

SEOV (AVG)

DEX -.100E-O3

5H1! -.312£+O7

SEX -.116E+10

. 3121-24-07 .Zsfizws . SIIBE-HJS .775E+09 .103E+IO
. 132E+09 . 389121-09 . 6471:1439 . 904E+O§ . 116E+10

 

Figure 34. Pattern 7A - Stress

 

AN

NODAL SOLUTI UII

my“ OCT 13 2008
SUB -1 15:59:23
Trim-1

usun (AVG)

RSYS-U

DHX - . 699E-O4
SEX -.699E-04

 

0 .1555-04 .3115-04 .466E-04 .6222-04
.7772-05 .233r-o4 .BBBE—O4 .5442-04 .699E-04

   

 

 

 

Figure 35. Pattern 7B - Displacement

57

 

 

IIDDAL SO LUTI OI

 

SHX I.563E+09

 

AN

OCT 13 2008
16:59:40

   

 

816537

.633E+06 ‘

. 376E+09
. 313E+09

126E+09 .251E+O‘3

. 180124-09 . 43851-09

. SOLE-+09
. 563E+O9

 

Figure 36. Pattern 7B - Stress

 

110th 50 LUTI Oll

STEP-1
SUB =1

(AVG)

SEX I. 102E—03

 

 

OCT 13 2008
17:00:24
.‘Kft. ‘ . WE
2272-04 .455E—04 .682E-04 3091-2-04
.3411'2—04 .568E—04 .7952-04 .102E-03

AN

 

Figure 37. Pattern 8A - Displacement

58

 

 

 

 

I
IIODAL SOLUTION

STEP-1
SUB =1
TIE-1
SEOV (AVG)
DHX -.102E-O3
SHIT -.109E+07
SHX -.110E+10

 

. 1091-307

 

 

 

OCT 13 2008
17:00:44
. 5“. . ..f
24SE+09 .4:38E+09 .732E+09 .97SE+09
. 366E+09 .610!+09 .853E+09 . 110E+10

 

Figure 38. Pattern 8A - Stress

 

IIODAI. SOLUTI 0N

USU'H (AVG)

DHX -.636E-04
SHX -.636E-O4

 

AN

0LT 13 2008
17:04: 58

 

 

d
.141E-04 .ZbJE-O‘l .42-IE-04 .ScsE-04

 

.707E-05 .212E-04 .353E-O4 .495E-04 .6365-04

 

Figure 39. Pattern 88 - Displacement

59

 

1
HOD“. 50 LUTI ON AN

STEP-1 OCT 13 2008
SUB -1 17:04: 39
TIRE-1

SEOV (AVG)
DEX -.636E-04
SHN -.173E+07
SEX -.310E+09

 

;. ’ mus ‘
.173E+07 .7UZE+08 .13954-09 .207E+09 .27SE+09
. 36024-08 . 104E+09 . 173E+09 . 24lE+09 . 310E+09

 

 

 

Figure 40. Pattern 88 - Stress

Analysis of Simulation Results

An interesting observation from the displacement diagrams is that for the A types
of designs, the maximum displacements occur in the inner fibers while they occur in the
outer fibers for every other design styles like B,C, D, E, etc. Also, another observation is
that as the inner circle becomes smaller (as the number of skip increases), the stresses in
the inner fiber decreases. This can be explained from the simulations in the varying
thickness outer shroud simulations. So as the fibers pass through the centre, the stresses
become less.

Efiect ofSkip
Recall from Chapter 8 the winding patterns included a skip number that denotes

the number of slots the fiber “skips” before it exits the mandrel. One can make the

60

generalization that the higher the skip number, the smaller the inner hub area of the
impeller is. The skips were varied for the different impellers and the stresses were
studied. Simulations were done until the number of points on the circumference of the
impeller was 24. The results plotted in Figure 41 through Figure 43 below indicated that
the stresses reduced as the number of skips increased. For convenience, the skip numbers
are divided into low (6-9), medium (10-14) and high (15 and above). The stresses
decrease monotonically as the number of skips increase. It can also be shown that the

stresses flatten out and do not increase or decrease very profoundly after skip 2.

 

No. of points on circumference = 6 - 9
1600 1‘
1400 -+f .
l A
1200 ‘1 i,__.
A 1 ' o 6.
g 1000 i x 1. 75
E i I ‘
tn = X 9.
£3 600 J . t;
m E x
400 4; , - X
200
0 x 7 __ f 7 i7s___ a 77s _ __fi
0 1 2 3 4
Skip (n)

 

 

 

Figure 41. Stress vs. skip for patterns with 6-9 slots

61

 

 

 

No. of points on circumference =10 - 14
1200 1
' A
I
1000 1
i .
7; 8°01 1’ ao
. 0
CL 1
a E l i ' In 11:
3 6001 I iA121
g I13I
m 400 4, O . 14i
1 Q ‘ 11--.]
200 1
!
o L_ s- H _ _ s___ _ -fiw _,
0 1 2 3 4 5 6
Skips(n)

 

 

 

Figure 42. Stress vs. skip for patterns with 10-14 slots

For the 10 through [4 patterns, the maximum skips are 5. Again, the stresses
show the same behavior as in the 6 through 9 patterns. The last case shown here is the
case where the numbers of points on the impeller are from 15 to 18. Almost all of the
points show similar behavior as the cases shown in Figure 41 and Figure 42. However,
as the skip is increased, or patterns like 18 and 19, the stress does not monotonically
increase or decrease with skip. This may be due to the fibers getting close to each other

and having non linear interaction effects.

62

 

No of points on circumference 15 - 18
1200 1
1
1000 I Q
I . .-—.—7.
A 800 1 - - J. 15
g l I ‘ l l 16
E I '
z 600 _J . O I . J l 17;
in I . ‘ I 18
g J t - g 1
‘0 400 J ° 9, “ n [3.19;
J A % I
: I
200 J
0 ‘T "' ‘fi '_ F —-i ‘ ‘ —— ' " T— —" Y'—— ‘T— ‘ "T—“Wfi' ‘——1
o 1 2 3 4 5 6 7 8 9
Skip (n)

 

 

 

Figure 43. Stress vs. skip for patterns with 15-18 slots

Eflect of Fiber Thickness

In plotting the effect of varying fiber thickness, the results were again
documented with reference to the number of slots in the winding patterns. Fiber
thicknesses of both the blades and the outer shroud of the impeller structure were selected
as l and 2mm. All the thickness simulations are not shown here because of space issues.

However, the general behaviors of the curves are the same.

