Flow Modelling of Scroll Compressors and Expanders

Karagiorgis, G. (1998). Flow modelling of scroll compressors and expanders.
(Unpublished
Doctoral thesis, City University London)
City Research Online
Original citation: Karagiorgis, G. (1998). Flow modelling of scroll compressors and expanders.
(Unpublished Doctoral thesis, City University London)
Permanent City Research Online URL: http://openaccess.city.ac.uk/7573/
Copyright & reuse

City University London has developed City Research Online so that its users may access the
research outputs of City University London's staff. Copyright © and Moral Rights for this paper are
retained by the individual author(s) and/ or other copyright holders. All material in City Research
Online is checked for eligibility for copyright before being made available in the live archive. URLs
from City Research Online may be freely distributed and linked to from other web pages.
Versions of research
The version in City Research Online may differ from the final published version. Users are advised
to check the Permanent City Research Online URL above for the status of the paper.
Enquiries
If you have any enquiries about any aspect of City Research Online, or if you wish to make contact
with the author(s) of this paper, please email the team at publications@city.ac.uk.
FLOW MODELING OF SCROLL
COMPRESSORS AND EXPANDERS
by
George Karagiorgis
A Thesis submitted to
The City University
for the Degree of
Doctor of Philosophy
August 1998
Department of Mechanical Engineering and Aeronautics

The City University
Northampton Square, London EC1 V OHB.
To my family
TABLE OF CONTENTS
PA GE
CHA PTER 1: INTRODUCTION .
1.1 Introductory Remarks .................................................................
1.2 Present Contribution .................................................................. 2
1.3 Thesis Layout .......................................................................... 3
CHA PTER 2: LITERATURE SURVEY .............................................. 5
2.1 Introduction ............................................................................ 5
2.2 Compressors ........................................................................... 5
2.2.1 Aerodynamic Compressors ..................................................... 6
2.2.1.1 Centrifugal Compressors .................................................... 7
2.2.2 Positive Displacement Compressors ........................................... 7
2.2.2.1 Reciprocating Compressor .................................................. 7
2.2.2.2 Screw Compressor ............................................................ 8
2.2.2.3 Scroll Compressor ............................................................ 9
2.2.2.3.1 Working Process ......................................................... 10
2.2.2.3.2 Scroll Members .......................................................... 12
2.2.2.3.3 Principle Of Operation ................................................... 12
2.2.2.3.4 Analytical Model ......................................................... 13
SuctionProcess ................................................................... 14
CompressionProcess ............................................................. 17
DischargeProcess ................................................................. 18
Leakages............................................................................ 19
2.3 Two Phase Expanders ................................................................ 22
2.3.1 Scroll Expanders .................................................................. 24
2.3.2 Lyshoim Total Flow Expander .................................................. 25
Ill
PA GE
2.3.2.1 City University And Lyshoim Expander .................................. 27
2.3.3 Two Phase Turbines .............................................................. 27
2.4 Conclusions ........................................................................... 29
CHAPTER 3: OUASI ONE DIMENSIONAL FLOW MODELING ............ 31
3.1 Introduction ............................................................................ 31
3.2 Theory ................................................................................... 32
3.2.1 Assumptions ....................................................................... 32
3.2.2 Conservation Equations .......................................................... 32
EnergyGain Due To The Gas Inflow ............................................... 33
EnergyGain Due To The Gas Outflow ............................................. 34
HeatTransfer ........................................................................... 35
WorkSupplied ......................................................................... 35
3.2.3 Gas Leakage ........................................................................ 35
3.2.4 Oil Injection ........................................................................ 37
3.2.5 Computation Of Fluid Properties ............................................... 38
3.3 Results Of The 1-D Flow Modelling ................................................ 39
3.3.1 Scroll Compressor ................................................................. 39
3.3.2 Scroll Expander .................................................................... 53
3.4 Applications Of The Scroll Compressor - Expander .............................. 58
3.4.1 Cycle Proposed .................................................................... 58
3.4.2 The Proposed Model ............................................................... 59
3.4.3 Expander Characteristics ......................................................... 60
3.4.4 Compressor Characteristics ...................................................... 60
3.4.5 Combined Characteristics ........................................................ 61
Iv
PAGE
CHAPTER 4: GEOMETRY AND GRID GENERATION ........................ 64
4.1 Introduction ............................................................................ 64
4.2 Geometry ............................................................................... 64
4.3 Theory of Grid Generation ........................................................... 68
4.3.1 Algebraic Methods ............................................................... 69
4.3.2 Partial Differential Equation Methods ......................................... 69
4.3.3 Adaptive Grids .................................................................... 71
4.3.4 Moving Boundary Grids ......................................................... 72
4.3.5 Computational Grids For Complex Geometry Domains ..................... 73
4.3.5.1 Multi Block Grids ............................................................ 74
4.3.5.2 Unstructured Grids ........................................................... 74
4.4 Development Of Grid Generation Code ............................................ 75
4.4.1 Two Block Grid Generation Code .............................................. 75
4.4.2 Four Block Grid Generation Code ............................................. 80
4.4.2.1 Implementation of Transfinite Interpolation .............................. 80
4.4.3 Mid-Planes Based Grid Generation Code ..................................... 85
4.4.3.1 Implementation Of Non Uniform Distribution Functions ............... 89
4.4.3.2 Suction-Part Of Compression Process .................................... 93
4.4.3.3 Remainder Of Compression Process ....................................... 94
4.4.3.4 Discharge Process ............................................................ 95
4.4.3.5 Inclusion Of The Inlet Pipe ................................................. 95
4.4.3.6 Use Of The Code ............................................................. 98
CHAPTER 5: THREE DIMENSIONAL FLOW MODELING ................... 115
5.1 Introduction ............................................................................ 115
5.2 Theory ................................................................................... 115
5.2.1 Governing Equations ............................................................. 116
5.2.2 Mathematical Behaviour of Partial Differential Equations .................. 117
V
PA GE
5.2.2.1 Hyperbolic Flows . 119
5.2.1.2 Parabolic Flows ............................................................... 119
5.2.1.3 Elliptic Flows ................................................................. 119
5.2.3 Numerical Solutions To Partial Differential Equations ..................... 120
5.2.3.1 Finite Difference Method ................................................... 120
5.2.3.2 Finite Element Method ...................................................... 121
5.2.3.3 Finite Volume Method ...................................................... 122
5.2.3.4 Comparison Of The Discretization Techniques .......................... 122
5.2.4 Pressure / Velocity Coupling Procedures ..................................... 124
5.2.4.1 Simple Algorithms ........................................................... 125
5.2.4.2 Piso Algorithms .............................................................. 126
5.2.4.3 Pressure / Velocity Coupling Techniques For FEMs .................... 127
5.2.4.4 Rhie-Chow Interpolation .................................................... 128
5.2.5 Advective Term Methods ........................................................ 128
5.2.6 Time Stepping Procedures ....................................................... 132
5.2.7 Solution Techniques For The Linearized Equations ......................... 133
5.2.8 Commercially Available Codes ................................................. 134
5.2.8.1 CFX-FLOW3D (Code Overview) .......................................... 135
5.3 Results Of The 3-D Flow Modelling ................................................ 141
5.3.1 CFD Analysis ..................................................................... 141
5.3.2 Thermodynamic Analysis ........................................................ 147
5.3.2.1 Expander Characteristics .................................................... 147
5.3.2.2 Compressor Characteristics ................................................. 149
CHA PTER 6: CONCLUSIONS A ND FUTURE W ORK ........................... 216
6.1 Conclusions ............................................................................216
6.2 Future work ............................................................................218
VI
PA GE
A PPENDICES.
A PPENDIX A : GEOMETRY ........................................................... 220
A.1 Introduction ...........................................................................220
A.2 Geometry Of The Scroll Device ...................................................220
A.2.1 Basic Geometry ..................................................................220
A.2.2 Contact Conditions ...............................................................228
A.2.3 Starting Points - Inner Circular Arcs ..........................................233
A.2.4 Length Of Device Spirals ........................................................245
A.3 Working Chambers and Working Volumes ................................... 256
A.3.1 An Overview ....................................................................... 256

A.3.2 Suction Chamber and its Working Volume ...................................263
A.3.3 Compression Chamber and its Working Volume ............................277
A.3.4 Discharge Chamber and its Working Volume ................................286
A PPENDIX B: GOVERNING EQUATIONS ........................................ 288
B .1 The Continuity Equation .............................................................288
B .2 The Momentum Equation ............................................................291
B .3 The Energy Equation .................................................................298
B .4 The Equation Of State ...............................................................303
APPENDIX C: PROGRAM LISTING ................................................ 305
C. 1 Hyperbolic Tangent Distribution ...................................................305
C.2 Trigonometric Tangent Distribution ................................................307
C.3 Transfinite Interpolation .............................................................309
C.4 Geometry ..............................................................................311
C.5 Command Language (CFX4) ........................................................320
REFERENCES & BIBLIOGRAPHY ................................................. 330
VII
A CKNO W LEDGEMENTS
I would like to express my sincere appreciation to my parents and family, and

especially to my wife Spyroula for her support and understanding during the difficult
times of this project.
I also would like to thank Dr. D. N. Papadakos (Imperial College) for allowing me to
use his powerful computing facilities without which this project would never have
reached this stage.
Finally, I would like to thank my supervisor Prof. I. K. Smith.
VIII
DECLARATION
I grant powers of discretion to the University Librarian to allow this thesis to be

copied in whole or in part without further reference to me. This permission covers
only single copies made for study purposes, subject to normal conditions of
acknowledgement.
Ix
A BSTRA CT
The use of scroll compressors in refrigeration systems is becoming increasing popular.

However the development and use of a scroll machine as an expander to replace a
throttle valve has never been reported. This thesis describes the work carried out to
develop a two phase scroll expander for use in vapour compression refrigeration
systems. By this means the power requirements of the system would be decreased and
the refrigeration effect increased.
A computer simulation of the geometry of the scroll device was first developed. This
was attached to a package of subroutines developed to analyse compressor and
expander thermodynamic performance to obtain estimates of both scroll expander and
compressor performance.
The results of the expander compressor analyses were used to estimate the
performance of a scroll expander driving a second scroll compressor in a sealed unit
used in a large industrial chiller of 4MW thermal output.
By this means the power requirement for the main compressor was reduced by
approximately 10% while, due to the expansion in the expander being more nearly
isentropic than in the throttle valve, the refrigeration effect was simultaneously
increased by nearly 2%. This lead to an overall improvement in coefficient of
performance of nearly 13.5%. Further investigation of the system and more specially
on the scroll device appears to be justified.
A 3D simulation was then performed by the use of a CFD package called CFX4. This
gave an excellent understanding on the flow 'inside' the machine. From the flow
patterns obtained it is obvious that the scroll device behaviour appears to be excellent
as compressor but not so good as an expander. This is mainly due to the huge
recirculation during the filling process and due to the significant pressure drop of
approximately 2 bar (Max) associated with it.
x
Chapter 1
Introduction
1.1 Introductory Remarks

Studies in fluid flow analysis are generally conducted using either experimental or
numerical-computational approaches. From the theoretical point of view fluid
dynamics are highly developed, however the mathematical formulations are highly
complex and can only be solved analytically in very few simplified cases. On the
other hand advances in numerical analysis and computer hardware/software have
made feasible the solution of the relevant equations using numerical techniques.
This thesis is concerned with the development of a an efficient scroll expander
which can operate with two phase flow, and the modeling of its working
conditions.
During the last few years, numerical flow analysis techniques have become more
and more important in machine design. A key factor in obtaining a realistic
solution is mesh generation, considered as a basic problem requiring a sustained
effort. In fact, the interaction between the mesh characteristics and the development
of it of the solver calculation are closely linked as far as stability, convergence and
definition of the solution are concerned.
The study was carried out using Computational Fluid Dynamics (CFD) techniques.
The advantages of successful computational procedures compared with actual
experiments are significant. Computational studies are relatively cheap to perform.
Furthermore the parameters defining the geometry or the operating conditions of
the device can be varied at will and hence the required optimum can be achieved.
Scroll expanders present definite advantages over other machines. In addition, the
reduced number of parts, the ability to cope with two phase working fluids and the
inherent reliability proven in compressor mode are major reasons for its wider use
in energy recovery. The fast growing mass production of such units as compressors
is also likely to contribute to a lowering of the production costs for expander
applications.
Chapter 1 Introduction
As is the case for all rotary volumetric machines, reverse motion is inherently
possible and, therefore the conversion of scroll compressor to scroll expander can
be envisaged. Smith's investigation for this type of machine indicated that the
machines most applicable for the expansion of a two phase flow are the Lysholm
twin screw expander and the scroll expander. As far as the first type is concerned,
work was initiated by Smith at the Centre for Compressor Technology, City
University London in late 1982.
In order to perform a thermodynamic analysis of the scroll compressor - expander

first we have to describe the physical behavior of the machine in terms of
mathematical equations. Functions of working volume, sealing lines, touching
conditions etc. in terms of crank angle, (i.e. time) form the foundations for the
analysis.
1.2 Present Contribution

The objective of this research project was to design and develop a two phase
expander to be used as a substitute of the throttle valve in a vapour compression
refrigeration system, and to model the thermodynamics and fluid mechanics of this
machine when operating in the compression mode. The major contributions of the
work presented in this thesis are:
1. A parametric study giving all the mathematical equations needed for the
modeling of the scroll compressor-expander. This includes the derivation of
mathematical expressions giving the working volumes, the sealing lines, the
touching conditions, the thickness and the lengths of the arms, etc. as functions
of the various design parameters (such as a and N ) and the crank angle 8
2. Implementation of the geometry developed into a Quasi One Dimensional Flow
package called SCORPATH. This gave some estimates of the performance of
the scroll compressor and expander operating under different working
conditions.
3. Design and development of a grid generation program that can handle the
complexity of the working volumes of the scroll device. Aspect ratios,
orthogonality, cell distortion and smooth distribution are some of the difficult
issues involved in the transient, structured, body fitted grid generation package
developed. This is a major contribution to the development and understanding
of these type of machines.
4. Coupling of the Transient Grid Generation Code developed with a
Computational Fluid Dynamics (CFD) package called CFX4. Implementation
2
of the resulted executable yielded thermodynamic results, as well as an in depth

understanding of the flow patterns within the scroll device. Consequently
visualization software tools can be used to improve the overall behavior of the
device.
5. Finally a test case was devised and studied in which the scroll expander
substituted the throttle valve in a Carnot refrigeration cycle.
1.3 Thesis Layout

The thesis is structured as follows:
In chapter 2 previous efforts to develop two-phase expanders and compressors are

examined. Emphasis is placed on the work done on scroll compressors mainly
because knowledge on this topic is essential for the design and development of
scroll expanders. It is also demonstrated that the scroll device is the most
appropriate machine for the requirements of the present study. The existing, very
limited amount of work that has been done up to now on scroll expanders, is also
presented.
Chapter 3 describes the computation of thermodynamics and flow process in a

scroll compressor-expander based on the work of N. Stosic. It also contains a
discussion of the results from the Quasi 1-D flow modeling. The importance of the
relationship between build-in volume ratio and build-in pressure ratio and the use
of the correct pressure ratio is clearly defined. The significant point is that the
build-in volume ratio is a constant for any given scroll compressor or expander
design but the pressure ratio is not, and is a function of both the gas being handled
and the rotational speed. The linking of a scroll expander to a scroll compressor in
a sealed 'Expressor' unit is also described in this chapter. This may be used both in
new systems and for retrofit applications in place of a throttle valve. The results of
computer simulations of Expressor performance in a large industrial chiller are
shown.
In the first part of chapter 4 we present the necessary equations needed for the
analysis. As will be shown in section 4.4, these equations form the basis for the
development of the grid generation code. The derivation of the equations was
necessary since it is not based on any publish data, however it is partially based on
the patent of the inventor of the machine. Full analysis of the geometry is given in
appendix A. Then follows a survey on the different methods of grid generation and
finally the design and development of a grid generation program that can handle
3
the complexity of the working volume of the scroll device is analyzed. Aspect
ratio, orthogonality, cell distortion, and smooth distribution, are some of the
difficult issues involved in the transient structured body fitted grid generation
package developed.
In chapter 5 we give the mathematical basis for a comprehensive general purpose

model of fluid flow and heat transfer from the basic principles of conservation of
mass, momentum and energy. This leads to the governing equations of fluid flow,
the so called Navier - Stokes equations. The governing equations are complex non
linear partial differential equations, which can be solved using numerical methods.
Discretization methods, advection schemes, time differencing schemes and
pressure / velocity coupling algorithms are discussed. Commercially available CFD
software packages are reviewed and CFX - FLOW3D is described in detail. Flows
inside scroll compressors - expanders are turbulent, time dependent, compressible
and for the case of the expander two phase. Furthermore these flows are driven by
the wall motion and, therefore, are characterized by moving boundaries. In the last
section of this chapter the results of the three dimensional modeling are shown.
Chapter 6 contains some concluding remarks and suggestions for possible future
research.
4
Chapter 2
Literature Survey
2.1 Introducflon
Since its invention by Perkins in 1834, there has been no major change in the
layout of vapour compression refrigeration plant. However, design aspects such as
refrigerating controls, amount of superheating, sub-cooling, etc. have been
modified over the years. All these modifications were related to production cost,
ambient conditions and other manufacturing related factors. The need to develop
more efficient compressors considering all aspects such as noise, vibration, weight,
efficiency as well as the losses due to throttling, still exists.
In this chapter we examine previous efforts to develop two-phase expanders and

compressors. Emphasis is placed on the work done on the scroll. It is also
demonstrated that the scroll device is the most appropriate machine for the
requirements of the present study. The existing, very limited amount of work that
has been done up to now on scroll expanders, is also presented.
2.2 Compressors
Compressors are generally divided into two different categories or types based on
the principle on which they operate. These types are positive displacement and
aerodynamic, the latter type is sometimes called dynamic. Figure 2-1 shows a chart
of common compressor types separated according to the above categories.
An area where confusion arises surprisingly often is in the specification of the

"duty" required from an industrial compressor. Different industries have different
conventions, some call for volume flow, others mass flow, and in the refrigeration
industry the "duty" is rated in the heat units absorbed by the refrigeration plant.
Over and above the basic "duty" specification, there is the question of
performance, tolerances and guarantees which can also be a subject of
misunderstandings.
The convention for rating compressors in the refrigeration industry is quite

Chapter 2 Literature Review
different from the convention used in the case of air, gas or chemical applications.
Compressors in the refrigeration industry are rated on the basis of their
refrigeration capacity at specific evaporating and condensing temperatures. This is
largely a function of mass flow.
Compressors
Positive displacement Aerodynamic
Reciprocating Rotary Centrifugal Axal Ejector
One shaft Two shaft Three haft
Piston Diaphragm Sliding Liquid Scroll Roots Twin Stepped Tooth Zmmern
vane ring screw rotor
Figure 2.1 Classification of common industrial compressors
22.1 Aerodynamic Compressors

In a turbine the working fluid enters at a high pressure and acquires increased
kinetic energy as it expands through a ring of fixed nozzles to a lower pressure.
The stream of fluid then undergoes a change of momentum as it flows trough
passages between blades attached to the rotor, and the component in a direction
tangential to the circle of rotation produces the output torque at the shaft.
This series of events is reversed in aerodynamic compressor. Input torque from

some external source imparts a change of momentum to the working passage
between the rotor blades. Having acquired an increased velocity, the fluid then
slows down with an accompanying rise of pressure while flowing through a ring of
fixed diffusers.
There are two main types of aerodynamic compressors, distinguished by whether

the flow through the machine is in the axial or radial direction. In these machines
the fluid undergoes a continuous steady-flow process and the speed of flow is very
high. For this reason comparatively small machines can handle large mass flows
and large work transfers, and the processes can be assumed to be adiabatic (Rogers
6
&Mayhew 1980).
2.2.1.1 Centrifugal Compressors

The first commercial centrifugal compressor used in refrigeration service was
promoted by Willis Carrier in 1920. Since then the centrifugal compressor has
become the dominant type of compressor in large installations.
Centrifugal compressors serve refrigeration systems in the range of 200 to 10,000

kW of refrigeration capacity. Evaporating temperatures in multistage machines
may extend downwards to -i00 C, although one of the major uses of the
compressor is the chilling of water to about 6° C to 8° C in air conditioning
systems.
Centrifugal compressors are similar in construction to centrifugal pumps in that the

incoming fluid enters the eye of the spinning impeller and is thrown by centrifugal
forces to the periphery of the impeller. Thus the blades of the impeller impart high
speed to the gas. From the impeller the gas flows either into diffuser blades or into
a volute, where gas momentum is lost and its pressure increases. The centrifugal
compressor may be manufactured with only one wheel if the pressure ratio is low,
although these kind of compressors are generally multistage. Centrifugal
compressors operate with adiabatic efficiencies of 70 to 80 percent.
2.2.2 Positive Displacement Compressors

In this type of machine gas is drawn into the compressor, it is trapped within it and
is then positively displaced out of it by the movement of one or more components.
The term "trapped" is used to indicate that the quantity of gas involved in the
compression processes, is not affected by either the suction or discharge cycles of
the machine. These types of compressor can develop at the discharge port exactly
that pressure which is necessary to discharge the gas from the machine, subject of
course to the power available and the physical constrains of the compressor design.
2.2.2.1 Reciprocating Compressors

Reciprocating compressors have been in existence for a very long time. The basic
principle of a piston or reciprocating compressor is that gas is drawn into a cylinder
by the movement of a piston. Originally, the piston's movement leaves a free space
and the gas flows into this to fill it. The piston then reverses the direction of its
7
movement and forces the gas back out of the cylinder. Valves are fitted to both
inlet and outlet which ensure that the gas enters and leaves by different ports.
From the action of the piston compressor it is clear that this device belongs to the
type of positive displacement machines. The air or gas is sucked in, compressed
and then pushed out, that is displaced by the movement of the piston. The higher
the pressure of the system into which the gas is to be discharged the greater the
force required to move the piston during the compression-discharge cycle and
hence the greater the input power of the device. Consequently, the power input to a
reciprocating compressor increases with its discharge pressure. The maximum
possible discharge pressure of a piston compressor is dictated by the available
power, subject to the mechanical and thermal limitations of the design, and is not
in any way affected by the speed.
By construction, at the end of the compression/discharge stroke, there is still some

space left between the piston and the top of the cylinder. This is the so called
clearance volume and the gas trapped in there is at discharge pressure. Due to this
pressure, if at the end of the compression/discharge stroke the force at the back of
the piston is released the gas in the clearance volume expands pushing the piston
backwards some distance until the forces on either sides of the piston become
equal. Two points arise from this situation.
Firstly, the expansion of the gas does useful work in moving the piston backwards,
so recovering much of the compression work put into the gas in the clearance
volume.
Secondly, no additional gas is drawn into the cylinder until the clearance volume
gas has expanded down to the suction pressure. After this point is reached, further
movement of the piston reduces the pressure in the cylinder below suction pressure
and suction flow past the inlet valve commences. Thus the volumetric efficiency of
a compressor with clearance volume is directly affected by the clearance volume.
2.2.2.2 Screw Compressors

The so called "twin" screw compressor is the type of compressor which has been
developed for the widest list of applications and has the largest range of available
sizes. Compared to some other types, for example piston compressors, the history
and therefore technology, of screw compressors is relatively young. The screw
8
Chapter 2 Lilerature Review
compressor was developed in the 1930s and first became popular for refrigeration
service in Europe in the 1960s.
The screw compressor consists essentially of two meshing rotors in a casing. The
male rotor drives the female rotor in a stationary housing. The refrigerant enters the
one end of the compressor at the top and leaves the other end at the bottom. At the
suction position of the compressor a void is created into which the inlet vapor
flows. Just before the point where the interlobe space leaves the inlet port, the
entire length of the cavity or gully is filled with gas. As the rotation continues the
trapped gas is moved circumferentially around the housing of the compressor.
Further rotation results in meshing of the male lobe with the female gully,
decreasing the volume in the cavity and compressing the gas. At a certain point into
the compression process the discharge port is uncovered and the compressed gas is
discharged by further meshing of the lobe and the gully. A major feature of this is
that there is zero clearance volume.
2.2.2.3 Scroll Compressors

After the 1970's energy crisis, conservation has been the most emphasized topic for
air conditioners. The demand for air conditioners with higher efficiency have been
required from the distributors to meet the customer's needs concerning energy
savings. Under such a situation, there came a strong demand for higher efficiency
compressors, because the improvement of the compressor efficiency is the most
economical method in spite of technological difficulties. Thus most compressors
manufactures have been eagerly improving conventional compressors. However,
some of them turned their attention to scroll devices.
The concept of scroll devices goes back to the beginning of this century. The scroll
machine was invented by Leon Greux, a French scientist in June 1905 and was
patented on 3rd October 1905. It is currently attracting the designers attention
because of its high efficiency and smooth operation. One of the main reasons why
scroll machines were not fully developed or substantially improved for such a long
period, was the lack of precise production techniques needed for the construction
of such machines. However designers started looking at these kind of devices
again in the early 1970's.
The scroll compressor was not commercially successful until 1976. This was
primarily because of sealing and wearing problems which placed severe limitations
on the efficiencies, operating life and pressure ratios attainable. Such sealing and
9
wear problems were of both radial and tangential types. However, since then a lot
of work is been done and many problems were overcome.
Scroll compressors have a lower level of noise arid vibration, than rolling piston
rotary compressors, and because of this, they are used extensively as air conditioner
refrigeration compressors.
Compared to the other types of compressors, the inherent advantages of the scroll
compressors are:
• fixed build-in volume ratio making it valveless and more reliable as well as
adaptable to variable speed,
• large inhaling hole and nearly continuous inhaling making it have small
pressure drop,
high volumetric efficiency,
. low noise.
However, designing and manufacturing this machine is very difficult. The early
1970's, research and development work on the scroll compressor has been carried
out steadily until now. Practical application however, was not achieved until
recently mainly for the following reasons
very high precision tooling is required for the production of the scroll
compressor,
• a complicated process is required to assemble the parts involved in the
compression mechanism.
2.2.2.3.1 Working Process
The main components of the scroll compressor are shown in Figs. 2.2a and 2.2b.
There are the fixed scroll, orbiting scroll, anti-rotation coupling, crankshaft and
crankcase. The outside suction port and the outside discharge port are in the fixed
scroll. The inside suction ports are located at the vane profile termination, the
inside discharge ports are located at the vane profile start. The area of the inside
suction ports varies periodically with the crankshaft rotation. The gas is inhaled
from the outside suction port then gets into the suction pockets tangentially through
the inside suction ports. After compression it leaves the displacement volume
tangentially through the inside discharge ports, then enters the central pockets.
Finally it leaves the compressor axially through the outside discharge port.
10
Dtse Sucton port
e Port
:horge port
Sictior Port
Figure 2.2a The ports
DPI VEI
SCROLL
AL
JOURNAL
- BEARINC
OF
'IOTOR
SHAFT
Figure 2.2b Scroll compressor rotating mechanism
11
The working process can be divided into five stages: steady suction, suction
closing, compression, initial discharge, and steady discharge. In the steady suction
stage, the displacement volume increases from zero to a maximum, the crank angle
range is broad and the suction time is long. In the suction closing stage, the volume
begins to decrease gradually, but the inside suction ports are not closed off, so the
gas returning phenomenon arises and the pressure of the closed pockets is higher
than the nominal suction pressure, this increases the volumetric efficiency. In the
compression stage, leakage has the main influence on the compressor efficiency. In
the initial discharged stage, the compressor starts to discharge forcibly, the gas
scurries through the inside discharge ports when the pocket pressure is different
from the central pockets. In the steady discharge stage, the gas leaves the
compressor but the effective area of the outside discharge port decreases because of
the orbiting scroll interference. So the velocity increases. A detailed survey of
scroll compressors follows.
2.2.2.3.2 Scroll Members
Conventional scroll members are mirror imaged pairs comprising a single

appropriately shaped spiral vane, of uniform thickness and height, protruding from
an end vane plate. Spirals of the involute type are most commonly used,
particularly the involute of a circle. The basic parameters defining the scrolls are
displayed on fig.2.3. They are the base circle radius, rb; orbiting radius, r 0 ; vane
height, H and vane thickness, T.
Vane thckness T
Wrap angle
Vane
Height
H
jose circie radius
Figure 2.3 Basic geometric parameters
2.2.2.3.3 Principle Of Operation
The members are appropriately phased, and eccentrically mated, so as to form

various pockets bounded by the vanes and their respective end plates, as shown in
12
fig.2.4 Most commonly one scroll is fixed and the other is made to orbit by a crank
mechanism. Through relative motion of the members the pockets, initially opened
to the surroundings, are first formed, sealed off, progressively moved inwards
while reducing in volume, and finally merged into a common discharged volume.
Conversely simultaneous rotation of the members about their respective geometric

centre yields an equivalent relative motion of members. However the first scheme
only was considered in the present study.
The inherent advantages of such a scroll machine are valveless compression

virtually constant volumetric efficiency, nearly continuous suction and discharge
flow, smooth operation, low relative velocity and potentially low unbalance.
Suction
gas
(a) (b) (c)
(d) (e) (f)
Figure 2.4 Working volumes in a scroll compressor
2.2.2.3.4 Analytical Model
To model the scroll suction compression and discharge process, descriptions of

volume in the pair of compression pockets and area open for suction and discharge
flow as functions of time are required. These geometric relationships are more
conveniently described as functions of scroll orbit angle, or crank angle.
13
The volume in a pair of scroll pockets for a typical scroll compressor is shown in
fig.2.5 from the start of suction (SOS) through the end of discharge (EOD). The
process from the start of compression (SOC) to the start of discharge (SOD) is
approximately a closed compression process; only leakage or porting for some
pneumatic operations such a back chamber, prevent it from being truly sealed.
Soc
0
>
C)
E
/ N _
1
SOS
Anqie
EOD
Figure 2.5 Volume in a pair of typical scroll pockets from SOS to EOD.
However as far as the "closed" process is concerned, by the time the inner wrap tips
open at SOD, the control volume in the pocket is no longer "closed", and the
problem of defining a boundary to the control volume in the open regions must be
resolved.
SUCTION PROCESS
The suction gas flow in a scroll compressor is driven by the opening and closing of
the pair suction pockets. The pair of suction pockets referred to are shown
crosshatched in fig.2.6 for a typical scroll compressor geometry. The volume in the
suction pockets, and the inlet cross - sectional area into the pockets varies with the
crank angle. During most of a scroll orbit cycle, the volume in the suction pockets
increases and causes gas to be pulled into the pockets. But near the end of the orbit
cycle, the volume in the pockets begins to decrease until they are abruptly closed
off at the end of one complete cycle. (Note that the end of the suction process
coincides with the start of the closed compression process)
14
To adequately model the scroll suction process, the flow dynamics into the control
volume of the suction pockets must be considered. The instantaneous mass of
suction gas contained within the suction pockets control volume can be defined by
the time differential continuity equation.
---(m) = ri (2.1)
Where : m, = mass of gas in suction pockets control volume

= mass of gas into or out of suction pockets
FL--'
Sc
Orbiting
scro
Figure 2.6 Typical pair of scroll suction pockets
The instantaneous mass flow rate of suction gas entering or leaving the control
volume can be described using the steady one-dimensional, isentropic flow
equation.
msp =C D pA Sp 2(h Up -h dO ) (2.2)
Where : C D = Discharge coefficient for flow into or out of suction pocket

p = Density of gas
A = Suction pocket cross sectional area
= Enthalpy of gas upstream or suction pocket inlet
h do = Enthalpy of gas downstream
15
Also, the first law of thermodynamics on a time rate basis is applied to the suction
pockets control volume.
v V (2.3)
[h +—+gz] 1 —,m OUt
[h+—+gz] out
2
Where: = Energy contain within the control volume
For all equations it is assumed that gas properties are uniform throughout each
defined region, e.g. properties are uniform throughout the suction pockets. In
applying the above equation it is reasonable to assume that the following terms are
negligible and can be eliminated: heat transfer, boundary work and kinetic and
potential energies. One might question why the boundary work term can be
neglected since the suction pockets volume changes considerably during each
cycle.
= - (2.4)
Where: Psp = Pressure in suction pockets

'sc = Pressure in suction chamber
Vp = Volume in pair of suction pockets
The reason is that the pressure difference across the suction pockets control volume
boundary is quite small, consequently the product in the work term is negligible
compared to the enthalpy terms. What is left then is the classic filling process
energy equation where it is assumed the time rate change of energy in the control
volume equals the time rate change of internal energy.
=mh (2.5)
Where: u, = Internal energy of gas in the control volume

To utilise eq. (2.2) for the mass flow rate into or out of the suction pockets the
conditions in the suction pockets and the suction chamber surrounding the orbiting
scroll must be known. Immediately after the start of the cycle, the volume in the
pocket is still quite small. During this early part of the suction process it can be
assumed that quasi - static filling of the suction pockets takes place so that at any
instant in time the volume in the pockets is completely filled with gas. This
assumption allows the analysis to be initiated as conditions in the suction pocket
can then be defined. With the conditions in the suction pockets and suction
16
chamber specified, the mass flow rate into the pockets can be computed. The
continuity equation then determines the instantaneous mass in the suction pockets
so that the conditions in the pockets can be defined for that time increment. This
process is continued until the end of the suction cycle.
What results from using this scroll suction model (Nieter J.F. & Gacne D.P. 1988)
is the ability to predict the pressure rise in the suction pockets that occurs as the
volume in the pockets is reduced near the end of the suction process. This increase
in suction pockets gas pressure indicates that early compression takes place before
the displacement volume of gas is sealed off at the start of the actual closed
compression process (S.O.C). Besides predicting the early compression of suction
gas, this suction process model can also produce a continuous function for
instantaneous mass flow rate at suction which is important for accurate prediction
of suction pressure pulsations in the manifold system.
COMPRESSION PROCESS
The compression process pertains to "closed" compression after the outer wrap tips
seal off at SOC and continues until the inner wrap tips open at the SOD. Farther
compression effectively continues after SOD, through the discharge process until
EOD. Therefore the following relationships that are used to model the compression
process are actually applied from SOC to EOD as well. For all process equations, it
is assumed that gas properties are uniform through each region, e.g. properties are
uniform throughout the compression and discharge pockets.
The instantaneous mass of gas contained within the control volume for a pair of
compression pockets can be described by the differential continuity equation.
= 0 0
(m-mout) (2.6)
dt
Where: m = Mass flow rate of gas in the volume
Leakage's are also included in the above equation, however a detail analysis about
leakage's follows later on in this chapter.
The gas state in the control volume during the compression process normally is
modeled by one of two approaches: either using the polytropic process, or using the
first law of thermodynamics (energy conservation). The polytropic process model
17
is quite often a good approximation to employ for displacement compressors. It

simply uses the relation:
1 ip
= p01i
_R_ (2.7)
P SOC
Where: Subscript C: control volume
The greatest difficulty with this model is obtaining an accurate value for the
polytropic exponent , a good approach is to measure II in a laboratory
experiment. However the first law of thermodynamics on a time rate basis can be
applied to the control volume using:
° [
I
v2
u )=Q-W+rn.
CC rh
+---+gz —m Out h++z] (2.8)
dt in[
Out
This approach is more difficult that the polytropic model. The general energy
equation given must be reduced to usable form, typically a differential equation of
gas temperature in the control volume. Further reduction of the general form
involves obtaining a number of partial differentials relating enthalpy and pressure
to temperature and specific volume. These are trivial for the ideal gas properties,
but for real gas properties are more difficult to compute, in such cases the
polytropic approach would be completely inappropriate. Finally, the greatest
obstacle using eq.(8) is in obtaining reasonably accurate values of heat transfer Q.
Models used for heat transfer in positive displacement compressors have been
fairly documented for reciprocating piston types, but much less so for other types,
and almost not at all for scroll compressors.
DISCHARGE PROCESS
The instantaneous mass flow rate of discharge gas exiting or back-flowing into the
control volume can be described using the steady, one dimensional isentropic flow
equation for a nozzle.
rnD (0) = p A C 2gd(h - h d O ) (2.9)

D D D
Where: A D = Discharge port area

CD = Appropriately chosen flow coefficient
18
As stated previously, the relationships used to model the state of gas in the control
volume during the discharge process are the same as those used during the "closed'
compression process.
Gas pulsation's in compressors manifolds have a significant effect upon suction and
discharge processes. Modeling this interaction between the discharge flow process
and the manifold pressure pulsation's can be accomplished in a number of ways.
Probably the most powerful and flexible of these approaches is the transfer matrix
which is performed in the frequency domain.
In the transfer matrix approach, pressure pulsation's are modelled by combining the
steady state acoustical impedance description of the manifold with an acoustical
source: the oscillatory gas flow in or out of the port. Since the mass flow rate
through the discharge port is a periodic function of time or crank angle, it can be
represented by a finite Fourier series.
LEAKAGES
Minimizing leakage from the compression chamber is a significant subject in any
type of compressor.
Compression chambers of the scroll compressor are the spaces configurationaly

formed between the fixed scroll wrap and the orbiting scroll wrap, so there occurs
clearance through the fine clearances from the higher pressure chamber to the
lower pressure chamber, as shown in fig.2.7. There are two kinds of leakage, one is
the radial flow through the axial clearance, and the other one is the tangential
leakage flow through radial clearance. In the scroll compressor, leakage from the
higher to the lower pressure pockets is the largest factor governing both the
volumetric and adiabatic efficiencies.
1-D APPROACH
Considering only the leakage between the tips and bases, flow rates that include
frictional effects are determined using (Gaillat J. L. 1988):
rn = A bt j 2 Lp p D h /(fL) (2.10)
Where: A bt = Leakage flow area between tips and bases

= Inter pocket pressure difference
19
= Upstream density
Dh = Hydraulic diameter
f = Friction factor - Moody
L = Inter pocket tip/base leakage path
To solve this equation iteration is required since the friction factor f, is dependent
on the Reynolds's number, which in turn is dependent upon mass.
Tip ciear Radial clearance
Lower Higher
pressur Pressure Higher
Lower
pocket ocket pressure pressure
pocket pocket
Tangential
leakage
Rodial leakage
Figure 2.7 Leakages in a scroll machine
Flank leakage is modeled as a converging nozzle. With a known pressure

difference and flank clearance, an assumption of isentropic steady-state
compressible flow yields.
r = KAf PUPU - Rp ] (2.11)
Where: K = Contraction coefficient - empirical

Af = Leakage flow area between vane flank surfaces
= Upstream pressure
R = Down stream to upstream pressure ratio
20
The deficiency of the 1-D inviscid model is that viscous effects, while important,
are neglected. On the other hand, purely 1-D viscous flow (stokes flow) ignores the
convection effect, which may be significant when the pressure gradient is large.
To fully understand the flow through clearances, an approach based on

computational fluid dynamics (CFD), made by Yong H. 1994 is presented here.
The aim was to calculate the detailed flow field, including velocity, pressure and
mass flow across the clearance and assessing the threshold as to when a quasi 1-D
model is sufficient enough to predict the mass flow. The set-up of the problem is
illustrated in fig.2.8 One dimensional models or quasi 1-D models can be a good
approximation only when the ratio /L is very small.
Ri H P2
Ti
Figure 2.8 Tip clearance geometry
2-D APPROACH
Two dimensional Navier Stokes equations are used to calculate the flow field. The
flow is assumed to be compressible, viscous and laminar. The ideal gas law is
employed as the equation of state. The system of partial differential equations
governing this type of flow is described as follows in non-dimensional form for
Cartesian co-ordinate.
1 Ev Fv1
_5_
(2.12)
a2=yi
Where: Q = [P, p, ps,, e]T

E=[ p2+p , u(e+p)]T
F = [p , Pu Pv2P v(e+p)]T
p is density, u and v are velocity components, e is total energy, P is pressure, a is

the speed of sound and y is the ratio of specific heats. Reynolds's number Re is
defined as: Re = p 1 a1 H/ji 1 . The viscous terms are:
21
0 1 r o 1
txx tyx (2.13)
E txy
=[
+ vt y + Pr'(Y - 1)_i a2 utxx + v txy + tp(y —l)'
=[ x]
1u
Where shear stress: = + etc.
and Pr is Prandtl number. The viscosity follows Sutherland's law:
r
J__L2iJ Too+110 (2.14)
[ -r]
jt 00 T+110
A finite volume method was used to discretize eq.(12) so that a system of algebraic
equations for five unknown variables (p, u, v, p, e) were derived. With the
appropriate upstream boundary conditions where total pressure and temperature
were specified and downstream conditions where static pressure was set, the
equations are integrated to steady state using Mac-Cormac method; which is
second order accurate. Upstream and downstream locations are defined as some
distance up and down the step (about 10 H).
2.3 Two Phase Expanders

A vapour compression refrigeration system is characterised by the direct two-phase
expansion of the entire flow, from the exit of the condenser to the entrance of the
evaporator. Fig.2.9 illustrates the cycle on a T-S diagram.
With reference to basic fundamental thermodynamic principles it can be stated that

this process (expansion) has the ability to produce a specific amount of work.
However since the fluid expands entirely in the two phase area, it is necessary to
design and develop an expander which could accept a two phase flow of liquid and
gas at reasonable efficiency and reliability.
22
T
2
NSER
ENSION PRESS ON
y2RATOR
Figure 2.9 Refrigeration cycle with isentropic expansion
Up to date a lot of work is been done to develop two phase expanders that could be
used to substitute throttle valve and recover some power from the expansion
process in a refrigeration cycle. Two phase expanders such as biphase turbines,
and screw expanders have been developed, however efforts to develop and produce
scroll expanders started recently, and not much information is available.
Instead of throttling the refrigerant to produce cold liquid, the refrigerant could be
expanded in a two phase expander. The work conventionally lost in the irreversible
throttling process could be recovered as shaft power, which could in turn assist in
driving a second compressor (or the main compressor ). The electricity
consumption for refrigeration could be reduced by approximately 10% also the
coefficient of performance will be increased by 13%. Apart from the above
application two phase expanders could be used in low grade heat recovery
(Trilateral Flash Cycle).
It is a well known fact that conventional turbines and reciprocators are not suitable
for the expansion of hot liquids in this manner since liquid droplets erode turbine
blades and reduce the aerodynamic efficiency of the turbine, while they wash the
lubricating oil off the cylinder walls of reciprocating expanders and so promote
wear and seizure of the mechanism. Therefore some alternative methods have been
investigated for this purpose. These are:
1. Scroll expander
2. Lysholm total flow expander
3. Specially designed two phase turbines
23
2.3.1 Scroll Expander

In recent years, scroll machinery has been put into practice as compressors, taking
advantage of their higher efficiency and their low noise/vibration. However the
scroll principle can be applied also to the expansion process.
Unfortunately not much work is been done on scroll expanders. However the very
small and limited amount of work that has been done is concerning only the single
phase flow.
In 1988 Yanagisawa et. al. published a paper (in Japanese) with the title "study on
fundamental performance of scroll expander". Equations of volume pressure and
output torque were derived. Moreover, practical losses which occur in the scroll
expander were analysed theoretically. It was also shown that the leakages through
axial clearances increase the flow rate of the expander and decrease the torque
greatly. An experimental scroll expander was also developed and showed
reasonable performance and recorded its maximum efficiency of 75%.
In 1992 Morishita et. al. used the scroll principle to develop a scroll expander to
use it in the Otto cycle spark ignition engine. The theory and experience of the
scroll compressor was extended to an engine with a scroll compressor and a scroll
expander. The scroll compressor compresses the air fuel mixture, and the scroll
expander is used to get power from the heat of the combustion. The flow is
continuous and uni-directional. Which is different from that of the reciprocating
engine. Although the Brayton cycle engine is possible with a scroll compressor and
expander, the temperature of the scroll expander could be extremely high due to
the continuous combustion. The requirement of the scroll material was becoming
very severe. The spark ignition engine was therefore studied in their report, based
on the air standard Otto cycle.
PRINCIPLE
The scroll engine Morishita recommends consists of two main components, the
scroll compressor and the scroll expander. The elements were assumed to have the
same geometric dimensions and to be synchronised via a proper mechanism. After
several degrees of rotation, the compressed air-fuel mixture discharge is to
commence. The volume of the combustion chamber (innermost) of the scroll
compressor is minimum at this angle, actually is zero. The discharge port of the
scroll compressor and the combustion chamber of the scroll expander were
24
connected via a discharge valve, the compressed air-fuel mixture is transferred to

the combustion chamber of the scroll expander. The volume change rate of the two
connected chambers is the same during the transfer process. The discharge valve is
then closed this corresponds to the end of the compression process.
The air fuel-mixture is ignited in the combustion chamber of the scroll expander.
The combustion is assumed to take place instantly. The pressure of the combustion
chamber of the scroll expander increases suddenly. The orbiting scroll of the
expander is therefore driven by the high pressure of the combustion gas. The gas
pressure decreases during expansion. This corresponds to the expansion process.
The expansion ends when the gas reaches the outermost chamber. The gas is then
exhausted from the expander.
Another paper on the scroll expanders was published by the Swiss Federal Institute
of Technology on 1994. That paper describes a 1 to 3.5 kW hermetic scroll
expander-generator modified from a standard hermetic scroll compressor, an
organic Rankine cycle test facility build to test up to 10 kW, and a set of
experimental results using HCF134a in the dry vapour domain. Peak overall
isentropic efficiencies in the range of 63% to 65% for speeds of rotation varying
2400 and 3600 rpm were reported.
2.3.2 Lysholm Total Flow Expander