63

 

 

1600 J

1400

l

1200

___L__

1000 1
800 1
600 <

400 ~

Maximum Von Mises Stress (MPa)

200 1

 

 

 

Stress vs skip

 

Fflhckness = 1mm)J
11201199535299.11

Skip(n)

 

Figure 44. Effect of fiber thickness for 7 slot pattern

 

 

 

 

 

Maximum Von Mises Stress (Mpa)
c»
o
o

 

 

 

 

,
i

do 9(thickness = 1mm)
1 - 9(thickness = 2mm)

 

 

 

 

 

 

J 400 J ' - °
1 a
200 ‘
1 1
J 0 J I r 1 "‘ F T
1 0 1 2 3 4 5
J Skip

 

 

 

 

 

 

Figure 45. Effect of fiber thickness for 9 slot pattern

64

FAA _.__ ___. ___________________

Stress vs skip
J o 16(thickness = 1m):
800 1 :1§<£hi<i093§:2m)

 

‘w

700 1
600 4'

.-
I

500 ~ 0

 

400 '1 0 °
300 1
200 r
100 1

Maximum Von Mises Stress (MP3)

 

 

 

 

 

O 1 2 3 4 5 6 7 8
Skip(n) j

 

 

 

Figure 46. Effect of fiber thickness for 16 slot pattern

Figure 44, Figure 45, and Figure 46 show that the stresses decrease as the
thickness of the fibers are increased. In order to explain this phenomenon, two cases
have to be considered. The first one is based on the variation of stresses in an annulus by
changing the inner diameter. As can be shown from the Figure 32, the tangential stress is
the predominant stress component when the annulus is subjected to a rotation. The total
stresses (vector addition of the radial and tangential stress) follows more or less along the
lines of the tangential stress. It can be shown that as the annulus is made thicker, the total
stresses decrease in the annulus. The decrease of stresses in the different patterns can be
attributed to this. Also,‘there is a pull on the outer shroud by the blades, which is more as
the thickness is increased, leading to less stresses in localized areas.

Thus, in conclusion, thickness plays an important role in the determination of

stresses in the fibers. Clearly, a low thickness is desirable for lower stresses, however,

65

during the manufacturing process, the actual thickness of the fibers manufactured are
pretty low (around 1mm). Even when coated with resin, the thickness does not increase a
whole lot. An alternative to this is the stacking of fibers together in the force direction to
reduce the stresses. Also, an increased fiber concentration at the areas where the stresses

are highest may help in reducing the stresses.

66

CHAPTER 10 : REALIZATION OF DESIGN

Rapid Prototype

The first physical models of the woven impellers were created on a rapid
prototype machine. Using the Unigraphics”! file, the machine can create physical
prototypes by arranging thin layers of plastic or 3-D printing. An example of one of
these models is shown in Figure 47. Here a woven impeller with a central hub is shown

with its electrical stator.

 

Figure 47. Rapid prototype of woven impeller with electric stator

Hand Winding

The simplest construction of a woven impeller is one wound by hand. Although
the automated manufacturing of the impeller is one of the most important advantages of
this design, initial prototypes were constructed simply by winding Kevlar thread around

the mandrel. This hand winding can be shown in Figure 48, Figure 49, and Figure 50.

67

 

Figure 48. Dry hand-wound prototype to be used in integrated motor

The initial hand winding was done with dry fiber to obtain a prototype
demonstrating the pattern design. This dry winding is shown in Figure 48. In addition to
the pattern model, this prototype was also used in an integrated motor demonstration
discussed later. The wet winding method where dry fiber tows are guided through a resin
bath is shown in Figure 49 and its finished winding pattern is shown in Figure 51. The
resin for this prototype was cured at room temperature using a self-hardening resin.
Hand winding using a pre-preg tow is shown in Figure 50, as well as its finished result

before curing in Figure 52. This model was heat-cured in an oven and its mandrel

removed.

68

 

Figure 50. Hand winding using pre-preg fiber

69

 

Figure 52. Finished result of pre-preg hand winding before heat curing

70

Automated Winding

Perhaps the most novel part of the impeller design is its design for automated
manufacture. Although this filament winding method is an existing and proven
technology for other parts such as pressure vessels and golf shafts, it is to the knowledge
of the author that this method has not been employed for compressor impellers.
Moreover, for the sake of convenience as well as to save additional costs, the
commercially available winding machines like the one shown in Figure 26 was not used
to create initial mechanically wound impellers. Instead a CNC mill that was purchased
from CNC MastersTM was modified for the specific manufacture of compressor wheels.
A photo of this set-up can be shown in Figure 53 and Figure 54. The main reason for this
decision was that the CNC mill was approximately 10 times less expensive than
purchasing a winding machine. In addition to this, the mill could also be used to
manufacture mandrels as well as other mill duties that arise in the turbomachinery

laboratory at Michigan State.

71

 

Figure 53. Photo of CNC winding set-up

Modifying the CNC Mill

To modify the CNC mill for winding impellers, several additions had to be made.
The first was to simply make the machine portable throughout the lab. The CNC mill
came as a small table-top mill with an optional stand to be mounted to a shop floor.
Since the stand was designed to be attached to the floor, it was made of light-weight
material. To make the mill portable throughout the lab, caster-wheels were required to be
installed on the bottom of the stand. The lightweight stand beneath the heavy mill made
the machine top-heavy and prone to tipping. To alleviate this, a large, heavy slab of steal

was attached to the bottom of the stand and caster-wheels installed. The result was a

72

sturdy, versatile machine that could be easily be wheeled throughout the lab as shown in

Figure 54.

 

Figure 54. Portable CNC winding machine

The CNC stage itself needed no modification. An add-on tum-table axis was
purchased from CNC MastersTM and along with the x-y-Z stage and the spindle, the mill
provided a 5 axis CNC stage. A spool was installed to mounts on the side of the machine
along with a container for a resin bath, including guiding rollers for filament winding.

These modifications are shown in Figure 55.

73

 

Figure 55. Winding machine with fiber spool and resin bath added

Next, the fiber thread needed to be guided throughout the mandrel in various
winding patterns. This was done using a sewing machine type needle also referred to as
the leader. Although the basic idea of leader method remains the same, it has had several
improvements. The initial leader was made of small copper pipe. The pipe was secured
in the spindle’s chuck, a hole was cut into the side for the thread to enter, and the thread
was then lead out the bottom end of the pipe to be guided around the mandrel. Although
this method served as a good initial attempt, it lacked in accuracy because of the freedom
of the thread to move laterally within the copper pipe. The leader has evolved into a
more accurate tool by using a very thin tube, bending it to the angle of the mandrel slots

and securing it again to the spindle’s chuck. This tooling method is shown in Figure 56.

74

 

Figure 56. Photo of winding with leader tool

Challenge of Fiber Creep

One of the challenges that arouse from the initial attempts at automated winding
was the phenomenon that was named fiber creep. This occurred mainly during dry
winding attempts. When the fiber thread was guided around the outside of the mandrel
and the leader was raised up and over the side of the mandrel, the previously laid fiber
tended to “creep” up the side.

One cause of this fiber creep was the lack of friction between the outside of the
mandrel and the fiber thread. This was especially prominent during dry winding attempts
without the resin because the resin causes the thread to “stick” to the mandrel. The initial
method to address this challenge was by attaching a high-friction material to the outside

of the mandrel. This high friction material came in the form of sandpaper. As shown in

75

Figure 57, sandpaper strips were attached to the outside of the mandrel prior to dry
winding. Although this method worked to keep the fibers in place while winding, it did
negatively affect the geometry of the final impeller because it added additional thickness
in some areas of the mandrel. Also, when the post-process resin was applied, the

sandpaper was difficult to remove.

 

Figure 57. Sandpaper applied to outside of mandrel for added friction

In addition to applying a high-friction material to keep the fibers in place, another
solution was explored to re-position the fibers if they crept up. To do this, an electronic
actuator was placed on the outside of mandrel that would “poke” the fibers back into
position at a time-controlled rate. Although fiber creep reduces significantly while wet-

winding, it is still a phenomenon that needs additional observation.

76

 

Figure 58. Winding shown with actuator to re-position fibers after fiber creep

Another cause of fiber creep is the tension created by the leader itself. With
limited to no tension control, cases arise when the tension is so large that the leader
“pulls” the fiber up and away from its intended position on the outside of the mandrel.
Tension control is a major factor in creating any accurate filament wound structure [28].
To enhance fiber placement, a tension control method is being researched. This may
include force transducers, torque controlled motors on the fiber spool, or even friction

material in the rollers or spool.

The Patterns

Since several of the hand wound and rapid prototype models were constructed in
the 8B pattern, the initial automated windings were also done using this pattern and an
example of this can be shown in Figure 59. However, two patterns were wound that had
no inner diameter as described in Chapter 8. These were patterns 8C and 6B. To fix a

small shafi to the center of these patterns, a shat was positioned in the center of the

77

pattern while the winding was completed. The winding code had to take this into account
by leading the thread slightly around the small shaft. A photo of this winding can be
shown in Figure 60. Several prototypes were created on mandrels of 5 inch diameter, as

well as with mandrels with 2.375 inch diameter.

 

Figure 59. Automated, wet wound prototype of pattern 8-B

78

 

 

 

 

S tructu

dUHng

StrUCtUr
high Sp
Compre
must b,

velociti

 

Figure 60. Winding pattern 8C including a shaft

Structural Spin Test

To demonstrate that the woven impeller prototypes can withstand the forces
during operation, a structural spin test was performed. The initial challenge of a
structural spin test is obtaining a mechanical drive to spin the prototype impeller at very
high speeds. As mentioned, the primary motivation of this research is to effectively
compress water vapor for mechanical refrigeration. To achieve this, very high tip speeds
must be attained. These tip speeds are in excess of 350 m/s. Rotational or angular

velocities are related to tip speeds by the equation

79

Where to is the angular velocity in radians per second, v is the tip speed or

circumferential velocity, and r is the radius of the impeller. Most drives have speeds

rated in terms of revolutions per minute or RPM. Angular velocities in rad/s can be

written in terms of RPM by the equation

30
n: —

(10)

Where n is the rotational velocity in RPM. From these equations,(9) and (10), one can

find the rotational speed of the drive needed to reach 350 m/s for a given impeller radius.

As mentioned above in the Mechanical Winding section of this chapter, the larger

impellers were 5 inch diameter.

52,634RPM to achieve 350 m/s tip speed.

Table 4. Spin test inquiries and quotes

 

 

 

 

 

 

 

 

 

 

Company Quoted work Cost

BALCO Perform dynamic balancing and spin/burst testing $2500
on one piece

Test Devices Inc. Conduct spin test with engineering consultation $20,000

Hines Industries Obtain dynamic balancer or conduct balancing No quote
test. Suggested a static toy plane prop balancer

Hi-Tek Balancing Company only does balancing No quote

Walter Gear Drives Obtain a speed increasing gearbox rated 50,000- $50,000
90,000 RPM

Barbour Stockwell Obtaining test equipment $500,000

 

 

80

For this radius, a drive must spin the wheel at

For the given budget of this project, this is easier said than done. While many
high speed drives exist as well as companies who conduct structural spin tests, this
hardware and services are extremely expensive as outlined in Table 4. To address this
issue, several methods of a high speed spin test are currently being investigated.

However, data for only one method has been collected so far.

High-speed Air-turbine Grinding Tool

One relatively inexpensive method of driving the woven impeller is with a high-
speed air-turbine grinding tool. An air-turbine tool was purchased for $473.77 and was
rated at 65,000 RPM with a power rating of 0.5 hp. The tool was mounted in a test stand
and surrounded with plexi-glass and wire mesh for safety in the event of a structural

failure. Pictures of this set up can be shown in Figure 61 and Figure 62

 

Figure 61. Photo of spin test using air-turbine (front View)

81

 

Figure 62. Photo of spin test using air-turbine (rear view)

When the tool was tested with no load using the compressed air-line available in
the turbomachinery laboratory at MSU, rotational speeds of 57,000 RPM were confirmed
using an optical tachometer. However, when the woven wheel was coupled to the tool,
speeds of only around 4,000 RPM were reached as shown from trials 1 and 2 from Table

5.

Balancing and Vacuum

From the vibrations and air flow of the first trials of the spin test, the need for
balancing and vacuum were evident. Balancing is an important challenge in the design of
any rotating machinery as unbalances can be attributed to power losses as well as
mechanical failure. To combat this, a flexible shafi was coupled to the shafi. This is

shown in the left photo of Figure 63. In addition to the flexible shaft, a static balancing

82

of the test impeller was conducted. Impeller imbalances were identified and alleviated by
adding additional resin, or in some cases lead tape to light areas of the impeller. A photo
of this balancing can be shown in the right photo of Figure 63. Although the shaft and

static balancing did alleviate some of the vibrations, it did not greatly change the speed

reached by the impeller driven by the air-turbine.

 

Figure 63. Flexible shaft coupled to wheel (left) and static balancer (right)

To reduce impeller work, a clear plastic food container was placed around the
impeller, as shown in Figure 64. This reduced the airflow created by the impeller and
hence the work done by the impeller. This did increase the effectiveness of the air-
turbine test as shown by the later trials in Table 5. A top speed of 6,518 RPM was
achieved at which the impeller structurally withheld. At this point it was concluded that
the air-turbine did not have the sufficient power, as it was rated, to rotate the woven

impeller at high the speeds for a structural spin test.

83

 

Figure 64. Airflow prevention chamber for spin test

As noted before and shown in Figure 64, blocking the airflow with a plastic
container gave way to larger speeds. This was achieved by reducing the work done on
the air by the impeller. To eliminate this work completely, and therefore increasing the
effectiveness of the drive even more, a vacuum chamber must be employed. A schematic

illustration of such a vacuum test chamber can be shown in Figure 65.

to vacuum pump

. I w friction plastic
‘ .
1

flexible

 

 

 

 

 

 

 

 

 

1' '1

Figure 65. Design illustration of vacuum test chamber

84

Table 5. Air-turbine spin test trials

 

 

 

 

 

 

 

Trial Maximum Speed Reached (RPM) Structurally withheld?
1 3403 Yes
2 4812 Yes
3 5140 Yes
4 5211 Yes
5 6529 Yes
6 6518 Yes

 

 

 

 

 

RC Racecar Nitro-motor

Another method of driving the woven impeller in a structural spin test is by the
use of a radio controlled racecar engine. These little engines are relatively inexpensive
and advertise high speeds and power. To confirm these claims an engine was purchased
from Larry’s Performance RC ’3 in Sterling Heights, Michigan. The name of the model is
the Ax 28 Spec 1 and the manufacturer advertises that it can reach speeds of 34000 RPM
and that it is rated at 1.157 hp. A test bench was also purchased from a local hobby shop
to break-in and max out the engine. A preliminary test set-up of this experiment is shown

in Figure 66. This test is currently being conducted.