The idea to use a Lysholm twin screw machine as a direct two-phase expander (of
geothermal brines) stems from the patents of Sprankle [1973][1976]. The Lysholm
design consists of a pair of parallel rotors with meshing helical lobes which form
compression or expansion chambers in the casing. This configuration was patented
in Sweden by Lysholm in 1934. The design has found its main use as a heavy duty
air compressor known for its reliability. The use of the Lysholm design as an
engine has been suggested at several times, including the suggestion for its use in
geothermal energy conservation. Lysholm [1967] had earlier proposed his
machines for dry gas expanders.
The Lysholm twin screw expander is a positive displacement machine which

operates by the direct expansion of fluids. With the aid of fig.2.l0 the principle of
operation of the expander can be described as follows. The pressurized fluid flows
through the throttle T and enters the high pressure pocket formed by the meshed
rotors, the rotor case bores and the case face. This pocket designated by A in the
25
figure, is mostly hidden by the rotor lobes, but it can be seen in the plan section
view. As the rotors turn, the pocket elongates, splits into a V, and moves away
from the inlet port to form the region designated by B. With continued rotation, the
V lengthens, expanding successively to C,D and E as the point of meshing of the
rotors appears to retreat from the expanding fluid. The expanded fluid at a low
pressure is then discharged into the exhaust port. Energy is extracted from the fluid
in the piocess. Engine geometry determines the theoretical volumetric expansion
ratio, which is the ratio of the final to the initial chamber volumes.
EXPANDING
ROTORS
Plan seciion view Side section view
Figure 2.10 Layout of Lysholm screw Expander
Testing of the Lyshoim engine has since been conducted by the Lawrence
Livermore Laboratory (LLL), by the Jet Propulsion Laboratory (JPL), and by the
department of Mechanical Engineering at University of California, Berkely. The
unit tested was an air compressor with 152.4 mm rotors modified by Sprankle. A
maximum efficiency of 55% was observed for a power output of 30 kW at 5000
rpm. Latterly a smaller version of this (130 mm rotors) was build from which a
maximum expander efficiency of 53% was obtained at 47 kWe output (Steidel
[19821). Despite extensive experimental work, there had been a general lack of
theoretical analysis on the Lyshoim screw expander. Steidel proposed an empirical
model using curve fitting. More recently Taniguchi et al [1988] developed an
analytical method for estimating the performance of a two phase screw expander.
The operation of a screw expander was simulated by the expansion of a fluid
between two pistons with appropriate leakage paths. Good agreement was
obtained against the experimental results from a prototype machine with a
26
diameter of 81.6 mm and a length of 135 mm. However, the various frictional
losses and the two phase interaction were not accounted in the procedure.
2.3.2.1 City University And Lyshoim Expander

Besides the work world wide, a research programme on two phase expanders was
initiated by Smith at City University, London beginning in late 1982. This was
intended primarily as a means of improving power recovery from geothermal
resources. The work concentrated on the development of twin-screw expanders
using suitable organic fluids operating in a closed cycle system.
A unique test rig was constructed and on it twin screw expanders attained adiabatic
efficiencies of more than 70% at power as low as 25 kW when operating in the
two phase mode. Ii addition to the experimental work, a programme of analytical
studies was developed from earlier work on the computer modeling of screw
compressor performance carried out at the university of Sarajevo by Stosic.
Smith and Stosic concluded from their investigation that the failure of the earlier
investigators to get good results was entirely due to the design of such machines
with inappropriate build in volume ratios. However, they also examine the
possibility to replace the throttle valve in the refrigeration cycle with a screw
expander. They obtained an overall adiabatic efficiency for the expander, including
- bearing, seal and timing gear losses, to be of the order of 70% over a large range of
operating conditions.
2.3.3 Two Phase Turbines

During the same period, work was initiated at JPL and LLL to seek an alternative
efficient two phase expander system aimed at overcoming the restrictions imposed
by the positive displacement type machines. Austin [1973] and his geothermal
group at the LLL proposed a two-phase impulse turbine for geothermal application
in which the two phase brine from geothermal wells would be passed though the
turbine. It was hoped that the water drops would be small enough for the steam
and water to behave as a homogeneous fluid and give these turbines efficiencies
comparable to those of vapour turbines. However the detail calculations on drop
sizes and trajectory showed that the liquid drops in two phase turbines would be
too large to follow the gas phase, and that the liquid would impinge on the blades
and form a thin liquid film. The friction drag of this film would be so high that the
27
efficiency of single-stage impulse turbines would be limited to little more than

50%.
As a result of low efficiencies achieved by two phase impulse turbines a rotary

separator turbine was developed by Biphase Energy Systems [1982]. The basic
idea was to replace the throttle valve with a separator system in the traditional
flashed steam cycle system. In place of the throttle valve, the geothermal fluid is
expanded in a two phase nozzle with an adiabatic efficiency of around 90%. The
high velocity liquid and vapour so formed are then separated in a rotary drum. The
steam is then passed to the steam turbine, while the liquid is either used in a liquid
turbine to produce additional power or pressurised in a diffuser to reduce the re
injection power. This rotary separator turbine was investigated and tested in the
laboratory and at three geothermal resources. The machines were reported to
operate well and a power gain of some 20 -25% more than that from a single flash
system was achieved. A maximum adiabatic efficiency of the separator system of
35% was measured although it was predicted that improved designs with up to
50% efficiency were attainable in the future. It was thought that the impact and
other losses in the rotary separator would limit the turbine efficiency as severely as
the liquid friction in blade turbines.
Further work on a specially designed two-phase turbine was attempt by Elliot

[1982] at the JPL using either steam for total flow or organic fluids. The two phase
turbine utilised a uniform mixture of liquid and gas accelerated in a two-phase
nozzle which is similar in design to that used in the rotary separator turbine. The
mixture acts directly on an axial flow or tangential impulse turbine. Based on this
concept he predicted an upper limit efficiency of about 65% for two phase
turbines, and probably only attainable with organic working fluids. This prediction
could not be justified by the extensive experimental program carried out on both
JPL and LLL, which showed that this efficiency was not attainable and a
maximum efficiency of only 55% was achieved using organic fluids. However it
was found that the JPL two-phase nozzle study calculated velocities which are
within 2 - 4 percent of measured values and the nozzle showed good performance
over a wide range of operating conditions.
Carrier [1994] introduced in their chiller plant a device using a high efficiency two
phase nozzle, followed by separation of the liquid and vapour components for
independent power recovery. This biphase turbine is introduced as a throttle valve
replacement. The adiabatic efficiency recorded is 50% and the turbine is linked
28
directly to the main compressor to reduce the power output to the electric motor.
Gains in COP of 3 - 5% are stated to be possible by this means. See figure 2.11.
BPHASE
TURBINE
Figure 2.11 Refrigeration cycle, Carrier
As it was said above, Carrier has linked the expander with the compressor and
therefore the expander run at the same speed as the compressor. This might result
in a not so good working performance of the expander for the case where the
system will run at off design conditions. This thesis uses a completely different
way to recover the work from the expander.
2.4 Conclusions
Since the expansion process in the vapour compression cycle occurs entirely in the
two-phase region, a mechanical expander which can operate with two phase
mixtures and operate at high efficiency is the most important part of this study,
because the success of the whole project depends upon this. As it can be seen from
the literature review the need to design and develop a machine like this still exists.
As is the case for all rotary volumetric principles, reverse motion is inherently
possible and, therefore the conversion of scroll compressor to scroll expander can
be envisaged. Smiths investigation for this type of machine indicated that the
machines most applicable for the expansion of a two phase flow are the Lyshoim
twin screw expander and the scroll expander.
29
Scroll expanders present definite advantages over potential volumetric machines.

In addition, the reduced number of parts, the ability to cope with two phase
working fluids and the inherent reliability proven in compressor mode are major
reasons for its wider use in energy recovery. The fast growing mass production of
such units as compressors is likely to also contribute to a lowering of the
production costs for expander applications.
TWO PHASE
EXPANDER
Figure 2.12 The proposed cycle
What is recommended by this project is instead of coupling the expander with the
main compressor, introduce a second smaller compressor that will run in parallel
with the main one. All the work that will be produced from the expander in terms
of shaft power will be used to drive this new compressor which in turn will
contribute to the mass flow rate. This design appears to be more complicated than
the one Carrier uses, but the main reason is the performance of the expander at off
design conditions. The cycle is illustrated above (figure 2.12).
30
Chapter 3
Quasi One Dimensional
Flow Modeling
3.1 Introduction
In this chapter we describe the computation of thermodynamics and flow process

in a scroll compressor and a scroll expander with reference to Stosic's work. This
chapter discusses also the results from the Quasi 1-D flow modeling. The
importance of the relationship between built-in volume ratio and pressure ratio and
the use of the correct pressure ratio is clearly defined. The significant point is that
the built-in volume ratio is a constant for any given scroll compressor or expander
design but the pressure ratio, which is a function of the gas being compressed or
expand is not. Speed also affects the relationship between pressure and volume
ratios.
Therefore, while it would be convenient to identify a compressor - expander by its

build-in pressure ratio since this is the basis of the correct selection for efficiency,
it is not practical to do so unless the compressor - expander is always going to
operate on only one gas. This is the case with air compressors so that typically
these use the pressure ratio for identification, whereas gas or refrigeration
compressors normally use the built in volume ratio. The feature of the "built-in
volume ratio" has been treated on the basis that it is a constant value in scroll
machinery. While this is true in the majority of cases, it is possible to incorporate
into the compressor - expander features which enable it to be varied. (such as
controlling the opening - closing of the discharge port)
The linking of a scroll expander to a scroll compressor in a sealed 'Expressor' unit

is also described in this chapter. This may be used both in new systems and for
retrofit applications in place of a throttle valve. The results of computer
simulations of Expressor performance in a large industrial chiller are shown.
Chapter 3 Quasi One Dimensional Row Modefing
3.2 Theory
The equation set consists of the conservation equations for energy and mass
continuity together with a number of algebraic equations defining various
phenomena, which accompany the fluid suction, compression and discharge. The
mathematical model employs the differential kinematics relationship which
describe3 the instantaneous operating volume and its change with the rotational
angle, as well as the equations of conservation of the mass and the energy for the
adopted control volume. These are applied for each phase of the process that the
fluid is subjected to: suction, compression and discharge. The solution of the
equation set is performed numerically by employing the Runge-Kutta 4th order
method.
The model takes also into account the following effects:

• The working fluid can be any ideal or real gas or liquid gas mixture of known
properties (internal thermal energy , enthalpy).
• Inevitable leakage of the working medium which occur at any stage of the
process through axial or radial clearances is taken into account.
• Account can be taken of the effect of liquid such as oil, water or refrigerant
injecting during suction, expansion or discharge for lubricating, sealing or
cooling purposes.
3.2.1 Assumptions
Some assumptions need to be made in order to ensure an efficient computation,
these are:
• The fluid flow is assumed to be quasi-one dimensional.
• Gas or gas-liquid mixture inflow or outflow through the compressor suction or
discharge port was assumed isentropic.
• The leakage of the fluid through the clearances is assumed to be adiabatic.
3.2.2 Conservation Equations

The main difference of this model (comparing with other models) is the use of the
non-steady flow rather than the steady flow energy equation. Stosic proved that
this practice is computationally beneficial in evaluating the properties of real
fluids, as compared with the conventional methods using enthalpy as the
dependent variable.
32
Non-Steady Flow Energy Equation: =rnhJ_(rnh)+Q+W (3.1)

[JV
Where: 0 Crank angle.
h = h(0) Specific enthalpy.
in = m(0) Mass flow rate.
Q = Q() Heat transfer between fluid and sarroundings.

dv
W = P -i-- Thermodynamic work supplied to the gas.
ENERGY GAIN DUE TO THE GAS INFLOW [mh]

The energy gain due to the gas inflow into the working chamber is represented by
the product of the mass intake, and its average enthalpy. As such the energy inflow
varies with the crank angle. During the suction period gas enters the working
volume bringing the average gas enthalpy which dominates in the suction
chamber. While this process is taking place, a certain amount of the compressed
gas leaks into the working chamber through the axial and radial clearances. The
mass flow rate of the leaking gas as well as its enthalpy are determined on the
basis of the gas leakage equations.
Total inflow enthalpy is also affected by the amount of enthalpy flow into the
working chamber in the case of oil injection. Therefore if oil is injected into the
working chamber, the oil mass inflow together with its enthalpy must be included
in the inflow terms.
The mass inflow rate consists of:

m.
in =msuc+ml.g+m.oil (3.2)
Where rn Suction mass flow rate.
rn = Mass flow rate due to leakage gain.

m 01 = Mass flow rate due to the injection of oil.
The velocity of the oil depends on the amount of oil injected. The velocity of the
leaking gas will be analysed later on this chapter. Now as far as the suction fluid
velocity is concerned the following equation is used:
(0 = - h1 ) (3.3)
33
Chapter 3 Quasi One Dimensional Row Modeling
Where t is the suction coefficient, assumed to take a value of t =0.6 While "1"
and '2" denote conditions upstream and downstream of the considered port.
The fluid total inflow component consists of the following components:

(mhj =m,uc h suc )+(rn i.g h i.g J+(rn oji hoi) (3.4)
ENERGY GAIN DUE TO GAS OUTFLOW [mh]
The energy loss due to the gas outflow from the working volume is defined by the
product of the mass outflow and its average gas enthalpy. During the compression
this is the gas which leaks through the clearances from the working chamber into
the neighbouring chambers at a lower pressure, where as in the case of delivery
this is the gas entering the discharge plenum.
The mass outflow rate consists of:
m 0 = m d S + m11 (3.5)
where m di, = Discharge flow rate.
= Mass flow rate due to leakage loss.
The discharge fluid velocity can be calculated by the use of equation 3.3 but with
the appropriate discharge coefficient p.. Therefore the fluid total outflow enthalpy
consists of:
(mhj _.jmdS hdiS)+1m11 h 11 J (3.6)
Each of the mass flow rates satisfies the continuity equation:
=V pA (3.7)
Where: V [mis] denotes fluid velocity

p [kg/rn3 ] fluid density
A [m2] the flow cross section area and
The mass continuity equation of the model is:
34
Chapter 3 Quasi One Dimensional F/ow Modeling
0
rdml
= m. —m 0
(3.8)
HEAT TRANSFER [Q]
The heat exchange between the fluid and the compressor/expander tip and casing
(and through it to the surrounding) due to the difference in temperatures of gas and
casing and spiral surfaces is accounted for by the heat transfer coefficient
evaluated from the expression:
Nu = 0.023 Re° 8
(3.9)
Here the surface over which the heat is exchanged, as well as the wall temperature
depend on the crank angle. The characteristic velocity for Re number is computed
from the local mass flow and the cross sectional area.
WORK SUPPLIED [WI
The thermodynamic work supplied to the gas during the compression process is
represented by:
W =P1 (3.10)
dt
Where : P = Pressure in control volume

dv/dt Working volume
3.2.3 Gas Leakage

The leakages in a scroll machine affect the total flow rate and therefore play an
important role because they influence the process by affecting the mass flow rate
(and therefore the volumetric efficiency), and the work efficiency. Two types of
leakages according to their direction with respect to the working chamber exists:
gain and loss leakages. The loss leakage leave the working chamber towards
neighbouring chamber with lower pressure. On the other hand gain leakages come
from the neighbouring working chamber which has a higher pressure.
The fluid flow through the clearance is taken into account for the computation of
the leakage velocity. The process is essentially adiabatic (Fanno flow). The present
model treats only the gas, no attempt is made to account for leakages of gas-liquid
35
mixture. The clearance gap is assumed to have a rectangular shape and the mass
flow of the leaking fluid is expressed by the continuity equation:
=IL1V 1P1A g =1/hA g (3.11)
where: = V p -The mass velocity of the leaking gas

Ag = 1g cg -The clearance ("gap") cross sectional area
-The leakage clearance length (sealing line)
ög -The leakage clearance width ("gap")
ji (Re,Ma) -The "discharge" coefficient.
The computation of the leaking gas velocity follows from the momentum equation
which accounts for the fluid wall friction:
2
dp
dx
COdO) +— +f ---- (3.12)
I 2Deg
Where: f(Re,Ma) is the friction coefficient (dependent on Re and Mach number)

Deg is the effective diameter of the clearance gas (Deg 2g)
dx is the length increment
From the continuity equation (and assuming that T constant to eliminate gas
density in terms of pressure) the equation can be integrated in terms of pressure
from high pressure side (position "2") to the lower pressure side (position "I ") of
the gap to yield:
1 D2 D2
YI[d] = W IPI (3.13)
= RT2[C+2;n(%)]
-s--- characterizes the flow resistance (3.14)

Where = +
Deg
Lg = clearance length in the leaking flow direction
f = friction factor and
= local resistance coefficient
can be evaluated for each clearance gap in function of its dimensions and shape
and flow Re and Ma numbers. The full procedure would require incorporation into
model of the friction and drug coefficients in terms of Re and Ma number for each
36
Chapter 3 Quasi One Dimensional Flow Modeling
type of clearance. However Stosic recommends a value of 0.6.
3.2.4 Oil Injection

Injection of oil or other liquids for lubrication, or sealing purposes, modifies
substantially the thermodynamic process in a scroll machine. A convenient
parameter to define the injected oil mass flow is the oil to gas mass ratio, mojl/mgas
In addition to lubrication major purpose for injecting oil is the cooling of the gas.
In the present model the heat exchange with the gas is determined from the
differential equation for the instantaneous heat transfer between the surrounding
gas and an oil droplet. Assuming that the droplets retain a spherical form, with a
prescribed Sauter mean diameter ds, the heat exchange between the droplet and
gas can be expressed in terms of a simple cooling law:
Q 0 = h ø A ø (Tg1 - T0 ) (3.15)
Where: A0 is the droplet surface A0 = ( ds2 it)

ds2 is the mean Sauter diameter of the droplet and
h0 is the heat transfer coefficient determined from the empirical expression
Nu 2+0.6 Re° 6 Pr° 33 (3.16)
The exchanged heat must balance the rate of change of heat taken or given away
by the droplet in the unit time.
dT
= mC 11 = m0C0. (3.17)
dØ
Where Coil is the oil specific heat and the index 'o" denotes oil droplet. The rate
of change of oil droplet temperature can be expressed now as:
di!;= hoAo(Tgas_To)
(3.18)
dØ m0C01w
A simplified integration of the equation between the two time (angle) steps yields
the new oil droplet temperature at each new time (angle) step:
37
= Tgas + kTOb
T° (3.19)
1+k
Where: Tob is the oil droplet temperature at the previous time step k is the non-
dimensional time constant of the droplet, k = 'rw / A4 with t = m0 Coil / h 0 A0
being the real time constant of the droplet. For the given mean Sauter diameter ds
the non-dimensional time constant takes a form:
= m00011w =
k (3.20)
h 0 A 0 zlØ 6 h 0 LIØ
The obtained droplet temperature is further assumed to represent the average

temperature of the oil, i.e. Toil T0 , which is further used to compute the
enthalpy of the gas-oil mixture.
Because the inclusion of a complete model of droplet dynamics would

unnecessarily complicate the computer code, Stosic developed and used the above
described simplified approach. The inputs parameters are only the mean Sauter
diameter of the oil droplets, ds and oil properties - density, viscosity and specific
heat.
3.2.5 Computation Of Fluid Properties

For the ideal gas, the internal (thermal) energy of the gas oil mixture is given by:
=
= mRTgas +(mT).1 Pv +(mcvT) (3.21)
y-1 '—i
Where R is the gas constant. The pressure and temperature of the fluid in the
working chamber can be explicitly calculated by the help of the equation for oil.
(y-1)(1+k)u(mT).1 (3.22)
Temperature: T = (1 +
k)mR + (m ).
PROPERTIES OF REAL GASES

For the case of a real gas the situation is more complex. Many gases differ in their
properties from ideal gases. The above theory is based on the compression of ideal
gases. In the application of industrial compressors allowances have to be made
where possible for the differences between real and ideal gases.
Real gases do not behave exactly in the manner of ideal gases, and in some
38
instances the differences can be quite significant, particularly when the pressure
and temperature are in the region of the critical conditions.
Accurate equations of state are inconvenient for everyday calculations because they
lead to unwieldy expressions. They are most used for the preparation of tables and
diagrams of properties. When tables are not available, and for conditions where the
perfect gas laws are too inaccurate, correction factors may be applied to the perfect
gas relations. Thus the equation of state, for example, can be written as
z=-p-v-- (3.23)
RT
where Z is the compressibility factor. The value of this factor depends upon the
gas and it is also a function of the pressure and temperature. The combination of
Z çb(p, T) and the above equation is the complete characteristic equation.
3.3 Results of the 1-D Flow Modeling
As it was stated earlier in this study the Quasi One Dimensional Flow Modeling
was performed with the aid of an advance commercial package called
"SCORPA TH", written by Professor N. Stosic. The performance characteristics
over a wide range of speed and other operating conditions have been analysed.
3.3.1 Scroll Compressor

The function of the compressor is to raise the pressure of the vapour leaving the
evaporator to a value at which its corresponding saturated vapour pressure is high
enough to enable heat to be rejected to its surroundings by means of a condenser.
Because in a scroll compressor the opening of the delivery port is determined

entirely by the geometry of the machine, there is a certain build - in volume ratio
and a corresponding build in pressure ratio for a particular gas. If the pressure in
the working chamber has not reached the pressure in the delivery pipe when the
port is uncovered, gas flows back into the working chamber. On the other hand, if
the pressure inside the working chamber at the instant of opening exceeds that of
the delivery pipe, gas rushes out suddenly at an unresisted expansion. In either
event there is a loss of efficiency.
39
Build In Volume Ratio 2.5:1 Pressure Ratio 5:1

2.OE+06
1 .8E+06
1.6E+06
E 1.4+06
1.2+O6
i.OE-i-06
U,
8.OE+05
6.0+05
4.OE+05
2.OE+05 iii iii ii liii
0.0000 0.0003 0.0005 0.0008 0.0010 0.0013
Volume (m'3)
Figure 3.1 Pressure - Volume diagram for condition where system pressure ratio is
much greater than build in Volume ratio. Speed 3000 RPM. Working fluid R134A,
dry vapour.
Ideally the build - in pressure ratio should match the actual pressure ratio for
maximum efficiency but this is not possible in practice because a compressor in
any given installation may have to operate at different pressure ratios, depending
on the evaporating and condensing temperatures. Nevertheless there is usually one
pressure ratio which corresponds to average or design conditions and the machine
can be designed for this.
In figures 3.1 to 3.4, the pressure-volume and pressure-angle diagrams

corresponding to the conditions of greater and smaller pressure ratio, than the build
in volume ratio are shown. The dramatic increase in scroll pocket pressure shortly
after the start of discharge due to the strong back flow is apparent in figures 3.1 and
3.2. This back-flow effect results in an increase in compressor power over what
would be required if the compressor pressure ratio matched the system. The area
located under the curve and above the dashed line in figure 3.1 represents this
increase in compression power. The total compression power is proportional to the
enclosed pressure-volume area.
40
Build In Volume Ratio 2.5:1 Pressure Ratio 5:1

1.8E+06
1.6E^06
' 1.4E-i-06
1.2E-i-06
1.OE-i-06
8.0E--O5
G)
6.OE-i-05
4.OE+05
2.OE-i-05 - liii liii -I-I II liii liii
-400 -200 0 200 400 600
Angle (Degrees)
Figure 3.2 Pressure - Angle diagram for condition where system pressure ratio is
much greater than build in volume ratio. Speed 3000 RPM. Working fluid R134A
dry vapour.
In figures 3.3 and 3.4, the pocket pressure after the start of discharge increases
more gradually due to normal flow and over-pressure before peaking and then
slowly decreasing back to discharge pressure. There is no increase in power due to
back flow, only the increase in power due to over-pressure. Generally speaking the
area between the discharge pocket curve and the discharge pressure curve for both
cases ( greater-lower pressure ratios) represents the increase in compression power
due to over-pressure.
As was said above there is an average pressure ratio that corresponds to the build in
volume ratio of the machine. Figures 3.5 and 3.6 illustrates this. As it can be seen
from these figures there is not any major decrease or increase on the pressure after
the discharge port is open. This is because there is more pressure equilibrium
between the discharge pipe and the chamber.
During the suction period which is quite long some fluctuations on the pressure
were recorded, this is due to the fact that some gas might flow backwards due to
the pressure difference caused from the opening of the suction port. Now as far as
the compression is concerned figures 3.5 and 3.6 shows that the process which
takes long appears to be very smooth. This is because the momentum term pV 2 is
small since gas density is not large.
41
Chapter 3 Quasi One Dimensional F/ow Modellng
Build In Volume Ratio 2.5:1 Pressure Ratio 1.5:1

1.OE+06
9.OE+05
8.OE+05
. 7.0ErO5
z
c 6.OE+05
-
5.OE+05
G)
4.OE+05
3.OE+05 _________ _______ _________ _________
2.OE+05 iii ill Il-f-- Ill liii

0.0000 0.0003 0.0005 0.0008 0.0010 0.0013
Volume (mA3)
Figure 3.3 Pressure - Volume diagram for condition where system pressure ratio
is less than build in Volume ratio. Speed 3000 RPM. Working fluid R134A dry
vapour.

1.OE+06
9.OE+05
8.0E+05
7.OE+05
z
w 6.OE+05
5.OE+05
- 4.OE^05
3.OE+05 _________
2.OE+05 iii iii iii iii iii

-400 -200 0 200 400 600
Angle (Degrees)
less than build in Volume ratio. Speed 3000 RPM. Working fluid R134A dry
vapour.
42

1 .2E^06
1 .OE+06
c.1
8.OE+05
6.OE+05
C,)
w
4.OE^05
___-__
2.OE+05 liii liii I i-+-- liii iii
O.00E+OO 2.50E-04 5.00E-04 7.50E-04 1 .00E-03 1 .25E-03

Volume (m"3)
Figure 3.5 Pressure - Volume diagram for condition where system pressure ratio is
the same with the build in Volume ratio. Speed 3000 RPM. Working fluid R134A
dry vapour.

1 .OE+06
8.OE+05 ///•
__
E
i
I-
6.OE^05
C,)
C,)
Q)
- 4.OE+05
2.OE+05 II liii I I II I II I
-400 -200 0 200 400 600

Angle (Degrees)
the same with the build in Volume ratio. Speed 3000 RPM. Working fluid
R134A dry vapour.
43
The relationship between build-in volume ratio and build-in pressure ratio and
individual gas properties is also important. The physical situation is that a scroll
compressor has a built-in volume ratio. It draws in volume of gas V 1 and reduces
this internally to a smaller volume V 2. The build-in volume ratio is therefore the
ratio of these, i.e.
v-v1
R- V2
However, we have already seen that the optimum efficiency occurs when the
pressure ratio is approximately the same with the built-in volume ratio. The
significant point is that the build-in volume ratio is a constant for any given
compressor design but pressure ratio is not. Z is a function of the gas being
handled. For an isentropic compression process we have
P2 j'1
Pl t\V2)
where y is the isentropic index of the gas concerned, i.e. C//CV . The isentropic
index varies quite widely for different gases, typical examples at atmospheric
pressure and temperature being: [O'Neill, P. A. 1993 Industrial compressors pp.
309]
Fluid I
Propane 1.14
Ammonia 1.29
Air 1.4
Helium 1.63
Thus if a compressor has a build-in volume ratio of 3.5:1 the build in pressure ratio
with the above gases would be:
Fluid Pressure ratio

Propane 4.17
Ammonia 5.03
A ir 5.77
Helium 7.71
If however, we are dealing with a polytropic process then the equation will become
P2 =(v117
P1 v2)
and the pressure ratio will depend on the value of i. Raymon, L., et. al. 1988
44
Chapter 3 Quasi One Dimensxw/Fk,w Modeling
Build In VolunE Ratio L5:1 Speed 3000 RPM

I .60E+
- Pressure Ratio 1.5: 1
I.40E+06
1.20E-'-06
Pressure Ratio 2.6: 1
I.00E+06
ci 8.00E+05
-Pressure Ratio 3.5: 1
m
6.00E+05
4.00E+05
ZOOE+05
0.00E+00 ------- -+--i--+--i-- I I I I --+--+--+- -i--+--

0.00E+00 250E-04 5.00E-04 7.50E-04 1.00E-03 1.25E-03
Volune (MA3)
Figure 3.7 Pressure - Volume diagram for different pressure ratios. Working fluid
R134A dry vapour.
Build In Volume Ratio 2.5:1 3000 RPM

I .60E+06
1.40E+06
I .20E-I-06 Pressure Ratio 3.5: 1
1.00E+06 Pressure Ralio3.O:1

8.00E+05
.00E+05
-Pressure Ratio 1.5: 1
4.00E+05
2.00E+05
0.00E+00 ----+---+---+---i-- I I -f---
-400 -200 0 200 400 600

Angle (Degrees)
Figure 3.8 Pressure - Angle diagram for different pressure ratios. Working fluid
R134A dry vapour.
45
Figures 3.7 and 3.8 shows influence of operating pressure ratio, on the pressure
curve. Under the ideal condition the discharge pressure changes in a stepwise
manner at the start of the discharge process. At higher pressure ratio operation, the
discharge pressure rises smoothly as an extension of the compression process and
the over- compression becomes small as the pressure ratio increases. On the other
hand over-compression becomes very large at an operating pressure ratio lower
than the build-in pressure ratio.
Figure 3.9 shows relationships between adiabatic and volumetric efficiencies for
different pressure ratios. The pressure ratio at which adiabatic efficiency peaks
depends on the built-in volume ratio as noted earlier. An optimum operating
pressure ratio moves to low pressure ratio with increasing leakage. The reason is
that the leakage loss becomes large as the operating ratio increases. The
characteristic of volumetric efficiency versus pressure ratio is quite different. The
back leakage increases a little with increasing pressure ratio but only to a small
degree. There are also leakage and fluid injection effects which greatly affect the
choice of volume ratio for a given pressure rise.
Idiabatic Efficiency Volum etric Efficiency 1

100
90
80
70
(I
60
so
40
30
0 1 2 3 4 5
Pressure Ratio
Figure 3.9 Efficiencies - Pressure Ratio. Speed 3000 RPM, Build In Volume
Ratio 2.5:1. Working fluid R134A dry vapour.
46
As the scroll compressor is a positive displacement machine, the discharge

pressure is not dependent on compressor speed, however Rotational Speed can be
used to control the capacity of the machine. Ideally this would be true only if there
is no leakage. Figure 3.10 shows the influence of the rotational speed on the
pressure curve. As the rotational speed increases, the discharge pressure increases.
On the other hand, pressure during suction process becomes slightly low but
pressure during compression is almost the same. This is due to the leakage, which
is influenced by the rotational speed. Leakage takes place between the radial
(flank) and tangential (tip) clearance. Both kinds of leakage raise the delivery
temperature, because the vapour which is throttle back to a lower pressure is
subsequently recompressed adiabatically. A sudden rise in delivery temperature is a
sign of excessive leakage. The presence of oil is an important factor in achieving
good sealing
The influence of leakage on the scroll compressor performance is mainly indicated

in the compression power and the volumetric efficiency, which is the ratio of the
volume flow rate to the theoretical delivery. The leakage through tip clearance,
increase the pressure during the compression process but lessens somewhat the
Build In Volume Ratio 2.5:1

1 .20E+06
5000 RPM
—4000 RPM
1 .3000 RPM
1 .00E+06
H -2000 RPM
8.00E+05
N
C
z
6.00E+05
U,
U,
4
0.
4.00E+05
2.00E+05
0.00E+00
-400 -200 0 200 400 600
Angle (Degrees)
Figure 3.10 Pressure - Angle diagram for different rotational speed. Working
fluid R134A dry vapour.
47
over-compression during the discharge process. On the other hand, the leakage
through flank clearance, accelerates the pressure rise during a later part of the
compression process and increases the succeeding over-compression. When both
the leakages are considered, the compression pressure increases.
Figure 3.11 shows that the volumetric efficiency increases with increasing
rotational speed though it decreases at extremely high speed operation. Figure 3.12
shows that power and mass flow rate also increase with speed. Figures 3.13 and
3.14 shows how the mass flow rate and the power of the compressor change when
system is working at off design conditions.
Increasing of rotational speed increases an over-compression loss at low pressure

ratio operation but mocleTates a toss t. kg,b. we atic
operation, which leads to a higher optimum operating pressure ratio. Therefore,
dimensions of scroll wrap must be selected such as it has a maximum efficiency at
an actual operating pressure condition by taking into account of rotational speed
and leakage.
Most refrigeration systems reject heat to the atmosphere, and the ambient
conditions change throughout the year. Process refrigeration plants that operate
year round are particularly subjected to a wide range of condensing temperatures.
The response of a scroll compressor to changes in condensing temperature is
analyzed here. Firstly the behavior of the machine was examined by varying the
condensation temperature and keeping fixed the evaporating temperature. This
forces the compressor to operate at off design conditions, as far as the operating
pressure ratio is concern.
Figure 3.15 shows the changes in the adiabatic and volumetric efficiency for a
scroll compressor as a function of condensing temperature, for an evaporating
temperature of -1OC. As the temperature increases the volumetric efficiency drops
off. On the same figure the increase of the pressure ratio due to the increase of the
condensation temperature, and therefore pressure is shown. The remaining
important characteristic is the power, shown on Figure 3.16. The work of
compression in kJ/Kg increases as the condensing temperature increases.
48
C/pter 3 Quasi One DimensArzaI Flow Fi*de/ing
Aabatic Effiency Vumetric Ethciencyl

100
80
.- 60
0
a,
0
40
20
0 -t--
0 1000 2000 3000 4000 5000 6000

Speed (RPM)
Figure 3.11 Efficiencies - Speed. Working fluid R134A dry vapour.
power Mass Flow Ra
140 3
120
2.5
100
., a,
a,
80
60
40
20
0 0
0 1000 2000 3000 4000 5000 6000
Speed (RPM)
Figure 3.12 Power I Mass Flow Rate - Speed. Working fluid R134A dry vapour.
49
Chapter 3 Quasi One Dimensional F.bw Modeling
3 - ______________ ______________ ______________ ______________ 1 00 %
II1
2 .5
_ 7__
7
1iII_______ II
_
ii
0 1000 2000 3000 4000 5000 8000
Speed (RPM)
Figure 3.13 Mass Flow/Rate-Speed. Working fluid R134A dry vapour. Different
load conditions
180 100%
85%
70%
[—Load 50%
140
120
100
' ::
40
20
0 --------- --
0 1000 2000 3000 4000 5000 6000
Speed (RPM)
Figure 3.14 Power! Mass Flow Rate - Speed. Working fluid R134A dry vapour.
Different Load conditions
50
Chapter 3 Quasi One Dimensknal Flow Modeling
-----Volumetric Efficiency Adiabatic Efficiency Press
100 20
90 _____ _____ 18
80 16
70 _____ _____ _____ _____ _____ -- 14

60 - ___ ___ 12
•i3 50 -- ____ - 10
C-..------.
40 ___ ___ ___ ___ -- 8
30 --------- 6
20 4
10 ____ ____ ____ ____ --• 2
0 .';:; II; --b - _I I ---I - 0
0 10 20 30 40 50 60 70 80 90
Condensation Temperature (Degrees C)
Figure 3.15 Efficiencies / Condensation Temperature. With fixed evaporating

Temperatures and therefore varying pressure ratios. Working fluid R134A dry
vapour.
LPower Adiabatic Work Of CompressE

140 70
120 60
C,
1 00 50
C
0
80 40
I-
60
40 20.
0
20 10
0- I I I;;;- --1 I I -- - - 0
0 10 20 30 40 50 60 70 80 90
Figure 3.16 Power-Work of Compression / Condensation Temperature. With

fixed evaporating Temperatures and therefore varying pressure ratios. Working
fluid R134A dry vapour.
51
Figures 3.17 and 3.18 shows the behavior of the machine for different condensation
temperatures but fixed operating pressure ratio. This was achieved by varying both
the condensing and evaporating temperatures. As it can be seen from figure 3.17
the volumetric efficiency remains approximately the same, and the adiabatic
efficiency appears to be better than the one in figure 3.15. Figure 3.18 shows that
the adiabatic work of compression remains approximately the same. The Indicated
power curve follows a similar trend to that of figure 3.16 but with match greater
range.
A few comments on the significance of the trends in Figs. 3.15 to 3.18 follow.
With more moderate differences between the condensing and evaporating
temperatures the expectation is that the power required by the compressor will
increase with an increase in condensing temperature. The refrigerating capacity
always decreases with an increase in condensing temperature. Another important
characteristic which is not shown on the graphs is the coefficient of performance,
which decreases monotonically as the condensing temperature increases.
From the standpoint of power and efficiency, a low condensing temperature is

desirable; thus condensers should use the coldest air or water available.
Volumetric Efficiency Adiabatic Efficiency Press R]
100 8
90 7
80
6
70
0
;60 5=
50 4 0
0.
2
20
10
0 0
0 10 20 30 40 50 60 70 80 90 100
Figure 3.17 Efficiencies / Condensation Temperature. With fixed pressure ratio.