85

 

Figure 66. Test bench for RC racecar nitro motor

High-speed Router Tool

Also being currently performed, is a spin test using high-speed router tools. Two
routers were obtained from www.cpooutlets.com. One is a Bosch PRZOEVSK and is
rated for speeds up to 35,000 RPM and a power of 1 hp. The other is a Milwaukee 5625-
20 and is rated for speeds up to 22,000 RPM and power of 3.5 hp. It is envisioned that
the drives in these routers can be coupled to a woven impeller in a test stand to perform a

structural spin test.

Integrated Motor

As mentioned in Chapter 8, one of the most novel aspects of the presented
technology is the integration of a motor into the woven impeller design itself.

Preliminary realizations of two configurations were conducted.

86

induction Motor

Chapter 8 presents the simplicity of designing an induction machine to be
integrated with the woven impeller. To carry out physical models of this design, several
induction motors were either purchased or generously donated to MSU by Lansing
Electric Motors, Inc. These motors were taken apart and their rotors removed. The inner
diameters of the motor stators were measured, and mandrels were created to fit within
them with somewhat generous clearance for proof of concept. To create an integrated
motor using this configuration, the mandrel needed only to be made of conductive
material. For its electrical conductivity and its ease of machining, aluminum was
selected to machine the mandrels. Two woven impellers were wound and configured to
Spin in the electrical stators with two different bearing configurations.

The first configuration was done using a shafi and bearings that fit the motor
housing. This configuration can be shown in Figure 67. This motor housing was useful
because it was already fitted for commercially available bearings and fitted concentric
with the center of the stator. This eliminated the need for manufacturing axel struts or
supports that had to be made concentric with both the motor-stator and shaft. The electric
stator was connected to an AC frequency inverter that could change the frequency of the
electrical current supplied by the laboratory. The inverter drives the woven impeller by

its integrated motor in this configuration at 6,000 RPM and structurally withheld.

87

 

Figure 67. Prototype woven impeller with integrated motor in an induction machine

The second bearing configuration was constructed using a larger mandrel that was
professionally machined out-of—house. This mandrel was machined with a slightly
smaller outer diameter at the ends than the rest of the piece as shown in Figure 56, Figure
59, and Figure 68. This was done to fit plates that acted as bearings at the ends of the
motor stator as shown in Figure 69. Low friction plastic called Delrin® was also fitted to
the outside of the aluminum mandrel shown in Figure 69. This woven impeller was

driven by its electric stator at 10,000 RPM and also structurally withheld.

88

 

Figure 69. Woven impeller with its integrated motor shown still (left) and spinning (right)

Permanent Magnet Motor
To realize the design of an integrated motor using a permanent magnet approach,

initial experiments have been conducted using iron powder and the epoxy resin. Because

89

of the hazardous effects of inhaling air-bome iron particles, a fume hood has been created
and installed in the laboratory and can be shown in Figure 70 (left). Initial experiments
using magnets shown in Figure 70 (right) indicate that an iron powder mixed with an
epoxy resin has the potential to be magnetized and thus used in a permanent magnet

motor configuration.

 

Figure 70. Puma hood for experimentation with magnetic powders (left) and iron-
powderlepoxy mixture experiments (right)

A Brief Cost Analysis

High performance impellers used in current water vapor chillers have been priced
at roughly $50,000 per wheel [16]. As mentioned in Chapter 6, Current State of the Art,
this is a common reason that hinders the application of R718 units in the US. A brief cost
analysis has been performed to demonstrate the economic advantages of a compressor
using a woven composite impeller.

o For the cost analysis of the woven impeller the ”consumed” materials were the resin

and Kevlar fiber. Taking into account the cost of resin needed for each wheel, plus

90

taking into account some waste losses or ”scrap", it is estimated that each wheel
takes approximately 60 mL of equal parts resin and hardener. Both resin and
hardener were purchased at about $30 per liter. Hence the resin or matrix part

attributed to each wheel was about $1.80.

For the Kevlar portion of the wheel, the cost of a 6,000 yard spool of Kevlar was
approximately $46.00. Using the computer code calculation as well as measured
amounts for one wheel, it was estimated that each wheel consumes approximately
700 yards of Kevlar fiber, including waste. At this rate the Kevlar cost of each wheel

would be about $5.37.

A real life expectancy of the modified CNC machine used for this particular winding I
application is not yet available. Assuming conservatively that it last for about 8,000
wheels. Since the machine itself along with a PC to run the software cost

approximately $9,000, the working hardware for each wheel would be $1.13.

Since the system is fully automated, it is estimated that a technician could be trained
to oversee the manufacture of the impellers on average about 1 hour per 20 wheels.
At a rate of $50 per hour, the labor cost of each wheel would be about $2.50 per

wheel.

The concluding cost of each wheel would then be $10.80. This is extraordinarily less

than the $50,000 cost for the wheels of the current state-of-the-art R718 chillers.

91

CHAPTER 11 : CONCLUSION

As described by this thesis, the advantages of the use of R718 are vast. These
include environmental and economic benefits. Using traditional compression methods in
conjunction with R718 and its unique characteristics, results in'systems that are not
economically feasible in the residential sector. To address this, an alternative compressor
configuration is described along with its novel design for manufacture and integrated
drive. The configuration is an axial, multi-stage, counter-rotating compressor with a
woven composite impeller.

The woven impeller manufacturing method allows for several desirable
characteristics. In addition to low-cost and rapid manufacturing, this design allows for a
multitude of shapes, number of blades, curved blades, inner to outer diameter ratios, and
outer shrouds. Perhaps the most novel characteristic of the design is its ability to include
an integrated motor. This involves the inclusion of electrical components in the wheel to
eliminate the need for a separate drive, essentially making the impeller the electric motor
itself. The two electric motor configurations explored were the induction machine and
the permanent magnet motor.

The design and manufacture of the woven composite impeller was realized on an
inexpensive modified CNC mill using Kevlar® fiber with a polymer matrix. These
impellers were inexpensive to prototype and testing of these composite wheels is
currently being pursued. This testing includes structural spin tests as well as its
performance in a compression cycle.

As mentioned in Chapter 1, the primary purpose for the development of a woven

composite impeller is envisioned to be incorporated in an economically priced chiller

92

using R718. However, this technology may be employed in other applications beyond
the immediate motivation of this research. These applications may include chillers using
traditional refrigerants, as well as applications in automotive, aerospace, train,
propulsion, power generation, pumping, ventilation, and many others.