Working fluid R134A dry vapour.
52
L11Power Adiabatic Work Of Compression

1
250 100
90
200 80
-,
70 .
150
60 .
50 .
100 E
Afl 0
u
0
30
a-
50 0
2 3.
10
0 0
0 10 20 30 70 80 90 100
40 50 60
C ondensation Temperature (Degrees C)
Figure 3.18 Power-Work of Compression / Condensation Temperature. With

fixed pressure ratio. Working fluid R 1 34A dry vapour.
3.3.2 Scroll Expander
As was said earlier, the function of a compressor is to admit the fluid from a low
pressure region, compress it and deliver it to a high pressure region. An expander
carries out the reverse processes with the object of producing work.
In section 3.3.1 the importance of the correct build in volume ratio for a scroll
compressor was analyzed. In the case of the expander the build in volume ratio is
also extremely important. If the pressure in the working chamber has fallen below
the pressure in the delivery pipe, (over-expansion) when the port is uncovered, gas
flow back into the working chamber. As can be seen from fig 3.19 the delivery
port opens before the pressure in the working chamber has fallen enough, (under-
expansion) this results in an unresisted expansion, which in this case corresponds
to the waste of a certain amount of energy.
53
1 .00E+06
9.O0E+05
8.00E+05
'
5. 0oE.-
4.00E+0
3.00E+O5
2.00E+05
-100 0 100 200 300 400 500
Angle (Degrees)
Figure 3.19 Pressure / Angle. Speed 3000 RPM. Working fluid R134A wet
vapour.
The most distinctive feature of the results is the shape of the pressure-volume
diagram, of which a typical result is shown in figure 3.20. As can be seen the
initial filling process, which is normally assumed to take place at approximately
constant pressure, is in fact associated with a small pressure drop and therefore
some expansion. This pressure drop is due to the acceleration of the fluid as it
enters the varying space between the spirals. Theoretically this pressure drop must
be larger than the pressure change associated with gas discharge from compressors
through the same area in the reverse direction. This is because the fluid entering
the suction pocket of the expander has a very high liquid content and therefore a
much greater density. However figure 3.20 shows that the pressure drop during the
suction process of the expander is smaller than the change of pressure during the
discharge process of the compressor.
However it must be noted that the pressure-volume characteristics of wet vapours

are different to those of dry vapours and these lead to the need for a larger build-in
volume ratio for scroll expanders.
54
1 .00E+06
9.00E+05
8.00E^05
7.00E+05
6.00E+05
0,
(a,
5.00E+05
0.
4.00E+05
3.00E+05
2.E-i-05 I I I I I I I I I I I I I I -I----I------- I I I I I I I I
0.0000 0.0005 0.0010 0.0015 0.0020 0.0025 0.0030

Volume (m'3)
Figure 3.20 Pressure / Volume diagram of a two phase expansion process. Speed
3000 RPM. Working fluid R134A wet vapour.
Figures 3.21 and 3.22 shows some performance characteristics of the expander
with respect to rotational speed change. The adiabatic efficiency is shown to
increase together with the speed, a similar behaviour is recorded for the mass flow
rate and the power. As can be seen adiabatic efficiencies are of the order of 70%
over a wide range of operating conditions. The reason for the comparatively high
efficiency at low speeds is that leakage, which is then relatively large is in the
same direction as the bulk flow and hence its effect on performance is not so
adverse. From fig. 3.21 we can see a huge drop of the Volumetric efficiency with
speed. This is because the momentum term associated with filling becomes very
large at high speeds.
Figures 3.23 and 3.24 shows how mass flow rate and power change, over a wide
range of load conditions.
55
Chapter 3 Quasi One Dimensional Fbw Modeing
Adiabatic Efficiency Volumetric Efficienj
100
90
80
70
60
40
Ui
30
20
10
0 I I I I I -i--I I I I I I I I
0 1000 2000 3000 4000 5000 6000

Rotational Speed (RPM)
Figure 3.21 Efficiencies / Speed. Working fluid R134A wet vapour.
110
100
90
80
70
60
30
20
10
0 -I--I-I I I I I I I -I-I I I I I I I I- I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
0 1000 2000 3000 4000 5000 6000
Figure 3.22 Power / Speed. Working fluid R134A wet vapour.
56
Chapter 3 Quasi One DfrnensknalFk,w Modeling
50 -Load 100%
-Load 75%
-Load 52%
45 - ______ ______ ______
-Load 22%
40
35
0)
.
;30
25
2
U 20
U)
U)
15
10
0
0 1000 2000 3000 4000 5000 6000
Speed (RPM)
Figure 3.23 Mass Flow Rate/Speed. Working fluid R134A wet vapour. Different
Load conditions.
90 -Load 100%
-Load 750%
-Load 52%
80 -Load 22%
70
60
50
30
20
10
0 _I
0 1000 2000 3000 4000 5000 6000
Speed (RPM)
Figure 3.24 Power / Speed. Working fluid R134A wet vapour. Different Load
conditions.
57
3.4 Applications Of the Scroll Compressor - Expander
The concept of improving the coefficient of performance of vapour compression

refrigeration cycle systems by utilising the throttle expansion process to generate
mechanical power is well known. It has the advantages not only of its potential use
to reduce the power requirement to the main compressor but also, by expanding
more nearly reversibly, of increasing the refrigeration effect per unit mass of fluid
passing through the evaporator. However, until relatively recently its practical
embodiment has not been followed with much interest by the refrigeration and air
conditioning industry. The main reasons for this lack of interest were:
I) Gains in COP produced by improved compressor performance and reduced
heat exchanger temperature differences were more cost effective and significant.
II) Efficient means of recovering power from two-phase expansion were not
available.
At present, the situation has changed in that:
a) The continuing gains obtained from improved compressor performance and
reduced heat exchanger temperature differences have reached a stage where
further advances are diminishing in significance and escalating in cost to the
point where such improvements are no longer cost effective.
b) As a result of the relatively high efficiencies in modern compressors, the gains
achievable from two-phase expanders, which are relatively constant, have
become more significant.
3.4.1 Cycle Proposed
To make use of the work produced from the expander in terms of shaft power and
therefore reduce the power requirements of the compressor, the scroll expander
needs to be coupled with the compressor. This might create some performance
problems with the expander. Since it is well known that the mass flow rate of the
refrigerant varies, mainly because it depends on the load, the scroll expander must
not be coupled directly to the main compressor because this will force the expander
to run at the same speed with the compressor. The result of this will be a not so
good working performance for the expander when the system will operate at off
design conditions. To overcome this problem and allow the expander to adjust its
speed for any instantly mass flow rate (which depends on the load) it was decided
to coupled the expander with a second compressor.
The second compressor will work in parallel with the main one, it will be driven
58
Chapter 3 Quasi One Dimensional Flow Modefing
only by the work produced from the expander and it will contribute to the system in
terms of mass flow rate, The whole cycle is illustrated below.
TWO PHASE
EXPANDER
Figure 3.25 The proposed cycle. ('Expressor')
3.4.2 The proposed Model
A study was performed on a standard vapour compression refrigeration cycle using

refrigerant R134a. Although several working fluids, as for example R22 or propane
etc. could be used for the desired applications, HFCI34a was chosen for ease of
use, non flammability and mainly because of global environmental acceptance. The
cycle was operating on the following conditions:
Evaporating temperature: 5°C
• Superheating : 1OC
• Condensing temperature : 35°C
• Sub cooling : 0°C
Parameters Main compressor Second compressor Expander
Width (m) 0.5 0.12 0.1

Height (m) 0.062 0.02 0.0165
Thickness (m) 0.00006 0.00003 0.000025
Speed (RPM) 2200 3000 3000
Mass flow rate (kg/s) 23.44 1.55 25.1
59
Chapter 3 Quasi One Dimensional Flow Modefing
Lii --
Figure 3.26 Scroll machine.
3.4.3 Expander Characteristics
These are shown in figures 3.21 to 3.24. As may be seen overall adiabatic
efficiencies, which include bearing, seal and timing gear losses, are of the order of
70% over a range of operating conditions. For the design conditions of this model
the power output of the expander is approximately 52 kW.
3.4.4 Compressor Characteristics
These are shown in figures 3.11 to 3.14. Since leakage in compressors opposes the
bulk flow, its effect on idiabatic efficiencies should be greater than in expanders.
However as it was said earlier in this Chapter the pressure-volume characteristics
of wet vapours are different to those of dry vapours and these lead to the need for a
larger build in volume ratio for the expander. The larger build in volume ratio for
the expander implies that the leakage flow is relatively larger. Hence the superior
compressor efficiency at higher speeds.
60
3.4.5 Combined Characteristics
The expander and compressor characteristics may be combined to show how the
proposed model will perform over different rotational speeds. Figures 3.27 to 3.29
show these performance. Figure 3.27 shows how the mass flow through the second
compressor compares with that through the evaporator and hence the main
compressor. From figure 3.28 it may be seen that the expander will be able to drive
the compressor to produce an increased mass flow rate at speeds above 3000 RPM.
As far as the performance gains are concern, firstly it should be noted that if the
scroll expander is used (as proposed) then the mass flow through the expander will
be increased by the extra mass flow it induces in the coupled compressor.
Secondly, the inclusion of the expander has the additional advantage of making the
expansion process more nearly isentropic and hence increases the evaporation
effect per unit mass flow.
The refrigeration effect was increased from 157.177 kJlkg to 160.082 kJlkg, 1.8 1%
increase. The power requirements of the main compressor were decreased from
923.199 kW to 827.266 kW, 10.39% decrease. Now as far as the coefficient of
performance is concern it was increase from 4.29 to 4.87 which corresponds to an
increase of approximately 13%. (This is only given as an example at one operating
point).
This study has demonstrated that an Expressor unit comprising a scroll expander
driving a scroll compressor in a sealed unit is a viable and stable device for use as a
throttle valve replacement in large vapour compression chiller systems which can
produce improvements in the Coefficient of Performance of the order of 13% at the
design operating conditions.
Since it is not connected mechanically to the main compressor, an Expressor may

be readily incorporated into a chiller unit even as a retrofit device.
61
Chapter 3 Quasi One DimenskrnalFkw Modeling
50
45
a 40
p5 ___ __
30
U- _______ _______ _______ _______ _______
25
20
I-
U
Ui
5
0 I I I I I II I I I I -I I I I I I I I I I I I
0 0.5 1 1.5 2 2.5 3
Second Compressor Mass Flow Rate (kgls)
Figure 3.27 Combined Characteristics, Mass Flow Rate.
Expander Second Compressor
140
120
100
80
60
40
20
0
0 1000 2000 3000 4000 5000 6000
Figure 3.28 Combined Characteristics, Power.
62
Chapter 3 Quasi One DimenskwialFkw Modeling
Expander Second Compressor

80
70
60
50
40
;30
< 20
10
I(IIIt----h IIIIt±+ liii -±---FH- IIIiH+±

1000 2000 3000 4000 5000 6000
Figure 3.29 Combined Characteristics, Efficiencies.
63
Chapter 4
Geometry And Grid
Generation
4.1 Introduction
In order to perform a thermodynamic analysis of the scroll compressor - expander
first we have to describe the physical behavior of the machine in terms of
mathematical equations. Functions of working volume, sealing lines, contact
conditions etc. in terms of crank angle, (i.e. time) form the foundations of the
analysis. In the first part of this chapter we present the necessary equations needed
for the analysis. As will be shown in section 4.4 these equations form the basis for
the development of the grid generation code. The derivation of the equations was
necessary since it is not based on any published data. However it is partially based
on the patent of the inventor of the machine. Full analysis of the geometry is given
in appendix A
During the last few years, numerical flow analysis techniques have become
increasingly important in machine design. A key factor in obtaining a realistic
solution is mesh generation, considered as a basic problem requiring a sustained
effort. In fact, the interaction between the mesh characteristics and the development
of the solver are closely linked as far as stability, convergence and definition of the
solution are concerned. Therefore section 4.4 of this chapter deals with the design
and development of a grid generation program that can handle the complexity of
the working volume of the scroll device. Aspect ratio, orthogonality, cell distortion,
and smooth distribution, are some of the difficult issues involved in the transient
structured body fitted grid generation package developed.
4.2 Geometry
The basic elements in a scroll compressor are the identical scroll wraps whose
quality is vital to the compressor's performance. Geometrically speaking the wraps
are usually made on the basis of an involute of a basic circle (bC).
Chapter 4 Geometry A nd Grid Generation
As can be seen from the literature survey, there is almost nothing published on the
geometry of the scroll compressor. However even in the very few papers published
so far on this topic, only basic principles are stated. Therefore it was decided to
develop our own unique geometry based on the patent of the innovator of the scroll
engine Leon Greux [1905]. A detailed analysis of the geometry needed for the
design of a scroll machine is shown in appendix A, however some of the basic
equations are shown here:
The involute spiral of a circle is an appropriate curve for a scroll compressor

because of its simplicity of construction. if one unwraps a string from a circle,
keeping the string taut, the end of the string will trace out an involute spiral. The
envelope, formed by orbiting (revolving without rotating) an involute about a
separate origin, is in itself an involute spiral which differs from the original
involute by its starting angle. To build a spiral of this type both the inner and outer
curves must be known.
The curves C1 CFQ , C2 CF/, C3 C00 and C4 C01 , defining the inner and
outer surfaces of the fixed and orbiting spirals of the compressor are shown in
figure 4.1. The meaning of the subscripts is : FO Fixed Outer, Fl = Fixed Inner,
00 = Orbiting Outer, and 0/ = Orbiting Inner. The analytical expressions
describing the curves CCFQ , C2 CR , C3 C00 and C4 C01 . are:,
CI=CFQ : = r''
y1 ) sin(,)— ,cos(,)J'
(cos(co+y)+cosin(co+y)')
C2 C : f lt2')=
Y 2)
(4.1)
(r(cos()+q,sin(q))+dcos(0)\
C3 C00 : 1x 3 =
y3) r(sin() - cos()) + d sin(e)J'
= ( r(cos(ço + y) + sin( + y)) + d cos(0)')

C4 CO3 :
y4) r(sin( + )— cos( + ))+ dsin(8))'
It is now assumed that the curves C1— 0 0 and C4 CO3 have been constructed in
such a way that they never intersect each other, but there exist points at which they
are tangent. The same assumptions are also made for the pair of the curves C2— CFI
and C3 C00 . At these touching points the following relations hold
65
Chapter 4 Geometry And Grid Generation
ço1 =2k7r+-+O, ço3=2kir++O,
= jr+ ço1 — 1 ' q,2 = ir+ — 7, ^ 0 (i=1,2,3,4), d =r( g - 2'). (4.2)
O<y=constant<r/2, kE\W , OE9L
Clearly, the relations between the angles q, q and 0 refer to the contact
conditions of the curves C/ CFQ and C4=C01 , while the relations between the
angles and 0 refer to the touching conditions of the curves C2 CF, and
c3=coo.
flfl
3 'W
Figure 4.1. Generating circles and involutes.
The derivation of the working volumes leads to a number of formulae. For

example the volume of the suction chamber only is described by four different
equations depending on the relative position of the spiral walls of the device.
Therefore it has been decided not to put all those equations here. As already stated
a mathematical analysis is shown in appendix A.
66
Figure 4.2. 3-D View of the Scroll Machine using the involute of a circle
New types of curves named as "The Involute of a Basic Square" and " The
Involute of a Basic Triangle" have been recently studied [1992] in an attempt to
improve or replace the conventional one. The involute of a basic square can be
obtained when unwrapping a string from a square as shown in figure 4.3.
R 22
R31 _4-./ P21
P32\ P12
R41
R4
Figure 4.3. Formation of involute of square, wrap design.
It is obvious from the formation process that, the curve consists of several quarter
circles, that is, its curvature changes every quarter circle. Something quite similar
happens with the involute of the basic triangle.
Besides the types of curve mentioned above an Archimedes spiral can also be used
for the design of scroll machinery. However when using this type of curve extreme
caution must be taken in order to avoid any problems regarding contact conditions.
(i.e. leakages)
67
4.3 Theory Of Grid Generation

One of the main problems in computing numerical solutions of Partial Differential
Equations (PDE) is the generation of the mesh for the solution domain. The grid
has to be well-constructed otherwise it may lead to instability or lack of
convergence. Numerical grid generation has now become a fairly common tool in
the numerical solution of PDE on arbitrarily shaped regions and several
commercial packages are available (Thompson, 1988). A comprehensive survey of
procedures and applications has been given by Thompson eta!. (1982; 1984).
Body-fitted curvilinear coordinate systems, used in modern CFD codes generated

to maintain coordinate lines coincident with the boundaries, were introduced to
extend the capabilities of Finite Difference Methods (FDMs) to deal with complex
geometries and assist in the imposition of boundary conditions without the need for
special procedures at the boundaries. Motion of the boundaries is also possible, the
coordinate system adjusting to follow variations developing in the evolving
physical solution. In any case, the numerically-generated grid allows all
computations to be done on a fixed square grid in the computational field which is
always rectangular by construction. However, the cost of the greater flexibility
afforded by the use of body-fitted grids and general coordinates is an increase in
complexity of the equations to be solved, due to the non-linear coordinate
transformation.
The first step in the generation of the grid is the transformation of the physical
space (complex-geometry domain) into a computational domain (simple-
rectangular domain), where all numerical algorithms, finite difference, finite
volume or finite element, are implemented. Numerical grid generation techniques
may be roughly classified into three categories:
(a) Complex variable methods.
(b) Algebraic methods.
(c) Differential Equation Techniques.
Complex variable methods are restricted to two dimensional problems and so their
applicability is very limited. These techniques have the advantage that the
transformation used are analytic or partially analytic as opposed to the other
methods that are entirely numerical. Algebraic and differential equations
techniques can be used for complicated three-dimensional problems. In the next
sections we present the main approaches to the generation of meshes for structured
grids. Then, special techniques for gridding complex geometries will be discussed.
68
4.3.1 Algebraic Methods

Algebraic mesh generation methods determine the coordinate positions by
interpolating among boundaries and / or intermediate surfaces using known
functions of sets of data points. The main advantages of algebraic mappings are
that they are direct and the metrics of the transformation can be computed
analytically. However, since inherent smoothing properties are missing,
discontinuities can arise in the mesh.
Several interpolation methods are used, and are classified as unidirectional and
multidirectional methods. In the unidirectional interpolation method, the
interpolation, in one curvilinear coordinate direction only, occurs between points.
These points can, however lie on boundary (or interior) curves of surfaces, and in
this sense the unidirectional interpolation can be considered to be between these
curves or surfaces. Several interpolation techniques are available (Lagrange,
Hermite, Splines, Tension Splines, B-Splines etc.), the most general procedure
being the 'multi-surface procedure (Eiseman. 1979) of which Lagrangian or
Hermitian interpolations are special cases. This procedure is constructed from the
interpolation of a specified vector field, defined from piecewise-linear curves
determined by the boundaries and successive intermediate control surfaces. These
vectors are taken to be tangential to the coordinate lines intersecting the surfaces,
so that integration of this vector field produces the position vector field for the grid
points.
Multi-directional interpolation involves interpolation among functions of curves or

surfaces. The most general method used is the 'transfinite', described in detail by
Rizzi and Erikson (1981). The interpolation generates the mapping by combining
unidirectional interpolation with Boolean sums, and matches the function at any
number of points.
4.3.2 Partial Differential Equation Methods

These techniques are the most highly developed for generating grids. The grid
points are obtained by the solution of a set of partial differential equations, which
can be hyperbolic, elliptic or parabolic. If the coordinate points are specified on
the entire closed boundary of the physical region, the equations must be elliptic,
while if the specifications is only a portion of the boundary the equations would be
parabolic or hyperbolic.
69
The use of elliptic PDEs to generate grids has been treated extensively by
Thompson (1974). This procedure transforms the physical space into the
computational space where the mapping is controlled by general Poisson-type
systems. The generation system guarantees a one-to-one mapping for boundary
conforming curvilinear coordinate systems on general closed boundaries. The
advantages of using this technique are to generate a smooth grid, complex
boundaries being easily treated, without overlap of grid lines (extremum
principles). However, grid point control on the interior is difficult to achieve, and if
boundaries are changing in time, (as in the case of a scroll compressor / expander)
since the grid has to be computed for each time step, large amounts of computer
time may be consumed.
The most simple elliptic partial differential system is the Laplace equation:
V2 tj =0 (i=1,2,3)
where ' is the coordinate vector in the computational domain.
This system exhibits an extremum principle and considerable smoothness. Control

of the coordinate line distribution in the field can be obtained by generalizing the
elliptic generating system for the Poisson equations:
V2 ' = P (i=1,2,3)
in which the 'control functions' P , can be fashioned to control the spacing and
orientation of the coordinate lines. In the absence of the control functions the
coordinate lines will tend to be generally equally spaced away from the boundaries
regardless of the boundary point distribution. Variations of elliptic systems are
noted in Thompson et al. (1982; 1984).
A number of different algorithms have been used for the solution of these
equations, including point and line SOR, ADI (Alternating-Direction-Implicit-
Technique) (Thompson et al., 1982; Thompson, 1984). The convergence can be
accelerated by using multiple grid iteration (Forsey and Billing, 1988). For general
configurations, point SOR is certainly the most convenient to code and has been
found to be rapid and dependable, using over-relaxation, for a wide variety of
configurations. Since the system is nonlinear, convergence depends on the initial
guess in iterative solutions. Algebraic grid generation procedures may be used to
generate this initial guess. Transfinite interpolation generally produces a more
70
reliable initial guess than unidirectional interpolation does because of its reduced
skewness.
A fundamental development of the hyperbolic generation system was made by

Starius (1977). Hyperbolic systems are applicable to domains in which the inner
boundary is specified but the surrounding outer boundary is arbitrary, and therefore
of interest only for use in calculation on physically unbounded regions where the
precise location of a computational outer boundary is important. The parabolic
system can be applied to generate the grid between the two boundaries of a doubly-
connected region with each of these boundaries specified. The grid control is
generally weak and difficult compared with the elliptic methods. However the
weakest method in the grid control is the grid generation method. Both parabolic
and hyperbolic grid generation systems have the advantage of being generally faster
than elliptic generation systems, but, as noted, are applicable only to certain
configurations.
Orthogonality is one of the most desirable properties in grid generation techniques.

Orthogonal coordinate systems produce fewer additional terms in transformed
partial differential equations, and thus reduce the amount of computation required.
This also makes the application of boundary conditions more straightforward, and
permit a greater degree of vectorization on supercomputers. A severe departure
from orthogonality will introduce truncation error in 'different expressions'.
Orthogonality in three dimensions is difficult to achieve, and only exists when the
coordinate lines on the bounding surfaces follow lines of curvature, i.e., lines in the
direction of maximum or minimum curvature of the surface. It is possible,
however, to have the system locally orthogonal at boundaries, and/or to have
orthogonality of surface coordinates. Since a part of the truncation error decreases
as the grid becomes more orthogonal, it is of interest to generate grids which are
'nearly-orthogonal'. Various generation procedures of orthogonal systems are
surveyed in Thompson (1980).
4.3.3 Adaptive Grids

Adaptive grids are generated with the object of directing the distribution of grid
points so that a functional relationship on them may represent the physical solution
with sufficient accuracy. As the physical solution develops, grid points move,
concentrating in regions of large gradients in the solution as they emerge.
Movement of grid points is coupled to their neighbours, and cannot be too far or
71
Chapter 4 Geomefry And Grid Generation
too fast, otherwise oscillations may occur.
Different techniques for constructing adaptive grids have been proposed. Brackill
and Saltzman (1980) have developed a technique for constructing an adaptive grid
using a variational approach. It offers a powerful method for constructing
computational grids; however, a considerable effort must be expended in solving
the equations which govern the grid generation. Many applications of adaptive
grids require grid motion along one coordinate. The equidistribution scheme is a
minimization process applied in a one-dimensional form, and the grid points are
distributed so that the weight function is equally distributed over the field. A
number of applications of the use of one-dimensional equidistribution are cited in
the survey on adaptive grids given by Thompson (1984). The grid speed scheme
developed by Hindman and Spencer (1983) also incorporates the idea of
equidistribution. The grid speeds are established by differentiating the steady grid
equation, and solving the resulting equation. To advance the grid to the next time
level the grid speeds are integrated. Several other approaches are discussed by
Thompson (1984).
Once adaptivity criteria have been formulated, it is necessary to move points

around the domain so that the flow solution has a minimum error. Three basic
strategies may be employed to do so: (i) redistribution of a fixed number of points,
(ii) local refinement of a fixed set of points, (iii) local increase in algorithm order.
Solutions to PDEs obtained using numerical methods in conjunction with adaptive

grids show significant improvements in either accuracy or resolution. Oscillations
associated with cell Reynolds number and with shocks in fluid mechanics
computations have been shown to be eliminated with adaptive grids. The adaptive
grid is most effective when it is dynamically coupled with the solution, so that the
solution and grid are solved together in a single continuous problem. The
development and application of adaptive grids have been surveyed by Eiseman and
Eriebacher (1987).
4.3.4 Moving Boundary Grids

Moving Boundary procedures have recently been developed by a number of
research groups and are included in the most advanced commercially available
CFD codes. Moving boundaries are a chief feature of the flow in the scroll machine
where the fluid flow is driven by the motion of the wall.
72
Details of the application of moving boundaries procedures to the Navier-Stokes

equations are given by Demirdzic and Peric (1990), and Hawkins and Wilkes
(1991).
4.3.5 Computational Grids For Complex Geometry Domains

There are three basic types of computational grid in current use:
(i) Single-block, fully structured grids,
(ii) Multi-block grids,
(iii) Unstructured grids.
Single-block fully structured grids, discussed above, are traditionally used in Finite
Difference Methods (FDMs). As shown in figure 4.4 they transform the physical
space into a rectangular (or hexahedronal in 3-D) computational space. In complex
geometry domains, to have accurate solutions of the flow, special techniques are
required for gridding, such as multi-block and unstructured grids.
Yl
1Y2
xl
-,
(a)
4-
x2
xl X2
Yl
t2
x2
(b)
xl
Figure 4.4 Samples of a single-block grid. (a) Physical domain. (b) Computational
domain
73
4.3.5.1 Multi-Block Grids

The multi-block approach is a very popular method of applying structured grids to
complex geometrical regions. The physical domain is segmented into sub-regions
which constitute contiguous blocks. Each block has to be transformed into a
hexahedron, so that several local curvilinear systems are constructed and connected
together. However, the global distortion of the grid is relatively less accentuated.
Grid points at block interfaces have to be treated so that points at the common edge
of any two adjacent blocks are continuous and so that slope continuity may be
enforced.
Multi-block grids are very powerful in that they allow the use of a wide range of
mesh topology for a given configuration and different coordinate systems are
allowed in each block. They are restricted to the grid generation technique -
algebraic or partial differential equations. Although the concept of a multi-block
mesh is very attractive from a mesh generation point of view, additional
complexities are introduced in the solution procedures, that involve a slow-down of
the calculation time.
Examples of three-dimensional multi-block grids applied to complex geometry are

given in the literature by Weatherill and Forsey (1984), and are noted in the survey
given by Thompson and Warsi (1983). Multi-block grids have been implemented
in the latest Releases (3.12, 3.2, etc.) of CFX-FLOW3D.
4.3.5.2 Unstructured Grids

Unstructured meshes, traditionally used in FEM, are generally composed of
triangles andlor arbitrary quadrilaterals in 2-D, and tetrahedra and for hexahedra in
3-D. Therefore the number of cells surrounding a typical node of the mesh is not
constant. Techniques for the generation of unstructured meshes are discussed in
detail by Peraire et. Al. (1990). It was found that to have an adaptive mesh,
triangulation is necessary to allow easy transition between small and large
elements.
The unstructured nature of the mesh requires a local coordinate system for each
element. The resultant mesh, therefore, is poorly ordered and less amenable to the
use of vectorization algorithms. Also larger computer time storage is required in
comparison with structured grids. However, unstructured meshes are more efficient
in complex geometry domains, and offer the possibility of incorporating adaptivity.
74
In order to minimize computer time requirements, renumbering strategies for

unstructured grids have been proposed.
4.4 Development Of Grid Generation Code

In this subsection we analyze the development of the grid generation code, which is
used for the requirements of this project . The area for which we need to create the
grid, is the working chamber of the machine (see figure2.4). Since this area
depends on the boundaries of the working volume, and since these boundaries are
changing with time, it can be concluded that we need a technique to move the mesh
to its new position with respect to the movement of the boundaries. What we have
done is to re-calculate the vertices. At this point we have to make it clear that we
are not doing re-meshing, we are using some analytical mathematical equations
that we develop to re-calculate the coordinates of the vertices. The topology is
maintained the same.
At the beginning we tried to use commercial grid generation packages (Patran,

Femgv, Sophia etc.). One of the limitations we found with these packages is their
inability to handle complex moving geometries, if the geometry is fixed in space
then these packages will produce a good mesh. If however, movement is involved
then problems arise.
4.4.1 Two Block Grid Generation Code

After we took the decision to develop our own grid generation code the first step
was to describe mathematically the working volumes. As can be seen from figure
2.4, the working volumes are mainly described by two curves each. Equations 4.1
and 4.2 (touching conditions) define these curves. The next step was to decide
about the number of blocks. At the first attempt we used a two block grid, the first
block was used for the one working chamber and the second one for the other
chamber. The first thought was that we would select a certain number of points
along each curve and then interpolate between the points to find intermediate
vertices. The problem was how to subdivide each curve. One idea was to use
uniform distribution, If we used this option then we would not have any idea how
the points on the one curve were related to the points on the other. Therefore we
thought that since the curves are similar to each other (both are involutes of a
circle), maximum orthogonality would be obtained if we found where the curves
have the same gradient (parallel) and fit straight lines between them. With this
75
method we only needed to subdivide the one curve. The idea was that we would
select a point on one curve, then calculate the gradient of the curve at that point and
then use a numerical technique to find which point on the other curve has a
gradient as close as possible to the first one. Note also that there might be more
than one point that would satisfy this condition (case of same gradients), therefore
a second restriction was imposed. This was that the second point must give the
minimum distance between the two curves. Once we knew these two points then
we could fit a straight line to connect them, or perform a linear interpolation to
calculate intermediate points. The number of the elements would depend on the
number of intermediate points and on the number of points selected on the first
curve. This logic was supposed to be true for all the different processes of the
machine (suction, compression, discharge). A major problem was the point where
the two working chambers approached each other (in space). When these two come
in contact their vertices should coincide, otherwise errors will occur in the solver.
This problem was overcomed by calculating the vertices only on the one site and
forcing the vertices of the other block to take the same values.
After spending a considerable amount of time, the program was ready. The grid
movement was successful, the user would need to specify the number of
subdivisions on the curve, the number of intermediate points, the number of
subdivisions on the Z axis, the time, and the rotational speed which is needed for
the calculation of the crank angle with respect to time. The grid results were very
disappointing the biggest problem was the skewnes of the grid especially on the
discharge process. At the end of the discharge process overlapping occurs. This
overlapping results in the calculation of negative volumes. The results of this first
attempt are shown in figures 4.5 to 4.15
76
Figure 4.5 Crank angle 2000
Figure 4.6 Crank angle 300°
77
78
Figure 4.9 Crank angIe 6000
79
Figure 4.12 Crank angle 860° Figure 4.13 Crank angle 900°
Figure 4.14 Crank angle 950° Figure 4.15 Crank angle 10000
4.4.2 Four Block Grid Generation Code

Having identified the problems from our first attempt we decided to make some
changes. The first was that we would use four blocks instead of two. Initially the
two blocks would be more or less equal, but as the domains move inwards the
block on the inside begin to expand where as the other one continues its shrinking
mode. This coincides with the beginning of discharge as shown in figure 4.24.
Then, some elements will be compressed and others will expand. This would give
the flexibility needed to control the vertices within the discharge chambers and
therefore avoid overlapping (which results in negative volumes). The second and
biggest change was the implementation of transfinite interpolation. This would
control the case where we do not know the intermediate points and we want the
intermediate curvature to be as close as possible to the boundaries i.e. the discharge
chamber.
4.4.2.1 Implementation of Transfinite Interpolation

The problem of generating a curvilinear coordinate system can be formulated as a
problem of generating values for the Cartesian coordinates in the interior region
80
from specified values on the boundaries. This, can be done directly by interpolation
from the boundaries, and such coordinate generation procedures are referred to as
algebraic generation systems. As was mentioned in section 4.3.1, these systems are
divided into two major categories, Unidirectional and multidirectional
interpolation. Transfinite interpolation falls in the second category.
In two directions we may write a linear Lagrange interpolation function

individually in each curvilinear direction:
r(,n) = cbnJ) r(,n)
ni
r(,n) = iírnjjJ T('flm)
This interpolation is now called 'Transfinite' since it matches the function on the
entire boundary defined by =O and =I in the first equation, or by n=O and n=J in
the second. The final form of the transfinite interpolation is shown below:
N
r(,n)=Øn(Jr(n,n)+ Y itrn1JT(uirn)
n=1 m=1 (4.3)
. t
n=im=1
øn1'm T(n,h1m )
The transfinite interpolation form above, interpolates the entire set of intersecting
arbitrary curves. The interpolation function defined above with N=M=2, using the
Langrange interpolation polynomials as the blending functions, is termed the
transfinite bilinear interpolant and is the one used in this project. With N=M=3,
this form is the transfinite biquadratic interpolant. Other immediate candidates for
the blending functions are the Hermite interpolation polynomials and the splines.
The results are shown on figures 4.16 to 4.26 As it can be seen there is some
improvement but not to an acceptable level. Initially we thought of subdividing
both curves and using a faster distribution on the one curve so that the pairs of
points will face each other as closely as possible. This was to be done by
implementing a geometric or other series that would control the distribution.
Eventually this proposal was abandoned, mainly because this would not guarantee
the required orthogonality and it would also need numerical opimisation.
81
82
83
84
4.4.3 Mid-Planes Based Grid Generation Code
After careful consideration of the problem it was decided to rewrite the code based
on a completely new procedure. The new idea was to find a curve that would be
located somewhere in the center of the two boundary curves. This curve would be
something like a mid-plane, and instead of subdividing the one or both the
boundary edges we would march along this mid-plane (see figures 4.27, 4.28). First
we selected a point on the mid plane. Then we drew a straight line that passed from
this point and intersected the one curve of the working volume perpendicularly.
Then a second line would be drawn, in this case the line passing from the same
point on the mid-plane and cutting the other curve of the domain perpendicularly as
shown in figure 4.29. By this method we ensured that the elements close to the
curves would have maximum orthogonality. Therefore now the objective was
firstly to find this mid-plane and secondly to find the points on the curves that
would satisfy this condition. After spending some time, two different numerical
85
techniques were found for the mid-plane and the points on the curves respectively.
Eventually, these techniques were not used mainly because they led to excessive
computational time. The next step was then to find analytical functions that would
satisfy the requirements specified above. After spending a considerable amount of
time, some mathematical equations were developed for the mid-plane, and for the
points needed for the lines. These are:
1COS + + — 7 JS1flq_\ -
XMP FO OI 2 2 ) I
=
YMP_FO_OIJ
Jcos_ sinO l
2 2 ) J
(4.4)
( XMP 00 1 COS +1 + -2 JsinQ+ ( -2 ) I

=
YMP 00_F!)
Jcos+ sino
2 2 ) )
where MP_FO_OI is the mid plane between the fixed outer and orbiting inner
curves, and MP_OO_FI is the mid plane between the orbiting outer and fixed inner
curves
Therefore now that we had three points we could either connect them with two
lines or, assume that these points are lying in the circumference of a circle, find the
center of the circle and then fit a circular arc to connect the points. In this case the
interpolation would be performed along the arc as shown in figure 4.30. In the
program the user has the option either to use straight lines or circular arcs. The
latter appears to produce better grids in terms of smoothness.
86
Figure 4.27 Scroll geometry and mid-planes. Crank angle 17d.
Figure 4.28 Working volumes and mid-planes. Crank angle 170°.
87
Figure 4.29 Working volumes, mid-planes and samples of lines.

starting from points on mid-planes and finishing at the boundary
curves forming angles of 900. Crank angle 1700.
Figure 4.30 Working volumes, mid-planes and samples of circular

arcs passing from mid-planes and finishing at the boundary curves
forming angles of 90°. Crank angle 170°.
88
Unlike the previous two versions of the program, this method is not the same for
all the processes. The general idea is the same but the program is divided into four
parts. The first part covers all the suction process of the compressor and a large part
of the compression process. It stops at the point where the edges of the spirals start
sliding along the circular arcs (see appendix A). The second part covers the rest of
the compression process. The third part is for the discharge process and finally the
last part is concerned with the inclusion of the discharge pipe in the grid. However
before we proceeded to analyze each part of the code, we had to analyze the
implementation of a non uniform element distribution. This option together with
the transfinite interpolation and the mid-planes are the 'heart' of this final and
eventually successful attempt to create a robust, transient, body fitted, structured
grid generation package, with maximum orthogonality and aspect ratio within
acceptable limits specifically developed for scroll machinery.
4.4.3.1 Implementation of Non Uniform Distribution Functions

With the above approach (mid-planes) the biggest problem i.e. departing from
orthogonality was overcome. The next step was to find a way to control the
element distribution. Eventually two different mathematical models were
implemented. These are, hyperbolic function distribution and circular function
distribution (Thompson 1985).
In general, interpolation between r 1 at =O and r2 at =I can be written as
r () q'(/I)r2 + [1_ q(/I)ri ] (4.5)
where q' can be any function such that (0)0 and 'p (l)=l. The linear
polynomial case is obtained here with q(/I) i/I. The function q in this form
may contain parameters which can be determined so as to match the slope at the
boundary, or to match interior points and slopes.
The interpolation function, q, in this form is often referred to as a 'stretching'

function, and the most widely used function is the exponential:
- exp (a c/I) —1
(4.6)
exp(a)-1
where a is a parameter that can be determined to match the slope at a boundary.
89
Thus. Since from equation 4.5
r= - r1)q' (4.7)
we can determine a from the equation
r2 —r1 a
(re) (4.8)
1 I exp(a)_1
with (r) 1 specified.
The truncation error is strongly affected by point distribution, and studies of

distribution functions have been made in that regard. The exponential, while
reasonable, is not the best choice when the variation spacing is large, and
polynomials are not suitable in this case. The better choices are the hyperbolic
tangent and the hyperbolic sine. The hyperbolic sine gives a more uniform
distribution in the immediate vicinity of the minimum spacing, and thus has less
error in this region, but the hyperbolic tangent has the better overall distribution.
(M. Vinokur, 1983, J.F. Thompson, 1983). These are implemented as follows
(following M. Vinokur, 1983), with the spacing specified at either both ends, or a
point in the interior, or a point distribution on a curve.
Let the arc length, s, vary from 0 to 1 as varies from 0 to I: s(0)=0, s(I)=1. Then
let the spacing be specified at =0 and —I:
s(0)=L1s 1 , s(I)= z1s2 (4.9)
The hyperbolic tangent distribution is then constructed as follows. First,
1' B = ,jz1s 1 s 2 (4.10)
Then the following nonlinear equation is solved for 8:
sinhö 1
(4.11)
8 B
The arc length distribution then is given by
90
u()
s()= (4.12)
A+(1_A)u()
where
+ tanh[8(/I - )11
u() - { (4.13)
- tanh (-- 8)
f
If this is applied to a straight line on which r varies from r 0 to r1, we have for the
point locations:
r(4)= r0 +(r1 —r0)s (4.14)
The points are then located by taking integer values of :
=O,1,2.... . ,I
Clearly, the arc length distribution, s(), is here the function of q of equation 4.5.
Note that B is essentially the ratio of the specified spacing to the linear spacing,
1/1. If B is greater than unity, i.e. if the specified spacing exceeds the linear
spacing, the hyperbolic functions all revert to circular functions in all the relations
of this section.
With the spacing zl s specified at only —_O, the construction proceeds as follows.
First B is calculated from
B=14 (4.15)
and eq. 4.11 is solved for 8. The arc length distribution then is given by
tanh{48(/I -1)]
s()=1+ (4.16)
tanh(8)
With the spacing specified only at =I the procedure is the same, except that
relation 4.13 is replaced by
= tanh(8/I)
s() (4.17)
tanh(8)
If the spacing zl s is specified at only an interior point s = o, B is again calculated
91
from 4.15, and then Sis determined as the solution of
2
IB cosh8-1+1/ci
1+1—I = (4.18)
sinhö
The value of at which s = c is obtained by solving the nonlinear equation,
sinh S
X - tanh1 (4.19)
5 1/a+cosh8-1]
1+sinh{S(—x)/I]1
s() - (4.20)
- sinh(öx/1)
This last distribution is based on the hyperbolic sine. From this, a distribution
based on the hyperbolic sine with the spacing specified at one end can be derived.
Here B is evaluated from 4.15, and then Sis determined as the solution of,
sinhS 1
(4.21)
S B
s() sinh(5/I)
(4.22)
sinh S
if the spacing is specified at — 0. With the specification at 4=!, the distribution is
sinh{S(1 -
s()=i- (4.23)
sinh S
In the code, the above method is implemented in such a way that the following
three options are available:
I) Both initial and final divisions are specified.
II) Only initial division is specified.
ifi) Only final division is specified.
However, the program is optimized and there is no need for the user to choose
between these options. The only thing that the user has to define is whether the
distribution is uniform or non uniform.
92
It has to be noted at this point that the functionality and the behavior of all the
mathematical equations used in this program were first checked using Matlab.
Matlab is a mathematical package that offers good graphics. The main reason for
using Matlab was the advantage of viewing the behavior of the curves first, and
then deciding whether or not to implement them into the code.
4.4.3.2 Suction - Part of Compression

Since the two working volumes are identical we will analyze only the one. In this
new approach the grid is made up of four blocks with two blocks in each working
chamber. From the figures that follow it can be seen that each working volume
appears to have only one block. In fact there is a second block but it is so small it
can not be seen at this stage. The importance of this 'hidden' block will become
apparent later on. (i.e. for the rest of the compression process). This part of the
code is not so complicated as the other parts. First we calculated the grid on the
'hidden' block using transfinite interpolation and then we calculated the grid on the
other block using the mid-planes technique. At this point it should be stated that
another 'sub-block' was introduced within block 3 as shown in figure 4.31.
Actually this is not a block by itself, but forms a part of block 3 and is used to put
additional smoothness in the connection between the two blocks.
Figure 4.31 Blocks structure during compression process. Crank angle 82d3.
93
No matter how small are the elements of the 'hidden' block, the use of non-
uniform distribution functions discussed above ensures that the increase of the size
of the elements on the vicinity will be as smooth as possible. In fact at the edge of
contact between the two blocks (block 3 and 4) the elements on each side either
have exactly the same size or there is a maximum difference of 1.2 between them.
The grid for this process is shown in figures 4.37to 4.50
4.4.3.3 Remainder of Compression

This part of the grid is more complicated than the first one. In this case the so
called 'hidden' block begins to expand in a non uniform way. For convenience this
'hidden' block will be called block 4 for the rest of this chapter. The expansion
should be as smooth as possible in order to avoid oscillations on the solution. In
this case the edges the hidden block 4 are shown in figure 4.32
Edge 1 B4 Edge 4 B4
Edge 2 B4
Block 4
Edge 3 B4
Perpendicular
from Mid plane
to Orbiting Out?A
Part of Block 3
Perpendicular
from Mid plane
to Fixed Inner\ Mid plane
Remainder of Block 3
Figure 4.32 Blocks structure during compression process. Crank angle 82cP.
The size of edge 4B4 remains the same until the discharge process. Even though
the size remains the same, because it is the clearance between the two curves, the
relative position will change according to the crank angle. The sizes of the edges,
94
Chapter 4 Geomefry A nd Grid Generation
1B4, 2B4 and 3B4 were found experimentally. It has to be noted here that this was
the most time consuming procedure in the development of the code. This is mainly
because without an analytical solution, everything was optimized experimentally.
Once the four edges were clearly identified and their mathematical definition
known we continued with the non uniform subdivision of the edges and then used
tranfinite interpolation to calculate vertices in the interior of block 4. The next step
now was to construct the mesh for part of block 3. In this case the edges were
found experimentally and this was also very time consuming. Then the edges were
non uniformly subdivided and finally, transfinite interpolation was used to
calculate points on the interior. For the remainder of block 3, the non uniform
distribution was employed again, and the interior vertices were calculated using the
mid plane technique. The grid for this section is shown in figures 4.51 and 4.52.
4.4.3.4 Discharge Process

In this case, edge 4B4 comes in contact with the corresponding edge of block2. In
order to avoid any problems, we have to make sure that the vertices in each side
match. The way we did this was to calculate the vertices only on the edge 4B4 and
then to force the other edge to have these values. This process is more or less the
same as the previous one. There are however two main differences. These are: (a)
the edge 4B4 is expanding and (b) part of block 3 disappears very slowly and very
smoothly. It mast be reemphasised that the size, type, etc. of the edges of block 4
and those of the part of block 3, were found and optimized using the extremely
time consuming technique of trial and error. This was the biggest problem we
faced. The grid for the case of block 4 and part of block 3, if there is any, are
calculated using the transfinite interpolation, and non uniform distribution, where
as the grid of the remaining of block 3 was created using the combination of mid
plane technique and non uniform distribution. The results for this section are
shown in figures 4.53 to 4.55.
4.4.3.5 Inclusion of the Inlet Pipe

The development of this code was the biggest challenge of the project. The next
step i.e. the inclusion of a pipe into the grid was also a great challenge. The pipe is
located at block 4 and is adjacent to the edge 1B4. The problem here is not how to
constrain a pipe into the grid. The main problem in this case is that the pipe is fixed
in space. Everything (elements) around the pipe can move but the grid of the pipe
95
has to remain stationary. This pipe is of great importance, because using this pipe
we will specify the boundary conditions needed for the analysis.
In order to do this, Block 4 was divided into four parts. The one part is the pipe,
and the other three are the parts around the pipe. Once the user set the flag for the
pipe ON then the program is waiting for the diameter of the pipe. Everything is
automatic. The user has to specify only the diameter, height and the number of K
subdivisions, if the pipe diameter is too big, for the size of the machine then the
program returns a warning, requesting a smaller diameter. Once the diameter is
within acceptable limits then the code calculates the position of the pipe, and then
the number of elements of the pipe. The number of cells for the pipe is not
determined by the user. if the grid of the pipe is too coarse then there is an indirect
way to add more elements in there. To do this the user can increase the number of
the elements of block 4, and this will automatically increase the number of
elements in the pipe. The four parts of block 4 are shown below (figures 4.33 to
4.36). The difficult part of this section was once again the specification of the
edges. The combination of transfinite interpolation with non uniform distribution
was the grid generation tool for this case. The final results for this section are
shown in figures 4.56 to 4.64.
Figure 4.33 Grid of Parts 1 and part 3 of Block4 Crank angle 925°, Diameter of
pipe 0.02 m. (Transfinite interpolation)
96
Chapter 4 Geometry A nd Grid Generatk'n
Figure 4.34 Grid of Part 2 of Block4 Crank angle 925°, Diameter of pipe 0.02 m.
(Transfinite interpolation)
Figure 4.35 Grid of Part 4 of Block4 Crank angle 925°, Diameter of pipe 0.02 m.
(Transfinite interpolation)
Figure 4.36 Grid of Block4 Crank angle 925°, Diameter of pipe 0.02 m.
97
4.4.3.6 Use of the Code

Before we proceed with the final description of how to use the code, we have to say
something about how the grid is constructed in three dimensions. As is known, the
height of the scroll compressor-expander is constant, therefore the grid constructed
in two dimensions (x,y) can be simply extruded in the Z-axis. Please note that the
constant height of the machine does not permit us to solve the problem by a two
dimensional approach. The reason is because the flow "in" (case of expander) or
"out" (case of compressor) is coming from the Z-axis (pipe).
To run the program there are two main options. The first one is to call it from CFX
as a sub-routine (the code is implemented into CFX4), and the second one is to run
it independently. Let us first analyze the case where the program is running
independently. In this case there is an input file in which the user can specify all the
parameters needed for the generation of the grid. The first thing the user has to
specify is the size of the machine. This is done by giving the base circle radius, the
thickness of the walls, the height of the walls, the clearance and the length of the
spirals. Then the crank angle must be given. The crank angle is the parameter that
will indicate whether the working volumes are in the Suction, Compression, or
Discharge stage. The next step is to specify the number of elements in each block.
Note that blocks which share the same edge should have the same number of
elements along that edge. This is done by specifying the number of subdivisions
along each axis (I,J,K).
Once the process of subdivision is completed, the type of interpolation for the
interior points has to be specified. There are two options as shown in figs 4.37 to
4.50. These are either the use of linear interpolation along a straight line or
interpolation along a circular arc. Next is the decision about the type of
distribution. Options are: Uniform distribution or Non uniform distribution. Finally
comes the option of the pipe. This option is not available for any crank angle. This
is obvious because we cannot have a pipe in the grid during the suction, or
compression processes. If, however, the user specifies a pipe by mistake at those
conditions, its requirement is ignored. At the end of the run an output file called
'GROUT' is created. The grid calculated and stored into file 'GROUT' is only for
that specific crank angle the user defined at the input file. To view the results we
have our own post processor Fortran program, using GINO Libraries.
For the case where the program is called from CFX there is only one major change.
98
That is the crank angle is related to real time and speed. In this case the user has to
specify the rotational speed of the machine, the time step (needed for the
simulation) and only the initial crank angle, all the other parameters are the same as
for the case where we run the program independently. For example if we have ten
time steps, the program will be called ten times. Each time the program is called a
new crank angle is calculated (because of the increase in time) from the relation
between speed and time. As a result of this a new grid for that specific time step is
created, and given to the solver. This process continues until the last time step.
99
Figure 4.37 Grid 150*20* 1. Crank angle 1800, Non Uniform Distribution, Circular arcs
Figure 4.38 Grid 150*20* 1. Crank angle 180°, Uniform Distribution, Straight lines
100
101
Figure 4.41 Grid 150*20*1. Crank angle 3300, Non Uniform Distribution, Circular arcs
102
Figure 4.44 Grid 150*20*1. Crank angle 450°, Uniform Distribution, Straight lines
103
Figure 4.46 Grid 150*20*1 . Crank angle 540°, Uniform Distribution, Straight lines
104
105
106
Chapter 4 Geome fry A nd Grid Generation
Figure 4.52 Grid 150*20* 1. Crank angle 850°, Non uniform Distribution, Circular Arcs
107
Figure 4.53 Grid 150*20*1. Crank angle 900°, Non Uniform Distribution, Circular arcs
Figure 4.54 Grid 150*20* 1. Crank angle 950°, Non Uniform Distribution, Circular arcs
108
Figure 4.55 Grid 150*20* 1. Crank angle 1000 0, Non Uniform Distribution, Circular arcs
Figure 4.56 Grid 150*20* 1. Crank angle 930°, Non Uniform Distribution, Circular arcs,
inlet pipe diameter 0.02m
109
Inlet pipe diameter 0.02m
Inlet pipe diameter 0.02m
110
930 0 , Non Uniform Distribution,