It was a goal of the presented research to show the enabling technology that will
make R718 a viable alternative to traditional refrigerants. This has been accomplished by
showing the thermodynamic advantages of water as a refrigerant, as well as addressing
the challenges of R718 with alternative compressor technology, namely an axial, multi-
stage, counter-rotating compressor configuration that uses a woven composite impeller.
The concept of designing for manufacture, and producing this woven impeller has been
proven to be feasible and economical, and it is evident that additional efforts may soon

make this technology available to the mass US. market.

93

CHAPTER 12 : RECOMMENDATIONS FOR FURTHER STUDY

In continuing the research presented in this thesis, the author has several
recommendations for further study provided additional funding. The first is to continue
with the accurate control of winding parameters. The first of this is undoubtedly the
accuracy of the fiber placement while winding the woven impellers. This can be
accomplished by a custom—built winding machine using a more precise CNC stage. In
addition to this, a feed-back control system is favorable to observe and control the
accuracy of the winding process.

Alternate winding methods are also an intriguing option that should be explored.
This may include the use of several fiber tows interwoven like used in a sewing machine
instead of using a single tow as previously discussed. Another winding concept has been
developed by using two leader tools to “lay” the fiber in a straight line across the mandrel
and into two slots. Methods like these as well as alternate tooling methods should be
realized given time and funding.

As can be referenced in [25] and [28], parameters such as controlled tension and
resin application are very important to composite characteristics. Because of this, it is
recommended that a tension and resin control system be utilized during the winding
process. The tension control could be as simple as a friction device to create a resisting
torque on the fiber spool, or it is also recommended that tension be controlled by electric
motors installed on the spool dictating the fiber delivery rate as well as torque. A system
like this would also benefit from a feed-back control loop. For the resin control, it is
recommended that the research continue with an approach currently being explored using

a syringe system with accurately controlled actuators.

94

For testing the woven impeller, efforts should be made to continue to establish a
structural spin test. This includes the use of the RC engine and high-speed routers
discussed previously. Although additional funding may tempt the use of expensive
outsourced spin testing, in-house high-speed capability is favorable, and will inevitably
become a necessity. Initial experiments indicate the usefulness of a dynamic balancer
along with a vacuum chamber to conduct spin tests and these items are currently being
constructed. As always, when working with structural integrity experiments, all safety
precautions should be met. These include a vacuum chamber capable of withstanding
bursting composite parts.

As mentioned in the preface, this work was a broad proof of concept and not a
specific optimization of a particular component of the technology. Having established
that these principles can, in general, be employed to enable the motivating concept of
R718, the ever-important, specific exploration of individual aspects of this research can
now be conducted. These include the optimization of blade angles, surface features,
pattern selection, motor design, bearing configurations, and such.

In terms of motor design, the concept an induction machine was generally proven
as conveyed earlier. However, additional exploration should be given to the concept of
the permanent magnet motor design. This includes the research of various magnetic
materials as well as their integration into the woven impeller. In addition, research
should also be conducted on the methods of magnetizing iron elements once integrated in
the impeller.

In terms of enhancing the refrigeration cycle thermodynamically, studies

presented in [4] indicate the usefulness of introducing a condensing wave rotor to the

95

cycle. The 3-port condensing wave rotor (CWR) pressurizes, de-superheats, and
condenses the refrigerant vapor in one dynamic process. Therefore it substitutes for 3
subsystems: intercooler, one compressor stage, & condenser. This greatly reduces the
size and cost of a chiller. In addition, it can enhance the energy efficiency of a R718

chiller by additional 10% [4].

96

Point 1
Temp.[°C]
Pressure[Pa]
entropy[kJ/kg.K]
h[J/kg]

Point 2

Tempis.[°C]
Pressure[Pa]
entropy[J/kg.K]
h2[kJ/kg]

Point 3

Temp.[°C]
Pressure[Pa]
entropy[kJ/kg.K]
h[kJ/kg]

APPENDIX A. EXAMPLE OF CYCLE CALCULATION

5
3497511
1724.701
4015572

26.04513
21.66858

571792
1724.701
4160412

20
571792
1097.614
227857.1

Point 4
Temp.[°C] 5
Pressure[Pa] 349751.1
entropy[kJ/kg.K] 1100.217
h[kJ/kg] 227857.1
4015572
. _ . __ 2068277
COP '
qL[kJ/kg.K] 173700
wnet[kJ/kg.K] 14483.98
Polytropic Efficiency
alfa 0.034556
npol= 0.7

R134a _

Point 1-Superheat Jr “a
Super Heat= 5

349.7511 Pressure[Pa] 3497511

1.724701 entropy[kJ/kgK] 1724.701

401.5572 h[J/kg] 4015572
h2i[kJ/kg] 411696

571.792 nis= i on

1.724701

416.0412
Point3-Subcooling 'I; 7“ g:
Temp.[°C] 20

571.792 Pressure[Pa] 571792

1.097614 entropy[kJ/kg.K]

227.8571 h[kJ/kg] 227857.1

349.7511

1.100217

227.8571

0.107993 1724.700744

1724.701 1024.612775
K=y= 1.035793
Tl 1.634854
nis= 0.697448

97

Point 1
Temp.[°C]
Pressure[Pa]
entropy[J/kg. K]
h[J/kg]

Point 2

Tempis.[°C]
Pressure[Pa]
entropy[J/ kg. K]
h2[kJ/kg]

Point 3

Temp.[°C]
Pressure[Pa]
entropy[kJ/kg.K]
h[kJ/kg]

Point 4

Temp.[" C]
Pressure[Pa]
entropy[kJ/kg.K]
h[kJ/kg]

L COP
q L[J/ kg . K]
wnet[J/kg. K]

COP=

5
871 .8345517
9026.899498
2510734169

120.1702265

2336.56155
9221 .405519
2726725791

20
2336.56155
296.2975647
83861.7474

5.000000003
871 .8345517
302.1724583

83861 .7474
21006.72659
2510734169

2426872421
2159916224
r 5,13%;- .{vvb‘i‘j-ziyfln‘f’ .

Ii ran»

Polytropic Efficiency

 

alfa 0.245283019
npol= 0.7
Calculation of K=y
T= 630206522
P1= 1604.198051
Cp= 1877.695135
= ‘ ‘ 481.
Cv= 1416.195135
K=v= 1 .325873172

 

R718

Point 1-Superheat 1:”

Super Heat=
0.871835 Pressure[Pa]
9.026899 entropy[kJ/kg.K]
2510.734 h[J/kg]

h21[kJ/kg]
2.336562 nis=
9.221406
2726.726 Ah:

4.47.44. :9; .,_.,L.._1.—;

BE
5
871.8345517

9026.899498
2510734169

2726725791

1703962971592

215991.6224 [kJ/kg]

Point 3-Subcooling ET .571

Temp.[°C]
2.336562 Pressure[Pa]
0.296298 entropy[kJ/kg.K]
83.86175 h[kJ/kg]

0.871835
0.302172
83.86175
0.025246
76.20553
9026.899

K:Y:

nis=

62.58511
1604.198

98

20
2336.56155
296.2975647
83861 .7474

1.325

2.680051559
0.662971597

APPENDIX B: EXAMPLE OF VISUAL BASIC CODE USED IN CYCLE

Sub IsoCOPwater()

Col = 20
COP = Range("F7")
Var = 1

Do While COP <= (Range("F7") + 8 * Range("F8"))

Row = 3

Range("C3") = 0

Range("C4") = Range("C3") + 5

Cells(Row — 2, Col + 1) = "COP="

Cells(Row - 1, Col + l) = COP

Do While Range("C3") < 360 And Abs(Range("C4")) < 360

Worksheets("All").Range("Cl3").Goalseek _
Goal:=COP,

ChangingCell:=Worksheets("All").Range("C4")

If Var = 1 Then Cells(Row, Col) = Range("C3") -
Range("F12")
Cells(Row, Col + l) = Range("C4") - Range("C3") — 2 *

Range("F12")
Range("C3") = Range("C3") + 1

Row = Row + 1

Loop

Var = 0

COP = COP + Range("F8")
Col = Col + 1

99

Loop

End Sub

Sub IsoCOP()
Clean
IsoCOPwater
Col = 8

COP Range("F7")

Do While COP <= (Range("F7") + 8 * Range("F8"))

Row = 3

Range("C3") = Range("F4")

Range("C4") = Range("C3") + 5

Cells(Row - 2, Col + 1) = "COP="

Cells(Row - 1, Col + l) = COP

Do While Range("C3") < Range("F5") And Abs(Range("C4")) <

Range("F5")

Worksheets("All").Range("Cl2").Goalseek _
Goal:=COP,
ChangingCell:=Worksheets("All").Range("C4")

RowMax 2 Row

If Abs(Range("C4")) < Range("F5") Then Cells(RowMax, 8) =
Range("C3") - Range("F13")

If Abs(Range("C4")) < Range("F5") Then Cells(Row, Col + l)
= Range("C4") - Range("C3") - 2 * Range("F13")

Range("C3") = Range("C3") + 1

Row 2 Row + 1
Loop

Cells(Row — 2, Col + 1) = Range("Gl")

100

Cells(Row - 3, Col + l) = Range("Gl")

COP = COP + Range("F8")

Col = Col + 1
Loop
End Sub

Sub MaxCOPLineRefrigerant()
MaxCOPLineWater

Col = 9

Do While Col < 18

Row 2 3

Maxlift = 0

Do While Row < 200
If Cells(Row, Col) > Maxlift Then
Maxlift = Cells(Row, Col)

Male = Cells(Row, 8)

End If

Row 2 Row + 1

Loop

Cells(Col + 15, 2) = Male
Cells(Col + 15, 3) = Maxlift

Col 2 Col + 1
Loop

End Sub

101

Sub MaXCOPLineWater()
Col = 21

Do While Col < 30
Row 2 3

Maxlift = 0

Do While Row < 350
If Cells(Row, Col) > Maxlift Then
Maxlift = Cells(Row, Col)

MaXTl = Cells(Row, 20)

End If

Row 2 Row + 1

Loop

Cells(Col + 13, 2) = Male
Cells(Col + 13, 3) = Maxlift

Col 2 Col + 1
Loop

End Sub

Sub Intersection()
IntersectionCells
Col = 32

Do While Col < 42

Var 10000

Row = 3

102

Do While Row < 200

If Abs(Cells(Row, Col)) < Var And Abs(Cells(Row, Col)) <> 0
Then

Inter = Cells(Row, 20)

Inter2 = Cells(Row, Col — ll)
Var = Abs(Cells(Row, Col))
End If

Row = Row + 1

Loop
Cells(Col + 13, 2) = Inter
Cells(Col + 13, 3) = Inter2

Col 2 Col + 1
Loop

End Sub

Sub IntersectionCells()

RRow 3

Coll 32

Do While Cells(RRow, 8) <> Range("T3")
RRow = RRow + 1

Loop

Do While Coll < 41

RRRow = 3
hold = RRow
Roww = hold

Do While RRRow < 200

103

Cells(RRRow, Coll) = Cells(RRRow, 20 + Coll — 31) -
Cells(Roww, 8 + Coll — 31)

RRRow = RRRow + l

Roww = Roww + 1
Loop

Coll = Coll + 1
Loop

End Sub

104

APPENDIX C. COMPLETE PATTERN DESIGN TABLE

Number

of Point Pattern
5 1, skip 1
5 1, skip 2
5 2. skip 1
5 2. skip 2
5 3. skip 1
5 3, skip 2
5 4. skip 1
5 4. skip 2

6 1. skip 1
6 1. skip2
6 1. skip 3
6 2. skip 1
6 2. skip 2
6 2. skip 3
6 3. skip 1
6 3. skip2
6 3. skip 3
6 4, skip 1
6 4. skip2
6 4. skip 3
6 5, skip1

6 5. skip2
6 5. skip3

7 1. skip 1
7 1. skip 2
7 1. skip 3
7 1. skip 4
7 2. skip 1
7 2. skip 2
7 2. skip 3
7 2. skip 4
7 3. skip 1
7 3. skip 2
7 3. skip 3
7 3. skip4
7 4, skip 1
7 4. skip 2
T 4. skip 3
7 4. skip 4
7 5. skip 1
7 5. skip2
7 5. skip 3
7 5. skip4
7 6. skip 1
7 6. skip 2
7 6. skip 3
7 6. skip4

Not
Possible Shape N
S-A

5—A
5-A

x

S-A
5—A
S-A

i

S

1’

6-A

I

6—A

7-A

x

7-A
7-A

X

dildo Notes

Sharp turns

Sharp turns

Sharp tums

0.5 Sharp turns

0 Stackvup in the center

0.5

0.5 Sharp turns

0 Stack—up in the center

0.5

Solid outside for 3 separate
sixths ofthe circuit, normal
0 inside. stack-up in the center

0.62349
0.22252
0.22252
0.62349 Sharp turns
0.62349
0.22252
0.22252
0
0.62349
0.22252

0
0.62349 Sharp turns
0.62349
0
0.22252
0.62349 Sharp turns
0

0.22252
0.22252
0.62349 Sharp turns
0.62349
0.22252
0.22252
0.62349 Sharp turns

[05

@@§

6-A

99

6—B

so

APPENDIX C. COMPLETE PATTERN DESIGN TABLE

Number

of Points Pattern
5 1. skip 1
5 1. skip2
5 2. skip 1
5 2. skip 2
5 3. skip1
5 3, skip 2
5 4. skip 1
5 4, skip 2

6 1. skip 1
6 1, skip 2
6 1. skip 3
6 2. skip 1
6 2, skip 2
6 2. skip 3
6 3. skip 1
6 3. skip 2
6 3. skip 3
6 4. skip 1
6 4. skip 2
6 4. skip 3
6 5. skip 1

s 5. skip 2
6 5. skip 3

7 1. skip 1
7 1. skip 2
7 1. skip 3
7 1. skip 4
7 2. skip 1
7 2. skip 2
7 2. skip 3
7 2. skip 4
7 3. skip 1
7 3. skip2
7 3. skip 3
7 3. skip4
7 4. skip 1
7 4. skip 2
7 4. skip 3
7 4. skip 4
7 5, skip1
7 5. skip 2
7 5. skip 3
7 5. skip 4
7 6. skip 1
7 6. skip2
7 6. skip 3
7 6. skip 4