Figure 4.59 Grid 150*30* 1. (Refined mesh) Crank angle
Circular arcs, Inlet pipe diameter 0.02m
Figure 4.60 Grid 150*30* 1. (Refined mesh) Crank angle 960°, Non Uniform Distribution,
111
Figure 4.61 Grid 150*30* 1. (Refined mesh) Crank angle 9900, Non Uniform Distribution,
Figure 4.62 Grid 150*30* 1. (Refined mesh) Crank angle 930°, Non Uniform Distribution,
Circular arcs, Inlet pipe diameter 0.01 (smaller pipe diameter)
112
Figure 4.63 Grid 150*30*1. (Refined mesh) Crank angle 960°, Non Uniform Distribution,
Figure 4.64 Grid 150*30*1. (Refined mesh) Crank angle 990°, Non Uniform Distribution,
113
Figure 4.65 Blocki mesh: 150*30*20, Block2 mesh: 30*30*20, Block3 mesh: 150*30*20
Block4 mesh: 30*30*20, Pipe mesh: 10*10*30, Total number of elements 219000.
Crank angle 930°, Non Uniform Distribution, Circular arcs, Inlet pipe diameter 0.02m
114
Chapter 5
Three Dimensional
Flow Modeling
5.1 Introduction
In this chapter we give the mathematical basis for a comprehensive general purpose
model of fluid flow and heat transfer from the basic principles of conservation of
mass, momentum and energy. This leads to the governing equations of fluid flow,
the so called Navier - Stokes equations. The governing equations are complex non
linear partial differential equations, which can be solved using numerical methods.
Discretization methods, advection schemes, time differencing schemes and
pressure I velocity coupling algorithms are discussed. Commercially available CFD
software packages are reviewed and CFX - FLOW3D is described in detail.
Flows inside scroll compressors - expanders are turbulent, time dependent,

compressible and, for the case of the expander, two phase. Furthermore these flows
are driven by the wall motion and, therefore, are characterized by moving
boundaries. In the last section of this chapter the results of the three dimensional
modeling are shown.
5.2 Theory
All of CFD, in one form or another, is based on the fundamental governing
equations of fluid dynamics, the continuity, momentum, and energy equations.
These equations speak physics. They are the mathematical statements of three
fundamental physical principles upon which all of fluid dynamics is based: namely
the conservation of mass, momentum and energy.
• Mass is conserved.
• Newton's second law, F=ma.
• Energy is conserved.
Chapter 5 Three Dimensional Row Modefing
5.2.1 Governing Equations

The equations that result from applying the Conservation of Mass law to a fluid
flow is called the continuity equation. The Conservation of Momentum law is
nothing more than Newton's second law (F= ma). When this law is applied to a
fluid flow, it yields a vector equation known as the momentum equation. The
Conservation of Energy law is identical to the First Law of Thermodynamics, and
the resulting fluid dynamic equation is named the energy equation. In addition to
the equations developed from these universal laws, it is necessary to establish
relationships between fluid properties in order to close the system of equations. An
example of such relationships is the equation of state, which relates the
thermodynamic variables pressure p, density p, and temperature T. The derivation
of all the above equations is shown in Appendix B.
5.2.1.1 The Continuity Equation
Physical principle: mass is conserved
(5.1)
Equation (5.1) is a partial differential equation form of the continuity equation. It

was derived on the basis of an infinitesimally small element fixed in space. The
fact that the element was fixed in space leads to the specific differential form given
by equation (5.1), which is called the conservation form. The forms of the
governing flow equations that are directly obtained from a flow model which is
fixed in space are, by definition, called the conservation form.
5.2.1.2 The Momentum Equation
Physical principle: F= ma (Newton's second law)
Equations (5.2a) to (5.2c) are the x, y, and z components respectively, of the

momentum equation (in conservation form). They are called the Navier - Stokes
equations in honor of two men, the Frenchman M.Navier and the Englishman G.
Stokes, who independently obtained the equations in the first half of the nineteenth
century.
116
Chapter 5 Three Dimensional Row Modeling
d(pu) + V dr + +d + (5.2a)
. ( pUv ) = pf
dt dx dx dy dz
d(Pv)+v.(pV v)=_J + x y + Y Y (5.2b)

dt dx dx dy dz
and
d(pw) ___
di drY Zdr
___ (5.2c)
+V (pwV )=- + x z +_+pf
dt dz dx dy dz
5.2.1.3 The Energy Equation
Physical principle: Energy is conserved
Equation (5.3) below is the conservation form of the energy equation written in
terms of total energy e + V 2/2.
a [ ( v1 r ' V 2" ] . d dT d ( dT'

=pq+— Ik--I+
d xL dx)
+-_(kit d(up) - d(vp) - d(wp) + d(ur) + d(ui) + d(ur)
dz' dz) dx dy dz dx dy dz
+ d(vr) + d(v) + + d(w) + d(wr) + d(w)
+pf •V
dx dy dz dx dy dz
5.2.2 Mathematical Behavior Of Partial Differential Equations

The partial differential equations for continuity, momentum, and energy in fluid
flow have a certain mathematical behaviour. This behaviour might be different
from one case to another, depending, for example, on the local Mach number of
the flow. The same equations may have different mathematical behavior depending
on whether the flow is locally subsonic or supersonic. The behavior may also be
different depending on whether the flow is steady or unsteady. Any differences in
mathematical behavior of these equations reflect different physical behavior as
well. There are two separate techniques for determining the classification of partial
differential equations: Gramer's rule, and the eigenvalue method. Both these
methods lead to the same results.
117
Chapter 5 Three Dimensional Flow Modefing
Quasi-linear second order partial differential equations in two independent

variables can be classified into three types: hyperbolic, parabolic, and elliptic. This
distinction is based on the nature of the characteristics, curves along which
information about the solution is carried. Every equation of this type has two sets
of characteristics.
In the hyperbolic, the characteristics are real and distinct. This means that
information propagates at finite speeds in two sets of directions. In general the
information propagates in a particular direction so that one datum needs to be given
at an initial point on each characteristic; the two sets of characteristics therefore
demand two initial conditions. If there are lateral boundaries, usually only one
condition is required at each point because one characteristic is carrying
information out of the domain and one is carrying information in. There are
however exceptions to this rule.
In parabolic equations the characteristics degenerate to a single real set.

Consequently, only one initial condition is required. At lateral boundaries, one
condition is needed at each point.
Finally, in the elliptic case, the characteristics are imaginary or complex so there
are no special directions of information propagation. Indeed, information travels
essentially equally well in all directions. Generally, one boundary condition is
required at each point on the boundary and the domain of solution is usually closed
although part of the domain may extend to infinity. Unsteady problems are never
elliptic.
These differences in the nature of the equations reflect in the methods used to solve
them. It is an important general rule that numerical methods should respect the
properties of the equations they are solving.
The Navier -Stokes equations are a system of non-linear second order equations in
four independent variables. Consequently the classification scheme does not apply
directly to them. Nonetheless Navier -Stokes equations do posses many of the
properties outline above and many of the ideas used in solving second order
equations in two independent variables are applicable to them but care must be
exercised.
118
5.2.2.1 Hyperbolic Flows

To begin, consider the case of unsteady inviscid compressible flow. A
compressible fluid can support sound and shock waves and it is not surprising that
these equations have an essentially hyperbolic character. Most of the methods used
to solve these equations are based on the idea that the equations are hyperbolic and,
given sufficient care, they work quite well.
For steady compressible flows, the character depends on the speed of the flow, if
the flow is supersonic, the equations are hyperbolic while the equations for
subsonic flow are essentially elliptic.
It should be noted however, that the equations for a viscous compressible flow are
still more complicated. Their character is a mixture of elements of all of the types
mentioned above; they do not fit well into the classification scheme and numerical
methods for them are difficult to construct.
In terms of CFD, the computation of flow fields that are governed by hyperbolic
equations is set up as "marching" solutions. The algorithm is designed to start with
the given initial conditions, say in the y axis, and sequentially calculate the flow
field, step by step, marching in the x direction.
5.2.2.2 Parabolic Flows

Information travels only downstream in these equations and they may be solved
using methods that are appropriate for parabolic equations. Parabolic equations,
like hyperbolic equations, lend themselves to marching solutions.
5.2.2.3 Elliptic Flows

When a flow has a region of recirculation i.e. flow in a sense opposite to the
principal direction of flow, information may travel upstream as well as
downstream. As a result, one cannot apply conditions only at the upstream end of
the flow. The problem then acquires an elliptic character. This situation occurs in
subsonic (including incompressible) flows and makes solution of the equations a
very difficult task. The major mathematical characteristic of elliptic equations is
that a disturbance is felt everywhere throughout the domain, therefore the solution
must be carried out simultaneously at all the points of the domain.
It should be noted that unsteady incompressible flows actually have a combination
119
of elliptic and parabolic character. The former comes from the fact that information
travels in both directions in space while the latter results from the fact that the
information can only flow forward in time. Problems of these type are called
incompletely parabolic.
5.2.3 Numerical Solutions To Partial Differential Equations

We have seen in the sections above that the equations governing the motion of
fluids are partial differential equations. The transformation of these equations to
what is known as an algebraic equation it is called discretization. In essence
discretization is the process by which a closed-form mathematical expression, such
as a function or a differential or integral equation involving functions, all of which
are viewed as having an infinite continuum of values throughout some domain, is
approximated by analogous (but different) expressions which prescribe values at a
finite number of discrete points or volumes in the domain.
Analytical solutions of partial differential equations involve closed - form

expressions which give the variation of the dependent variables continuously
throughout the domain. In contrast, numerical solutions can give answers at only
discrete points in the domain, called grid points. The discretization methods fall
into three main categories: (a) Finite Difference method, (b) Finite Element method
and (c) Finite Volume Method.
5.2.3.1 Finite Difference Method
This is the oldest method for numerical solution of the partial differential equations
believed to have been introduced by Euler in the 18th century. It is also the easiest
method to use for simple geometries.
The starting point is the conservation equation in differential form. The solution
domain is covered by a structured grid. At each grid point, the differential equation
is approximated by replacing the partial derivatives by approximations in terms of
the nodal values of the functions. The result is one algebraic equation per grid
node, in which the variable value at that and at a certain number of neighbor nodes
appears as a number of unknowns.
Taylor series expansion or polynomial fitting is used to obtain approximations to

the first and second derivatives of the variables with respect to the coordinates.
When necessary, these methods are also used to obtain variable values at locations
120
Chapter 5 Three Dimensional Flow Modefing
other than grid nodes (interpolation).
On structured grids, the finite difference method is very simple and effective. It is
especially easy to obtain higher - order schemes on regular grids. The restriction of
simple geometries is a significant disadvantage.
5.2.3.2 Finite Element Method

The domain is broken into a set of discrete volumes of finite elements that are
generally unstructured. The distinguishing feature of finite element methods is that
the equations are multiplied by a weight function before they are integrated over
the entire domain. In the simplest FE methods, the solution is approximated by a
linear shape function within each element in a way that guarantees continuity of the
solution across element boundaries. Such a function can be constructed from its
values at the corners of the elements. The weight function is usually of the same
form.
This approximation is then substituted into the weighted integral of the

conservation law and the equations to be solved are derived by requiring the
derivative of the integral with respect to each nodal value to be zero; this
corresponds to selecting the best solution within the set of allowed functions (the
one with a minimum residual). The result is a set of non-linear algebraic equations.
An important advantage of finite element methods is the ability to deal with

arbitrary geometries; there is an extensive literature devoted to the construction of
grids for finite element methods. The grids are easily refined; each element is
simply subdivided. Finite element methods are relatively easy to analyze
mathematically and can be shown to have optimal properties for certain type of
equations. The principal drawback, which is shared by any method that uses
unstructured grids, is that the matrices of the linearized equations are not as well
structured as those for regular grids making it more difficult to find efficient
solution methods.
A hybrid method called the control-volume-based finite element method should

also be mentioned. In it, shape functions are used to describe the variation of the
variables over an element. Control volumes are formed around each node by
joining the centroids of the elements. The conservation forms in integral form are
applied to these control volumes in the same way as in the finite volume method.
121
Chapter 5 Three Dimensional Flow Modeling
The fluxes through control volume boundaries and the source terms (associated
with the creation or destruction of ) are calculated element-wise.
5.2.3.3 Finite Volume Method

The third, and probably most popular, numerical discretization method used in
CFD is the finite volume method. This method is similar in some ways to the finite
difference method, but some implementations of it also draw on features taken
from the finite element method. The finite volume method was developed
specifically to solve the equations of heat transfer and fluid flow and is described in
detail by Patankar (1980).
The finite volume method uses the integral form of the conservation equations as
its starting point. The solution domain is subdivided into a finite number of
contiguous control volumes (CV), and the conservation equations are applied to
each CV. At the center of each CV lies a computational node at which the variable
values are to be calculated. Interpolation is used to express variable values at the
CV surface in terms of the nodal (CV center) values. Surface and volume integrals
are approximated using suitable quadrature formulae. As a result, one obtains an
algebraic equation for each CV in which a number of neighbor nodal values
appear.
The finite volume method can accommodate any type of grid, so it is suitable for
complex geometries (a structural orthogonal grid is recommended, whenever
possible). The disadvantage of finite volume methods compared to finite difference
schemes is that higher than second order methods are more difficult to develop in
3D. This is due to the fact that the finite volume approach requires two levels of
approximation: interpolation and integration.
5.2.3.4 Comparison Of The Discretization Techniques

There are, several differences between the three methods and these include the
following:
• the finite difference method and the finite volume method both produce
numerical equations at a given point based on the values at neighboring
points, whereas the finite element method produces equations for each
element independently of all other elements. It is only when the finite element
equations are collected together and assembled into the global matrices that
the interaction between elements is taken into account.
122
. the finite element method takes care of derivative boundary conditions when
the element equations are formed, and then the fixed values of variables must
be applied to the global matrices. This contrasts with the other two methods
which can easily apply the fixed-value boundary conditions by inserting the
values into the solution, but must modify the equations to take account of any
derivative boundary conditions.
There are, however, several common features. These features are that each method:
produces equations for the values of the variable at a finite number of points
in the domain under consideration.
• requires that we know the boundary conditions of the problem.
• can produce explicit or implicit forms
Finite difference methods are based on the substitution of difference equations for
the partial derivatives in partial differential equations. These difference equations
link the values of variables at a set of points to the derivatives and so a grid of
points is used throughout the spatial domain. The finite difference method requires,
that the grid points is topologically regular. This means that the grid must look
cuboid in a topological sense.
Finite elements produce the numerical equations for each element from data at
known points on the element and nowhere else. Consequently, there is no
restriction on how the elements are connected as long as the faces of neighboring
elements are aligned correctly. This means that the faces between elements should
have the same nodes for each of the adjoining elements. This flexibility of element
placement allows a group of elements to model very complex geometry.
Algorithms that have been developed using the finite volume method have tended
to use a regular grid to take advantage of the efficiency of computation, just like
the grids used with finite difference methods. Recently, however, to enable
calculations to be carried out in complex geometries, algorithms have been
developed with the finite volume method that can utilize irregular finite element-
like meshes. Both finite element and irregular mesh finite volume programs pay a
computational overhead for this geometrical flexibility. This flexibility slows down
the programs considerably.
123
5.2.4 Pressure I Velocity Coupling Procedures

The process of discretization of the governing differential equations lead to a set of
non -linear, coupled algebraic equations, which are solved using iterative methods.
One of the most popular early techniques for solving the 2-D incompressible
Navier-Stokes equations is the V orticity-Stream-Function method (Gosman et al.,
1969). The main attraction of this approach is the fact that pressure does not appear
in the formulation; however difficulties in specifying the vorticity at boundary
conditions at sharp corners may cause problems in obtaining a converged solution.
A change of variables is made replacing the velocity components with the vorticity
and the stream function i. Using these new dependent variables, the two
momentum equations can be combined to lead to the parabolic vorticity transport
equation and an elliptic PDE in the form of a Poisson equation. This allows the
separation of the mixed elliptic-parabolic PDE into one paiabo%ic. and ont
that can be solved with any standard iterative method such as SOR (Succesive
Over-Relaxation). In order to determine the pressure, an additional equation,
referred to as a Poisson equation for pressure is solved.
The extension of the vorticity-stream function approach to three-dimensional

problems is complicated by the fact that a stream function does not exist for a truly
three-dimensional flow. However, using a velocity-potential vector, it is possible to
have a vector Poisson equation and a vector vorticity transport equation, which
must be separated into three parabolic and three elliptic PDEs. As a result, it does
not appear that the vorticity-stream function approach offers any advantage over
the primitive-variable approach when solving a 3-D problem (Roache, 1972). The
latter uses the incompressible Navier-Stokes equations in the primitive-variable
form, and then non-dimensionalizes them.
The artificial compressibility method of Chorin (1967) is one of the early

techniques proposed for solving the incompressible Navier-Stokes equations in the
primitive-variable form. An artificial compressibility is introduced which vanishes
when the steady-state solution is reached and makes the differential equations a
mixed set of hyperbolic equations which can be solved using a standard time-
dependent approach.
The most common primitive-variable approach, however, involves the use of a

Poisson equation in place of the continuity equation in order to separate the
pressure effects into a single equation. This technique was first used by Harlow and
124
Welch (1965) in conjunction with their Marker-and-Cell (Mac) method, for solving
the incompressible Navier-Stokes equations for transient flows. Approximate
initial values are given to the pressure gradients to find velocities from the
momentum equation and the pressure from the Poisson equation. This procedure is
repeated until the solution converges. The Alternating-Direction-Implicit (ADI)
scheme is applied to the momentum equations (Ghia et al., 1981) and a SOR
method is used to solve the Poisson pressure equation.
5.24.1 Simple Algorithms

All the above methods converge slowly, suffered from stability problems and were
very difficult to use in 3-D flows. In the early 1970s a pressure-correction
technique referred to as SIMPLE (Semi-Implicit Method for Pressure-Linked
Equations) procedure was developed by Patankar and Spalding (1972). The
SIMPLE procedure marked a significant improvement and is currently used, in one
of its many forms, in all general purpose CFD codes.
This procedure is based on a cyclic series of guess-and-correct operations to solve

the governing equations and is, with its variants SIMPLER and SIMPLEC, the
most widely used in solving for the pressure field. It was first proposed for
parabolic flows, and then was soon adapted to general 3-D elliptic problems. The
velocity components are first calculated from the momentum equations using a
guessed pressure field. The pressure and the velocities are then corrected so as to
satisfy continuity. This process continues until the solution converges. The actual
pressure p at the generic iteration step is written as
p = p 0 + p'
where p0 is the estimated (or intermediate) value of pressure and p' is the pressure
correction. Similarly in 2-D
u=uO +u,
v =v o+v ,
where u0 and v 0 are estimated values, and u' and v' are the velocity corrections.
Using these equations and approximate forms of the momentum equations, we

obtain a pressure correction which is a Poisson equation in the pressure correction
p' and the estimated velocity vector. Once the first guess of the pressure is made
and the velocities are then estimated by the momentum equations, the pressure-
125
correction is solved to find p', then p and finally u, and v using the approximate
form of the momentum equations. These new values are then used as estimated
values and the process is repeated until the solution converges.
Because the pressure correction equation tends to overestimate the value of p' in
certain cases, the rate of convergence was found to be unsatisfactory. Under-
relaxation techniques were used to improve the convergence properties. A new
technique was then proposed by Patankar (1981), called SIMPLER (SIMPLE
Revised), where initially the velocity field is guessed. Velocity corrections are
computed in the same manner as in SIMPLE, but a complete Poisson equation is
used to compute the pressure. With this procedure the need of under-relaxation is
greatly reduced and a convergence solution is obtained in a fewer iterations,
although more computational effort is involved per iteration. The original
implementation of this scheme required the use of staggered grids in order to
prevent a numerical instability known as 'checkerboarding', which occurred on a
single grid because the velocities were insensitive to small oscillations in the
pressure solution, and vice versa.
The SIMPLEST method (Spalding, 1980) is based on an explicit treatment of

convective terms and implicit treatment of diffusive terms in the momentum
equations. The SIMPLEC method, proposed by Van Doormal and Raithby (1984),
uses consistent under-relaxation for the momentum and pressure corrections. This
method has been shown to allow faster convergence in prcssure gradient-dominant
and drag forces-dominant problems.
5.2.4.2 Piso Algorithms
An alternative pressure-correction scheme was proposed by Issa (1985)as a

Pressure-Implicit with Splitting of Operators (PISO) technique. In order to remove
completely the need to iterate between pressure and velocity equations, a second
pressure-correction equation is solved in each iteration, much like SIMPLER
yielding pressure and velocity fields that are second-order accurate in time. This
allows its use as a non-iterative method (one cycle per time step), but it requires an
accurate solution of the linearized equations.
Although this method appear to be preferentially indicated for problems where

conservation of mass and momentum are particularly important, it does require
pressure and velocity equations to be solved to tight convergence tolerances, and in
126
practice the scheme is often implemented in an iteration loop similar to that used
for SIMPLE. It was proved to be faster than SIMPLE for transient problems in
which the flow field varies markedly at each step, but not for 'smooth' time-
dependent flows (Ciofalo and Collins 1988). At last, in the PISOC method similar
modifications to SIMPLEC are applied.
A review of pressure-velocity coupling algorithms is given by Latimer and Polland

(1985). The SIMPLE-like and PISO-like methods have all been implemented in the
previous versions of CFX-FLOW3D (single-block versions). Only SIMPLE and
SIMPLEC methods are available in the more recent multi-block versions of CFX-
FLOW3D. In fact, these are the only ones that have been proved to be robust for
more general purpose applications.
5.2.4.3 PressureNelocity Coupling Techniques for FEMs

As in FDM, early calculations using FEM circumvented the pressure/velocity
coupling problem by using the vorticity-stream function formulation. Techniques
in use to date for handling the coupling between velocity and pressure are the
Penalty Method, the Lan grange Multiplier Method or a combination of these, and
the PA LM Method (Smith, 1985). In other cases either the momentum and
continuity equations are solved directly or a SIMPLE-like iterative procedure is
devised.
The Penalty Method, first presented by Zienkiewizc (1977), uses a penalty

parameter which, multiplied by the pressure, is introduced in the right-hand side of
the continuity equation, and should be zero for the continuity condition. This
parameter is very small so that the continuity equation is nearly satisfied. The
pressure, given as the ratio of the divergence of the velocity vector and the
vanishing penalty parameter, is substituted into the momentum equation, thereby
eliminating pressure. A general discussion of the penalty method is given by Reddy
(1982), while Baker (1984) has described the use of the penalty method in three-
dimensional parabolic flows. In the Lagrange Multiplier Method, the pressure is
considered as a linear function or piecewise constant over quadratic or linear
velocity elements respectively (Gresho et al., 1980). That is, the approximation for
pressure has to be of lower degree than that used for the velocity.
Solution existence can be only guaranteed for those velocity/pressure element

nodal arrangements that satisfy the Babuska-Brezzi condition (Babuska and Aziz,
127
1982; Brezzi, 1974). That is, on a triangular element it is required to have velocity
on more than each vertex (for example also on the medians), and pressure at more
than at the centroid (for example, at three internal point centroids).
5.2.4.4 Rhie-Chow Interpolation Method

In order to prevent a 'checkerboard' pressure field that may develop when velocity
components and pressure are located at the same grid point and the FDM are
applied, a common approach has been the use of 'staggered grids'. In this case
pressures and scalars are defined at the centroids of the control volumes, while the
velocities are defined at the centroids of the faces to which are normal. However,
this approach may become quite cumbersome in the presence of non-orthogonal,
body-fitted grids.
The Rhie-Chow algorithm (Rhie, 1981; Rhie and Chow, 1983) allowed the
required components on the cell faces to be computed from the values of velocity
and pressure at the cell centroids in non-staggered grids without including
checkerboarding. It works on 3-d body fitted grids, and allows a natural application
of SIMPLE-like algorithms. The Rhie-Chow algorithm has been incorporated into
CFX-FLOW3D since its release 2 version; and it is also implemented into most
last-generation computer codes.
An improvement to this algorithm has been carried out within the last versions of
CFX-FLOW3D to allow applications to be treated where large pressure gradients
occur which balance strongly varying body forces. Details of this formulation are in
the user manuals (Anonymous, 1994a; 1994b).
5.2.5 Advective-Term Methods

In recent years, a significant amount of research effort has been directed toward
discretization of the combined convection (transport due to fluid flow) and
diffusion (transport due to variations of 0 from point to point) fluxes. Inaccurate
modeling of these terms can seriously degrade overall accuracy and stability of the
solution. In many cases, it is almost impossible to refine the grid sufficiently so that
the numerical errors will reduce to acceptable levels. Thus, it is essential to have a
convection-diffusion formulation that leads to a stable and accurate solution with
grids of modest fineness. Whereas many methods have been formulated, there
exists no clear consensus about a preferred method. The more accurate schemes
128
tend to be less robust and slower.
th order to help subsequent discussion, some of the more common models will be
described. We shall consider a simple situation of steady one-dimensional flow
where only convection and diffusion terms are present. If 0 is the variable advected
at a constant velocity u, then, integrating the relevant advection term over the
control volume shown in figure 5.1 gives
JU___=U(0e — 0w)
As values of 0 are only known at the center of each control volume, a linear profile
can be taken to estimate cbe and . The result is the Central Difference Scheme
(CDS):
Oe = (0 - and cbs, = j (Op - Ow)
The factor ½ arises from the assumption of the interfaces being midway. This
scheme is formally second-order accurate, but can lead to spurious spatial
oscillation ('wiggles') of the solution (Roache, 1972) or physically non-realistic
solutions unless the cell Peclet number (Pe = p u A x/F, where Llx is the cell
width, and F is the diffusion coefficient) is less than 2. There has been some
misunderstanding in the literature as to the nature of these oscillations; however, it
is now accepted that they are related to an accuracy problem rather than to a
stability problem (Vreugdenhil, 1989). It is also relevant that, when CDS is used
and the cell Peclet number is high, the linearized governing equations lose diagonal
dominance. This makes the CDS algorithm lose its 'robustness', and very small
under-relaxation factors, or very small time steps, are needed to obtain
convergence.
ww w P E
. S S .
4ww
w e
Figure 5.lTypical control Volume
129
An unconditionally stable but rather 'crude' approximation is to assume that the

value at the face is the same as the nodal value upstream of the face. This technique
is the basis of the Upwind-Difference Scheme (UDS). It was first put forward by
Courant et a!. (1952) and subsequently by Gentry et al. (1966). This scheme
introduces a false 'numerical diffusion' and produces a result that is actually the
solution to a different problem.
The exponential scheme which uses the exact solution profile would give a highly
desirable behaviour, but at very high computational cost and without being able to
be applicable to more general situations. An easy-to-compute scheme that has the
qualitative behavior of the exponential scheme is the Hybrid Difference Scheme
(HDS). It was developed by Spalding (1972), and applies a central difference
scheme for the Peclet number range — 2 ^ Pe ^ 2 and an upwind difference
outside this range by ignoring diffusion. This advection model is used as a default
in many FDM programs. It is again first-order accurate and cannot eliminate false
diffusion errors, but it is slightly better than upwind differencing because second-
order central difference will be the result across streams and in regions of low flow.
Since it is the grid Peclet number that decides the behavior of the numerical
schemes, it is, in principle, possible to refine the grid (i.e., to use smaller zix) until
Pe is small enough (<2 ) for the central differencing scheme to yield a reasonable
solution. In most practical problems, however, this strategy requires excessively
fine grids, which are usually not feasible on economic grounds.
A large number of differencing schemes have been proposed to prevent numerical

diffusion while preserving robustness and avoiding 'wiggles'. They include the
following approaches:
(a) Higher-order upwind schemes, based on the use of more than one grid point
value in the finite difference expression of the advective fluxes, such as the
second-order upwind, third-order upwind (QUICK), or the related 'power law'
schemes (Patankar, 1980) (PLDS), 'Hermitian Polynomial' schemes (Glass and
Rodi, 1982), and 'spline' methods (Rubin and Graves, 1975).
(b) 'Skew' schemes, such as the 'skewed upwind' (SUS) and the 'skewed upwind
weighted' (SUWDS) schemes (Raithby, 1976), the 'mass flow weighted'
(Hassan et al., 1983), the 'vector upstream' (Lillington. 1981), 'bounded skew
upwind' (BSUDS) (Syed and Chiappetta, 1985), and 'flux bending' schemes
130
Chapter 5 Three Dimensional F/ow Modeling
(Gosman and Peric, 1985): all take some account of the angle at which the fluid
crosses the coordinate lines.
(c) Methods based on a locally exact solution of the governing equations, such as
the 'finite analytic' method (Chen et al., 1981), and the 'locally analytic'
scheme (LOADS) (Wong and Raithby, 1979). These formulations do give
stable and accurate results but are rather complex and expensive for regular
use.
(d) Modified-central differencing schemes, such as CONDIF (Runchal, 1987), and

NONDIF (Hedberg, 1989).
Among these we give some details of advection schemes available in CFX-

FLOW3D. The Higher-order Upwind scheme (HUM) represents a more accurate
upwind scheme, which uses an additional nodal value further from the face.
Assuming equal spatial increments, Ø4, is given by:
3 1
=w---Øww
A formally more accurate scheme than HUV is the quadratic upwinding scheme
known as QUICK (Leonard,1979). This is an upwinded scheme which is third-
order accurate for the advection and second-order accurate for the diffusion terms.
This scheme uses two upstream points and one downstream point, which makes it
slightly unstable:
Ø =- cbp 3
4 8
The higher order upwinded schemes can suffer non-physical overshoots in their
solutions. For example, turbulent kinetic energy can become negative. The CCCT
scheme is a modification of the QUICK scheme which is bounded, eliminating
these overshoots. The scheme has:
\\
(3 (3
=
where a is a parameter that depends on the curvature of the variable 0 . Full details
for the calculation of a are given by Alderton and Wilkes (1988).
131
CONDIF is a modification of central differencing in which the coefficients are

written in such a way as to be diagonally dominant, and upwind differencing is
used at any points at which the solution may have a non-physical overshoot. Again
full details are given by Alderton and Wilkes (1988).
A survey of comparative studies on advection schemes is given by Patankar (1988).

While some reviews are in favour of the QUICK scheme, others report over- and
under-shoots, oscillations and convergence problems, and conclude in favour of
HUW or even UDS. Also, higher-order schemes imply a significant increase in
computational complexity. Bums and Wilkes (1987) handled this problem using
the 'deferred correction approach', that is, they included the extra terms on the
right hand side of the equation in the source term (associated with the creation or
destruction of 0).
Convection-diffusion formulation in Finite Element methods have followed a

similar pattem, and various comparable upwinding schemes have been proposed.
Because of non-linearity of the advection terms, Newton-Raphson or Picard
iteration methods are commonly used. A control-volume formulation was
introduced by Baliga and Patankar (1980), who used a shape function based on the
direction of the local velocity vector. In the Galerkin formulation of the finite-
element method, a streamline upwind procedure was employed by Brooks and
Hughes (1982).
5.2.6 Time Stepping procedures

In steady-state problems, the generic transport equation for the variable 0 can be
written as:
F(cb)=S0
In time-dependent problems this equation becomes:
The time derivative of the flow variable 0 can be discretized with a degree of
implicitness ' (0 < ^ 1)
1(n+i
1t'
132
Chapter 5 Three Dimensional Flow Mode fing
which can be written in the same form as the steady-state equation (5.31):
F(çb'') = S0
Thus the same basic algorithms can be used, though some extra storage is required
for the values of the variables at step n. For a degree of implicitness =1.0 we
have, for example, the fully implicit backward time differencing scheme; for ,'
=0.5 there is the time-centered scheme of Crank-Nicolson and for =0.0 the
scheme is fully explicit. The fully implicit and the Crank-Nicolson schemes are
unconditionally stable in the mathematical sense, while the explicit method is
subjected to the Courant stability criterion, e.g., the Courant number Cr ^ 1,
(where C r = c/(zi x/zI t), with c as celerity of propagation in the analytical
solution and LI x the width of the smallest cell).
In Finite Element Methods, Crank-Nicolson or backward Euler time stepping may

be used. However, more general time-splitting techniques are becoming popular,
some of which have recently been reviewed by Glowinski and Pironneau (1992).
5.2.7 Solution Techniques For The Linearized Equations

The discretizationflinearization techniques described above lead, at the generic
inner iteration (SIMPLE iteration), to very large systems of linear equations
representing either scalar transport equation or the pressure correction equation,
having the general form:
Ax= b
In the pressure correction case, the coefficient matrix is symmetric. In all cases, if
the 'deferred correction technique', described above, is used the matrix A contains
only seven non-zero diagonals (in 3-D problems) and is efficiently stored as a
N x 7 array, N being the overall number of control volumes in the grid.
To solve such a large system only iterative methods, such Gauss-Seidel and
Successive Over-Relaxation (SOR) methods, are applicable. These consist in
guessing and solving the equations respectively until the iterative procedure
converges.
Whereas the Gauss-Seidel (point-by-point, line-by-line or ADI) methods have been

commonly used for the solution of linear algebraic equations, these methods
133
converge very slowly, especially when a large number of grid points are involved.
For large three dimensional problems, it has been found that a more reliable
technique is the Strongly Implicit Procedure (SIP) proposed by Stone (1968). This
technique has proved to be the best in the solution of the pressure correction
equation having a high degree of accuracy. It applies a 'factorization strategy',
which consists in replacing the original sparse matrix A by a modified matrix A +T,
such that it can be decomposed into upper and lower triangular matrices. The
matrix T has to be selected so that the ADI procedure can be applied.
Recently, new techniques have been proposed, some of which are extensions of the
SIP method. Lin (1985) has proposed the use of three free parameters to accelerate
the convergence of Stone's method. In the methods formulated by Schneider and
Zedan (1981) and Peric (1987), the five diagonal matrix in Stone's method is
replaced by a nine-diagonal matrix. As a result, these modified methods (Modified
Strongly-Implicit, MSI) are applicable to the discretization formulations which lead
to nine-diagonal matrices. They have also been shown to be more efficient when
applied to file-diagonal systems.
5.2.8 Commercially Available Codes

Computational Fluid Dynamics has been used for numerical prediction of fluid
flows and heat transfer since the 1960s. With the development of computer
hardware, computational techniques have been vigorously proposed, tested and
refined to the point that these are recognized as cost-effective and convenient
means of obtaining detailed data for complex physical situations. Many powerful
CFD programs are available based on the three above mentioned (section 5.2.3)
numerical methods with varying degrees of modeling capabilities. Among these,
commercially available codes are: CFX-FLOW3D, STAR-CD, PHOENICS,
FLUENT, ASTEC, FIRE, FEAT, FIDAP, P/FLOTRAN, NEKTON.
Grid generation and post-processing packages are often included. Pre and post
processing are also available from specialized companies and referred to as
visualization software, such as CAD, PATRAN, FEMVIEW, UNIRAS and AVS.
The power and flexibility of the CFD packages available today are such that there
is now little need to develop purpose-build programs for specific problems.
Vector-processing and parallel processing capabilities of new supercomputers have

made some impact both on computer programming and on the development of
134
numerical algorithms that have been specifically designed to achieve a high degree
of vectorization and speed up computational time. A recent survey of a number of
CFD codes has been presented by Wolfe (1991) and by Collins and Ciofalo (1991).
(a) Finite Volume Codes. TEACH (Gosman and Pun, 1974) can be considered the
'ancestor' of many CFD codes based on FVMs. It was developed in the early
1970s at Imperial College, and could handle problems of 2-D flows; it
employed a k-e turbulent flow model and the SIMPLE algorithm. Several last
generation codes, such as PHOENICS, CFX-FLOW3D, STAR-CD and
ASTEC, can be considered as 'offsprings' of TEACH. These codes solve 3-D
flows in laminar and turbulent regimes, with steady and time dependent
conditions, in complex geometries treated by body-fitted grids. PHOENICS and
CFX-FLOW3D are based on structured grids, while STAR-CD and ASTEC
can use unstructured grids. Details of CFX-FLOW3D, which was used
throughout this research, are given in the next section.
(b) Finite Element Codes. FEAT, developed at the Berkeley Nuclear Laboratories,
is a 'classic' finite element code for CFD and heat transfer problems. Also
FIDAP is widely used. In addition to these, most general purpose FE packages
such as MSC/NASTRAN, ADINA/ANSYS, NISA-il and ABAQUS, include
CFD modules although they are less advanced than specialized CFD codes. A
review of FE methods for CFD problems is given by Lohner (1987).
5.2.8.1 CFX-FLOW3D: CODE OVERVIEW
CFX-FLOW3D, also known as HARWELL-FLOW3D up to a few years ago, is a

general purpose thermofluid-dynamic computer code born, as other computer
codes of the same family, in the early seventies and developed from the research
work at Imperial College. The original version of FLOW3D was limited to simple
staggered and rectilinear grid, Cartesian or cylindrical coordinates, (Wilkes et al.
1985, Jones et a!. 1985, 1986). HARWELL-FLOW3D was marketed with the
release 2 of the code (Burns et al., 1986, 1987). A body fitted coordinate system
was implemented in a structured non-staggered grid, and the Rhie-Chow algorithm
(Rhie and Chow, 1983) was used to prevent 'chequerboarding'. The
pressure/velocity coupling algorithm was a SIMPLE-family algorithm. Linear
equations solvers were used (Kightley and Jones, 1985), and higher order advection
treatment schemes (including HUW and QUIVK) (Thompson and Wilkes 1982).
Adaptive gridding, combustion models, coupling with radiation codes, some two-
135
hapter 5 Three Dimensional Flow Modeling
phase flow capabilities, and new advection treatment schemes were introduced,
like CCCT and CONDIF.
Description Of The Code
The suite of CFX-FLOW3D programs consists of a number of modules: (a) Pre-

Processing Modules, or Geometry and Grid Generators; (b) Interactive Frontend;
(c) Solution Module of CFX-FLOW3D; (d) Post-Processing, or Graphics Modules.
The geometry and Grid Generators may be used to define the finite difference grid.
The grid coordinates are dumped to disk in a form readable by the frontend.
Through the use of a flag in the Frontend, the program is informed that the grid
information is to be read from the disk. Use of these modules is optional. The grid
can also be defined within the Frontend. The geometry and Grid Generators include
the interactive grid generator CFX-BUILD. Body-Fitted Grids are generated using
transfinite interpolation.
The Frontend takes the input specification of the problem and converts it from a
form convenient for the user into a form designed for efficient execution. Detailed
error checking is performed. Facilities are available to provide a database which
calculates the physical properties of some common fluids. The problem is specified
in a single data file using the Command Language, that is a set of English-like
commands, subcommands, and associated keywords. In the Interactive Frontend,
this data file is constructed automatically via a series of displays on the screen.
User-defined Fortran routines may be included for features that are too complex to
be described using the Command Language.
The solution module solves the discretized representation of the problem. It

receives the information in a form that permits maximum efficiency to be obtained
on different types of computers including vector processors. The solution Module
has only a few output facilities- for example, for printing and dumping the solution
to disk files.
The Graphics Modules produce the main graphics output, interrogating the disk
files written by the solution module. Interfaces to other post-processing packages
have been constructed, and there are a number of post-processing options available
for various workstations. Use of these modules is optional, and described in the
ENVIROMENT User Guide.
136
In the present work no use was made of the pre-processing facilities such as the
interactive grid generator (CFX-BUILD former SOPHIA), since this so far does not
allow for transient grids automatically. In fact, it would have been necessary to
construct the grid for each time step. Use of the Fortran user subroutine USRGRD
was found to be more flexible to define a time dependent grid. Problem
specification was made using the Command Language file.
The main features of the code are summarized in table 5.1
Computational Domain
The generic computational domain in 2D is shown in figure 5.2
Porous
region
Symmetry plane
-1
Inlet Thin surface _________
outlets
I/
Walls
Figure 5.2 Generic CFX-FLOW3D 2-D computational domain and structures

which can be defined within it.
Multi-block grids are constructed by patching together a number of simple,

rectangular grids to form a grid that is not restricted to be topologically rectangular.
Topological features within the domain of each block are described using patches
(Boundary condition surfaces and interblock boundaries). The generic block grid
include NI, NJ, NK volumes along the 'intrinsic' directions I, J, K respectively.
The grid is defined by specifying the coordinates of the grid vertices in physical
space, arbitrarily oriented.
The generic control volume, is shown in figure 5.3. The corners of the volume are
numbered, and 'intrinsic' directions are indicated following the 'compass rose'
convention. In order to invoke boundary conditions the control volumes lying on
the perimeter of a grid block, considered outside the solution domain, are added
137
(automatically) to constitute dummy control volumes. The interior control volumes

are called active control volumes. Two dimensional problems are treated as three-
dimensional with two symmetry or periodicity planes. If cylindrical coordinates are
used, a special treatment is implemented to deal with the singularity created by the
fact that the axis is included in the computational domain.
(I,J+1,K+1
(I+1,J+1,K+1)
(I,J+ 1 ,K)
(I+1,J,K+1)
(1+1 ,J,K)
(I,J,K)
Figure 5.3 Generic control volume. Indices of corners are shown
A summary of CFX-FLOW3D commands and keywords is reported in the manual.