Not

Possible Shape N dildo Notes
5—A Sharp turns
5—A

5—A Sharp turns
S-A
5A Sharp turns

5
@2

5—A

6—A 1 0.5 Sharp turns

0 Stack—up in the center

9 S

0.5
6-A 1 0.5 Sharp turns
x 6—A
6-8 0 0 Stack-up in the center

6—A 1 0.5
Solid outside for 3 separate
sixths ofthe circuit. normal
0 inside. stack-up in the center

3

7A 1.5 0.62349 6-8
7-5 0.5 0.22252
7—B 0.5 0.2252
7-A 1.5 0.62349 Sharp turns
7—A 1.5 0.62349
7—B 0.5 0.2252
7-8 0.5 0.2252
x 0
7—A 1.5 0.62349
78 0.5 0.2252
x 0
7-A 1.5 0.62349 Sharp turns
7—A 1.5 0.62349
x 0
7—8 0.5 0.2252
7-A 1.5 062349 Sharp turns
0

743 0.5 0.22252

75 0.5 0.22252

7—A 1.5 0.62349 Sharp turns
7-A 1.5 0.62349

7-8 0.5 0.2252

7—B 0.5 0.22252 78
7-A 1.5 0.62349 Sharp turns

so

105

94.
94.
94.
94,
94.
95,
95.
95,
95.
95,
95.
96.
96.
96.
96,
96.
96.
97.
97.
97,
97.
97.
97.
98.
98.
98.
98.
98.
98.

10 1,
10 1.
10 1.
10 1.
10 1.
10 1.
10 1.
10 2,
10 2.
10 2.
10 2.
10 2.
10 2.
10 2.
10 3.
10 3.
10 3.

10 3.
10 3.

sMpZ
sMp3
sHp4
shpS
sHp6
sMp1
sMpZ
sHp3
sHp4
sHpS
sHpG
sMp1
sMpZ
shp3
sHp4
sMpS
sHpS
shp1
sHpZ
sMp3
sMp4
sHpS
sHp6
sHp1
shp2
sHp3
sMp4
sMpS
sHpB

skip 1
skip 2
skip 3
skip 4
skip 5
skip 6
skip 7
skip 1
skip 2
skip 3
skip 4
skip 5
skip 6
skip 7
skip 1
skip 2
skip 3

skip 4
skip 5

$§§$ 86

10—A

1 0—A

1 0-D

10-C

10—D
10-C

1.5
0.5
1.5
2.5
1.5

0.5
1.5

2.5

0.5
0.5

2.5
1.5
0.5
1.5
2.5

2.5
1.5

0.5
1.5

-- N w -I (a)

‘0

0.5
0.17365
0
0.5
0.76604 Sharp turns
0.76604
0.5
0
0.17365
0.5 SC
0
0.76604
0
0.17365
0.17365

0
0. 76604 Sharp turns
0

0.5

0.17365

0

0.5
0.76604 Sharp turns

0.76604

0.5

0

0.17366

0.5

0

0.80902
0
0
0
0.30902
0

0.80902 Sharp turns
0

 

0
0
0 Stack-up in the center
0

0.58779
0
0
0
0.30902
Solid outside for 5 separate
tenths of the circuit, normal

0 inside. stack—up in the center
0.30902 103

107

10 3.
10 3.
10 4.
10 4.
10 4.
10 4.
10 4.
10 4.
10 4.
10 5.
10 5.
10 5.

10 5.
10 5.
10 5.
10 6.
10 6.
10 6,

10 6.
10 6.
10 6.
10 7.
10 7.
10 7.

10 7.
10 7.
10 7.
10 7.
10 8.
10 8.
10 8.
10 8.
10 8.
10 8.
10 8.
10 9.
10 9.
10 9.

10 9.
10 9.
10 9.
10 9.

shp6
sup?
sHp1
shp2
sHp3
shp4
shpS
sHpG
sfip7
sHp1
shp2
sHp3

,skip4

shpS
sMpG
sHp7
sHp1
sHpZ
sfip3

.skip4

skip 5
skip 6
skip 7
skip 1
skip 2
skip 3

sHp4
sHpS
sHp6
shp7
sfip1
sHpZ
sHp3
shp4
shp5
sMp6
sHp7
sHp1
shp2
sHp3

skip 4
skip 5
skip 6
skip 7

10—A

1043

104B

108

10—A

10—C

10—C

10—D

10-A

10—C

10A

(.0

N

N

w

d

u.

U

N

N

w

—I

.5

N

(a)

d

U

0
0.80902 Sharp turns
0
0.58779
0
0 Stack—up in the center
0
0.58779
0
0.80902 100
0
0.30902
0
0.30902
0
0.80902 Sharp turns
0
0.58779
0
0 Stack—up in the center 10-0
0
0.58779
0
0.80902
0
0.30902
Solid outside for 5 separate
tenths of the circuit. normal
0 inside. stack-up in the center
0.30902
0
0
0
0.58779
0
0 Stack-up in the center
0
0
0
0.80902
0
0.30902
Solid outside for 5 separate

tenths of the circuit, normal
0 inside. stack-up in the center

0
0.80902 Sharp turns

108

11L
11L
111
111,
111.
11L
111
111
112
112
112
112
112
112
112
112
113
113
113
113
113
113
113
113
115
114
114
114
11A
115
115
115
115
115
115
115
115
115
115
115
115
115
115
115
115
115
115
115

sHp1
shp2
shp3
sHp4
sHpS
sHpG
sHp7
shpa
shp1
shp2
shp3
sHp4
sMpS
sHp6
sHp7
sHpB
shp1
sfip2
shp3
sHp4
smpS
shp6
smp7
shpa
sHp1
sHpZ
smp3
5Hp4
sMpS
sHpS
sHp7
sHpB
sHp1
shpz
sHp3
sHp4
shpS
sHp6
sMp7
sMp8
sMp1
sMpZ
sHp3
sMp4
sMpS
sHp6
shp7
sHpa

35

15
05
05
15

35
35
25
15
05
05
15
25
35
35

0&“25
0&Mfl5
on1su
014231
014231
041flu
(1&M86
0&“258mmphmm
0&“25
0&M86
04190
014231
01MB1
04EM2
06RW6
0.64125 Sharp turns
08M25
06RW6
041au
01MB1
01431
04HM2
0&Mfli
0.84125 Sharp turns
0&"25
0&M86
0415a
014231
014231
0415a
065W6
0.84125 Sharp turns
0&"25
asses
oauuz
014231
014231
04150
069W6
0.84125 Sharp turns
0&“25
ussuw
041sm
014231
014231
04flM2
065M6
0.84125 Sharp turns

109

 

 

1LC

110

117,
117.
117.
117,
117,
117,
117.
117.
118.
118.
118.
118,
118.
118.
118.
118.
119.
119,
119.
119.
119.
119.
119.
119.