The schematic flow chart for the code is shown in figure 5.4. the 'inner' iterations
of the linear solvers, are nested in the loop of the 'outer' iterations, which in turn
are nested in the time stepping ioop. The 'outer' iteration process is stopped either
completely or for the current time step of a transient problem, when the chosen
convergence criteria are satisfied. The criteria are for example: maximum number
of outer iterations, the tolerance on a residual, the CPU limit. In the tolerance on a
residual criterion, iterations are stopped when the overall residual mass source Em
(the sum of the absolute mass source residual in all control volumes) falls below a
predicted value SORItvIAX.
138
Problem description
'I,
Initialization
N=N+1 Time
1
Outer iterations
N iteration = N iteration + 1
1.
linearization
'I,
N110 = N110 +1 Inner iterations
Solution of
linearized equations
Res <MaxR NO
YES
NO
rt ' N, max NO
yES
END
Figure 5.4 Schematic flow chart of CFX-FLOW3D.
139
Table 5.1
Main features and capabilities of CFX-FLOW3D, Release 4.2
1. Physical Models and Geometry
Physical Space Coordinates: Cartesian / Cylindrical
Type of Flow: Laminar / Turbulent
Steady / Unsteady
2-Dimensional I Three Dimensional
Incompressible / Weakly Compressible, Fully Compressible
Isothermal / Non Isothermal
Forced / Natural Convection
Buoyant: Boussineq approximation of fully compressible
Multiphase flow: multi-fluid model I homogeneous model
Boundary Conditions: Solid Walls
Symmetry or Periodicity Surfaces
Inlet I Outlet Faces
Pressure Boundary
Mass Flow Boundary
Turbulence Models: k-E for high Reynolds number /
Low Reynolds number model
Higher order turbulence models:
Algebraic Reynolds Stress model (ASM)/
Differential Reynolds Stress model (RSM)/
Differential Reynolds Flux model
Combustion Models: Two gaseous model 5:
eddy break-up model / mixed-is-burnt model
Particle Transport Models: Langrangian
Rectangular / body-fitted
Multi-block
Rotating coordinate system
Adaptive Grids
Other Options: Solids / Thin Walls / Porous Regions
Heat Conduction in Solids
Radiative Heat Transfer (via RAD3D code)
Chemical Kinetics for single-phase chemical reactions
Additional Scalar Transport Equations for fluid mixture
2. Numerical Methods
Finite Volume Discretization: multi-block unstructured, non staggered grid
Time - Differencing: a
Fully Implicit Backward Euler Differencing
Crank Nicolson (central) differencing
b
Pressure / Velocity Coupling: SIMPLE / SIMPLEC
Advection Algorithm Schemes: Central /Upwind / Hybrid / HUW

QUICK / CONDIF / ICCG
Linear Equation Solver: Line Relaxation I
Preconditioned Conjugate Gradient (ICCG)
Full field Stone's Method I
Block Stone's Method
Algebraic Multi-Grid
a
This option should not he used with transient gridding
b
PISO and PISOC are available on the new Release
140
5.3 Results Of The 3D Modeling

In this study the three dimensional flow modeling was performed by the use of a
finite volume CFD code known as CFX-FLOW3D. In Chapter four the design and
development of a grid generation package, that can handle the complexity of the
grid, for a scroll compressor - expander was analyzed. The next step was to
implement this transient grid generation package with CFX-FLOW3D. The final
stage before running the code was to write the Command Language that described
the physics of the problem. The Command Language is shown in Appendix C.
In solving fluid flow problems we need to be aware that the underlying physics are
complex, and the user must have skills in a number of areas. Prior to setting up and
running a CFD simulation there is a stage of identification and formulation of the
flow problem in terms of the physical and chemical phenomena that need to be
considered. Typical decisions that might be needed are whether to model a problem
in two or three dimensions, to exclude the effects of ambient temperature or
pressure variations on the density of an air flow, to choose to solve the turbulent
flow equations etc. To make the right choices requires good modeling skills,
because in all but the simplest problems we need to make assumptions to reduce
the complexity to a manageable level whilst preserving the salient features of the
problem in hand.
Flows inside scroll compressors - expanders are turbulent, time dependent,

compressible and for the case of the expander two phase. Furthermore these flows
are driven by the wall motion
When the solver runs it produces a large amount of data that has to be analyzed.
The analysis can be divided into two major categories, (a) CFD analysis, i.e. check
to see if the solution is reliable, and (b) Thermodynamic analysis i.e. use CFD as a
tool to improve understanding of the behavior of the thermodynamic system under
consideration. The following two sub-sections 5.3.1 and 5.3.2 discus the results
obtained from the CFD simulation with reference to the above two categories.
5.3.1 CFD Analysis
Three mathematical concepts are useful in determining the success or otherwise of

CFD algorithms: convergence, consistency and stability. Convergence is the
property of a numerical method to produce a solution which approaches the exact
solution as the grid spacing, control volume size or element size is reduced to zero.
141
Consistent numerical schemes produce systems of algebraic equations which can

be demonstrated to be equivalent to the original governing equations as the grid
spacing tends to zero. Stability is associated with damping of errors as the
numerical method proceeds. If a technique is not stable even round off errors in the
initial data can cause wild oscillations or divergence.
Engineers need CFD codes that produce physically realistic results with good
accuracy in simulations with finite (sometimes quite coarse) grids.
Performing the actual CFD computation itself requires operator skills of a different
kind. Specification of the domain geometry and grid design are the main tasks at
the input stage and subsequently the user needs to obtain a successful simulation
result. The two aspects that characterize such a result are convergence of the
iterative process and grid independence. The solution algorithm is iterative in
nature and in a converged solution the so-called residuals-measures of the overall
conservation of the flow properties are very small. Progress towards a converged
solution can be greatly assisted by careful selection of the settings of various
relaxation factors and acceleration devices. There are no straightforward guidelines
for making these choices since they are problem dependent.
Optimization of the solution speed requires considerable experience with the code
itself, which can only be acquired by extensive use. There is no formal way of
estimating the errors introduced by inadequate grid design for a general flow. Good
initial grid design relies largely on an insight into the expected properties of the
flow. A background in the fluid dynamics of the particular problem certainly helps
and experience with gridding of similar problems is also invaluable. The only way
to estimate errors due to the coarseness of a grid is to perform a grid dependence
study, which is a procedure of successive refinement of an initially coarse grid until
certain key results do not change. Then the simulation is grid independent. A
systematic search for grid-independent results forms an essential part of all high
quality CFD studies.
Figure 5.5 shows how the residuals change with iterations. Figures 5.6 to 5.10
show how residuals, pressure, and velocities (p, u, v, w) change with the number
of elements in the domain (grid dependency study). From figure 5.6 it can be seen
that the residuals do not change significantly for a number of elements higher than
60000. In figure 5.7 the pressure follows a similar trend to the residuals but in this
case the solution is stable for a number of elements higher than 40000. For the case
142
of u velocity which is shown in figure 5.8 stability appears to be insured for a

number of elements higher than 70000. As far as v, and w velocities are concern
they appear to be stable earlier, 30000 elements for v and 60000 for w.
As it can be seen from the figures the solution changes when the number of
elements is less than 70000 and it appears to be stable for number of elements
above that. Figure 5.11 show another important parameter that might affect the
solution, namely the length of the time step. It is obvious that the smaller the time
step the easier for the code to achieve convergence. However there is no need to
use extremely small time steps because then more computational time will be
needed. As can be seen from figure 5.11 the solution does not change for time steps
less than 5.55E-05 seconds. This is the time step we use for the 3D simulation, it
corresponds to a change of half a degree on the crank angle when the rotational
speed is 1500 RPM. For higher or lower rotational speeds the time step is adjusted
so as to reflect only half a degree change on the crank angle.
-UMOM -VMOM -WMOM -MAt
1.50E-02
1.43E-12
1.35E-02
1.28E-02
U) I .20E.02
w I.13E-02
I.O5E-o
0 9.75E-03
U) 9.00E.n
-J
8.25En
7.50E.o3
U)
6.75E..fj3
Lii 6.O0E..o
UI 5.25En
I.-
4.SOE.o3
-J
0 3•75E-.03
U) 3.00E..o3
2.25E3
I.50En
7.SOE.04
0.O0E^yJ
0 10 20 30 40 50 60 70 80 90 100
ITERA11ONS
Figure 5.5 Residuals - Iterations. Speed 3000 RPM.
143
1 .20E-O
1.1OE-03
C,,
(ti
1.00E-03
9.00E-04
E
(I)
U)
8.00E-04
0
C,,
0
7.00E-04
6.00E-04
0 20000 40000 60000 80000 100000 120000
Number of Elements
Figure 5.6 Residuals - Number of Elements. Speed 3000 RPM.
1.00E+06
9.60E+05
c'J
9.20E+05
z
U)
8.80E+05
U)
8.40E+05
8.00E^05
0 20000 40000 60000 80000 100000 120000
Number of Elements
Figure 5.7 Pressure - Number of Elements. Speed 3000 RPM.
144
1.1
1.02
U)
0.94
>,
0
0
a)
7
> 0.86
0.78
0. 7____ ______
20000 40000 60000 80000 100000 120000
Number of Elements
Figure 5.8 U velocity - Number of Elements. Speed 3000 RPM.
-2
-2.5
-.. -3
U)
E
>.'
35
0
a)
>
> -4
-4.5
-5
0 20000 40000 60000 80000 100000 120000
Number of Elements
Figure 5.9 V velocity - Number of Elements. Speed 3000 RPM.
145
-4
-8
C',
-12
0
ci)
>
-16
-20
0 20000 40000 60000 80000 100000 120000
Number of Elements
Figure 5.10 W velocity - Number of Elements. Speed 3000 RPM.
1.60E-03
1.40E-03
U,
cc 1.20E-03
1.00E-03
0
E 8.00E-04
C/)
. 6.00E-04
0
4.00E-04
2.00E-04
0.00E+00
0.00E+00 2.00E-05 4.00E-05 6.00E-05 8.00E-05 1.00E-04 1.20E-04
Time Step (sec)
Figure 5.11 Residuals - Time Step. Speed 3000 RPM.
146
5.3.2 Thermodynamic Analysis

After the CED analysis is completed the results can be used to improve
understanding of the behavior of the suction process of the scroll expander, and
discharge process of the scroll compressor. As was shown in section 5.3.1 the
system is said to be grid independent for 70000 elements or more. The results
shown in the figures of this section are obtained from a grid of 70000 elements.
In this section we use different types of graphic techniques to present the results.
These are contour plots, vector plots, and streamlines. The xy plots are not used in
this section mainly because they represent the variation of one dependent variable
versus another independent variable. A disadvantage of xy plots is that they usually
do not illustrate the global nature of a set of CFD results on one view. On the other
hand, contour plots, vector plots, streamlines and particle tracking do provide such
a global view.
5.3.2.1 Expander Characteristics

Figures 5.12 to 5.71 give an excellent understanding of the behavior of the suction
process of the scroll expander. The results cover rotational speeds from 1000 RPM
to 6000 RPM, the entry conditions are defined in section 3.4.2. What it is shown in
these figures is the result of the compressible turbulent and transient flow
modeling. It has to be noted here that in order to come to this stage the simulation
was performed in steps. The first step was an incompressible laminar transient
simulation, followed by an incompressible turbulent transient simulation, the
model used for the turbulence parameters was the high Reynolds k e. The next step
was to calculate changes of density. In this case density changes were calculated
from the equation of state but they were only affected by pressure because the
temperature was fixed. This simulation was called weakly compressible turbulent
transient flow. The final step (which is presented here) allows density calculation
to depend on the changes of both pressure and temperature. In all the above models
the simulations were performed for the rotational speed range of 1000 RPM to
6000 RPM (every 1000 RPM). It has to be mentioned at this point that the results
presented here are only the 1/10 of the total results obtained.
As far as the results are concerned it is very difficult, perhaps impossible, to

compare them with the results obtained from the quasi 1-D modeling (chapter 3).
In that case the expander show good overall working performance. However the
3-D modeling which is more advanced and much more accurate indicates that this
147
is not 100% true. A careful examination of the figures that show the pressure
contours will show that there are quite significant pressure differences within the
flow domain. At higher rotational speeds these pressure differences are higher. The
velocity vector graphs show that there are areas of re-circulation both in z and y-
planes. If we look at the stream lines figures we will see that the flow pattern is not
so good. At the beginning, the flow is coming in from the inlet pipe with a constant
velocity, it hits the bottom wall of the expander and as a result there is a build up
of pressure at that point. Then the flow goes in all directions hitting the side walls
while the velocity begins to decrease. After the flow hits the side walls, the velocity
decreases further and there also is re-circulation.
From this simulation it is obvious that the scroll expander will have major
problems that will affect its operating behaviour (specially during the suction
process). The reason for this is mainly the significant pressure drop that takes place
during the filling process. At higher speeds the pressure drop within the working
volume is of the order of 2 bar. From this modeling, and with the use of the
visualisation packages, we manage to get the feeling of how the fluid will behave
in such a machine. To give a better understanding of what we learn let us examine
the case of an expansion that takes place in a turbine or in a screw expander, and
then compare it with the scroll-expander. The reason for doing this is to identify
the major problem of the scroll expander. What happens in those cases is that the
flow 'in' is coming in a completely different way, angle of attack, and the machine
will also behave in a completely different way. The flow hits the blades of the
expander and as a result of this there is a movement of the blades. In our case the
fluid is coming from the top but the expected movement of the spirals is horizontal,
(90° degrees difference). Unlike the case of the screw expander the flow 'in' in a
scroll expander hits first the bottom of the machine (stagnation point occurs there),
which results in the build up of pressure there, and then the residual flow moves
the wall spirals. This is also the main reason for the fluid to swirl. Please note that
all these facts are taking place during the suction process of the scroll expander. It
looks like things will improve during the expansion process, but the results are not
clear in that case.
148
5.3.2.2 Compressor Characteristics

As for the case of the expander, the simulation of the discharge process of the
scroll compressor was also performed in steps, the same ones as for the case of the
expander. However the results in this case, which are shown in figures 5.72 to
5.143 indicate that the scroll compressor is an excellent positive displacement
machine. We have approximately uniform pressures within the domain. The
velocity vectors shows that there is a small area of re-circulation closed to the
outlet pipe, but this is something reasonable for this type of machine. The stream
line figures show that some swirling is taking place, mainly in the outlet pipe.
Close to the walls the velocity is relatively low and uniform. As the walls move
inwards the fluid velocity near the pipe increases and takes its peak value inside the
discharge pipe. The velocities within the pipe are not uniform and are higher on the
side of the moving spiral. Almost under the pipe we can see that there is some area
of re-circulation. This area increases as the flow approaches the inlet of the
discharge pipe. Now as far as the pressures are concerned it can be seen that
everywhere in the domain except from the area below the discharge pipe the
pressure is at its highest value and is also uniform. As the flow approaches the
discharge pipe from all sides small pressure drops begin, these pressure drops
continue smoothly until about the middle of the discharge pipe where they stabilize
for the rest of the outlet pipe.
The midplane (Z-plane=-O. 1) velocity vectors characterize the nature of the flow
within the entire volume. The flow vectors indicate the nature of the flow
depending upon the location. Three dimensional flow tends to move vertically
upwards as it approaches the central section of the volume which is directly below
the discharge port. The flow characteristics exhibited here possess a strong axial
component combined with an in-plane (horizontal) vortex component occurring in
all axial levels. The results show that this basic vortex flow pattern persists in the
area below the discharge port throughout the entire axial extent of the flow volume.
The calculations indicate that the flow passing through the discharge port is not
uniform.
149
Chapter 5 Three Dimensk'nalFkw Modeing
9.52821
1J 9.51361
9.49901
9.4844F
9.4698k
9.4551 F
I 9A405f
Figure 5.12 Scroll-Expander. Rotational speed 1000 RPM, Pressure contours,

Z-plane =-0.2, step100, 50 degrees (8.333m sec) after suction commences
I.4(j29--01
1.1691 F + 01
9.3530E+0O
7.0147E^0O
4.6765E^00 'N
2.3382FÔ0
I o+oo N \ \
- N
ft
- - - . '- I I I
_-\ Ill
- .---
I/Il
..-/ / I \
lIlt/I
---/11
. jff/f
- -_---. . - /
I I / ! / / I
- . ---------.. .
/ ri/I
/1. •-' ' '
_,r ti / (z r /I
I
• / V / / / / / / / / , - /
/////// / I \ /
I / / / / / / / t - - - - - / /
/////// / .
I
1 1 / / / / / \ -
-
- - - - ,
/11//I /
Figure 5.13 Scroll -Expander. Rotational speed 1000 RPM, Speed vectors,
Z-plane =-0.018, step= 100,50 degrees (8.333m sec) after suction commences
150
Chapter 5 Three DfrnensonalFk,w Modeling
9.5282E•
9.5136E
9.4990E
9.4844E
I9.4551E
9.444J5R

Z-plane =-0.1, step= 100, 50 degrees (8.333m sec) after suction commences
1.4029E±01
1.1691E±01
93OL
7.01 47E-4-O0
4.6765E±00
2.3382E-I-00 .-
I 0.00001:+0o --
I \ I
r II
- -. - -
• ,lf l
I'
S 5
111
1
I,
-.5-.--. .-.-.--.-.. . I
.5 . .- -.
---.---.---.--------.-• I,
,- / I
- -• - ..------------•---• -5- -
S' . .--.--.----. -5.-. _._,/
Figure 5.15 Scroll-Expander. Rotational speed 1000 RPM, Speed vectors,

Z-plane -0.1, step= 100, 50 degrees (8.333m sec) alter suction commences
151
Chapter 5 Three D/mensk.,naIF.bw Modehfr,g
9$282E4
95136E4
9.4990E-4
9.4M4E-4
9.4(,9E-
9.455 IE-I
I 9.441)5E-1

Z-plane -0.0025, steplOO, 50 degrees (8.333m sec) after suction commences
I.4029E±01
1.169 IF +01
9.3530E+00
7.0147E±O0
4.6765E^00
2.3382E-i-0O
; 0.0000E±00 S
1 1
I
I S
II
Ill 1•''
11111'
Ii

Z-plane =-0.0025, step=100, 50 degrees (8 333m sec) after suction commences
152
Chapter 5 Three Dknens*'nalFk'w F*deh'ig
9.5282E±
U 9.5136E+
9.4990R+
9.484411 ±
9.469811+
I 9.455111+
- 9.4405E±

Arbitraiy plane, step= 100, 50 degrees (8.333m sec) after suction commences
1.402
U 1.169
- 9.353
7.014
4.676
2.338
0.000

Arb itraiy plane, step= 100, 50 degrees (8.333m sec) after suction commences
153
Chapter 5 Three Dimensional Fk'w Modeing
12.3769
10.3141
8.2512
6.1884
4.1256
2.0628
I 0.0000
Figure 5.20 Scroll-Expander. Rotational speed 1000 RPM, Stream lines

steplOO, 50 degrees (8.333m sec) after suction commences
, 12.3769
10.3141
8.2512
6.1884
41256
2 . 0628
I0.0000
Figure 5.21 Scroll-Expander. Rotational speed 1000 RPM, Stream lines,

step=100, 50 degrees (8.333m sec) after suction commences
154
Chapter 5 Three Dnensk'nIF4bw I
. 9.5282E
9.5 136[
9.4990E
• 9.484-4E
9.4698E
9.4551E
I 9.4405E

1.41J29E+O1
1. 169 IEÔ1
7.0147EÔO
4.6765E-4—OO -
2.3382EÔO . - - N N N \ \ \
9353OE+(J()
,, ---j- _
- I I I / / I
Ill//I/I
/1/ tIf/fI//tI
- / /
__•/_•/// -7/1
1 ..• .. .
1
I/ / .-' v- - - -. - — - - - / / / /
/ / / / / / / J \ \ - - ' / /
/ / / / / / / , A - - ' /
\1//////i I
\I1//////sI\\
--.
/ / J
\\_\_ / •
S S
S
S 'S 'S .
-
-
-
-
' 'V
S S
-
S S S S S 'S

Z-plane =-0.01S, step=200, 100 degrees (16.66m see) after suction commences
155
Chapter 5 Three D/menskna/ Flow £*deIng
. 9.5282E
9.5136E
9.4990E
9.4844E
9.4698E
9.4551E
; 9.4405E

Z-plane =-0. 1, step=200, 100 degrees (16.66m sec) after suction commences
i4O29E--(jI
1.1691E+O1
93530E+OO
7.0147E-i-OO
/ /.. ///it11l
/1 -ii1f 111'
I
7171
I -
TI'
- / / / 7 1

Z-plane =-0.1, step=200, 100 degrees (16.66m sec) afler suction commences
156
Chapter 5 Three Dimensk,nalFkw MOGW ng
9.5282E
95136E
9.4990E
9.4844E
9.4698E
9.4551
! 9.4405E

Z-plane =-0.0025, step=200, 100 degrees (16.66m sec) after suction commences
l.4029E^(J1
U 1. 169 1EÔ1
9.353OE--UO
70147E+oO
4.6765E+OO -
23382E-i-OO
; O.00(JOE±UO - ' \ ,'
-..
\\\
'
\ \
\ \ \
\
I I ¶ I I I
S
'11IIIIII
-
.iIIIIIII
\-.' --.-.-.-
-_,,rtIIII I
: j7""
Figure 5.27 Scroll-Expander. Rtationa1 speed 1000 RPM, Speed vectors,

157
Chapter 5 Three Dimensional Fk'w ModeI'g
9.528:
U 9313
9.499
9.484
9.469
9.455
9.440

Arbitraly plane, step=200, 100 degrees (16.66m sec) after suction commences
1.4029E+01
U i.1691E+01
- 9.3530E+04)
7.0147E±0()
4.6765E ±0))
2.3382E+0()
I (I.(XNX)E +01)
- ..
I.. . .-

Aibitraiy plane, step=200, 100 degrees (16.66m sec) afler suction commences
158
Chapter 5 Three DimensA nalFk,w Modeling
14.5028
12.0856
9.6685
7.2514
4.8343
2.4171

14.502E
12 . 085
9.6685
7.2514
4.8343
2.4171
I 0.0000

StCp2OO, 100 degrees (16.66m sec) after suction commences
159
Chapter 5 Three Dimensional Fkw ModeNg
i 9.8562E+(
9.725OE+
9.5938E±I
Li 9.462xiE+
9.3314E^
9.2OO2+
I 9.0690E^

Z-plane =-0.2, step= l00, 50 degrees (2.777m sec) after suction commences
4.2091F+01
3.5075E^ 01
2.8060E-i-01
I I I I I
-, / / / I / /
/
/
t
/
/ / / / / / / / 1 .
\ -N--; - / / '
\////////1 t\ .
\///////I 4 t\ .
\///////I \
\ //////I
/
\ 1 1 / /
/ / I I \ \ - - - - - ,
--,
1 / /
/ / /
' 'A' 1
\i/ i /ii '(\ -
/ / 1 / I /
- - -

160
Chapter 5 Three Di,nenskrnaIFkw A*deg
9.8562E-
9.725013-
9.5938E-
9.4626E-
9.3314E-
9.2002E-
9(369013-

Z-plane =4).!, step= l00, 50 degrees (2.777m sec) afler suction commences
4.2O9lF+O1
3.5075E±01
2.8060E+01
2.104513+01
I .4o3oI+o1
7.0151E+0()
(1.0000E+00 _____.:__-:--.:._:7.... "\\
,7
I........... I\. .-.nhl111,,
........-
. -. -. -.---....----._------.... . . I I
I I I
i '--- .. - .----.-..--..----..-.--.--.. .
I I,/_._.-_.._._________/1Il II-.
/ / ..--------.----.--------. __._____,1
I II
/ . •.— _._____...__,,!1 IT '
/. .___,,pt1J

Z-plane =4).!, step=100, 50 degrees (2.777m sec) after suction commences
161
Chapter 5 Three Dimens4,naIFk'w /i*xlelfrig
. 9.8562E+
9.725013±
9.5933T3+
9.462ii+
9.3314+
9.200213-1
9.0690134

4.2(J91E+01
3.5075E^01
2.8060E+O1
' ---.-.--------.-.
jI
rtI1l••

Z-plae ='-0.0025, step=100, 50 degrees (2.777m sec) after suction commences
162
Chapter 5 Three Dimensional Fbw Modehig
9.8562E+05
9.7250E+05
95938E+05
94626E+OS
9.3314E+05
9.2OO2E±O
I 9.O9OE+O5
Figure 53S Scroll-Expander. Rotational speed 3000 RPM, Pressure contours,

Arbitrary plane, step= 100, 50 degrees (2.777m sec) after suction commences
42091€-4
33075E-1
2.8O6O13
2.1045E-
i 4)3OE-
7i)1 51 F.-
I O.o000u

Arbitrary plane, step= 100,50 degrees (2.777m sec) after suction commences
163
Chapter 5 Three DimensA na/FA bw Mode/rig
37.0303
30.8585
24 . 6868
18.5151
12.3434
6. 17 17
I0.0000

stejP 100, 50 degrees (2.777m sec) after suction commences
37.0303
30 .8585
24 . 6868
18.5151
12.3434
. 6.1717
I 0.0000
FigUre 5.41 Scroll-Expander. Rotational speed 3000 RPM, Stream lines,

SteP100, 50 degrees (2.777m sec) after suction commences
164
Chapter 5 Three D/niens*,nalFbw it*ic
9.8562E-I-
9.7250E-f
9.5938E+
9.4626E-l-
9.3314E+
9.20O2E
I 9.0690E±

4.2091E±O1
3.5075E-i-01
2.8O6O1±(j1
2.1045E-i-01
I .4030E-i-01 N \
7.015 1E+0O \ \ \
N N N
I O.0000E+tjO -.N\\\
II,!
/
111/
I
1 '
/
///////J I
j/////// / 1
It \\

165
Chapter 5 Three Dimensional Fbw Modeling
9.8562E-4-O
9.7250E±O
9.5938E+O
9.4626E-i-O
9.3314E±O
9.2002E+O
I 9.O69OEU

I 4.2Q91E±O1
3.5075E+Oi
2.8060F.±O1
2.1Q45E+O1
I .4030E±() 1
.0151E±OO - \ \ --
I O.0000E±OO . N \
...........
- s\\\\\\\'
• -. . .' ' ii 1111
,!1IlT1iI

166
Chapter 5 Three Dmensk,naIFkw Modeling
9.8562E±O.
II 9.7250E±O.
9.5938E±O
9.4626E+O
9.3314E+O
9.2002E+O
; 9.0690E±O

Z-plane =-O.0025, step200, 100 degrees (5.54m sec) after suction commences
- 4.2091EÔ1
- 3.5075F±O1
2.8060E±O1
2.1045E-1-01
i.4030E±O1
7.O151}.ÔO N \
I 00000F+OO \\
'.55.-..
• 'N --
'S..,., Ii
- I I

Z-plane =-0.0025, stcp=200, 100 degrees (5.54m sec) after suction commences
167
Chapter 5 Three DmenshnalFk'w Modeling
, 9.8562]
9.72501
95938]
9.4626]
9.3314
9.20021
I 9.0690
Figure 5.48 Scroll-Expander. Rtationa1 speed 3000 RPM, Pressure contours,

Aibitiaiy plane, step=200, 100 degrees (5.54m sec) after suction commences
. 4.2091E±OI TT
3.5075F+O1 '
2.8060F^01 j
2.1045E+O1
1t.44J30E+Oi 4.
7.0151E+O0
I 0.0000E+00
:jH ;:::.•.
- -../.,;i. H, ---.-t.-. ----------.----,
Mbitraiy plane, step=200, 100 degrees (5.54m see) after suction commences
168
Chapter 5 Three DimensK,nal Fbw Modeling
43.5978
36.3315
29.0652
21.7989
14.5326
7.2663
I0.0000

43.5978
36.3315
29.0652
21.7989
14.5326
7.2663
I 0.0000

169
Chapter 5 Three DIrnenskwz9IFA ,w f*ick'I/ng
1.084
1.032
I 9.275:
8.750:
8.2251
I 7.701:
Figure 5.52 Scroll-Expander. Rotational speed 6000 RPM Pressure contours,

Z-plane r..O.2, step= 100, 50 degrees (1.389m sec) after suction commences
8.4'28E+O1
7.0440E+W
5.6352i+O1
////// / / 1 - ,/
/////// / I
t ----/
////// / / 1
\ ///// / / I ---,/
\I / / / / / / 1 1
--/
\i / / / / / / I I
\: I I / / / 4 1 -,/
\I1 / /44 \
1/I

170
Chapter 5 Three Dirnensk'nalFkw Fkde¼ig
1.0849
1.1)325.
. 980'J'J
9.2753
8.7505
8.2258
I 7.7011:

Z-plane =-0. 1, step= 100,50 degrees (1.389m sec) after suction commences
8.4528+O1
U 7.0440E+(>1
I
5.6352.+O1
- - -. .__..._.________-/ I
___-.____..__._.__._-,/I1Il
- / ,
, ., ..-.-.--- ..--.-
t
Figure 5.55 Scroll-Expander. Rotational speed 6000 RPM, Speed vectors

Z-plane -0.1, step=100, 50 degrees (1.389m sec) after suction commences
171
Chapter 5 Three Dimenstnal Fbw Modellng
1.0849F
1.03251
9.80001
9.27531
8.75051
I 8.22581
7.70111

Z-plane =-0.0025, step= 100, 50 degrees (l.389m sec) after suction commences
S.42b±O1
7.04-40E+(il
4.2264E±O1
2.8176E+O1
1.4os1rô1
5.6352i+O1
ftlXiOOE+O() ' '
. .
I' 0
-
:\
00 _.00000____ -. I
/
.- - - 0 _0' ,
\ -.-------- -.
.--, It
I
- - - - -
--
•0 000 II
0-I
Figure 5 • 57 Scroll-Expander. Rotational speed 6000 RPM, Speed vectors,

Z-plane =-00025, step=100, 50 degrees (1.389m sec) after suction commences
172
Chapter 5 Three D/menskrnal Fk'w Modeling
1.0849E+06
1.0325E+06
9.8OOOF+O5
• 9.2753E+05
8.7505F.+05
8.2258E+05
I
7.70 11E^05

Ajj,itraiy plane, step= 100, 50 degrees (1.389m sec) after suction commences
• 8.4528E+
7.0440E+
5.6352E+
4.2264E-f
18176E f
1.4088E+
I ().(X)O(1E4

AEbitxaly plane, step= 100, 50 degrees (1.389m sec) after suction commences
173
Chapter 5 Three Dfrnenss ronalFk,w Modeling
74.4989
62.0824
49.6660
37.2495
24.8330
12.4165
I0.0000

step= 100, 50 degrees (1.389m sec) after suction commences
n 74.4989
62 . 0824
49.6660
37.2495
24.6330
12.4165
I0.0000

stCpl00, 50 degrees (1.389m see) afler suction conunences
174
Chapter 5 Three DinensknalFk,w odeling
1.0849E
1.0325E
9.8000E
9.2753E
I 8.7505E
8.2258E
7.7011E

8.4528E^01
7.1J440E+01
5.6352E+01
4.2264E+01
2.8176E+01 SS' \ \
1 .4088E+01 ss-SssS.sSs
I
O-0000E^ 00
S S
1// / // / :
/ /
I I / / / / / 1 \ \ - - - /
•/////// I \ \

Z-plane =-0.ois, step=200, 100 degrees (2.778m see) after suction commences
175
Chapter 5 Three D/mensk.rna/Fkw tIk?dellng
1.0849E
1.0325E
9.8000E
! 9.2753E
8.7505E
8.2258E
I 7.7011E

t452E+UL
7.0440E+OI
5.6352E+1
4.2264E-l-01
2.176EÛ1
1.4088E+O1 .-_ - \
; ()000(JE+(X)
\ \ ......\
S.S5'\\\
5-
. ._..\•'\\S
-5 S
: .,
- ........... S 7 7 ¶
'I1 11
- S -.
S.
•
.5
- - - , ,
1 1 1

176
Chapter 5 Three Dimensional Fbw A*idehfrg
1.0849E-
1.0325E-
m 9.8000E
9.2753E-
8.7505E
8.2258E-
I 7.7011E-

Z-plane = 0 0025, step200, 100 degrees (2.778m sec) after suction commences
452SE+Ol
7.0440E+O1
42264E+O1
2.8176E+OI
L4088E-.-01 \ ' \ ' '
5.6352E+O1
O.O(XUJE+OO \ \ \ \ \ \ \ \
- --.'
-- S \\ " \
\\ \
- S I \ \ \ \
I III
- 1 1111
1llIIII
7IIII1I'''
- III 1
I I 1
' 5' -' -.--.---.---
5.,7
I Ill

177
Chapter 5 Three Dinensk)rlalFkw Modeling
- 1.0849
- 1.0325
9.8000
9.2753
8.7505
8.2258
I 7.7011
Figure 568 Scroll-Expander. Rotational speed 60')') RPM, Pressure contours,

Arbitraiy plane, step=200, 100 degrees (2.778m sec) after suction conunences

Arbitrazy plane, step=200, 100 degrees (2.778m sec) after suction commences
178
Chapter 5 Three Dimensional Fkw fIk,de#ng
87 .3844
72.8203
58.2563
43.6922
29.1281
14.5641
I0.0000

step200, 100 degrees (2.778m sec) after suction commences
87.3844
72.8203
58.2563
43.6922
29.1281
14.5641
0.0000

step200, 100 degrees (2.778m sec) afler suction commences
179
Chapter 5 Three Di,nensAna/ F.bw M,de/ing
9.5188E
Li
9.50 13E
9.4926L
9.4838
9.4750E
I 9.4663E
Figure 5.72 Scroll-Compressor. Rotational speed 1000 RPM Pressure contours

Z-plane =-0.2, step=100, 50 degrees (8.333m sec) after discharge commences
L2668F.+01
1.055613+01
8.445113+00
6.3338E+0(J
4.222613+00
2.111313+00
I 0.000013+00 - - - -.
, -. - -. - - -. - -..--
\ '
I'
,_-/// ////__- I,
,,i/1/ lit,,'
,,jji17
/ / / / / / 1 /
I''''''
1\\\"..\li//i/I/I
itrti111 \\
/
1 1 \ \ -..-----....____
--- - - --------
t I I I
\ \ \ \ \ \ \ I \ \
\ \ \ \ \ \\\ \
\
\
\ \ \\\\\\
\ \ \\\\\ \ \
\ \\\\\\
'.' \ "\ '\
V
NNNNN
Figure 5.73 Scroll-Compressor. Rotational speed 1000 RPM, Speed vectors,

Z-plane =-0.018, step= 100, 50 degrees (8.333m sec) after discharge commences
180
Chapter 5 Three £nensA ,nalFkw Modeling
95188]
9.51011
9.50131
9.49261
9.48381
9.47501
I 9.46631
Figure 5.74 Scroll-Compressor. Rotational speed 1000 RPM, Pressure contours,

I.266R+Oi
1.0556E+(J1
8.4451E+(J(J
/
I IltlI\
( tIt(fllItl\ -.-.-------.-.---,--
I
lilt It
11111111
11111111
\ It I' ' ' '...._._.__.____

181
Chapter 5 Three Dimensional Fk'w Modeling
9.5188}
9.510fl
950131
9.49261
9.48381
9.47501
I 9.46631

Z-plane =-0.0025, step= l0O, 50 degrees (8.333m sec) after discharge commences
1.1J556E+(J1
8.4451E±OO
I, i
.-.,,S,SS -
I i\ -• --_ ___, - --
I \ \' -- - - -
I I S ' _____ - -
* * * * * s \ '- _ - -- -
* * * ' ' ' 5' •S_ _______ - - - -
' * * **,.%' '5 ---------- -
SS \ \ * \ \ \

Z-plane =-O.0025, step= 100,50 degrees (8.333m sec) after discharge commences
182
Chapter 5 Three Dinensknal Fkw Modekig
9.5188E+05
L 9.510111+05
9.501311+05
1 9.492611+05
9.483811+05
9.4750E+ 05
I 9.466311+05

p bitrary plane, step= 100, 50 degrees (8.333111 sec) after discharge commences
1.26(i8E+01
U 1.0556E+01
• 8.4451E+00
0.33381L+00
4.222611-i-OIl
2.111311±00
I 0.000011+00
1T

Arbitraiy plane, step= 100,50 degrees (8.333m sec) after discharge commences
183
Chapter 5 Three D!mensk,nal Fkiw Modeling
15.7513
13.1261
10.5009
7.8757
5.2504
2.6252
1 0.0000
Figure 5.80 Scroll-Compressor. Rotational speed 1000 RPM, Stream lines

step= 100,50 degrees (8.333m sec) after discharge commences
15.7513
r
13.1261
10.5009
7.8757
5.2504
2.6252
I 0.0000
Figure 5.81 Scroll-Compressor. Rotational speed 1000 RPM, Stream lines,

step= 100, 50 degrees (8.333m see) after discharge commences
184
Chapter 5 Three Dimensk,n/ Fkw A*deNng
15.7513
13.1261
10.5009
7.8757
5.2504
2.6252
I 0.0000

step= 100, 50 degrees (8.333m sec) after discharge commences
15.7513
13.1261
10.5009
7.8757
5.2504
2.6252
0.0000

185
Chapter 5 Three Diniensk,nalFkw Modeling
93188E±
9.5101E+
9.sal3E+
9.4926E+
9.4838E+
9.4750E-I-
I 94663+

Z-plane =-0.2, step =200, 100 degrees (16.66m sec) after discharge commences
1.2óôE+O1
I 0556E+()
6.3338E+OO
4.2226E+OO
2.1113F+OO
84451E+OO
O.0000E±OO - -, -: - -- - -
- - / _/ 7 7 -, - - -. - .- .-.
/ / / / / / / -. - I
4T' '
I I 1 1 1 I \ \
I 1 I I I 1 I I I
1 1 1 1 1 1 I \ \ N\NNNNN-----------
I I 1 1 1 1 I I \ 'N NN-N
I I I I I I i I N N NNNNNNNN NNNN
\ NNNNNNNNNNNNNNNNN
\ \\\\\NNNNNNNNNNNNNNNNNNN N
\\\\\NNNNNNNNNNNNNNNNNNNNNN
\ \ \ ANNNNNNNNNNNNNNNNNNNN /
\\ \ \ I \ N N N N N N N N N N N N N N N N N N
\\ \ I NNN N N NN N NNNN NNNNN /
\ \ \- N N N N N N N N N N N N N N N /
\\ \\ \N N N N N N N N N N N N N N N N/
\\

186
Chapter 5 Three Dimensional Fkw MdeIing
93188E+(
9.5101E+(
93013E+(
r 9.4926E+(
- 9.4838E+
I 9.4663E-*-

Z-plane =. 0. 1, steprr200, 100 degrees (16.66m sec) after discharge commences
1.2(,6SE±O1
I .O556i+() I
8.445 IE+OO
I IIII1III\\\\NN-.
III 1 I
I I I I I
\IIIIIIII
ç: :
111111

Z-plane =0.i, step=200, 100 degrees (16.66m sec) after discharge commences
187
Chapter 5 Three Dinenscna/Fk'w Modeling
i 93188Ei-
E&1 9.5101E+
93013E^
9.492lE+
I9.4750E+
9.4663F+

1.2668E+OI
- I.055(,F+Ol
8.4451E+OO
6.3338E+OO
4.222(E+U()
Iui ii - - -
I -_______________ -
It
TI, I liii'' \ \ N-_._... ._ - - -

Z-Plane =-0.0025, step'200, 100 degrees (16.66m sec) after discharge commences
188
Chapter 5 Three Dimensx'nalFbw Modeling
9.5 188E+U5
95101F.+05
Y.5O13F+05
9.4926E±05
9.4838F.+05
9.4750E+05
I 9.4663E+ 05

Arbitrary plane, step=200, 100 degrees (16.66m see) after discharge commences
1.2&,8E+O1
rn 1.0556E+O1
8.4451E+0O
C,.333SE+OO
4.2221,E-t-OU
21113I+OO
I O.0000Li+Oo

Arbitrary plane, step200, 100 degrees (16.66m see) after discharge commences
189
Chapter 5 Three Dimenskn9/ Fkw A*xg
13.0747
10.8956
8.7165
6.5374
4.3582
2.1791
I0.0000

step200, 100 degrees (16.66m sec) after discharge commences
13.0747
10.8956
8.7165
6.5374
4.3582
2.1791
I 0.0000

190
Chapter 5 Three Dfrnenskrnal Fbw Modehfhg
13.0747
10.8956
8.7165
p6.5374
4 .3582
2.1791
I 0.0000
A TIf I
\\ -.

step=200, 100 degrees (16.66m sec) after discharge commences
13.0747
10.8956
8.7165
6.5374
4.3582
2.1791
0.0000

step =200, 100 degrees (16.66m sec) after discharge commences
191
Chapter 5 Three Dimensknal Fkiw ModeiXsg
9.758E+(
9.6792F.+
9.5996E+
9.5200F. + I
9.4405E+
9.360913±
I 9.281313+

_ 3.fl83F.+01
3.1819E+01
2.5455E+01
1.9091E+01
I
1.272813+01
6.3638E+OO \\\ \
\\ \ \ I
/1 \\
11'/
, / I / / / .' -\
I!,! I I I \\,\.\ , , / / / / , .- -
1 1 I I \ \ \ / /
/ / 1
____________
\ \ I \ \ I ¶ I \\\\N' .-.......____.___ ____
\\\ \\\\\\\ \ \'
\ \ \ \ \ \ \ \ \ \\NNNNNNN
\ \ \ \ \ \ \ \ \ \ N N NNNNNNNNNNNNNN
\ \ \ \ \ \ \ \ \ AN N N N N N N N N N N N N N N
\\\\\\\\N(
\\\\\\\\\ \NNNNNNNNNNNNN/
\\\\\\\\ \NNNNNNNNNNN/
NNNNI
\NNNN
NNNN

Z-plane =-0.018, step= 100,50 degrees (2.777m see) after discharge commences
192
Chapter 5 Three skrnalFkw Modeling
9.75S8E+
9.6792E+
9.5996E ± I
9.5200F + I
9.4405E+
9.3609E+
I 9.28 13E+

Z-plane =-0. 1, step= 100, 50 degrees (2.777m sec) after discharge commences
3.4W3F±O1
3.1819E±O1
25455R+Ol
L9t)VIE+O1
12728E+O1
6.3638E+OO ' '
I 000n+oo
iii ill
I iiT1llll\\ - --------
71
7% \ 7777 \_\_-__ .-.------.--

77777' 1 I --------

Z-plane =-0. 1, step= 100, 50 degrees (2.777m sec) after discharge commences
193
Chapter 5 Three DimenskvialFbw Modeling
9.7588E-I-
U 9.6792F.+
9.5996E-i-
Li 9.5200E-f
9.4405E-f
9.3609E#
! 9.28

Z-plane =-0.0025, steplOO, 50 degrees (2.777m sec) after discharge commences
3.U3E-4-01
3.IS19E±O1
25455F+Ol
Iii
- / - -
III _____
I I I \ \\ ___ -
I' \ \- - ------ - -
II I I s ------------- - -
kIII I I ----------- - -
I S S S S S \I' _ SS__._ - -- -
S S S S S S S S S S - - -------------

Z-plane =-0.0025, steplOO, 50 degrees (2.777m sec) after discharge commences
194
Chapter 5 Three Dimensional Fk'w i% *deling
9.7588E+O
9.6792E+O
9.5996E+()
9.5200E-1-O
9.3609E+O
9.4405E+O
9.2813E+O
Figure 5.102 ScrollComp reSSOr. Rotational speed 3000 RPM, Pressure contours,
Arbitraiy plane, step= 100, 50 degrees (2.777m sec) after discharge commences
38183E+O1
rn 3.1SI9E+O1
L9091E+O1
12728E±O1
6.3638F±OO
2.5455E+O1
O.0000E+OO
I iI
E4
Arbitrary plane, step= 100, 50 degrees (2.777m sec) after discharge commences
195
Chapter 5 Three Dimenskna/Fk,w /i*xhng
47.2646
39.3872
31.5098
23.6323
15.7549
7.8774
I 0.0000

step 100, 50 degrees (2.777m sec) after discharge conunences
47.2646
39.3872
31 .5098
23.6323
15.7549
7.8774
I 0.0000

196
Chapter 5 Three D/mensAona/Fk,w McsdeAr,g
47.2646
r
39.3872
31 .5098
23.6323
15.7549
7.8774
I: 0.0000

47.2646
39.3872
31. 508
23.6323
15.7549
7.6774
0.0000

197
Chapter 5 Three D/mensvna/Fk'w A*cg
9.7588E+05
9.6792E+05
9.59961+05
L 9.5200U+05
9.4405E+05
9.3609E+05
I 9.281313+05

3.8183E+01
3.1819E+0I
2.545513±01
I .272E±01
6.3638E±0()
1.909113±01
0.000013 I 00
/ //
, / / / / / / / / //
If//fill ' ' ii
/ , / I I I I I 1 1 \ \ \\\. I)//
;,1fIIT11ll\ \
1 1 1 1 I 1 1 1 \ \\\\ N_.
I I I I I I \ \
1 I Il I Il I \\NN\NNN...........-----
I I \ I I I \\NNN\NNNNNNNN . --
\ \ \ \ \ \ \ \NNNNNNNNNNNNNNNNN
\ \ \ \ \ \ NNNNNNNNNNNNNNNN
\\\\\NNNNNNNNNNNNNNNNNNNNN
AN N N N N N N N N N N N N N N N N N N N /
\ \ \ I \N N N N N N N N N N N N N N N N N N N N /
\ \ NNNNNNNNNNNNNNNNNN/
NNNNNNNNNNNN/
NNNNNNNNNNNNNNN
NNNNNNNNNN
NNNNNNNNN

198
Chapter 5 Three DfrnenskrnaIFbwMockng
_ 9.758813+05
9.679213+05
j 9.599613±05
9.520013+05
9.440513+05
9.360913+05
I 9.281313+05

Z-plane =-0. 1, step=200, 100 degrees (5.54m sec) after discharge commences
3.183E+01
3.11I9E-1-0J
1.9091 E+U1
I .272KE-f-() I
6.3638E±0() -
2.5455E±01
(UXJ{)0E±0() - -- --. .-. - - ...-
- - ,_-__ ......-.---..-----
- / //_-'/ , . .
--------.----.-----'
,,/I/Il/•
,,,,,,tii
::1
I / / I / / 1 •\ I / /
, iitiill .. .
/ / - -
T ,f f !l1l11\ .---.
tilt II 1\\\\\N' ...

Ii 1 I I
tilt Ii I I' ''.'.'..
1111
\ \\ \ '
"\

Z-plane =-0.i, step=200, 100 degrees (5.54m sec) afler discharge commences
199
Chapter 5 Three Dimensional Fbw MxNng
9.7588E+05
9.679213+05
9.599613+05
H 9.520013+05
9.440513-4-05
9.360913±05
I 9.281313+05

Z-plane =-0.0025, step=200, 100 degrees (5.54m see) after discharge commences
3.8183E±O1
3.II9E-l-0i
2.5455E±01
I , ii' .-- -.-.-----. -

I\\\ .___.__________ -
II
II I I \ \\\ - - -
I\S N N.'..-.___.__- - -
Ii
1 I
S S S S S

Z-plane =-0.0025, step=200, 100 degrees (5.54m see) after discharge commences
200
Chapter 5 Three DiinensxnalFk'w Modefing
9.7588E+05
9.6792F+O5
• - 9.5996E+05
• 9.5200E+05
- 9.4405E+05
9.3609E+05
9.2813E+05

Axbiliaiy plane, step=200, 100 degrees (5.54m sec) after discharge commences
3.8183E+O1
3.1819J±O]
23455E±O1
L9O91J±O1
1.2728E+O
6.3638E+O(
I O.0000E-i-O(

Aibitrary plane, step='200, 100 degrees (5.54m sec) after discharge commences
201
Chapter 5 Three Dinienskr,al Flow Modeling
39.3890
32.8242
26.2594
19.6945
13.1297
6.5648
I0.0000

39.3890
32.8242
26.2594
19.6945
13.1297
6.5648
I 0.0000

steP200, 100 degrees (5.54m sec) after discharge commences
202
Chapter 5 Three Dynenskna/Fbw x/e/ing
39.3890
32.8242
26.2594
19.6945
13.1297
6.5648
I 0.0000

\
39.3890
32.8242
26.2594
19.6945
'•1
13.1297
6.5648

203
Chapter 5 Three DimenskrnalFkw A*ck*ng
1.0595E+
I 1.0276E+
9.9576F+
9.638E+
93200E±
9.0011E+
I 8.6823E+

Z-plane =-0.2, stepr= 100, 50 degrees (1.389m see) alter discharge commences
7.636(JE+U1
6.3633E+U1
3.81OE+O1
25453EÔ1
1.2727E-*-01
5i)907E+U1 \ \
O.0000EÔO - - -
\ \ \ I
\ \ \
\ \1
it/
I I I /
/ I / / / / / //N
I I I I ' -
ti/f l
I I III/
,1, 11/1 7
/ I III / / "
/ I I I 1 1 7
/ / / / / / / / - -
,T(77711 \\\ I
•__I/////
l7lll11\\ -------
7 ¶ 7 7 \ \\\N.____---------
\ 77171
-
\ \ \ \ \ \ \ \ \ \ \
\ \ \ \ \ \ \ \ '\ \ \ \ ' NN-'-.
\ \ \\\\\\\ \\ /
\\\\\\\ \ /

Z-plane =-0.018, step'100, 50 degrees (1.389m see) after discharge commences
204
Chapter 5 Three DimenshinalFkw Mode/fig
1.0595E+l
N 1.0276E+l
9.9576E±l
9.638E±
9.32OOE+
9.001 1E±
I 8.6823E±
Figure 5.122 Scroll-Compressor. Rtationa1 speed 6000 RPM, Pressure contours,

Z-plane =-0.1, step 100, 50 degrees (1.389m see) after discharge commences
7.6360E+01
6.3633 E-I- 01
5.0907E-4-01
I,,
'7 / I i"'-..___.__. /1 / / /
tiT1fl .
II 11 l\\
II ltil\\\' - ._____
I I I I I 1 I I I \ \ \ '\ - . - - - - - -
11111111 II _-_._.... _.__ --
\\\\\\\\\\I-.....
\ \\\\\\\\ \'-._....

Z-p1a =-0.1, slep=100, 50 degrees (1.389m see) after discharge commences
205
Chapter 5 T hree D/m ensv naIFk w 4*7c/a7g
r 1.0595E
'' 1.0276E-
9.9576E-
- 9.638812
9.3200E
9.001 1E•
I 86823E-

Z-plane --0.0025, step = 100, 50 degrees (1.389m see) after discharge commences
7636(JE+(J1
(,.3633E+01
:8i8uJ--01
2.5453E-I-U1
1.2727E-f-01
50701
O.0000E+00 -.
/ - / --
/ - /
/ ii T
t , II ------ - //------
1 --_ _____ -
III! I / I I - _- ------- -- -
I I I I S \ \ - - -
I / I I I I I' \ ----------- -- -
II I I I / I I ------------------ -
5'

Z-Plane =-0.0025, step=100, 50 degrees (1.389m see) after discharge commences
206
Chapter 5 Three Dimenskw,a/Fh,w Modeling
1.059513-
1.0276E-
9.957613-
9.638813 -
9.3200E -
9.001113-
! 6.682313-

Axbitraiy plane, step=100, 50 degrees (1.389m see) after discharge commences
7.6
6.3
I
5-oc
3.61
2.5
12
0.0(
FigUre 5.127 Scroll-Compressor. Rotational speed 6000 RPM, Speed vectors,

Axbitraiy plane, ster 100, 50 degrees (1.389m sec) after discharge commences
207
Chapter 5 Three Diinensiona/Fkw h*idefrig
94.5488
78.7907
63.0326
p:::::
2L.
15.7581 ( 1 -
1 0.0000 (
Figure 5.128 Scroll-Compressor. Rotational speed 6000 RPM Stream lines

94 .5488
78.7907
63. 0326
47.2744
31.5163
15.7581
I 0. 0000

208
Chapter 5 Three DA'nensAnaI Fkw Modeling
94 .5488
78.7907
63.0326
•47.2744
31.5163
15.7581
I 0.0000
Figure 5.130 Scroll-Compressor. Rotational speed 6000 RPM Stream lines

94.5488
78.7907
63 . 0326
47.2744
31.5163
15.7581
1 0.0000
Figure 5.13 1 Scroll-Compressor. Rotational speed 6000 RPM, Stream lines,

209
C hap t er 5T hr ee Dinensk'nalFk'w Mode/1sg
1.0595134
1.0276E4
9.9576134
9.6388E-+
9.3200E4
9.001 1E4
I 8.6823134

7.ô360E+OI
6.3633E+O1
3.13 18(JE±0I
2.5453E±O1
1.2727E±O1
5.(J907E±Oi
O.0000E±(JO
/ . . - _ - -
j
/ / / / / / 7 -,
I
, .'
r / , i I I / / ft / / //
/1 117111
1 1f Y '
1
I I I 7 1 1 7 1 1 \ \
I 1 7 1 1 1 1 1 \ \ \
Ill i
I I I ¶ I I I \ \\\ NNNNNN N
I \ I \ \ \ \ \ \\NNNNNNNNNNNNN
\\ NNNNNNNNNNNNN
\ \ \ \ \ \ \ \ N
\ \ \ \ \ \ \ N N NNNNNNNNNNNNN
N
\\\\\\NNNNNNNNNNNNNNNNNNNNNN
\ \ \ \ / NNNNNNNNNNNNNNNNNNNNJ
\\ \ \ / \NNNNNN NNNNNNNNNNNNNN/
\ \ \ \ NNNNNNNNNNNNNNNNNN/
\ \ \ \ NNNNNNNNNNNNNNNN /
NNNNNNNNNNNNNN
\ \ N NNNNNNNNNNN 7
NNNNNNNN

210
Chapter 5 Three DfrnenskvzeIFkw Modeling
1.0!595E
r 1.0276E
9.957613
9.6388E
9.320(JE
9.001 IE
; 8.682313

Z-plane =-0. 1, step=200, 100 degrees (2.778m sec) after discharge commences
l2 7.6360E+O1
6.3ô33E±O1
5.(J907E±O1
3.818(JE+OI
2.5453E±O1
Ii r I I 1 - .- ..- - - .- - -
III ii ¶\\\\\'.-_
I I ¶S\\\' -.
I
\ I 1 1 I
\jj,y

211
Chapter 5 Three Dinensk'nelFbw Model/rig
_ 1.0595E-f
1.0276E-F
9.957611-I
9.63881-I
9.32tXJE4
9.001 lEt
I 8.6t23E4

Z-plane =-0.0025, step200, 100 degrees (2.778m sec) after discharge commences
7.6360E+O1
o.3633E±O1
5.090711+01
L:7/T1IIIII\\
III -.-._ -------- -
I -
II
III' ' ' \\\\ -

212
Chapter 5 Three DinensionalFbw Modeling
1.0595H+06
1.0276E+06
9.9576E-4-05
9.638813+05
9.3200E+05
I 9.001IE+05
8.6823+05
Figure 5.138 Scroll-Compressor. Ritational speed 6000 RPM, Pressure contours,

Axbitraiy plane, step=200, 100 degrees (2.778m sec) after discharge commences
7.63(,OE±OI
814 6.3633E+Ol
5.0'O7E+Ol
3.81 80E+O1
2.5453E±0 I
1 .2727E-4-01
I 0.0000E+O()

AEb itraxy plane, step=200, 100 degrees (2.778m sec) after discharge commences
213
Chapter 5 Three DlmenskrnalF4bw Iikideling
78.7740
65.6450
52.5160
39.3870
26.2580
13.1290
10.0000

78.7740
65.6450
52 .5160
39.3870
26.2580
13.1290
I 0.0000

214
Chapter 5 Three Dfrnens4naIFbw Mode/hg
78.7740
65. 6450
52.5160
39.3870
26.2580
13.1290
I0.0000

78.7740
65.6450
52 . 5160
139.3870
26.2580
13.1290
I 0.0000

step=200, 100 degrees (2.778m sec) afler discharge commences
215
Chapter 6
Conclusions and
Future Work
This chapter presents some conclusions which can be drawn from the work done
and also a few suggestions for possible future work.
61 Conclusions
In this study, an analytical model for the geometry of scroll machines, and a grid
generation program with automatic moving mesh capabilities has been developed
and used together with a finite volume code to model the flow of a scroll
compressor - expander. The geometry was also attached to a package of
subroutines developed to analyse compressor and expander thermodynamic
characteristics based on a Quasi 1-D modeling approach. The proposed grid
generation program developed is capable of resolving the complexity of the
rotating working volumes of these machines.
The main findings can be summarised as follows:
1. The importance of the relationship between built-in volume ratio and pressure
ratio and the use of the correct pressure ratio is clearly defined. The significant
point is that the built-in volume ratio is a constant for any given scroll
compressor or expander design whereas the pressure ratio, which is a function
of the gas being compressed or expanded is not. Speed also affects the
relationship between pressure and volume ratios.
2. A scroll expander driving a scroll compressor in a sealed unit could be used as

a throttle valve replacement in large vapour compression chiller systems. Such
a device would be stable in operation and increase the Coefficient of
Performance.
3. The development of a grid to analyze motion in space and time, and the use of
non uniform distribution functions such as the Hyperbolic tangent distribution
Chapter 6 Conclusions
together with the transfinite interpolation has produced excellent results. The
non uniform distribution gives the flexibility to user to put fewer elements in
areas which are not so important or to concentrate elements in crucial areas so
as to allow the solver to capture the activity of the fluid with a higher accuracy.
Aspect ratio, orthogonality, cell distortion, and smooth distribution, are some
of the difficult issues involved in the transient structured body fitted grid
generation package developed. All these issues increase the speed of
convergence.
4. Flows inside scroll compressors and expanders are turbulent, time dependent,
and compressible. The flow inside the scroll expander is two phase as well.
Furthermore these flows are driven by the wall motion. An interesting feature
of the predicted flow patterns of the scroll expander, is the significant pressure
drop which takes place within the flow domain during the filling process and
the existence of re-circulation areas. It is clear that at these areas a stagnation
point occurs. However the important conclusion from this case is the fact that
these re-circulation areas are directly related to the geometry of the machine,
the movement of the boundaries, and the direction of the flow.
5. The significant finding for the case of the scroll compressor is that we have
approximately uniform pressures within the flow domain. There are also some
areas of re-circulation but unlike the case of the expander are related to the
discharge pipe. The calculations indicate that the flow passing through the
discharge port is not uniform. Another important finding is the swirling that
takes place, mainly in the outlet pipe. This swirling appears to be an extension
of the re-circulation which is underneath the discharge pipe.
6. From the results presented in Chapter 5, we can conclude that both qualitative
and quantitative results were obtained from the numerical predictions. These
results have indicated that modern CFD codes, such as CFX4, are probably
capable of simulating flows in scroll compressors and expanders. These codes
can be used in such a way that they can provide a powerful and reliable
comprehensive analysis tool for the design of such machines. However the
computational cost of this simulations is something that needs to be taken into
consideration.
217
6.2 Future work

The following areas, arising from the present work, are considered suitable for
further investigation:
1. The grid generation code has not been optimised to execute the minimum
number of operations. Therefore, further attempts will be necessary in order to
make this code more efficient for practical purposes. In this optimisation,
one should concentrate on the calculations of the non uniform distribution
functions.
2. It may be of interest to adjust the code in such a way that can be used in an
adaptive grid strategy in conjunction with the solver. When an adaptive grid is
used, the grids are clustered where sharp gradients are present such as in the
case of a shock. By concentrating the grids only at the locations where sharp
gradients are taking place, one can avoid the use of unnecessary grids at
locations where there is no need and therefore save on storage and
computational time.
3. A further recommendation would be to put the discharge pipe in different

angles. As it was said earlier some swirling is taking place inside the discharge
pipe for the case of the compressor. if we examine the flow patterns underneath
the pipe and the stream lines inside the pipe we will see that the swirling is due
to the direction of the flow underneath the pipe, therefore the use of the
discharge pipe in different angles might decrease the presence of this swirling.
Another pipe related recommendation would be the modeling of the flow
through the discharge pipe when an adjusting pressure valve is fitted on it. This
will give us an indication of how the flow underneath will be affected from the
pressure valve.
4. Now, after significant progress in both software and hardware technology, it is

possible to carry out coupled simulations such as fluid-structure interactions.
One way to do this is to solve the governing equations within one code, another
approach would be the use of two different codes one to do the fluids and the
other to do the structure calculations. The second approach will require a
coupling interface as well. The design and development of a scroll compressor
- expander require a detailed examination of structural behaviour and fluid
dynamics. In this study only the fluid side of the problem was investigated.
218
Therefore another recommendation would be a fully coupled analysis of fluid

structure interaction.
5. Finally, in this thesis 3-D numerical studies have modeled only the fluid
dynamics of the suction process of a scroll expander and the discharge process
of a scroll compressor. This was due to the prohibitive computational cost.
However, the expansion and discharge process of the scroll expander, and the
suction and compression process of the scroll compressor are still to be
modeled. As it is well known, the biggest and most difficult part of a 3-D
Numerical Simulation is the grid generation. In this case the grid generation as
well as the setup of the program have been developed, and the remaining task is
the provision of powerful computing facilities. Modeling all the processes of
the scroll compressor - expander in 'one run' is another recommendation that
can be done in the future.
219
Appendix A
Scroll
Compressor-Expander
Geometry
A.1 Introduction
In this Appendix the geometry of the scroll compressor-expancer is ana'yzed. Th
derived equations are general parametric equations which amongst others can be
used for
• The design and construction of scroll compressor-expander devices.

• The generation of boundary-fitted, orthogonal, block-structured,
transient grids used in CFD analysis.
• The determination of the working volumes of scroll devices as
functions of the crank angle 0 and the parameters ci and N
determining the lengths of the devices' spirals.
Al2 Geometry of the Scroll Device

A.2.1 Basic Geometry
In this section the equations defining the inner and outer surfaces of the fixed and
orbiting spirals of the compressor are derived. Figure (A-i) depicts the Cartesian
system X Y which is used in the treatment of the problem. The centre of a
generating circle with radius r, is located at the point
s=(J.
The meaning of the term "generating" will become apparent later in this section.
The line segment BP, shown in figure (A-i), is constructed so that it is tangent to
the generating circle at the points B, hence perpendicular to the radius B. Its
A ppendix A Scroll Compressor-Expander Geometry
length BP is equal to the length GB of the arc GB, as measured anti-clockwise,

that is
BP±SB, BPGB=r, q'Ô, 2,r>eÔ, E=constant, (A.1)
where by definition, qi is the angle GSB subtended by the chord GB and

E A SG is a given constant angle in the semi-open range [0, 2,r). Clearly, the
coordinates of the point B are given by
(rcos(+e)+S
B =I X , ço ^ 0, 2,r> e^ 0, s = constant. (A.2)
rs1n(+e)+S)
The curve C, also shown in figure (A-i), is defined as the geometric locus of the
points P satisfying the relations (A.1). From this definition and the geometry
shown in the figure, it is obvious that the parametric equations of the points P and
hence the curve C, are
Figure (A -i). General geometry of the involute.
221
A ppendi) A Scroll Compressor-Expander Geometry
(x" - (SD + DE S) - (r cos(p + e)+ BP sin(

e) + S, - +
.
y)EF—PF)rsin(co+e)—BPcos(ço+E)+S
(A.3)
(r cos( s) + rço sin(p + e) + S,
+
^ 0, 2,r> £ ^ 0, e = constant.
rsin(p+ e)—rçocos(ço+ e)+S)'
The starting point G of this curve is obtained form the above equations for = 0.
This is the only common point between the curve and the generating circle and its
coordinates are
G " (rcos(s)+S'
G (x (A.4)
YG) rsin(E)+Sj' 2>eÔ
The gradient m of the curve C at any of its points P, can easily be obtained from
equation (A.3). Differentiation of this equation once with respect to , yields
(dx\
C: qiÔ, 2,r>eÔ, e=constant, (A.5)

I dy I sin(ç+e))
and hence the gradient m is given by
m = tan(q + e), q ^ 0, 2r> £ ^ 0, e = constant (A.6)
This is equal to the gradient of the radius SB, a fact obvious from figure (A-i).
Consequently, the tangent L to the curve C at the point P is parallel to the radius
SB, and according to relations (A.1)
The tangent L to the curve C at a point P and the tangent BP to the

generating circle are perpendicular, while L is parallel to the radius SB, (A. 7)
that is SB I BP, L I BP, SB I L,
where the coordinates of the points B and P are given by the relations (A.2) and
(A.3) respectively.
A second Cartesian system XY' is defined so that its origin is located at the point
(cos(0)
S = —d d = constant, 0 E
sin(0)) ,
222
A ppendA ' A Scroll Compressor-Expander Geometry
and its axes are parallel to the axes of the X Y coordinate system as shown in
figure (A-i). The symbol R denotes the set of the real numbers.
The curves C1 CFQ , C2 CR , C3 C00 and C4 C01 , defining the inner and
outer surfaces of the fixed and orbiting spirals of the compressor are shown in
figure (A-2). The meaning of the subscripts is: FQ = Fixed Outer, F/= Fixed Inner,
00= Orbiting Outer, and 0/ = Orbiting Inner. The analytical expressions for these
curves are obtained from equations (A.1), (A.2) and (A.3) applied for different
values of the parameter e and different locations of the point S. The values of e
and S corresponding to each of these curve are given below
Cl CFQ : S= e=0,
C2=C, : S = e=y , 0<y= constant <,r/2,
(A. 8)
(cos(Jr+6)") e=2r,
C3 C00 :
(cos(,z +
: S =dj(0)J e=ir+y , O<y=constant<,r12,
where by definition, y is a constant angle, called thickness angle, and
0<r<d=constant, (A. 9)
Combining these equations with the relations (A.3), we obtain the equations
describing the curves CI— CFO, C2— CR , C3 C00 and C4 C01 . Thus,
(x1 = r 05(00))
CI CFQ :
ly1) sin(q)— cos(q.'))'
C2 C , : 1x2 = 1c05+nun1+1)
Y 2) sin(+ 7)— cos(+
(A.1O)
(r(cos(c9) + qsin(9)) + d cos(9)")
C3 =C00 : 1x3")
L y3 ) r(sin(cp) - cos()) + d sin(0))'
- (x4" (r(cos(p + y) + sin( + y)) + d cos(0)

C4— 0O3 : 1=—I
y4 ) r(srn(ço + ')— cos(q + y))+ dsin(0)
223
Appendix A Scroll Compressor-Expander Geometry
C:- C
3 'UU
Figure (A -2). Generating circles and invo lutes.
where q'Ô, O<y=constant<r/2, 8ER, O<r<d=constant.
The starting points of these curves, according to equations (A.4) and (A.8) are
IT," (cos(y)
C1 CFQ : G1 C2 CF/ : G2 = rI I
= 1\OJ sin(y))
(A.11)
+ d cos(0)"\ (r cos(y) + d cos(9)")
C3 C00 : G3 C4C01:
= - dsin(0) J' rsin(y)+dsin(0))
224
J
A ppendb A Scroll Compressor-Expander Geometry
From the relations (A.8), (A.9) and (A.1O) we can easily deduce the following
facts, important for the treatment of the present problem. The curves C1 CF0 and
C2 -CF/ are associated with a generating circle whose centre coincides with the
origin of the XY Cartesian system, while the curves C3 —000 and C4 CO3 are
associated with a second generating circle whose centre is located at the origin of
the X'Y Cartesian system. The parameter q appearing in the definitions (A.1O)
of the curves, is always measured anti-clockwise. However, for the curve C1 CF0
it is measured from the positive X semi-axis, for the curve C2 CR it is measured
from the line 0G2 , for the curve C3 C00 it is measured from the negative X
semi-axis and for of the curve C4 C01 it is measured from the line SG4 . Finally,
the angle 0 is measured anti-clockwise from the negative X semi-axis. All these
facts are shown in the figure (A-2).
The gradients m (i=1,2,3,4) of the curves C, (i_-1,2 ,3,4 ) for any given value of the
parameter q, are obtained from equation (A.6) upon introduction of the values of
the parameter e given in equations (A.8), thus
C1 C 0 : m1=tanq,
C2 CR : in2 = tan( + y), , ^ 0, 0< y = constant < ,r /2,

(A.12)
C3 C00 : m3=tanço,
C4=C01 : m4 =tan(co+y), p ^0, 0<y=constant<ir/2.
We now proceed with the determination of the distances between the pairs of the
curves C1 CFQ , C2 CR , and C3 C00 , C4 C01 . As it will be shown below, these
distances are equal throughout the lengths of the spirals. They are thus independent
of the value of the parameter .
By construction, the points P1 , P2 , P3 and P4 , shown in figure (A-3),

correspond to the same, arbitrary value of the parameter , say co. The segments
B,P (i=1,2,3,4), drawn from these points, are tangents to the generating circles at
the points B (i=1,2,3,4). The points Q1 and Q3 are the intersections of the curves
C1 and C3 with the produced segments 82P2 and B4P4 . Finally, the lines L1
(i=1,2,3,4), also shown in the figure, are the tangents to the curves C, (i=1,2,3,4)
at the points Q1 , P2 . Q3 , and P4 respectively.
225
Appendi)' A Scroll Compressor-Expander Geometry
CF C
-1
'-'3 '-'00
Figure (A -3). Generating circles, involutes and their tangents.
Clearly, the values of the parameter corresponding to the points Q 1 , P2 , Q3,

and P4 , are given by
P2 and P4 : ço=w, wÔ

(A. 13)
°1 and Q3 : =w+r, wÔ, O<y=constant<r/2.
Substitution of these values into equations (A .1O) yields the coordinates of the
226
above mentioned points. Hence
(XQI"1
= Icos(w + + ( w + y)sin(w + =
01 7,
YQ1) sin(w+ ) — (w+ )cos(+
• (xp2") 1cos(w+r)+wsin&+r) -
P2 . i i—ri . I,
YP2) sin(w + - U)coS(U) +
(A.14)
03 : 3=1ros++(+ S1fl +y +d of' 2=w+y,

kYQ3 ) r(sin(w + - (w + ) cos( w + + d srn(8))
p4 - Ir(cos(w + + asin(w + y)) + d cos(Of

y ) - r(sin( + - wcos(w + )) +dsin(0)J' =
where co ^ 0.
Furthermore, combining equations (A.13) with equations (A.12), it is easy to prove

that the lines L1 (i=1,2,3,4), 082 and SB4 are parallel, with gradients given by
m1 =m 2 =m 3 =m4 =tan(w+y),
On the other hand, the line segments B2Q 1 and B4 Q3 are perpendicular to the radii
0B2 and SB4 and hence they are also perpendicular to the lines L1 (1=1,2,3,4).
From the above discussion it is evident that B2Q1 and B4 Q3 are perpendicular to
the curves C (i=1,2,3,4), at the points Q 1 , P2 , Q3 , and P4 . Consequently, P2Q1
is the distance between the curves CCFO and C2 CR , while P4 Q3 is the
distance between the curves C3 C00 and C4 C01 . These distances, easily derived
from equations (A.14), are independent of the parameter q. They are equal and
depend only on the common radius r of the generating circles and the thickness
angle 7, that is,
Distance between C1 and C2 = P2 Qi=j( xp2 - x Ql ) 2 + ( y p2 - YQI)2 = r

(A. 15)
Distancebetween C3 and C4 = P4 Q3= \fp4_xQ3)2+(yp4_yQ3) =ry,
where 0< y = constant < r /2, 0< r <d = cons tan t.
227
It is worth noting that the distance between the curves is equal to the lengths of the
arcs G7G2 or G3 G4 , to which the thickness angle subtends. Furthermore, the
curves Cl — CF0 , C2 CR C3 C00 and C4 CO3 are perpendicular to the generating
circles at their starting points.
A.2.2 Contact Conditions
It is now assumed that the curves CI CFQ and C4 CO3 have been constructed in
such a way that they never intersect each other, but there exist points at which they
are tangent. The same assumptions are also made for the pair of the curves C2—CR
and C3 C00 . Two such touching points are illustrated in figure (A-4). Clearly, at
these points the touching curves (either C1 and C4 or C2 and C3 ) have the same
coordinates and the same gradients. However, the value of the parameter at
these points would be different for each curve. Denoting the values of q,
corresponding to the curves C1 (i=1,2,3,4 ), by (i=1,2,3,4 ) respectively and
introducing equations (A.1O) and (A.12), the above assumptions for the pair of
curves C1 and C4 are fully described by the following relations
rcos(ço1 ) + rço1 sin(q 1 ) = —r cos( 4 + 2') - rca4 sin(ca4 + y) - d cos(0),
rsin(q 1 ) - rca1 cos(q 1 ) = —rsin(,4 + 2') + rço cos (ca4 + y) - d sin(0), (A.16)
tan(ca1 )= tan(q'4 + 1)
while the corresponding equations for the pair of the curves C2 and C3 are
rcos(q 2 + 7)+rp2 sin(q 2 + y) = —rcos( 3 )—r 3 sin(q 3 )— dcos(G),
rsin(ca2 + y) - rq 2 cos (ca2 + 2') = —rsin(ca3 ) + rca3 cos( 3 ) - d sin(0), (A.17)
tan(p2 + y) = tan(ço3).
The above algebraic systems obey the conditions
Ô (i=1,2,3,4), O<7=constant<yr/2, OeR,

(A.18)
O<r<d = constant.
The last of equations (A.16) and (A.17) have the following general solutions
228
C1= C.o
Figure (A -4). Involutes and their innermost touching points.
229
Appendi) A Scroll Compressor-Expander Geometry
q 4 =(2k+1)7r+q'1 —y, kE\W , ç,, ç04Ô,
q 4 =2kir+q 1 -7, kE\W , q'1, qiÔ,

(A. 19)
4=—(2k+1)r+q'1—y, kE\W , ç,, q4Ô,
q 4 =-2kir+ç 1 —y, kEW , q,, ço4Ô,
and
q12 =(2k+1)7r+q 3 —y, kE\W , q'2, q'3Ô,
q,2 =2k,r+cp3 —y, kE\W , Q2, O3Ô

(A.20)
q'2=—(2k+1)r+ç3—y, ç2, q3Ô,
ço2 =-2kr+ço3 —, kE\W , q3^0,
where the symbol \W denotes the set of the whole numbers.
Introducing the solutions (A.19) into equations (A.16) and the solutions (A.20) into
equations (A.17), squaring and adding the resulting expressions, we obtain the
following relations
d=r[(2k+1)r—y],
c,2 =(2k+1),r+q3—y,
d=r[(2k+1)ff+y}, = —y—(2k+1)r,
q2=3—y—(2k+1)r,
d = rj4 + [2kir + 2,1 - = 2k,r + -

(A.21)
d=r,14+[2k,r+2c03_ ]2, = 2kr + (p3 —7,
d=rj4+[2coi_y_2k,r]2, q4=q,1— y-2kr,
d=r/4+[2q,3_y_2k,r]2 = ço3— y-2kr,
230
A ppendi)c' A Scroll Compressor-Expander Geometry
where ç01 ^0(i=1,2,3,4), 0<y<,r/2, O<r<d, kE\W .
The restriction imposed upon d from the last of the relations (A.18) implies that
the last four of the above solutions are unacceptable since they depend on either
or q 3 and hence such a d is not constant. In order to select one of the first two
solutions, we impose the following "extra" condition : the distance d must have
the minimum possible value satisfying the relations (A.16), (A.17) and (A.21).
Under this condition, the first of the above equations for k = 0 yields
d = r(,z - 7),
ço4 =ii+ço1 7, 2+c)37, (A.22)
q 1 ^0 (i=1,2,3,4), 0<y=constant<ir/2, 0<r<d=constant.
We must of course justify the imposition of the "extra" condition mentioned

above. Any other constant value of d given by equations (A.21), except the
minimum, produces spirals that not only touch but intersect each other as well. For
example the second of equations (A.21), for k = 0 furnishes
d = r(ir + 7),
q2_q372r,
^ 0 (i-1,2,3,4), 0< y = constant < r /2, 0< r <d = constant.
This value of d differs from the minimum given by equation (A.22) by 2ry which
according to expressions (A.15), is twice the thickness of the spirals. This increase
of d corresponds to a movement of the orbiting spiral away from the origin 0 by
2ry. Consequently, for d = r(Jr + 7) the spirals touch and intersect each other
simultaneously as shown in the figure of next page.
Having found the expression for the distance between the centres of the two
generating circles, we proceed with the determination of the relation between the
parameters ç, and 0 at the touching points. Substituting equations (A.22) into
equations (A.16) and (A.17), and after some algebraic calculations, we obtain
tan(ç,) - cot(0), tan( 3 ) = - cot(0),
4_ 71+77, 21 q3-7, qÔ (i=1,2,3,4), d=r(ir— y), (A.23)
0<y=constant<r/2, &e9.
231
L.1
Figure (A -5). The involutes touching internally.
These trigonometric equations have the following general solutions
or O=2k r++q 1 or
(A .24)
Jr Jr
ço3 =2k Jr+-+O or 9=2k Jr+-+q 3 or
232
AppendA- A Scroll Compressor-Expander Geometry
In order to select one of the above solutions, we impose the restrictions implied
from the relations (A.23) and figure (A-4), that is
O<ç0 1 <(p4 , O<ç'3 <ç 2 , (A.25)
and hence at the touching points the following relations hold
ço1 =2kr++O, co3=2kir++O,
o4 -2r+q'1 7, ço2 =ir+ço3 —y, qÔ(i=1,2,3,4), d=r(r—y), (A.26)
O<y=constant<2r/2, kE\W , OE9.
Clearly, the relations between the angles q' 1 , p and 0 refer to the touching
conditions of the curves CCFo and C4=C01 , while the relations between the
angles q, and 8 refer to the touching conditions of the curves C2 CR and
C3=coo.
Two other solutions satisfy equations (A.23) and the restrictions (A.25), namely
1=(2k+1)r++0, q,3=(2k+1),r+!+0, 4=r+q'1—y, q2=r+q3—y.
These however correspond to the internal touching points as shown in figure (A-5),
and are unacceptable.
A.2.3 Starting Points - Inner Circular Arcs
Under the conditions (A.22) and (A.26) not only the external curves touch with the
internal curves in pairs (C1 CFO, C4 CO3 and C2— CFI, C3—000 ), but for a
specific value of 8 the external curves CICFO and C3—000 touch externally as
well. This situation is illustrated in the figure (A-6). Clearly at the point of
touching these curves have the same coordinates and equal gradients. Denoting the
value of the parameter at this point by q' for the curve C1 CFQ and by c'3S for
the curve C3—000 , and introducing the relations (A.1O) and (A.12), the touching
conditions are fully described by the following relations
COS(q'1) + rço1 Sifl( (Pis) —r cos(q 3 )— rco3 sin(q 3 ) - d cos(0),
TSifl( (Pis ) - rço cos Q'is) TSjfl(3 ) + rço3 COS( 3 ) - d sin(0) (A.27)
tan(p) tan('3).
233
A ppendi)c- A Scroll Compressor-Expander Geometry
Y.
x.
C 3 C00
Figure (A -6). The trimming point of the external involutes.
These are, of course, subject to the conditions (A .18) and (A.22). That is
d=r(ir— y), ' s^ O , O<r<d=constant,

(A.28)
9E9, 0< y= constant<ir/2.
Given that - <2,r, the general solution of the last of (A .27) has the form
Pis = + 3S = q 1 + ,i, 1S = 3S
(A.29)
234
Substituting any of the first two of these solutions into equations (A.27), squaring
and adding the resulting expressions and after some rearrangements, we get
d = nt, '1S = '3S + = + 2r.
These relations do not agree with the first of the expressions (A.22) and they are
therefore unacceptable. On the other hand, combining the last of the solutions
(A.29) with equations (A.27) and (A.28), following steps similar to the steps
described above and rejecting negative values for and q35, we find that the
curves C1 CFQ and C3 =C00 touch externally when
is' IS3S'
30)
d = r(r— y), c 3S ^0, 0< r<d = constant,
OeR, 0< =constant<r/2.
The first of these relations implies the condition
d ^ 2r,
which is evident from the geometry shown in figures (A-2), (A-3) and (A-4). Note
that for d r(r— y), the above inequality furnishes (7r - 2) ^ y . Given that
, / 2> it —2, the condition 0 < = constant < it / 2 appearing in most of the
above equations, must be substituted by
0<y=constantîr-2. (A.31)
Evidently, for values of the parameter q' less than the critical angle
()2_i
= [(it_y)/2]2 —1, 0<y= constantît-2, (A.32)
cr
there exist values of the crank angle 9 for which the spirals intersect each other.
We are therefore forced to introduce one more condition while defining the curves
(i=],2,3,4), namely
'2
= 7t-7)
—1 0<=constantîr-2. (A.33)
- 2)
235
A ppendiv A Scroll Compressor-Expander Geometry
This condition, being a general one, holds true at the contact points of the curves C
(j=1,2,3,4)as well. Thus, in the case of the curves C1 CF0 and C3—000 , though
equations (A.26) imply that the values (p and q of the parameter (p at the points
of touching can lie anywhere in the semi-open interval [0, 00) as the crank angle
O varies in the range (_00, + 00), the above inequality restricts the range of
and (p3 to values greater than or equal to (per' thus
pt ,
( (p3 ^ 1S = (per = 2J —1, 0< y = constant ^ —2. (A.34)
Similarly, for the curves C2CFI and C4 C01 , according to equations (A.26) and
(A.34), the values (p and (p4 of the parameter (p at the points of touching satisfy
the relation
(pa ' 4 ^(p2S (p4S ((per +_7) 2 J —1
(A.35)
0<1 = constant^r-2.
The above two restrictions must hold regardless of the value of the crank angle 0.
It must be stressed here that for values of or (p4 less than 2s or the
curves C2 CF, and C4 CO3 do not have any contact points, since a drop in the
value (p or (p4 below 2s = (p = (Pcr + 7t — 2') implies a drop in the values of (p
or below (per• This of course, according to restriction (A.34), is unacceptable.
(p3
However, for values of the parameter (p in the closed interval [(

per' ((per + JT —
we require that -as the moving spiral rotates in a circular orbit round the fixed
spiral- the curves C2 CF/ and C4 CO3 are continuously in touch with the starting
points G3 and G1 of the curves C3 C00 and CI CFQ respectively. This is of great
importance in the design of the scroll compressor if leakage of the flow from the
discharge chamber to the last pair of compression chambers is to be avoided.
The above requirement can be satisfied if and only if the sections of the curves
C2 CF, and C4— 001 , defined for values of (p in the range cr ^ (p ^ ((per + It— y),
are replaced by the paths followed by the starting points of the curves CI CFO and
C3—000 as the scroll device is in motion. It will be proved below that these paths
are circular arcs. The relevant geometry is shown in figures (A-7a) and (A-7b).
236
A ppendk A Scroll Compressor-Expander Geometry
C ORB
Figure (A -7a). Replacement of internal involutes by circular arcs.
237
Figure (A-Th). Contact of spiral tips with inner circular arcs.
238
Appendik' A Scroll Compressor-Expander Geometry
The condition (A .33) alters the starting points of the curves, which are not any
more given by equations (A.11). Instead, equations (A .1O) upon introduction of the
critical angle yield the starting points or tips G1 and G3 of the curves C1CF0
and C3 C00 respectively, that is
(COS(cocr ) + cr Sifl (cocr )

C1 CF0 : = r1
S 1fl(cr) cr COS(cr )J'
(A .36)
(_ T(COS((Pcr) + sifl(q)) + d cos(7r +
C3 C00 : G3 = -
r(sin('Pcr) - 'Pcr COS ('Pcr)) + d sin( +0)
The starting points G2 and G4 of the curves Cf CR and C4— 0O3 will be discussed
later on in this sub-section.
The introduction of the values 'P2s = c°4S = ( + it - y) into equations (A .1O)

furnishes the coordinates of the points T2 and T4 shown in figures (A-7). At these
points the tips G1 and G3 touch the curves C2 CF/ and C4 C01 nearest to the
generating circles for the first time respectively, thus
(COS ('Pcr ) + ( q'cr + 'r - y)Sifl(cocr )'

C2 CF/ T2 =
Sfl('Pcr) - ('Pcr + it 7) COS('Pr
(A .37)
(T(COS('Pcr ) + ('Per + it - 7) S j fl(4)) + d cos(ir +
C4=C01 : T4 =
T (Slfl ('Pcr) - ('Per + - 7) COS ('Pcr )) + d sin(it +
The independent variable describing the motion of the spirals of the machine is the
crank angle 0 as it varies in the open interval (_00, + 00), Thus, according to
the second of equations (A.36), the motion of the tip G 3 , is a circle (hereafter
called CQRB ) with radius d and center located at the point QQRB whose
coordinates are
XQ ORB (COS(Pcr ) + 'Pcr sin(cOcr )

(A .38)
[YQ,O
=-d Sfl('Pcr) - cr COS ('Pcr )J•
By construction this circle passes through the points G 1 and T2 , and at the point T2
its slope is equal to the slope of the curve C2—CR . Indeed, subtracting the first of
equations (A .36) and (A .37) from equation (A.38), squaring and adding the
resulting expressions, rejecting negative Solutions and after some algebraic
manipulations, we obtain the distances GIQORB and T2QQRB
239
A ppendL A Scroll Compressor-Expander Geometry
GIQORB = 2r( 1+ q2 Cr) , T2 00RB = r(r - 7).
These expressions, according to equation (A.22) and (A.32) are equal to the radius
d of the circle CORB, that is
GIQQRB = T2QORB = d,
and hence, this circle passes through the points G1 and T2.
In order to determine the slope of the circle at the point T2 , we require the value
OIim2 of the crank angle 0 corresponding to this point. By definition, at T2 the tip
G3 touches the curve C2 CR for the first time and therefore 3 = P ' 3S = Pcr and
2S = (q3 + ir - y). Substituting these values into equation

(A.26) and
introducing equation (A.32), we get
O li2 = cr -
j -2' —1 - 0< cons tan t ^ r —2. (A. 39)
J
It is worth noting that according to this relation, 01jm2 lies in the interval
J24_
(A .40)
2011m2< 2
The slope of the circle CORB at any point of its circumference, as derived from the
second of equations (A.36), is given by
where 0E9.
tan(;ir+ 0)
Combining the above expression with the relation (A.39), we obtain the slope of
the circle CORB at the point T2
—1 I ,r
slopeof CoRe at T2 = -=---cot +cocrj=tancocr.
tan(+Oijm2 ) 2
On the other hand, for the curve C2—CF, at the point T2 we have
2S (qc + r - y), and according to equations (A.12) the slope of the curve
at this point is given by
slope of C2 at T2 = tan(q'cr + r) = tan(qcr).
240
A ppendb A Scroll Compressor-Expander Geometry
This is equal to the slope of the circle CQRB and guarantees geometric continuity of
first order for the circle and the spiral at T2.
Another useful parameter in the present analysis is the value OIjml of the crank
angle 0, attained when the tips of the curves ClCFo and C3 C00 are just in
touch. This situation is illustrated in figure (A-6). From this figure and the above
discussion, it is evident that G1 and G3 are in touch when ç, = = q and hence
8Iiml = + - /, ,
where i1 is the angle MOB easily determined from figure (A-6). Thus
(rqi,.
ii = arctanl = arctan(q).
r)
Combining the above two equations and making use of the definition (A.33), we
obtain
_____ (sr— 7)12

_arctan[[ (A.41)
2 ]
The upper and lower bounds of OIiml are easily deduced from this equation. Thus,
,J24
O<,r—arctan---- 2 <OIiml <ir+ <3n12. (A .42)

2
(/4J
It is now possible to replace the section of the curve C2 C, defined for values of
ç, in the range 'cr ^ ç ^ (Q'cr + 2r - 7), by the minor arc T2G1 of the circle CORS,
defined for values of in the closed interval [ 0iimi 0Iim2 + 27r] as shown in
figures (A-7a) and (A-7b). This replacement guarantees the required continuous
touching of the tip G3 of the moving spiral with the inner surface of the fixed
spiral. It also guarantees geometric continuity of the first order (smooth blending)
at the point T2 between the circle CORB and the curve C2 as defined by equations
(A .1O) for q ^ (q'cr + r - y) . Finally, the arc T2G1 meets the curve C1 at its
starting point G1 rather smoothly, a prerequisite for the manufacturing of the scroll
device. This way G1 coincides with the starting point G2 of the inner surface of
the fixed spiral.
241
A ppendik A Scroll Compressor-Expander Geometry
Before proceeding with the replacement of the innermost part of the curve C4C01,
it is necessary to obtain the equations of the curves C, (i=1,2,3,4), when referred to
the rotating X Y e Cartesian coordinate system. This is achieved by subtracting the
"instantaneous" coordinates
(SX
d1
sj - sin(0))'
of the origin S of the X Y ' system from equations (A.1O), which thus take form
(x1*'\ = çsin((p)) + d cos(6)J

Cl = CFQ . I i
y1 ) r(sin(q3)— qcos(qi))+ dsin(0)
(x2 * = [r(cos( + y)+ q'sin(q + y)) + d cos(0)J

I I
y2 r(sin( + y)— cos(qi + y ))+ d sin(0)
(A .43)
(x3* (cos(q)+qsin(qi)'\
C3 C00 . I J=— rI
y3 ) sin(q)—qcos(qi)
(cos(q+ y)+ qsin(q,+ 1)"

C4 C00 : *1=-ri
Evidently, the definitions of the curves C, (i=1,2,3,4 ) are independent of the

system of reference and hence the meaning of the parameters q' and 0 is not
altered in the new system of coordinates. The same is true for the thickness angle
' and the particular values (i=1,2,3,4), ç 5 (i=1,2,3,4), °Iimi and °ijm2•
Introduction of the critical value into the first and third of the equations (A .43)
furnishes the expressions for the tips G 7 and G3 in the X *Y * system of reference.
Thus
(r(cos(cocr ) + Pcr Sifl(cocr )) + d cos(0)"\

CI CFQ : G1 =
r(sin(qY ) — cr COS (c9cr)) + dsin(0)J'
(A.44)
(cos(q'cr)+ 'cr Sifl(cocr)')
C3 C00 : G3 =
- cr cOs(cocr))
Similarly, substitution of the values 'zs + r - y) into equations