11 10.
11 10.
11 10.
11 10,
11 10.
11 10.
11 10.
11 10.

shp1
shpZ
sMpB
shp4
shpS
shp6
shp?
shp8
shp1
shp2
shpB
sHp4
shpS
shp6
shp?
sHpB
shp1
sHpZ
shp3
sHp4
sHpS
shp6
sHp7
sMp8
sMp1
SMpZ
shp3
shp4
shpS
sMpG
sMp7
sHpB

11-A
11-B
we
11-0
11-0
114:
11-8
11-A
11-A
113
no
114)
11.0
no
11-B
11«A
11-A
11-8
114:
11-0
11-0
114:
11-8
11-A
11-A
1143
11c
11-0
11-0
no
118
11-A

3-5
2.5
1.5
0.5
0.5
1.5
2.5
3.5
3.5
2-5
1.5
0.5
0.5
1.5
2.5
3.5
3.5
2.5
1-5
0.5
0.5
1.5
2.5
3.5
3-5
2-5
1.5
0.5
0.5
1.5
2-5
3.5

0-84125
0.65486
0.41542
0.14231
0.14231
041542
065486
0-84125 Sharp turns
0.84125
0.65486
0.41542
0.14231
0-14231
0.41542
0.65486
0.84125 Sharp turns
0.84125
0.65486
041542
014231
0.14231
0.41542
0.65486
0.84125 Sharptums
0.84125
0.65486
0.41542
0.14231
0.14231
041542
065486
0.84125 Sharp turns

[10

REFERENCES

l. Dossat, Roy J. and Horan, Thomas J. Principles of Refi"lg€l‘all()ll- 5th ed. Upper
Saddle River, NJ : Pearson Education, Inc., 2002.

2. Brown, D. Summer 2006 Electricity Supply and Demand Outlook. S.l. : California
Energy Commission, 2006.

3. Cengel, Yunas A. and Boles, Michael A. T hermodynamics- An Engineering
Approach- 3rd ed. Boston, MA : McGraw-I—Iill, Inc., 1998.

4. Preliminary Study of a Novel R718 Compression Refrigeration Cycle Using a 3- Port
Condensing Wave Rotor. Amir A. Kharazi, P. Akbari, N. Mueller. s.l. : ASME Journal
of Engineering for Gas Turbines and Power, 2005, Vol. 127.

5. S .E. Wright, T. Yoshinaka, B. A. Le Drew, N. C. D’Orsi. The Efficiency Limits of
Water Vapor Compressors. s.l. : ARTI-21CR/605-10010-01, 2000.

6. Lachner, B., Nellis, G., and Reindl, D. The Use of Water Vapor as a Refrigerant
Impact. 31. : ARTI-21CR/611-10080-01, 2004.

7. Design of Compressors for Water as a Refrigerant. Mueller, N. 3.1. : ASHRAE
Transactions, 2001, Vol. 107.

8. Development of Large Capacity High Efficiency Mechanical Compression (M V C)
Units. Zimmerman, Z. s.1. : Desalination, 1994, Vol. 96.

9. Turbo Water Chiller with Water as Refrigerant. Wobst, E., Kalitzin, N ., Apley, R.
Purdue University : lntemational Compressor Engineering Conference, 2005. C 129.

10. Burandt, B. T echnische Information R718-Turbo-Kaltwassersatz. Dresden,
Germany : Institute fur Lufi- und Kaltetechnik (ILK), 2004.

1 1. Turbo chiller with water as a refrigerant. Albring, P., Heinrich, G. Aarhus,
Denmark : 11R, 1998.4.

12. The State of Art for Water Vapor Compressors and Cooling Plants Using Water as
Refrigerant. Madsboll, M. S., Minds, G., Nyvad, J., and Elefsen, F. Hanover: IIR,
1994.

13. District Cooling with Water as Refrigerant Combined with Energy Storage Systems.
Paul, J. 05150, Norway : IIF-IIR, 1998. 32, Bl,El,E2.

14. Water Vapour Compression: Application and Outlook. Albring, P., Burandt B.,and
Schenk, N. 6, 3.1. : K1 Luft Kéiltetech DE, 2002, Vol. 38.

111

15. New Age Water Chillers with Water as Refrigerant. KiihnI-Kinel, J. Geneva,
Switzerland : CERN, 1997.

16. B., Burandt. Planungsdaten R718-Turbo-Kaltwassersatz (R 7 I 8 chiller: data for
planning). Dresden, Germany : Institute fur Luft- und Kaltetechnik (ILK), 2004.

17. Wasser als Kt'iltemittel (water as a refrigerant). Germany : BINE Infomiationsdienst,
2003. http://www.bine.info/pdf/publikation/bi0803intemetxpdf.

18. Albring P., Miiller N., et al. T urbo—compressor impeller for coolant. Patent No.
5,464,325 United States, 1994.

19. —. Kc’iltemittel- Turboverdichter (Refrigerant turbo—compressor). Reg. -No.
941 085 02.9 European Patent Registration, 1994.

20. —. Radiallaufrad (Radial-flow impeller). DE 43 21 I 73 C2 Deutsche Patentschrifi
(German Patent), 1993.

21. Water Chilling and Ice Making with Water as Refrigerant. Paul, J., Jahn E.
Washington, USA : International Conference on Ozon Protection Technologies, 1996.

22. Aungier, R. H. Axial-Flow Compressors A Strategy for Aerodynamic Design and
Analysis. New York, NY : ASME Press, 2003.

23. Dixon, S. L. Fluid Mechanics and Thermodynamics of Turbomachinery- 5th ed.
Burlington, MA : Elsevier Butterworth-Heinemann, 2005.

24. A filament-wound structure technology Overview. Shen, Frank C. 8.1. : Materials
Chemistry and Physics, 1995, Vol. 42.

25. Chawla, Krishan K. Composite Materials- Science and Engineering- 2nd ed. New
York, NY : Springer Science and Business Media, Inc., 1998.

26. Simulation and Production of Wound Impellers. Eyler, Allen and Mueller, Norbert.
Berlin, Germany : ASME Turbo Expo, 2008. GT 2008-51310.

27. Preliminary Mechanical Analysis for Novel Woven/ Wound Composite Axial-
Compressor. Lahiri, A., Lindberg, B., Miiller, N. Seatle, Washington : ASME, 2007.
IMECE 2007-42490.

28. Influence of the filament winding tension on physical and mechanical properties of
reinforced composites. P.Mertiny, F. Ellyin. s.l. : Composites: Part A, 2002, Vol. 33.

29. A comparative study of water as a refrigerant with some. Mueller, N ., Kilicarslan,
A. 5.1. : lntemational Journal of Energy Research, 2005, Vol. 29.

112

             

   

TATE UNIVERSIT

MICHIGAN S Y LIBRARIES
1111111111 1111 IIIIIIIIIIIIIHIII
3 1 03 17

293 063 37