(A.43), yields the coordinates of the points T2 and T4 in the X *Y * coordinate
242
system, that is
(- r(cos(q) + ( ç + jr - Y ) S (
)) + d cos(0)"\
cr
C, :
C2 = F T2
r(sin(cr - cr + - y) COS ( c r )) + d sin(0)
-
(A.45)
( COS(cocr )+(co +r 7)S1fl(cr)'
C4 =C00 : T4 =
Sfl(cr) — (cr + — y) COS(cr )J
From the first of equations (A.44), it is at once obvious that as the crank angle 0
varies the relative motion of the tip G1 is a circle (hereafter called Crnx) with
radius d and centre located at point 0RX •
= (COS(c9cr)+ cr S1fl(9cr)
(A.46)
Y Q,FIx) Sfl(cr) - cr COS(cr
It will be shown below that this circle passes through the points G 3 and T4 and at
the point T4 its slope is equal to the slope of the curve C4 C01 . Subtracting the
second of equations (A .44) and (A .45) from equation (A.46), squaring and adding
the resulting expressions, rejecting negative values and after some algebraic
manipulations, we obtain the distances G3 QF,x and T4Q
, 1/2
G3 QR x =2r(1+ qrcr) = d, T4 QFIX =r(r- y)=d.
These are equal to the radius d of the circle CRX and hence, this circle passes
through the points G3 and T4.
The slope of the circle CF/X at any point of its circumference, as derived from the
first of equations (A.44), is given by
—1
where 0ER.
tan(0)
On the other hand, by definition, at T4 the tip G1 touches the curve C4 CO3 for
the first time and hence = and 4S = (q + r - y).
Consequently, the corresponding value of the crank angle 0, according to
equations (A.26) for k = 0 is given by
8 cr — 2 = °11m2
243
Combining the above two relations we get the slope of the circle CRX at the point
T4 , that is
—1 ( (r
slope of CRX at T4 = —cot1 c°cr --1 = cotl -- c'cr I = tan(q).
ta n(011m2 ) 2,i )
For the curve C4 CO3 at the point T4 we have qi = = (ccr + ir - y), and
according to equations (A.12), the slope of the curve at this point is given by
slope of C4 CO3 at T4 tan(q' 4 + y) = tan(q r + = tan(qicr).
The above two equations guarantee that the slope of the curve C4 at the point T4 is
equal to the slope of the circle CF/x and secures smooth blending between the
circle and the curve C4 CO3 at this point.
As in the case of the curve C2 , the section of the curve C4 CO3 defined for values
of in the range c'cr ^ q' ^ ('cr + r - '), is replace by the minor arc T4 G3 of the
circle defined for values of 9 in the closed interval [iimi' °11m2 + 2,r] as
shown in figures (A-7a) and (A-7b). With this replacement we secure the required
continuous touching of the tip G1 of the fixed spiral with the inner surface of the
moving spiral. We also secure geometric continuity of the first order (smooth
blending) at the point T4 between the arc T4 G3 and the curve C4 CO3 as defined
by equations (A.1O) for ç, ^ (q'cr + r - y). Additionally, the minor arc T4G3
meets the curve c3 c00 at its starting point G3 rather smoothly, a prerequisite for
the manufacturing of the spirals of the device. This way G3 coincides with the
starting point G4 of the inner surface of the moving spiral.
After the replacement of the innermost sections of the curves C2 CF, and C4CO3
by the arcs T2G1 and T4 G3 the tips G1 and G3 are in continuous touch with these
arcs for values of the crank angle 0 in the closed interval [0Iiml' 0lim2 + 2k].
Th the limiting case when 0 °Iiml' the two tips are in touch as depicted in figure
(A-b). At the other limiting case when 0 = 01im2 + 2,r, tips G1 and G3 are in
touch with the curves and C2 =C, at the points and 7 respectively as
T4
shown in figure (A-7b). Intermediate positions of the tips for values of the crank
angle 0 in the range °Iiml <0 < °Iim2 + 2,r, are shown in figures (A-i 1).
244
A.2.4 Length of Device's Spirals
This section completes the geometry and the design of the scroll compressor used
in the numerical investigations of the fluid flows presented elsewhere in this thesis.
The rotation of the orbiting spiral of the compressor is described by the crank angle
0. Hence, the geometry of every chamber completes a full cycle of changes
returning to its original configuration, as 0 continuously increases (or decreases)
by 2,r. The innermost touching points of the curves C (i=1,2,3,4) as given from
equations (A.26) when k = 0, are
= .+0, i=l,3, i3O

= ,r ++0 V, i=2,4, (A.47)
where by definition, the single-subscript notation of the angles q is related to their

double-subscript notation as follows
i=1,2,3,4, k E \W (A.48)
'j=k
The relations (A.47), upon introduction of the inequalities (A.34) and (A.35),
become
0=c9i,o—^per—, i=l,3,
(A.49)
0= - +7 ^ Pcr - 2' i=2,4.
These inequalities can be reduced further using the expression (A.39), that is
0^01im2 (A.50)
The maximum value of 0, as suggested by the form of equations (A.26) and the
above inequality, may not exceed the value of (°Ijm2 + 2r) and therefore
0Iim2
0< 61im2
+ 2,r. (A. 51)
For the sake of clarity the end, outermost points of the curves C (i=1,2,3,4 ) are
denoted by S, (i=],2,3,4 ), while the maximum values of the parameter ,
defining these points are denoted by w (i=],2,3,4) respectively. The locations of
the points S (i=1,2,3,4) are shown in the figure below.
245
s3 - -—
Figure (A -8). The angles co and the location of the points S.
Obviously, the angles w1 (i=1,2,3,4) can be selected arbitrarily and independently

of each other and hence the spirals of the machine can have unrelated and arbitrary
lengths. However, in practice for self explanatory reasons, the spirals are always
constructed to have equal lengths. In the present study, the choice of the maximum
angles co (i=1,2,3,4 ) is based on a specific well defined range of volume ratios.
Volume ratios and the reasons for selecting this particular range are discussed
246
extensively elsewhere in this thesis. The precise values of the angles co (i=1,2,3,4)
are determined by another important factor, namely the position of the end points
S, (i=1,2,3,4).
The pairs of the curves C, (i=1,3 ) describe the geometry of the external and
internal surfaces of the compressor's spirals. It is therefore necessary to select such
values for w (i=1,2,3,4 ) that the distances S7S2 and S3S4 are minimum and
equal to the constant thickness of the spirals. According to equations (A.15), this is
achieved when
s1s2=s3s4 = Ti. (A. 52)
Equations (A.34) and (A.35) and the definition of the angles w (i=1,2,3,4) imply
the following relation
cr ^q, ^w, O^j ^k, i=1,3, j,kE\W , (A.53)
where are the values of the parameters at the various touching points of the
curves C1 (i=1,2,3,4 ). The above condition, upon introduction of equations
(A.32) and (A.39), take the form
W, W3^_+Oiim2=41Cr>O (A.54)
According to relations (A.26) and (A.51) this condition furnish at least one pair of
touching points between the spirals as described by equations (A.1O).
In the most general case the angles W (i-1,3 ) must be allowed to cover all
possible values equal to or greater than the above minimum value, hence
(0 1, W 3=2N2r+a#(.I+Oiim2) (A.55)
where by definition N is an arbitrary whole number and the angle a obeys the
condition
0 ^ a < 2ir. (A.56)
For the sake of convenience, we introduce the parameter /3 defined by
/3 a + °1im2 (A. 57)
247
The range of /3 is deduced from the range of the parameter a. Given that
0 a < 2,r, /3 lies in the semi-open interval
°Iim2 /3 < ( 01im2 + 2r). (A.58)
Combining the definition (A.57) with equation (A.55), we obtain
(0 1, (0 = 2Nr + - + /3, /3 E [OIim2

0 2 + 2,r), N e \W .
11m
(A.59)
Assuming that the curves C2 and C4 are long enough, there exists a value O of
the crank angle 6 for which the suction process begins. At this instance the end
points S and S3 come in touch with the curves C4 and C2 respectively, and
hence equations (A.26) and (A.59) furnish
d2)i = 0)3 -
l,N3,N_2N++fl_2IC2+OS,
(A.60)
Jr
2,N c'4N =(2N^1),r+—+fl–y=(2k+1)r+-+O,-7,
2
kE\W , NE\W , O<y^r-2, GIim2^fl, Os<(Oiim2+2Jr)•
The functionality of the compressor under study is based on the existence of the
touching points between its spirals. Hence, the most appropriate choices for the
end, outermost points S2 and S4 of the curves C2 and C4 are the points
corresponding to the angles q'2,N and q'4,N given by equations (A.60). In this case
- (04 =(2N+1)Jr+--+fl–y=(2k+1)7r+ i +O5 –v,

(A. 61)
kE\W , NeW , O<yîr-2, Olim2^fl, Os<(Oiimz+2J
If the angles o. (i=1,2,3,4) were to be accepted as given by equations (A .59) and

(A .61) it would be impossible to construct the spirals of the compressor due to the
considerable difference in the length of their external (C1 , C3 ) and internal
(C2 , C4 ) surfaces. This difficulty can easily be avoided. The lengths of the curves
C1 and C3 are extended by adding r to the angles W and (03 given by equations
(A.60). Then equations (A.60) and (A.61) yield
248
(01= U)3_(2N+1)7r++fl—(2k+1)7t+2+e5,
2 (0 4 =(2N+1)7r++fl—y=(2k+1)2r+-+O—Y, (A .62)
ke\W , N e\W , O<y^r-2, OI im2^ IB 9s<(fuim2+2T)
The end, outermost points S1 (i=1,2,3,4 ) of the curves C (i=1,2,3,4) are obtained
from equations (A .1O) upon introduction of the relations (A.62). After some
algebraic manipulations the formulae giving the coordinates of the end points are
- r sin(/3) - (P cos(fl)
- - cos(fl) - (P sin(/3)J'
- sin(/3) - ((/) - ') cos(/3)

- -T [cos(fl) - ((P - 7)sin(fl)J'
(A . 63)
r(sin(fl) - (P cos(J3)) + d cos 0
-
r(— cos(/3) - (Psin(J3)) + d sin
- ( r(sin(fl) - ((P - y) cos( J3)) + d cos 0

- r(— cos(8) - ((P - )sin(J3)) + d sin
O<r ^d 12, O<y^r-2
where by definition
(P(2N+1)r+ir/2+J3. (A.64)
It is worth noting that according to equations (A .62)
and ' 2 04 Y (A .65)
The distances S7S2 and S3S4 obtained from the relations (A .63) satisfy the
requirement (A.52).
The relation connecting the angles 0 and /3 is deduced from equations (A .62) as
described bellow. The inequalities (A .51) and (A .58) guarantee that both angles 0
249
A ppendi)- A Scroll Compressor-Expander Geometry
and /3 are defined in the same semi-open interval [°Ijm2 OIim2 + 2ir). On the other
hand the relation (A.40) implies that the lower and upper bounds of this interval are
_4)112
—rtI2 and [(r2 ^ 3ir]1 2 . These bounds lie well inside the closed interval
[2,r, 2,z], and hence
- 2,r <—iv /2 ^ /3, o < (iir2 —4+ 3ir) /2 < 2ir. (A.66)
As a result of this condition, equations (A.62) can be valid if and only if k = N,

NE\W, thus
OS = fl °Iim2 ^ Os , /3< (011m2 +2ir), (A.67)
or according to definition (A.57)
= a + °Iim2 (A. 68)
If k and N where to be taken different, then either /3 or O would have been

absolutely greater than 2,r violating the condition (A.66).
It is important to note that the relation k = N was proved to be true only when
9 = O, and in this case N is the maximum value of k entering equations (A.26).
For any other value of 0 the maximum value of k, hereafter called K max is
determined from the restriction (A.53). This restriction, when combined with the
relations (A.26) and (A.62), furnishes
2(N - K max )ir ^ (0— Os ), OIimZ ^ O, 0< (Iim2 + 2ir), K max N E \W . (A. 69)
Given that N and Kmax are whole numbers, the above inequality and the relations
(A.40), (A.51), (A.57) and (A.58) yields the following conditional solutions for
K m in terms of N , depending on the value of the crank angle 0
K max = N, °s 0 ^ 011m2' K max N E \W ,

(A.70)
Kmax = N 1, (°Iim2 + 2ir) > 0 > 0,, Kmax N E \W .
From the definition of Kmax it is obvious that at any instance of the scroll device's
operation, there exist 2(K max + 1) pairs of touching points between the curves
C, (i=1,2,3,4). These are obtained from equations (A.26) for k = O12•••K m,5 . It
must be emphasised that these points do not include the touching points due to the
replacement of the innermost sections of the curves C2 Cfl and C4=CO3 by
circular arcs as discussed in the previous section.
250
The value of N is of paramount importance in the design of the scroll compressor-

expander. It determines the length of the spirals , the number of the pairs of the
touching points, and the number of the working chambers. It also controls together
with the angle a the maximum volume of the suction chamber and hence the
volume of the trapped fluid. In practice the selection of N is based on volume and
pressure ratio considerations as discussed elsewhere and is left to be decided by the
designer of the device. The minimum possible value of N is determined by the
requirement that the device has long enough spirals to be able to function as a
compressor. For N = 1, even when a= 0, that is /3= O2, the device has a pair of
touching points given by equations (A .26) and it works properly as a compressor.
This situation is depicted in figures (A-9) below
SI
52
D4
S3
Figure (A-9). Scroll device with N 1 and a=O at 0 =

0 = /3 = 011m2.
251
A ppendb( A Scroll Compressor-Expander Geometry
It will be shown below that even for N = 0 the device has long enough spirals to
function perfectly well as a compressor if the angle a exceeds a certain minimum
value amin . This is due to the replacement of the curves C2 and C4 by circular arcs
for values of q, in the interval [c'cr, Qcr + it - y] and to the resulting continuous
touching of the tips G 1 and G3 with these arcs for values of the crank angle 0 in
the closed interval [ OIim1 011m2 + 2ir] . Obviously, the device can function as a
compressor if and only if there are two pairs of touching points between the
surfaces of its spirals. The limiting case occurs when the two tips G 1 and G3 are
just in touch and the lengths of the spirals are such that there exists one extra pair
of touching points between the curves C1 (i=1,2,3,4) as shown in figures (A-b). In
this case, by definition 0 = °Ijml while the angles W2 and (04 correspond to the
outer touching points. Thus, according to equations (A.26), (A.57), (A .62) and the
definition of O, we have
= t9s = 0hmi ' i8 = /3min = a min + 0Iim2 N = 0'

(A .71)
= Co4 = 2,O 4,O
7t+7t12+Iimin 7=it+7T/2+011m1 Y
The second of these equations becomes
(A. 72)
i8min = 0Iiml
Combining this equation with the relations (A.39), (A.41) and (A.57), we obtain
flmin = °Iiml = cr - arctan + it,

(A .73)
3,r
am in = 0IjmI - °Ijm2 - - arctan cr
Note that according to relations (A.40), (A.42), (A.68) and (A.73), for N = 0 the
first of the equations (A .70) must be applied and therefore in this particular case
'max = N = 0.
A direct consequence of the replacement of the innermost sections of the curves

C2 =CFI and C4— 0O3 by circular arcs, is the increase of the pairs of touching points
between the device's spirals for values of the crank angle 0 lying in the open
interval ( Oijmi '0iimz + 2it). Obviously, this alters the number of chambers formed
by the spirals. Both the total number of touching points (hereafter denoted by T)
and the total number of chambers (hereafter denoted by CB) depend on the value
of the crank angle 0 and the design parameters N and a. The kind of dependency
252
of the actual values of T and GB on 0, N and a is obtained with the help of

equations (A.39), (A.41), (A.67) , (A.68) and (A.73), and it is given in tabular
form below. Using the above mentioned equations, it can be proved that the
relation between the angles amin and a dictates the relation between the angles 0
and OIiml and vice versa. Thus, the following reciprocal relations hold
a> a min > +- (011m2 + 2,r)> O > OIjml > °Iim2

a amin
= = (0 + 2,r)> 9 = °timl > °Iim2
a < a min O <0iimi (0iim + 2r)> 0Ijml > 0s > 011m2
These inequalities and the geometry of the scroll device guarantee the validity of
the following:
Table (A-i)
If and then
a> a mjn 011m2 < °Iiml 1max = N
T 2(K max +1)=2N+2
G B T + 1= 2N +3
O 0Iiml < -'max N
T 2(K max +1)+12t +3
CB TA +1= 2N +4
0Iiml <0 ^ Os Kmax = N
TP 2(Kmax+1)+2214
C B T +12N5
Os <O<(22r+9) Kmax N —1
Tp 2(K ma , +1)+221 +2
C fi = T +1= 2N +3
253
If and then
a amin 01im2 Ô<OIiml 0S Kmax = N
T = 2(K max +1) = 2N +2
G B = T +1= 2N +3
= = Os 'max = N
T =2(Kmax+1)+1=2N+3
GB = T +1= 2N +4
9Iiml = 0s <0< (2,r+ °Iim2) Kmax = N—i

T =2(K max +1)+2=2N+2
G B =T^1=2N-I-3
If and then
a < amjn 91im2 ^ 0^9 Kmax = N

T = 2(K max -i-1)=2N-i-2
C B =T+1=2N+3
OS< O < OIjmI Kmax=Nl
T =2(K+1)=2N
GB =T+1=2N+1
Kmax=Ni
T =2(K +1)+1=2N+1
GB = T +1 = 2N +2
°Iimj <O<(2Jr+91.2) K max Ni

T =2(K+i)+2=2N+2
GB =T+1=2N+3
254
AppendLv A Scroll Compressor-Expander Geometry
C 3 coo
C l = C0
Figure (A-b). Scroll device with N = 0 and a = am j n = 0Iiml - 011m2 at

00s /3min °Iiml
The above discussion concludes the necessary details required for the full
understanding of the geometry and the touching conditions of the spirals of the
scroll compressor-expander.
255
A ppendi)' A Scroll Compressor-Expander Geometry
A.3 Working Chambers and Working Volumes
A.3.1 An Overview
This section gives a detailed account of the geometry of the working chambers of
the scroll device -often called pockets- and concludes with the determination of the
working volumes as functions of the crank angle 0 and the other design
parameters such as N , a, y, etc.
For the sake of clarity and convenience the following definitions are introduced:
pressure inside the discharge chamber at the start of disharge cycle

pressure ratio of the compressor
pressure inside the suction chamber at the end of the suction cycle
volume of the suction chamber at the end of the suction cycle

volume ratio of the compressor
volume of the discharge chamber at the start of disharge cycle
As mentioned previously, the angles w (i=],2,3,4) can be selected arbitrarily and

hence the spirals of the machine can have unrelated and arbitrary lengths.
However, in practice the spirals are always constructed to have equal lengths which
can be determined from the required pressure ratio. When designing scroll
compressors-expanders, more often than not, volume ratios are preferred to
pressure ratios. This is due to the fact that pressure ratios are sensitive to factors
not related to the geometry of the machine, such as the physical properties of the
working fluid.
For any value of the crank angle 0 the scroll device has rotational symmetry of
1800 about the mid point of the segment connecting the centres of the two
generating circles. This symmetry is apparent in figures (A-i i) which show a plan
view of a scroll compressor for various values of 0. Furthermore, the geometry of
the device changes with period 2r as the orbiting spiral rotates around the centre
o of the fixed generating circle. The independent variable describing this
periodicity is the crank angle 0. The physical process that describes the motion of
the fluid through the device, though essentially periodic, repeats itself only after a
number of complete revolutions, say M, M E \W . Clearly, the number M
depends on the values of the parameter N and the angle a, introduced by the
relations (A .55) and (A.56).
256
A ppendi)c A Scroll Compressor-Expander Geometry
Figure (A -ha). Scroll device with N = 1 and a = 3000 - ° IimZ at

8 = 8 - 600.
257
A ppendk A Scroll Compressor-Expander Geometry
S3
Figure (A -llb). Scroll device with N = 1 and a = 3000 - °Iim2 at

0= 0 —120°.
258
Figure (A -tIc). Scroll device with N = 1 and a = 3000 - OIimZ at

0= 0 —180°.
259
A ppendA A Scroll Compressor-Expander Geometry
Figure (A -lid). Scroll device with N = 1 and a = OO° - 011m2 at

0 = 0 - 240°.
260
A ppendi A Scroll Compressor-Expander Geometry
Figure (A -lie). Scroll device with N = 1 and c = 3000 - jim2 at

8 = O - 3000.
261
CF0
C3= C0
Figure (A-hf). Scroll device with N = 1 and a = 300° - O im2 at

0=0- 3600.
262
Appendi A Scroll Compressor-Expander Geometry
Clearly, the number M depends on the values of the parameter N and the angle
a, introduced by the relations (A.55) and (A.56).
The geometry described in section A.2 allows the scroll device to operate either as
an expander or as a compressor. The selection of the plus or minus sign in the
above equation depends on the device's mode of operation. A clockwise rotation
means that the device operates as a compressor and equation (A.74) should be
taken with the minus sign. An anti-clockwise rotation means that the device
operates as an expander and equation (A. 74) should be taken with the plus sign.
The two modes of operation can easily be realised from the figures presented in
this chapter such as figures (A-1 1).
In practice it is important to be able to deduce the crank angle 0 from the rotation
angle 0. For this purpose two pieces of pseudo-code are given below
corresponding to the compressor and expander mode of the scroll device.
Compressor mode Expander mode
0=9 6=6)
while (6 < 61im2) while (0 ^ 011m2 + 2.ii)
ü= 6+ 2,r } { 0 = 0— 2,r }
A.3.2 Suction Chamber and its Working volume
When the scroll device operates as a compressor, the suction cycle begins at,
(9 O, ± 2nr and finishes after a full revolution of the moving spiral when
± 2n - 2r, n E \W . The suction is associated with a pair of working
chambers, sometimes called suction pockets. The volumes of these chambers
increase smoothly from zero at the start of the cycle to a maximum at the end of it.
The suction chambers are symmetrical and have the same geometry, shape,
volumes, etc. Therefore, without loss of generality, only one of these chambers
need to be studied, namely the chamber which is enclosed by the inner surface
defined by the fixed spiral C2 and the outer surface defined by the orbiting spiral
C3 , This chamber is clearly marked in figures (A-li). The reason for the selection
263
-Iw
A ppendi) A Scroll Compressor-Expander Geometly
of this chamber is simple, its outer wall, associated with the curve C2 , is fixed in
space when referred to the XY coordinate system.
The suction chamber under study is completely sealed and its geometry fully
defined by the curves C2 and C3 only when 6= O ± 2nr, n E \W . For any
other value of 0 this chamber is open at one end. In this case, in order to define
the volume of the trapped fluid, the chamber is notionally sealed with a straight
line connecting the end point S2 with a point F located on the curve C3 , as shown
in figures (A-il) and in detail in figure (A-12). The point F is selected in such a
way that coincides with the point S2 when 0= O ± 2nr, n E \W and the suction
chamber is fully closed. At this position the points S, S2 and F are collinear. We
decide to maintain this condition of collinearity for any value of the angle 8. Thus
F is the point at which the straight line L 1 , defined by the points S 1 and S2,
intersects the curve C3 nearest to the point S2 . From this definition it is evident
that the location of the point F on C3 is not constant but varies slightly with the
value of the rotation angle 8.
The volume of the fluid trapped into the suction chamber when a suction cycle
ends and a compression cycle is about to begin, is independent of the shape and the
position of the sealing line L 1 as far as the points S2 and F coincide at this instant
when, of course, 0 = O ± 2n,r, n E \W . In fact, just before the sealing of the
suction pocket the points S1. 2 and F are nearly collinear and hence the straight
line L 1 is the only natural selection. Moreover, this selection facilitates the
calculation of the suction volume since the line L 1 is fixed in space and
independent of the value of the rotation angle 0.
Figure (A-12) and equations (A.26) and (A.62) imply that the values of the
parameter qi corresponding to the points S2 and F when 9= O ± 2nr, n e
are W2 and q 51 respectively, where
cseai(&3_7r)2N7r++O,
(A.75)
N \W, 0 ^ y < I1 —2, °Jjm2 ^ O, <(0km2 + 2ir).
The value c'F of the angle cc, defining the position of the point F on the curve C3
when 0 ^ O ± 2n,r, n e \W , differs only slightly from the value of q 1 . It can
therefore be written as
= ccseai + (A. 76)
264
Si
S:
Figure (A -12). Details of suction chamber of a scroll compressor with N = 1
and a=300°-Oijm2 at O=O_14O0.
where the correction angle 8cF is within few degrees from zero and is a function of
the rotation angle 0.
The analytical expression relating 8q, with 0 is obtained from the definition of
the point F, as the point of intersection between the line L 1 and the curve C3 . By
construction, the line L 1 passes through the points S 1 and S2 whose coordinates
are given by the first two of equations (A.63). It is therefore fairly easy to prove
that the equation defining the line L 1 has the form
L 1 : x sin 8— y cos /3 = r, x, y E R, °Iim2 /3 < (91im2 + 27r). (A . 77)
265
The coordinates of the point
(xF
FEI I,
YF)
located on the involute C3 and the line L 1 , satisfy simultaneously the equation
(A.77) and the third of equations (A.1O). Thus,
X F sinfl— YF COSJ3 =
(A. 78)
(xF\1 - (r(coscoF + 4F Sill coF)+ d cosE)
F=
r(sincF—Fcos'F)+dsinO
These relations obey the conditions
0= 6± 2nir, n e \W , 61im2 <(9 + 2,r). (A 79)
Combining equations (A.78), (A .75) and (A.76), and after tedious and laborious
algebraic manipulations, we obtain the relation between the correction and the
rotation angles 8F and 0 in the form
F(öq F )= o,
where
F(8q F )rcos(87F )+r(2NJr+.ir/2+fl+8,)sinSq1 F +[dsin(0_fl)_r], (A .80)
0=t9±2nr, flE\W , iim2^fi<(9iirn2+2
The roots of this equation cannot be determined analytically. Numerical techniques

must therefore be employed in order to solved it with respect to Sq. Various fixed
point methods (secant, bisection, fixed point iteration) have been tried. However,
the Newton-Ralphson iterative algorithm was found to be the fastest and most
accurate. This method is based on the formula
( F(ö)
= ') , iE\W,
F,i+1 (A.81)
- dF(6co F )I d8F )'F=o'F
and if a prediction 5Fj for the correction angle 8q'F is available, a better
prediction can be computed.
Differentiation of equation (A .80) once with respect to 5 and introduction of the

resulting expression, together with the function F(&pF ). into the above relation
266
Appendi)c' A Scroll Compressor-Expander Geometry
furnishes the formula
r cos ( S F,I ) + r(2N,r + + /3 + S(PF,I)S1n(8(PF,) + d sin(E) - /1) - r

2
F,i+1 84F,
r(2N yr + - + /3+ 8c°F, ) cos(6q?F,I)
2 (A .82)
8= 0± 2n'r, i, n, N E \W , 01im2 /3' 0< (°Iim2 + 2,r).
The correction angle 8F is small compared to and its absolute value is very
close to zero. Consequently, an obvious selection for the starting value of the
iteration is = 0. After Sq is computed to the required accuracy, equations
(A.62), (A.75), (A.76), (A .78) and (A .82) allow the numerical calculation of the
coordinates of the point F
The line L 1 , as defined above, is not only passing through the points S I , S2 and F,
it is also tangent to the fixed generating circle at a point hereafter denoted by Q1
The coordinates of this point which is fixed in space, are given by
XQ1 (cos(w1)'\
=rI I
[YQJ] sin(w1))'
or, upon introduction of the first of equations (A .62)
XQ1 ( sin/i"
01 =rI I 011m2 ^ /3 < (81im2 + 27r)
—cosfl)' (A. 83)
[YQ1]
It can easily be proved by direct substitution that the point Q 1 is located on the line
L 1 and the circumference of the fixed generating circle. Furthermore, the gradients
of the radius 00 1 and the line L 1 , obtained from equations (A . 83) and (A . 77), are
equal to tan/I and - cot/I respectively. Consequently, 001 and L 1 are
perpendicular. Hence, L 1 is tangent to the fixed generating circle at Q as shown
in figure (A-12).
The equations of two more straight lines, namely L2 and L 3 , are required for the
determination of the volume of the suction chamber. Figure (A-l2) shows the
positions of the lines L, (i=1,2,3) relative to the generating circles and the curves
C, (i=1,2,3,4).
By definition, the line L2 passes through F and it is tangent to the orbiting

generating circle at a point hereafter denoted by Q3 Of course, from an external
267
point two tangents can be drawn to a circle. The ubiquity is resolved by selecting
the line L 2 so that it coincide with the line L 1 when the suction chamber is
completely sealed at & = O ± 2nir, n E \W . The coordinates of the touching point
are obtained from the definition of the involutes and the statement (A.7). Since
the location of F on the curve C3 is determined by the angle PF' it easily can be
shown that
(xQ3(rcos(coF+7r)_dcose"
o (rcoscoE+dcose
= YQ 1 - r SIfl(P F + r) d sin6 ) -
J I\T F + d sin 6
(A.84)
6=O ±2n, nE\W , Oiim2Ô<(Oiim2+27r),
Note that the angle q'. is adjusted by r. With this adjustment the angle defining
the point Q3 on the circumference of the orbiting generating circle is measured
anti-clockwise from the positive X semi-axis, as against which is measured
anti-clockwise from the negative X semi-axis.
The equation describing the line L 2 is deduced from the coordinates of the points F
and Q3 , given by equations (A.78) and (A.84) respectively. After some calculations
it can be shown that
L2 YSlHF + XCOS F =—(r+dcos(c F - 0)),

(A.85)
6=O±2njt, nE\W, 0Iim2^0<(0m2+27
The line L 3 is defined by any two pairs of points at which the curves C2 and C3
come in contact. It should be emphasised here that all here points are collinear. The
values of the parameter ç corresponding to the two outermost pairs of these
touching points, are obtained from the relations (A.26) and (A.51) and the
definition of Kmax . Thus,
3,Kmax =2KmaxJr+yr/2+O,
2,Kmax = (2K ma + 1)r + r /2+0— y,
3,Kmax-1 = 2( Kma 1)2+,r/2+0,

(A.86)
(21 ma - 1)r + ir / 2 + 0— y
01im2 Ô<(O1j2+p), O^y<r-2
268
Introducing these equations and equation (A .22) into the second and third of the
relations (A.1O), and after long algebraic manipulations on the resulting
coordinates, we obtain the following equation for the line L3
L 3 : xsine—ycose=r,
(A. 87)
6=O±2nir, nE\W , eiim2ô<(eiim2+27r).
This line is tangent to the fixed and orbiting generating circles at two points
hereafter denoted by °2 and Q4 respectively. Their geometry is shown in figure
(A-12). The coordinates of these points are given by
XQ2 (COS(2,Kmax +
Q2 7))
lfl(2,Kmax +
(YQ2J
+,r)—dcos9
X Q4 (TCOS(c03,Kmax
(A .88)
d sin 6)'
[ YQ4] r Sfl(3,Kmax + ) -
6=O±2nr, ne\W 0. <0<'I Iim2 +2.ir), O^y<,r_2,

, 1im2
or, upon introduction of equation (A .86)
I sin(9'
[YQi r_ 06J,
XQ4 " rsin6—dcosO" (A.89)
Q4
yQ4J_rCoSe_dsiflOJ
6=0±2nr, flE\W , °lim2 Ô<(Oiim2+27t).
As mentioned above, the points °2 and Q4 are located on the line L 3 and the
circumferences of the fixed and orbiting generating circles. This can be proved by
direct substitution of equations (A.89) into equation (A .87) and the equations of the
generating circles. Note also that the gradients of the radii 002 and SQ4 , obtained
from the foregoing equations, are identical and equal to - cot 0. On the other
hand, the gradient of the line L 3 , as derived from equation (A.87), is tan 0.
Consequently, 0Q2 and SQ4 are perpendicular to the line L 3 . Hence, L 3 is
tangent to the generating circles at the points 02 and 04 as shown in figure (A-
12).
269
The values of the angles q2Kmax and entering equations (A.88), have been
adjusted by y and r respectively. With these adjustments the angles defining the
points Q2 and 04 are measured anti-clockwise from the positive X semi-axis, as
they should.
Two more points are required for the determination of the working volume of the
suction chamber, namely the point 05 , defined by the intersection of the lines L1
and L 3 , and the point Q6 defined by the intersection of the lines L2 and L3.
Equations (A.77), (A .85) and (A .87) form two algebraic systems. These, when
solved with respect to x and y, yield the following coordinates for the points 05
and 05
T ')(COSfl_COS6")
YQ5) sin(e—fl))sinfl—sinO J'
(A .90)
1Q6") = - 1 cosc9[r+dcos(q1F - e)1—rsinc9F"\
I,
1\ YQ6) cos(qi - ) sin E.1[r + d cos(q - 0)] +r cos F)
9= 0± 2n,r, nE \W, 0 11m2 ^ 0, 13 < (°Iim2 + 2it).
In order to obtain the instantaneous working volume of the suction chamber as

function of the crank angle 0 and the compressor's design parameters, we require
the expressions for a number of surface areas. The description of these surface
areas and the derivation of their analytical expressions are given below.
From the definition of the points Q1 , Q2 , S , and the diagram in figure

(A-12), it is clear that the section TKflX S2 of the involute C2 is defined for values
of the parameter q lying in the closed interval [2,Kmax0)2L When constructing
this section, the straight line segment Q2 TKmaX moves from its original position to
the position Q1S2 , sweeping an area hereafter denoted by AF . This area is shown
in figure (A-12) hatched with vertical lines. In complete analogy, the section
TKmaxF of the involute C3 is defined for values of the parameter qi in the range
When constructing this section, the straight line segment Q4TKm8X

moves to the position Q3F thus sweeping an area hereafter called A0 . This area is
shown in figure (A-12) hatched with horizontal lines.
270
UP
A ppendi)' A Scroll Compressor-Expander Geometry
Y*
x*
Figure (A -13). Determination of surface area under involute
271
The determination of the surface areas AF and A0 is based on the definitions of the
curves C2 and C3 . Figure (A-13) portrays the orbiting generating circle together
with a section of the curve C3 starting from the tip G3 . The differential area dA,
also shown in the figure, is swept by a line segment tangential to the generating
circle, when the value of the parameter changes by dq. According to statement
(A.7), any line tangential to the generating circle is perpendicular to the curve.
Thus, the differential area dA approximated to the second order, is a triangle with
base and height equal to rqdço and r respectively. Consequently
(A.91)
dA = !(r2q,zdq,).
The integration of this relation with respect to q' from an initial angle E to a final
angle £2 furnishes the marked area A shown in figure (A-13). Thus
2co2dco= —r2(e2_
3 3
e1 ). ('A.92)
A =!Jr
2E
The area is completely equivalent to the areas AF and A 0 , the only difference been
the range of integration. By definition, integrating over the range [ '2,K max' w2 ] we
obtain the area AF, while integrating over the range [ p3K max' q] we get the area
A0 . Combining equations (A.62), (A.67), (A.75), (A.76) and (A.86), these areas
can be expressed in the form
AF T{02 - Q'2,Kmax} =
3
rJ[( 2 N + 1)2r + + - [(2Kmax +1)+_+8_] },
6 2
Ao = rfr - '3,Kmax}
(A.93)
3
1 —12K rna+—+O
! r2 [2N++f l+sF]
6 L
] }
NK m,flE\W , °lim2 Ô,/3<(Oim2+27\

O<y^7r_2
272
where Sq 1. is the correction angle obtained from the iterative algorithm (A.82).
The relations (A.67), (A .70) and (A .93) guarantee that equations (A .93) always
yield non-negative values for the surface areas A F and A 0 . For 0 = O ± 2n,r,
n \W , these areas attain their minimum values A F = = 0. It is worth noting
that the angle 0 entering the foregoing equations, is the crank angle and not the
rotational angle 0. This is in agreement with what is stated in section A.2, since
the areas A F and A 0 are part of the scroll compressor's geometry and only
indirectly related to the motion of the fluid through the device.
1?igure (A-14). Details of the regions DFand D0
273
A ppendbc' A Scroll Compressor-Expander Geometry
The determination of the instantaneous volume of the suction chamber cannot be

completed without the expressions for the surface areas of three more plane
geometric regions. These are, the region DF bounded by the straight line segments
Q1Q5 , QQ and the minor circular arc Q 1Q2 ; the region D0 bounded by the
straight line segments Q3Q6 , Q4 Q5 and the minor circular arc Q3Q4 ; and finally
the triangle FQ5Q6 . The regions DF and D0 are shown in figure (A-14), in
comparatively large scale.
The area A of the triangle FQ5Q5 is computed from the formula
A ={ S(S— a)(S-b)(S-c)} "2, (A.94)
where by definition a, b and c are the triangle's sides and S is the triangle's
semi-perimeter, that is
aQ5Q6 , bFQ5 , cFQ6 , SL(a+b+c). (A .95)
The lengths a, b and c can be obtained from the coordinates of the points F, Q5
and Q6 . The resulting expressions are extremely long and they are not be stated
here. For computational purposes suffice to state the following
211/2
az{(xQs_xQ6)2+(yQs_y06) j
2 1/2
b = {( x F - X Q5) + (YF - YQ5) , (A.96)
1 2 2 1/2
c= t( x F_ x Q6) +(yF—yQ6) }
where the x and y coordinates entering these equations are given by the
expressions (A .78) and (A.90).
Figure (A-14) implies that the surface area A DF of the region DF is the difference
of the surface areas of the quadrilateral 0Q1Q502 and the minor circular sector
Q10Q2 . Hence,
ADF = { area of quadrilateral 0Q1Q5Q2 - minor circular sector Q10Q2 }. (A.97)
274
U
Regardless of the value of the rotation angle 6, the quadrilateral 0Q1Q5Q2 is an

inscribed quadrilateral with angles 0Q1Q5 = 0Q
2Q5 r /2. Consequently,
Q
area of quadrilateral OQIQ5 2 = OQ Q1 Q = r 1°SL (A.98)
The angle Q1°2 is easily obtained from the geometry depicted in figure (A- 14),
thus
angle Q1 ÔQ2 = 2arctan°1Q5l'J. (A.99)
By virtue of this equation, the surface area of the minor circular sector Q10Q2 is
given by
surface area of minor sector Q1002 = r2 arctanl IQIQSDI (A. 100)

r)
In view of equations (A.98) and (A.100), equation (A .97) becomes
ADF = r{IQ1Q5I - r arctan[i2slJ}. (A.l0l)
The expressions for the equal distances and QZQSI obtained from the
coordinates of the points Q, Q2 and Q5 given by equations (A.83), (A.88) and
(A.90), have the form
(1— cos(E) -/3))

= 02051= sin(O—/3)
(A. 102)
Hence, equation (A.101) reduces to
ADF
Ii- cos(6) - 3 3)11
/) arc tan[1 - cos(E) - /
(A. 103)
= r2 t sin(E) - 6) - sin(E) - /3) ]
Completely analogous steps furnish the follow relations from which the surface
area ADO of the region D0 can be deduced with the help of the geometry shown
in figure (A-14)
ADO { area of quadrilateral SQ3Q6Q4 - minor circular sector Q3SQ4),
area of quadrilateral SQ3 Q6Q4 =r Q3061'

= 1S031103061
275
A ppend/k A Scroll Compressor-Expander Geometry
angle Q3Q4 = 2arctan°3Q6J
surface area of minor sector Q3SQ4 r2 arctanQ3Q6I"J
ADO = r{Q 3 Q 6 - r arctanQ6IJ}
(1 - cos(0 - /3))
Q3Q6 = =r
sin(0 - ,8)
Combing these equations, we obtain the expression for the surface area A DO of the
region D0 in the form
sin( F - 0) - arctan[1 - sin(q - 0)11

ADO = r2{1 - (A.104)
cos(qF — 0) cos(F — 0) IL
Figures (A-i 1) and (A-i2) show plan views of the scroll compressor-expander for
various values of the crank angle 0. A careful inspection of these figures reveals
that the base area A 5 of the suction chamber can be computed using the following
formulae
Table (A-2)
If then and
0= 0=05±2n7r, n€\W A 5=
AF - A0 = 0
0 <0 < O ±2nir< 0< 0 ±2nir

fl E
A F —A 0 —(AT+ADF—ADO)
0= 0 = O, ± 2nr, n E
AF - A0 + 2rd = 2rd
A 5=
O <O<(O +2,r) O ±2n,r< 0< O ±2nr+2,r
n E \W A F —A 0 +(A I +A DF — A 00)
276
WY .-
where, by definition O, 0 + it. Evidently, the surface areas entering the

equations in table (A-2) are given by (A.93), (A.94), (A.103) and (A.104).
The terms cos(q — 0) and sin(0 — c'F) appearing as denominators in the

foregoing expressions vanish for specific values of the rotation angle 0 and they
may cause computational problems while calculating the suction volume.
Specifically, for 0 = O ± nit, n E \W equations (A.67), (A.75), (A.76), and
(A .82) furnish
sin(0 - = cos(qF — 9)i = (A.105)

O=6,±n,r O=O±n,r
This implies infinite values for the distances QJ Q5 , Q3Q6 , Q5 Q6 , FQ5 , FQ

and the areas ADF and ADO. However, as the rotation angle 0 approaches the
values 0 = Os ± nit, n \W , the area (A T + A DF - ADO ) tends to finite values.
Indeed,
lim(AT + ADF - ADO ) = 2rd

(A.106)
0 —* ± (Zn + 1)ff
and
lim(AT +A DF A DO )— 0
(A. 107)
0 —* ± 2n,r
A.3.3 Compression Chamber and its Working volume
Following the suction process, the first compression cycle starts when
O = O - 2,r, 0= 0 ± 2nir, n\W and completes after a full rotation when
E
o = - 4,r, 0= 0 ± 2n,r — 2ir n E \W . If the design of the scroll device

allows for more than one compression cycle, the mth, mEZ such cycle starts
when 0 = 0 ± 2nir— 2mir, n E\W , mEZ
and completes when
0=O±2nir-2(m+1)ir, n€\W , m&Z ,where the symbol Z denotes the
set of the positive integers.
Due to the symmetry of the device each compression cycle is associated with a
pair of geometrically equivalent chambers. These chambers, regardless of the value
of 0, have the same shape and area. They also have rotational symmetry of 1800.
In the general case of the m th , mEZ compression cycle, the first of these
chambers is formed by the scroll device's surfaces defined by the curves C2 and C3
277
and it is sealed by two pairs of touching points corresponding to the following

values of the parameter q
3,Kmax—m+1 = 2(K max - m + 1),r + r /2 + 0,
= 2(K max - m + 1)r+ r+ r/2+ O— y,
3,Kmax—m = 2(K max - m)ir + ir/ 2 + 0, (A.108)
2,Kmax—m = 2(K max - m)ir+ 'r+ r/2+ 0— y,
Olim2^ O <(OIim2+ 2 ), O^y<,r-2, mEZ.
53
Figure (A -15). Details of a compression chamber
278
. r
The second chamber is formed by the by the scroll device's surfaces defined by the
curves C1 and C4 and it is sealed by the touching points associated with the
following values of the parameter q
1,Kmax-m+I = 3,Kmax-m+1' 4,Kmax-m+1 2,Kmax-m+1
1,K max-in = q'3,K max-rn ' 4,K max-rn = 2,K max-rn
A plane view of the scroll compressor is shown in figure (A-15), where one of the
outermost compression chambers is clearly marked. From these figure as well as
figures (A-il) and (A-12), it is clear that following steps similar to the steps
leading to equations (A.93), we obtain the expression for the base area A m of
either of the symmetrical compression chambers of the mth, m Z compression
cycle. This expression has the form
A,m = r 2 i$ ,Kma(_m+1 2d _2 1,(ma,m+1 2d} =

q2,Kmax-m J3,Kmax-m
{[2(K max_ m+1)2+7r+7rI2+O_Y ] - [2(Kmax—m)lr+lr+7r/2+9-7]3
(A. 109)
]3}
- [2(Km - m + 1)r + r /2 + + [2(Km - m)lr + .ir / 2 +
°Iim2^ O /3 <( OIl m 2 +2), O<y^.ir-2, mZ
The formulae given above hold only for cases where the compression chambers are
formed by surfaces fully described by the curves (A.10). However, there are cases
in which parts of the walls are formed by the innermost circular arcs defined by
equations (A.36), (A.38), (A.44) and (A.46), as shown in figure (A-16). Clearly
the equation (A.109) cannot be applied. This equation is not valid under the
condition 8Iiml ^ 0 < (Iim2 + 2ir). In these cases, in order to determine the base
area Arn mth, mEZ, knowledge of the following areas is required:
• The area AD1 enclosed by the curve C1—C Q , the innermost circular arc T2G1
and the straight line segment T2 T2 , shown in figure (A-16). In figure (A-17)
this area is hatched with horizontal lines.
• The base area AD2. This is the section of the device's fixed arm, enclosed by
the straight line segments 77 and AC and the curves CI CFQ and C2—CR . It
is shown in figure (A-16) and partially in figure (A-l7) hatched with slant
lines. By construction, the straight line segment AC is perpendicular to all
279
. r__
AppendA" A Scroll Compressor-Expander Geometry
curves C1 (i=1,2,3,4 ), while by definition, the point B is located on this

segment and is the first touching point between the curves CI CF0 and C4C01.
The values of the parameter corresponding to point B are given by
equations (A.26) and have the form ç = = it / 2+8 and
q 44,0 —ir+;ir/2+Oy.
. The area AD3 enclosed by the two generating circles and their external
tangents, shown in figure (A- 16) cross-hatched.
A03
(p,= (p10= It/2+ 0

and
(p4 (p40 1c-1-1r/2ê
(p3 it + IC /2 + e
Figure (A-16). Details of a innermost compression chambers.
280
• The area AD4 enclosed by the innermost circular arcs defined by equations
(A.36), (A.38), (A .44) and (A.46), for various values of the crank angle 0 in
the semi-open interval [Oijmi °11m2 + 2.ir),.
The determination of the above mentioned areas require knowledge of the sizes of
various angles and the lengths of a series straight line segments, which are
calculated below.
From figures (A-7a) and (A-17), the definition (A .36) of the tip point G 1 and the
definition (A .41) of the angle 0IjmJ'
it is at once obvious that
arctan cr' (A.11O)

X ÔF cr and XOG1 °Iiml - = c°cr -
and hence,
(A . 111)
G1 OFX OF— X ÔG 1 = arctan cr
C 1 = CF,
- r1
Figure (4-17). The base areas A DJ and A D2
281
On the other hand, according to equations (A.37), the point 7 is located on the
curve C2 = CR and the corresponding value of the parameter q is
= 2S = + q - y. Therefore, as suggested by equations (A .36) and (A.35),
the value of q corresponding to the point T, located on the curve CI CFQ , is
XÔH= +' = + cr
(A . 112)
92S
This equation and the definition of the curve CI— CFO furnish
HT2 =OH(X OH)= r(2r+ Pcr)' (A .113)
consequently,
-. * (H7*
HOT2 = arctanl I = arctan(,r + cr) (A . 114)
(OH)
The coordinates of the point H, obtained form this equation, are given by
(C0S(cocr+it)'\ (COScOcr\
H 1=—ri I. (A . 115)
YH)
[X
Sifl(cocr + it)) \Sfl cr)
The foregoing equation and equations (A .37) and (A .38) furnish the slopes of the
straight line segments T QORB and QORBH. These slopes are identical and equal
to - cot Therefore, the points T2', QORB and H are collinear. Similarly, from
equations (A .11O) and (A .112), it is cvident that the angle FOH is a straight angle
and thus the points F, 0 and H are also collinear. Furthermore, combining
equations (A .38) and (A .115), the distance HQORB is found to be
(A.116)
HQQRB = T,
hence
HQORB
HOQ ORB = arctan air) = arctan Cr' (A.117)
and
OORB T = OHQORB +HÔ QORB - + arctan cr (A . 118)
From figure (A-17) and equations (A .111) and (A .117) it is evident that the
triangles FOG1 and HOQORB are equal, that is
282
Appendl)( A Scroll Compressor-Expander Geometry
AFOG1 = LHOQORB . (A .119)
As a result of this relation, the areas of the plane figures HFGIPT2*H and
QORB G1 PTQORB are equal and hence figure (A-17) easily yields the following
relation
area of QORB G1 F'T QORB - area of HFG1 PT H=
(A . 120)
=area of HBFG1PTH+ 2
The area of the plane figure HBFG1 PTH entering the foregoing equation is
swept by the tangent to the fixed generating circle BP, as it moves from the
position FG1 to the position HT. It can therefore be calculated following steps
completely analogous to the steps leading to equation (A.92), where of course
E1 =XÔF= cr and e 2 =XÔH= r + q. Thus equation (A .120) takes the form
3 31
area of QORBGIPTQORB {(" q'cr) 'cr (A .121)
The base area A DJ, as figure (A-17) suggests, is obtained by subtracting the area
A DI _c of the circular sector QORB G1 T2 , centered at QORB, from the area A DI F of
the plane figure QORB G1 PTQORB ,that is
AD1 = ADI _ F - ADI_c. (A . 122)
The area A91_ of the circular sector, determined with the help of equation
(A .118), is given from
= 2 OQORB TZ r2
= - - + arctan
AD1_c cr} (A . 123)
2 2 2
Substitution of the relations (A .121) and (A .123) into equation (A .122) yields
r211
AD1 = + co) - cc + - arctan Pcr (A . 124)
The determination of the area A D2 is again based on the discussion leading to

equation (A.92). By definition , the values of the parameter corresponding to the
points A , B and C shown in figure (A-16), are given by
283
Q1O = + 0, =7r+-+0y ,
(A .125)
2A 2,Oh+OY Q3C_7+O.
On the other hand, according to equation (A .112), the values of the parameter q7
corresponding to the points T and T2 are
(p u. =X ÔH= 2S 7 7t + cr and 2T2 = 2S = r + cr - y (A.126)
respectively.
From figure (A- 16) it is evident that the area A D2 is the difference of the areas
swept by the tangents to the fixed generating circle HT2 and HT2 as they move
form their original positions to the positions DB and DA . Thus, according to
equations (A .92), (A .125) and (A .126)
1 ,r/2-4-G ,r12+O-y
1
A D2=- $r 2 co 2 dço_ $r2co2dco=
L ,r+q', J
(A .127)
2/ \3 3
F I(7 31
=— .fl _+0 I _+o_iJ
6 L2
The base area A D3 is the difference between the areas of a parallelogram and a
generating circle. The width of the parallelogram is equal to the distance between
the centres of the generating circles while its breadth is equal to their common
diameter. Thus
A D3 =r(2d— ) (A .128)
The area A D4 is the common intersection of two identical circles. From elementary
geometry is known that such an area is given by R 2 (co - sin w)12, where R is the
common radius of the intersecting circles and w is the angle to which the common
chord subtends.
The co-ordinates of the point G 3 when referred to a system of axes X Y shown in

figure (A-18), been parallel to X Y and centered at the point QORB, are obtained
from equations (A .36) and (A.38), thus
284
-\ I-'
-.3 L00
Cl= C x
Figure (A-18). Details of the innermost miniscus
G3 = 1 XG3 - IXQ0RB "1 d07r + O) = _d0S e (A.129)

Y G3 ) YQOIth) sin(,r+O)J sinO)
Consequently, the line segments QORB G3 and OS are parallel having the same
slope equal to - tan8. Referring to equation (A .41) and the figures (A-iS), (A-16)
and (A-17), we have
X 'ÔG1 =X ÔG1 = 2 cr - arctanq. (A.130)
and therefore
G1 ÔS=GI QOrb G3 = 0— - - arctan 'cr) = 0— 0Hm1'

(A . 131)
Oiimi - <( Olimi +2g ).
<0 -'
285
'r
A ppendbc- A Scroll Compressor-Expander Geometry
Furthermore equations (A.37) and (A.38) furnish QORB 7T2 = r(,r - y) = d while by
definition OS= d . Hence
QORBiZ=OS=d. (A. 132)
Application of the formula R 2 (co - sin w)12 mentioned above for R = d and
w =G QORB G3 0- 8Iiml' furnishes
A D4 =-{ (e 0iimi )S1fl(0Oiimi )}. (A. 133)
Having found the expressions for the areas A DI , A D2 and AD3 it is easy to
determine the area mtt, Z of the innermost compression chambers.
1 2 I 2
I
21r 1 ,r
=-
2d=—r I+—+0—y —[—+0- (A.134)
2 I
(2
I 6 2 [2 1}
ft
o
limi <0<(011m2 +27t)
Therefore,
(21—AD3)-2AD2-2AD1--AD4 _IA.D3AAAD4 (A .135)

2 2 2
A.3.4 Discharge Chamber and its Working volume
Following the compression process, the discharge cycle starts when

6 = 0± 2n,r— 2(m + 1),r, it e\W , mEZ , where the symbol denotes the
set of the positive integers, and completes when 6 0 ± 2nr - 2(m + 2)'r.
For values of the parameter 0 in the range °Jiml ^ 0 < ( 011m2 + 2.ir) the working
volume is given by equation
A D = -{(o Olimi ) sin(0 - Olimi )} (A.136)
Where as for the rest of the discharge process is given by the formulae
A D =(21—AD3)-2AD2-2AD1 (A.137)
286
(ç= lt+7t/2+O
(p4=lt,2+e-y / (P;-(P,0 it+it/2+9 i
(p3 It +It/2+ e
Figure (A -18). Determination of area of innermost chambers
287
U
A ppendix B
Governing
Equations
B.1 The Continuity Equation
Physical principle: mass is conserved
Consider the flow model shown on figure B.1 , namely, an infinitesimally small
element fixed in space, with the fluid moving through it. Here, for convenience we
adopt a Cartesian coordinate system, where the velocity and density are functions
of (x,y,z) space and time t. Fixed in this (x,y,z) space is an infinitesimally small
element of sides dx, dy, and dz, (figure B.!). there is a mass flow through this fixed
element, as shown in figure B.1. Consider the left and right faces of the element
which are perpendicular to the x axis. The area of these faces is dy dz. The mass
flow through the left face is (pu) dy dz. Since the velocity and density are functions
of spatial location, the values of the mass flux across the right face will be different
from that across the left face; indeed, the difference in mass flux between the two
faces is simply [[pu)/dx] dx. Thus the mass flow across the right face can be
expressed as { pu+ ['pu)Idx] dx } dy dz. The mass flow across both the left and
the right faces is shown in figure B.1. In a similar vein the mass flow through both
the bottom and the top faces, which are perpendicular to the y axis, is (pv) dx dz
and {pv+ [c'pv)Ic]dy } dx dz, respectively. The mass flow through both the front
and back faces, which are perpendicular to the z axis is (pw)dxdy and
{ pw+ pw)I&dz} dx dz, respectively. Note that u, v, and w are positive, by
convection, in the positive x, y, and z directions, respectively. Hence the arrows in
figure B.1. Show the contributions to the inflow and outflow of mass through the
sides of the fixed element. if we denote a net outflow of mass as a positive
quantity, then from figure B. 1 we have
Appendbc- B The Governing Equations
Net ouflow in x dire ction.
d(pu) d(pu)
pu+ dx dydz— (pu)dydzdx
= dy dz
dx
Net outflow in y direction:
d(pv) d(pv)
pv+ dy dxdz— (pv)dxdz= dxdydz (B.1)
dy dy
Net outflow in z dire ction:
[ d(pw) 1 d(pw)
Ipw+ dzldxdy—(pw)dxdy= dxdydz
L dz ] dz
Hence, the net mass flow out of the element is given by
[d(pu)d(pv)
_____ ____ d(pw) l
Net mass flow = I Idxdydz (B.2)
L v dy dz ]
The total mass of fluid in the infinitesimally small element is p (dx dy dz); hence
the time rate of increase of mass inside the element is given by
Time rate of mass increase= --(dx dy dz) (B.3)
The physical principle that mass is conserved, when applied to the fixed element in
figure B.1, can be expressed in words as follows: the net mass flow out of the
element must equal the time rate of decrease of mass inside the element. Denoting
the mass decrease by a negative quantity, the statement can be expressed in terms
of equations (B.2) and (B.3) as
[d(pu) + d(pv) d(pw)

]dxdY dz=_(dxdY dz)
dx dy + dz dt
(B.4)
dp [d(pu) + d(pv) + d(pw)1 =
or
dx dy dz ]
289
A ppendix B The Governing Equations
p = p (x,y,z,t)
y V = ui +vj +wk
U = u (x,y,z,t)
V = v (x,y,z,t)
w = w (x,y,z,t)
-I
I
x
t9(P v)
pv+ dy dxdz
z dy [
Ipw+
d(pw)
dzdxdy
d(pu)
dx dydz
Pu dy dz
pvuxaz
Figure B.1 Model of the infinitesimally small element fixed in space and a diagram
of the mass fluxes through the various faces of the element-for a derivation of the
continuity equation.
290
A ppendi)' B The Governing Equations
hi equation (B.4) the term in brackets is simply V (pV). Thus equation (B.4)
becomes
(B.5)
Equation (B.5) is a partial differential equation form of the continuity equation. It

was derived on the basis of an infinitesimally small element fixed in space. The
infinitesimally small aspect of the element is why the equation is obtain directly in
partial differential equation form. The fact that the element was fixed in space
leads to the specific differential form given by equation (B.5), which is called the
conservation form. The forms of the governing flow equations that are directly
obtained from a flow model which is fixed in space are, by definition, called the
conservation form.
B.2 The Momentum Equation
In this section, we apply another fundamental physical principle to a model of the

flow namely:
Physical principle: F= ma (Newton's second law)
The resulting equation is called the momentum equation. Unlike the derivation of
the continuity equation, where the model of an infinitesimally small element fixed
in space was used, here we will apply a different model. We will utilize the moving
fluid element model shown in figure B.2 because this model is particularly
convenient for the derivation of the momentum equations as well as the energy
equation (to be consider in section B.3). Momentum and energy equations can be
derived using the other model of flow shown in figure B. 1, each different model of
flow leads directly to a different form of the equations.
Newton's second law, expressed above, when applied to the moving fluid element
in figure B.2, says that the net force on the fluid element equals its mass times its
acceleration of the element. This is a vector relation, and hence can be split into
three scalar relations along the x, y, and z axes. Let us consider only the x
component of Newton's second law,
F = ma (B.6)
where F and a are the scalar x components of the force and acceleration
respectively.
291
Appendiv B The Governing Equations
Velocity
I components
uy
w ( dr
yx +_--dY ]dxdz rdxdy
dy
pdydz — Ip+"dx "dy dz
: ,j, :9x )
idydz *– -1 --_____________
+ -ELd xJ dy dz
a
+ d zJ dx dy
dx dz
Fluid elem
at time t =
z
V2
Same fluid element
at time t =
Figure B2 Infinitesimally small, moving fluid element. Only the forces in the x
direction are shown. Model used for the derivation of the x component of the
momentum equation.
292
A ppendi) B The Governing Equations
First, consider the left side of equation (B.6). We say that the moving fluid
experiences a force in the x direction. There are two sources of this force:
1. Body forces, which act directly on the volumetric mass of the fluid element.
These forces "act at a distance"; examples are gravitational, electromagnetic,
centrifugal, and Coriolis forces.
2. Suiface forces, which act directly on the surface of the fluid element. They are
due to only two sources: (a) the pressure distribution acting on the surface,
imposed by the outside fluid surrounding the fluid element, and (b) the shear
and normal stress distributions acting on the surface, also imposed by the
outside fluid "tugging" or "pushing" on the surface by means of friction.
Let us denote the body force per unit mass acting on the fluid element by f, withf
as its x component. The volume of the fluid element is (dx dy dz); hence,
Body force on fluid element acting in x direction = pf (dx dy dz) (B. 7)
The shear and normal stresses in a fluid are related to the time rate of change of the
deformation of the fluid element, as sketched in figure B.3 for just the xy plane.
The shear stress denoted by r in figure B.3a, is related to the time rate of change
of the shearing deformation of the fluid element, whereas the normal stress,
denoted by r in figure B.3b, is related to the time rate of change of volume of the
fluid element. As a result, both shear and normal stresses depend on velocity
gradients in the flow. In most viscous flows, normal stresses (such as i) are much
smaller than shear stresses and many times neglected. Normal stresses (say in
the x direction) become important when the normal velocity gradients (say du/dx)
are very large, such as inside a shock wave.
x x
(a) (b)
Figure B.3 Illustration of (a) shear stress and (b) normal stress
293
A ppendA ' B The Governing Equations
The surface forces in the x direction exerted on the fluid element are sketched in
figure B.2. The convention will be used here that ib denotes a stress in the i
direction exerted on a plane perpendicular to the i axis. On the face abcd, the only
force in the x direction is that due to shear stress, i dxdz. Face efgh, is a distance
dy above face abed; hence the shear force in the x direction on face efgh is
[r,+(dr1 9y ) dy 1 dxdz . Note the direction of the shear force on the faces abcd and
efgh; on the bottom face, i is to the left (the negative x direction), whereas on the
top face, -i-(c9,jdy)dy is to the right (the positive x direction). These directions
are consistent with the convention that positive increases in all three components of
velocity, u, v, and w, occur in the positive direction of the axes. For example, in
figure B.2, u increases in the positive y direction. Therefore, concentrating on the
face efgh, u is higher just above the face than on the face; this causes a "tugging"
action which tries to pull the fluid element in the positive x direction (to the right)
as shown in figure B.2. In turn, concentrating on face abcd, u is lower just beneath
the face than on the face; this causes a retarding or dragging action on the fluid
element, which acts in the negative x direction (to the left) as shown in figure B.2.
The directions of all the other viscous stresses shown in figure B.2 including 't.,
can be justified in a like fashion. Specifically on face dcgh, i acts in the negative
x direction, whereas on face abfe, + (drJdz)dz acts in the positive x direction.
On face adhe, which is perpendicular to the x axis, the only forces in the x
direction are the pressure force p dy dz, which always acts in the direction into the
fluid element, and r dy dz, which is in the negative x direction. In figure B.2, the
reason why 'ti, on face adhe is to the left hinges on the convention mentioned
earlier for the direction of increasing velocity. Here, by convention, a positive
increase in the u just to the left face of adhe is smaller than the value of u on the
face itself. As a result, the viscous action of the normal stress acts as a "suction" on
face adhe; i.e. there is a dragging action toward the left that wants to retard the
motion of the fluid element. In contrast, on face bed, the pressure force [p +
(/&)dx]dy dz presses inward on the fluid (in the negative x direction). And
because the value of u just to the right of the face bcgf is larger than the value of u
on the face, there is a "suction" due to the viscous normal stress which tries to pull
the element to the right (in the positive x direction) with a force equal to [z +
(c9rh9x)dx]dy dz.
With the above in mind, for the moving fluid element we can write
294
A ppendA B The Governing Equations
Net surface force in x direction = [ - + dx]dY dz
^YX
+ ' dxJ_rxx]dYdz+[[Tx dY )J_r,x]dXdZ (B.8)
+ r+ dz –r dxdy
L dz )
The total force in the x direction F, is given by the sum of equations (B.7) and
(B.8). Adding, and canceling terms, we obtain
+
YX dXXX
+ +d
F = [- 1dx dy dz + dx dy dz (B. 9)
[ dx dx dy dzj
Equation (B.9) represents the left-hand side of equation (B.6). Considering the
right-hand side of equation (B.6), recall that the mass of the fluid element is fixed
and is equal to
m = pdxdydz (B.1O)
Also recall that the acceleration of the fluid element is the time rate of change of its
velocity. Hence, the component acceleration in the x direction, denoted by a, is
simply the time rate of change of u; since we are following a moving fluid element,
this time rate of change is given by its substantial derivative. Thus,
Du
a (B.11)
Dt
Combining equations (B.6) and (B.9) to (B.l1), we obtain
Du dp d
XX+
___ ___
(B.12a)
p—=--+
Dt dx dx dy dz
Which is the x component of the momentum equation for a viscous flow. In a

similar fashion, the y and z components can be obtained as
Dv - dpd dr c9r
p - -
Xy YY ZY
+
(B. 12b)
-;;- dx + dy +
Dw dp dr dv
XZ+ Y Z +_+pf
+ (B. 12c)
Dt d z d x dy dz
295
Equations (B.12a) to (B.12c) are the x, y, and z components respectively, of the

momentum equation. Note that they are partial differential equations obtained
directly from an application of the fundamental physical principle to an
infinitesimal fluid element. Moreover, since this fluid element is moving with the
flow, equations (B.12) are in non conservation form. They are scalar equations and
are called the Navier - Stokes equations in honor of two men, the Frenchman
M.Navier and the Englishman G. Stokes, who independently obtain the equations
in the first half of the nineteenth century.
The Navier - Stokes equations can be obtained in conservation form as follows.

Writing the left-hand side of equation (B.12a) in terms of the definition of a
substantial derivative we have
Du c9u
p— =p---FpV •V u (B.13)
Dt dt
Also, expanding the following derivative,
d(pu) :9U+t9p
c9t dt
and rearranging, we have
du - d(pu) dp (B.14)
dt
Recalling the vector identity for the divergence of the product of a scalar times a
vector, we have
V . (puV) = uV . (pV )+ (pV ) . Vu

(B.15)
or pV.Vu.=V.(puV)–uV.(pV)
Substituting equations (B.14) and (B.15) into equation (B.l3).
Du - d(pu)
p--- dt .(pV )+ V . (puV )
dt
(B.16)
- d(pu) -
+ V . (PV )] + V (puv)
- at [dt
296
A ppendbc B The Governing Equations
The term in brackets in equation (B.16) is simply the left-hand side of the
continuity equation, hence the term in brackets is zero. Thus equation (B.16)
reduces to
Du a(Pu)V (V )
(B. 17)
iit = d t
Substitute equation (B. 17) into (B. 12a).
____ dr (B.18a)
dt dy dz
Similarly, equations (B.12b) and (B.12c) can be expressed as
dr dr
+V.(pv17)=-+
X)'
YY+_2+ (B. 18b)
and
d(pw) + V ___
+dr
YZdr
++pf (B.18c)
. (pwV ) =
at dz x dy dz
Equations (B.18) are the Navier - Stokes equations in conservation form. In the late
seventeenth century, Isaac Newton stated that shear stress in a fluid is proportional
to the time rate of strain, i.e., velocity gradients. Such fluids are called Newtonian
fluids. (fluids in which r is not proportional to the velocity gradients are non
Newtonian fluids). For Newtonian fluids, Stokes in 1845 obtained
? =(V.V)21u (B. 19a)
=%(V . V )2j- (B. 19b)
rzz =(VV)2ji (B. 19c)
r a du1 (B. 19d)

= r5 =tI — +
[ax
= =[du dw
ti + —
(B. 19e)
Ldz dx
297
1 dw dv 1
(B.19f)
ryz =
where u is the molecular viscosity coefficient and 2 is the second viscosity

coefficient. Stokes made the hypothesis that
which is frequently used but which has still not been definitely confirmed to
present day. Substituting equations (B.19) into (B.18), we obtain the complete
Navier - Stokes equations in conservation form.
B.3 The Energy Equation
Physical principle: Energy is conserved
In keeping with our derivation of the Navier - Stokes equations (i.e. the momentum
equation) in section B.2 we will use again the flow model of an infinitesimally
small fluid element moving with the flow (as shown in figure B.2). The physical
principle stated above is nothing more than the first law of thermodynamics. When
applied to the flow model of a fluid element moving with the flow, the first law
states that
Rate of change Net flux of Rate of work done

of energy inside = heat into + on element due to (B.20)
fluid element element body and surface forces
A = B + C
where A, B, and C denote the respective terms above.
Let us first evaluate C; that is. Let us obtain an expression for the rate of work done
on the moving fluid element due to body and surface forces. It can be shown that
the rate of doing work by a force exerted on a moving body is equal to the product
of the force and the component of velocity in the direction of force. Hence the rate
of work done by the body force acting on the fluid element moving at a velocity V
is
pf .V (dxdydz)
298
With regard to the surface forces (pressure plus shear and normal stresses),
consider just the forces in the x direction shown in figure B.2. The rate of work
done on the moving fluid element by the pressure and shear forces in the x
direction shown in figure B.2 is simply the x component of velocity u, multiplied
by the forces; e.g. on face abcd the rate of work done by r dx dz is ui dx dz,
with similar expressions on the other faces. To emphasize these energy
considerations, the moving fluid element is redrawn in figure B.4, where the rate of
work done on each face by surface forces in the x direction is shown explicitly. To
obtain the net rate of work done on the fluid element by the surface forces, note
that forces in the positive x direction do positive work and that forces in the
negative x direction do negative work. Hence, comparing the pressure forces on
face adhe and bcgf in figure B.4, the net rate of work done by pressure in the x
direction is
r ( 9(up) d(up)
I up —I up + dx I dy dz = - dx dy dz
dx dx
Similarly, the net rate of work done by the shear stresses in the x direction on faces
abcd and efgh is
d(ur ) d(ur )
u + dy - u dx dz = - dx dy dz
dy dy
Considering all the surface forces shown in figure B.4, the net rate of work on the
moving fluid element due to these forces is simply
[ d(up)d(ur)d(u1 y ) d(ur )1
+ Idrdydz
dx dx dy dz ]
The above expression considers only surface forces in the x direction. When the
surface forces in the y and z directions are also included, similar expressions are
obtained. In total, the net rate of work done on the moving fluid element is the sum
of the surface force contributions in the x, y, and z directions, as well as the body
force contribution. This denoted by C in equation B.20 and is given by
299
Appendix B The Governing Equations
Id(UP)+
dx dy dz ) dx dy dz
+ d(vr) + d(vr) + d(vr.)

(B.21)
dx dy dz
d(wr) d(w)d(wr)1
+ dx + dy dz ]
Note in equation (B.21) that the first three terms on the right-hand side are simply
V.(pV).
[ dur
ur + dyY Y
xd jdxdz u r dx dy
h
g
( d(ouP)]
zv. 1.uP+ dx
-iz -
e :ay I
up dy dz + dur dxJddz
dx
urdy dz ___
dy dz
[qi +3_dxJdydz
dx
[u1 + dzjdxdy
z
u dx dz
Figure B.4 Energy fluxes associated with an infinitesimally small, moving fluid
element. For simplicity, only the fluxes in the x direction ere shown. Model used
for the derivation of the energy equation.
B is the net flux of heat into the element. This heat flux is due to (1) volumetric
heating such as absorption or emission or radiation and (2) heat transfer across the
surface due to temperature gradients, i.e. thermal conduction. Define q as the rate
of volumetric heat addition per unit mass. Noting that the mass of the moving fluid
element in figure B.4 is pdx dy dz, we obtain
Volumetric heating of element = pq dx dy dz (B.22)
300
In figure B.4, the heat transfer by thermal conduction into the moving fluid element
across face adhe is q dy dz, where q is the heat transferred in the x direction per
unit time per unit area by thermal conduction. (The heat transfer in a given
direction, when expressed in dimensions of energy per unit area perpendicular to
the direction is called the heat flux in that direction.) The heat transfer out of the
element across face bcgf is [ + (t9'j/dx)dx]dydz. Thus, the net heat transfer in
the x direction into the fluid element by thermal conduction is
+-/--dxJ]ddz = ----dxdydz
[ -
Taking into account heat transfer in the y and z directions across the other faces in
figure B.4 we obtain
Heating of
= _[L +L + Z JdxdY dz (B.23)
fluid element by
thermal conduction
The term B in equation (B.20) is the sum of equations (B.22) and (B.23).
B (B.24)
The heat flux due to thermal conduction, from Fourier's law of heat conduction, is
proportional to the local temperature gradient:
dT . dT . dT
q=–k----- q=–k---- q=–k-----
cix cy c'Z
where k is the thermal conductivity. Hence equation (B.24) can be written
r a(
B = pq+-I k— 1+ (B.25)
L a dx) __[
dT dy dz dz )]
Finally the term A in equation (B.20) denotes the time rate of change of energy of
the fluid element. The fluid element has two contributions to its energy:
1. The internal energy due to random molecular motion, e (per unit mass). This is
the physical significance of the internal energy that appears in the first law of
301
thermodynamics.
2. The kinetic energy due to translational motion of the fluid element. The kinetic
energy per unit mass is simply V 2/2.
Hence, the moving fluid element has both internal and kinetic energy; the sum of
these two is the "total" energy. In equation (B.20), the energy in the term A is the
total energy, i.e. the sum of the internal and kinetic energies. The total energy is
e + V 2 /2. Since we are following a moving fluid element, the time rate of change
of the total energy per unit mass is given by the substantial derivative. Since the
mass of the fluid element is p dx dy dz, we have
V 2'
A _ p__
D e+__Jdxddz (B.26)
The final form of the energy equation is obtained by substituting equations (B.2 I),
(B.25), and (B.26) into (B.20)
D ( V 2'l . d ( dT d ( dT"l a ( dT
p— i e+— i =pq + — i k— i+ — i k— i+ — i k -
Dt 2) dx ' dx ) dy dy ) dz ' dz
- d(up) d(vp) d(wp) d(ur) d(ur,) d(ur)
(B.27)
dx dy dz dx dy dz
d(vr) + d(vr) + d(v) + d( w r) + d(wr) + d(wr) +pf.V
+ dx dy dz dx dy dz
This is the non conservation form of the energy equation; also note that it is in
terms of the total energy e +V 2 /2. Once again, the non conservation form results
from the application of the fundamental physical principle to a moving fluid
element.
The left-hand side of the equation (B.27) involve the substantial derivative of the
total energy D (e + V 2/2)/D t. This is just one of the many different forms of the
energy equation; it is the form that comes directly from the principle of
conservation of energy applied to a moving fluid element. This equation can be
readily modified in two respects as follows;
1. The left-hand side can be expressed in terms of the internal energy e alone, or
the static enthalpy h alone, or the total enthalpy h0 = ( I, + V 2/2) alone.
302
A ppendi)- B The Governing Equations
2. For each of the different forms of the energy equation mentioned above, there
are both non conservation as well as conservation forms.
Note that the manipulations required to change the non conservation form to the
conservation form change only the left-hand side of the equations; the right-hand
side remains the same. Equation (B.28) below is the conservation form of the
energy equation written in terms of total energy e + V 2/2.
d[ ( v 2l . d ( d
P1 e+— H+V. [[e+Jv ] =pq+k _I+
dt[ 2)] dx dx ) dy [ dy)
d i dT d(up) d(vp) d(wp) d(u r) d(ury )d(ur)
+— k— _____ (B.27)
dz dz ) dx - dy dz d d
+ d(v,) + d(v) + d(v) + d(wr) c9( wr ) d(wr)
+ ±p f V
dx dy dz dx dy dz
B.4 The Equation Of State
The motion of a fluid in three dimensions is described by a system of five partial

differential equations: mass, three momentum equations, and energy. In these
equations there are seven unknowns p, p, e, T, u, v, w . In order to close the system
of fluid dynamic equations it is necessary to establish relations between the
thcrmodynamic variables p, p, e, T as well as to relate the transport properties
1u,k to the thermodynamic variables. It is obvious that two additional equations
are required. These two additional equations can be obtained by determining
relations that exist between thermodynamic variables. Relations of this type are
known as equations of state. According to the state principle of thermodynamics,
the local thermodynamic state is fixed by any two independent thermodynamic
variables, provided that the chemical composition of the fluid is not changing
owing to diffusion or finite-rate chemical reactions. (John C Tannehill page 258).
A perfect gas is defined as a gas whose intermolecular forces are negligible. A

perfect gas obeys the perfect gas equation of state,
p =pRT (B.28)
where R is the gas constant. The intermolecular forces become important under
conditions of high pressure and relatively low temperature. For these conditions,
303
A ppendb B The Governing Equations
the gas no longer obeys the perfect gas equation of state, and an alternatively
equation must be used. An example is the compressibility factor Z (equation
(3.23) section 3.2.5), another example is the Van der Waals equation of state,
(P+ap2)[-!__bJ=RT (B.29)
where a and b are constants for each type of gas.
For problems involving a perfect gas at relatively low temperatures, it is possible to

also assume a calorically perfect gas. A calorically perfect gas is defined as a
perfect gas with constant specific heats. In a calorically perfect gas, the specific
heat at constant volume c, the specific heat at constant pressure c,,, and the ratio
of specific heats all remain constant, and the following exists:
e = cT h=cT (B.30)
For fluids that can not be considered calorically perfect, the required state relations
can be found in the form of tables, charts, or curve fits.
The coefficients of viscosity and thermal conductivity can be related to the

thermodynamic variables using the kinetic theory. For example, Sutherland's
formulas for viscosity and thermal conductivity are given by
T 312 T312
p=C1 k=C3
where C, - C4 are constants for a given gas. The Prandtl number
CII
Pr =
k
is often used to determine the coefficient of thermal conductivity k once p is
known. This is possible because the ratio c r /Pr, which appears in the expression
below is approximately constant for most gases.
C
k= ---6u
Pr
304
A ppendix C
Program Listing
C.1 Hyperbolic Tangent Distribution
* SUBROUTINE HYPERBOLIC TANGENT *

* DECEMBER1996 *
* This routine MUST be called ONLY when the argument BETA is LESS than *
* one. In this case the hyperbolic functions are used. See reference *
* Thompson *
* ARGUMENTS: *
* NSUBDIV INTEGER*4. On entry specifies the number of subdivisions *
* which is one less than the number of nodes. Unchanged on *
* exit. *
* ALPHA *
* BETA = REAL*8 . On entry specify the parameters defined by eqs *
* (A.6) and (A.7) respectively. Unchanged on exit. *
* Note that in the cases ICASE=2 or ICASB=3, ALPHA is not *
* used and can have any REAL*8 value *
* ICASE = INTEGER*4 . On entry has one of the following values and *
* corresdponding meanings: *
* 1 (one) if both initial and final division are specified. *
* See equationS (A.9) and (A.lO). *
* 2 if only the initial division is specified. See equation *
* (A.13) and finally *
* 3 if only the final division is specified. See equation *
* (A.14). *
* IERROR = INTEGER*4. On exit has one of the following values *
* 0 (zero) = no error in subroutine *
* 1 (one) = BETA passed to routine is greater or equal to 1 *
* or ICASE has a value different than 1, 2 or 3 *
* 2 Denominator in N-R method to small. *
* 3 No convergence achieved while solving for delta *
SUBROUTINE HYPERB(NSUBDIV,BETA,ALPHA,
A
ICASE,IERROR)
IMPLICIT DOUBLE PRECISION (D)
C IMPLICIT REAL*4(AH 2OZ), INTEGER*4(IN)
PARAMETER(MAXITERP=1000,EFZEROP=1 .OD- 1 8,TOLP=1 .OD-03,
AIDPHIP1000, ONE 1 .OD+00)
COMMON /ARRAY/ SXI(IDPHIP)
*
WRITE(*,*)' SUBROUTINE HYPERBOLIC TANGENT'
IERROR=0
IF(ICASE.LT.l OR. ICASE.GT.3) THEN
Appendb C Program Listing
IERROR= 1
RETURN
ENDIF
IF(BETA.GE.ONE) THEN
IERROR= 1
RETURN
ENDIF
* Solution of eq (A.8) with respect to delta, using Newton-Raphson

* method. Original value obtained by expanding hyperbolic sine.
* Correction is the -(Function/Derivative of Function)
ITER=O
CORRECTION=1 .OD+1O
DELTA=DSQRT(6.OD+OO*(ONEBETA)/BETA)
DO WHILE(CORRECTION.GT.TOLP.AND.ITER.LE.MAXITERP)
DERIV=BETA*DCOSH(DELTA)ONE
IF(DAB S(DERIV).LE.EFZEROP)THEN
IERROR=2
RETURN
ENDJIF
CORRECTION=(BETA*DSINH(DELTA)DELTA)fDERIV
DELTA=DELTA+CORRECTION
ITER=ITER+ 1
END DO
IF(ITER.GE.MAXITERP) THEN
IERROR=3
RETURN
ENDIF
SUBDIV=FLOAT(NSUBDIV)
PARONOM=DTANH(O.5D+OO*DELTA)
DO KSI=O,NSUBDIV
RATIO=FLOAT(KSI)/SUBDIV
IF(ICASE.EQ. 1) THEN
UKSI=DTANH(DELTA*(RATIOO.5D+OO))
UKSI=O.5D+OO*(ONE+UKSI/PARONOM)
SKSI=UKSJJ(ALPHA+(ONEALPHA)*UKSI)
ELSE IF(ICASE.EQ.2) THEN
SKSI=DTANH(O.5D+OO*DELTA*(RATIOONE))
S KSI=ONE+SKSI/PARONOM
ELSE
SKSI=DTANH(O.5D+OO*DELTA*RATIO)
SKSI=SKSI/PARONOM
ENDIF
* Array starts from one and not zero. Dimensions
* are 1 to (NSUBDIV+1) and not 0 to NSUBDIV
SXI(KSI+1)=SKSI
END DO
RETURN
END
306
AppendLv C Program Listing
C.2 Trigonometric Tangent Distribution
* SUBROUTINE TRIGONOMERTRIC TANGENT *

* DECEMBER 1996 *
* This routine MUST be called ONLY when the argument BETA is GREATTER *
* than one. In this case the hyperbolic functions are used. See ref. *
* Thompson *
* ARGUMENTS: *
* NSUBDIV = INTEGER*4. On entry specifies the number of subdivisions *
* which is one less than the number of nodes. Unchanged on *
* exit. *
* ALPHA *
* BETA = REAL*8 . On entry specify the parameters defined by eqs *
* (A.6) and (A.7) respectively. Unchanged on exit. *
* Note that in the cases ICASE=2 or ICASE=3, ALPHA is not *
* used and can have any REAL*8 value *
* ICASE = INTEGER*4 . On entry has one of the following values and *
* corresponding meanings: *
* 1 (one) if both initial and final division are specified. *
* See equations (A.9) and (A.10). *
* 2 if only the initial division is specified. See equation *
* (A.13) and finally *
* 3 if only the final division is specified. See equation *
* (A.14). *
* TERROR = INTEGER*4 . On exit has one of the following values *
* 0 (zero) = no error in subroutine *
* 1 (one) = BETA passed to routine is less or equal to 1 *
* or ICASE has a value different than 1, 2 or 3 *
* 2 Denominator in N-R method to small. *
* 3 No convergence achieved while solving for delta *
************************************************************************
SUBROUTINE TRIGON(NSUBDIV,BETA,ALPHA,
A ICASE,IERROR)
PARAMETER(MAXITERP= I 000,EFZEROP= I .OD- 1 8,TOLP= I .OD-03,
A IDPHIp l000,ONE=1 .OD+00)
COMMON /ARRAY/ SXI(IDPHIP)
WRITE(* , *)' SUBROUTINE TRIGONOMETRIC TANGENT'

IERROR=0
IF(ICASE.LT.1 .OR. ICASE.GT .3) THEN
IERROR=1
RETURN
ENDIF
IF(BETA.LT.ONE) THEN
IERROR=1
RETURN
ENDIF
* Solution of eq (A.8) with respect to delta, using Newton-Raphson

* method. Original value obtained by expanding hyperbolic sine.
* Correction is the -(Function/Derivative of Function)
ITER=0
CORRECTION= 1 .OD+ 10
307
A ppendi)c' C Program Listing
DELTA=DSQRT(6.OD+OO*(BETAONE)fBETA)
DO WHILE(CORRECTION.GT.TOLP.AND.ITER.LE.MAXITERP)
DERIV=BETA*DCOS(DELTA)ONE
IF(DAB S(DERIV) .LE.EFZEROP)THEN
IERROR=2
RETURN
ENDIF
CORRECTION=(BETA*DSIN(DELTA)DELTA)IDERIV
DELTA=DELTA+CORRECTION
ITER=ITER+1
END DO
IF(ITER.GE.MAXITERP) ThEN
IERROR=3
RETURN
ENDIF
SUBDIV=FLOAT(NSUBDIV)
PARONOM=DTAN(O.5D+OO*DELTA)
DO KSIO,NSUBDW
RATIO=FLOAT(KSI)/SUBDIV
IF(ICASE.EQ. 1) THEN
UKSI=DTAN(DELTA*(RATIOO.5D+OO))
UKSI=O.5D+OO*(ONE+UKSI/PARONOM)
SKSIUKSJ/(ALPHA,(ONEALPHA)*UKSI)
ELSE IF(ICASE.EQ.2) THEN
SKSI=DTAN(O.5D+OO*DELTA*(RATIOONE))
SKSI=ONE+SKSJJPARONOM
ELSE
SKSI=DTAN(O.5D+OO*DELTA*RATIO)
SKSI=SKSI/PARONOM
ENDIF
Array starts from one and not zero. Dimensions
are 1 to (NSUBDIV+1) and not 0 to NSUBDIV
SXI(KSI+1)=SKSI
END DO
RETURN
END
308
AppendA C Program Listing
C.3 Transfinite Interpolation
* SUBROUTINE PREPARE FOR TRANSFINITE INTERPOLATION *

* MARCH 1996 *
* This routine MUST be called ONLY when the definition of the edges is known and *
* the distribution of the vertices along the edges completed. (SEE common block) *
* ARGUMENTS: *
* IMAX = INTEGER*4 . On entry specifies the number of subdivisions *
* which is one less than the number of nodes on the I-Direction. *
* Unchanged on exit. *
* JMAX = INTEGER*4 . On entry specifies the number of subdivisions *
* which is one less than the number of nodes on the J-Direction. *
* XXX = 2-Dimensional Array. *
* On exit has the X-values of the boundaries of the BLOCK *
* YYY = 2-Dimensional Array. *
* On exit has the Y-values of the boundaries of the BLOCK *
C
SUBROUTINE PREPARE_FOR_TRANS (XXX,YYY,IMAX,JMAX)
PARAMETER (NI:= 1 00,NJ= 100)
C
C
COMMON /ARRAY/ X_EDGE 1 (NI),Y_EDGE1 (NI),X_EDGE2(NI),Y_EDGE2(NI),
A
X_EDGE3(NI),Y_EDGE3(NI),X_EDGE4(NI),Y_EDGE4(NI)
C
DIMENSION XXX(NI,NJ)
DIMENSION YYY(NI,NJ)
C
IMIN= 1
JMIN=1
C
DO 230 J=JMIN,JMAX
XXX(IMIN,J)=X_EDGE4(J)
YYY(IMIN,J)=Y_EDGE4(J)
230 CONTINUE
DO 235 J=JMIN,JMAX
XXX(IMAX,J)=X_EDGE2(J)
YYY(IMAX,J)=Y_EDGE2(J)
235 CONTINUE
DO 240 I=IMIN,IMAX
XXX(I,JMIN)=X_EDGE 1(I)
YYY(I,JMIN)=Y_EDGE 1(I)
240 CONTINUE
DO 245 I=IMIN,IMAX
XXX(I,JMAX)=X_EDGE3(I)
YYY(I,JMAX)=YEDGE3(I)
245 CONTINUE
RETURN
END
309
Appendix C Program Listing
* SUBROUTINE TRANSFINITE INTERPOLATION *

* MARCH1996 *
* This routine CAN be called ONLY from the subroutine PREPARE_FOR_TRANS *
* ARGUMENTS: *
* IMAX = INTEGER*4. On entry specifies the number of subdivisions *
* which is one less than the number of nodes on the I-Direction. *
* JMAX = INTEGER*4 . On entry specifies the number of subdivisions *
* which is one less than the number of nodes on the i-Direction. *
* XX = 2-Dimensional Array. *
* On entry has only the X-values of the vertices along the *
* boundaries of the BLOCK *
* On exit has all the X-values of the vertices of the BLOCK *
* YY = 2-Dimensional Array. *
* On entry has only the Y-values of the vertices along the *
* boundaries of the BLOCK *
* On exit has all the Y-values of the vertices of the BLOCK *
C
SUBROUTINE TRANS (XX,YY,IMAX,JMAX)
PARAMETER (NI= 1 00,NJ= 100)
DIMENSION xx(NI,NJ),yy(NI,NJ)
DIMENSION xxl(NI,NJ),yy 1(NI,NJ),xx2(NI,NJ),yy2(NI,NJ)
400 DO 410 I=1,IMAX

DO 410 J=1,JMAX
RI 1=FLOAT(I- 1 )/FLOAT(IMAX- 1)
RI2=FLOAT(IMAX-I)/FLOAT(IMAX- 1)
xxi (I,J)=RI1 *)Q((JMAX,J)+R12*)Q(( 1 ,J)
yy 1 (I,J)=RIi *yy(4AX,J)+RJ2*yy( 1 ,J)
410 CONTINUE
DO 420 I=1,IMAX
DO 420 J=1,JMAX
RJ1=FLOAT(J- 1 )IFLOAT(JMAX- 1)
RJ2=FLOAT(JMAX-J)IFLOAT(JMAX- 1)
XX2(I,J)=RJ1 * (XX(I,JMAX)-XX1 (I,JMAX))+RJ2*(XX(I, 1 )-XX 1(1,1))
YY2(I,J)=RJ 1 *(YY(J,JyJYy1 (I,JMAX))+RJ2*(YY(I, 1 )-YY 1(1,1))
420 CONTINUE
DO 430 I=1,IMAX
DO 429 J=i,JAX
XX(I,J)=XX1 (I,J)+XX2(I,J)
YY(I,J)=YY1 (I,J)+YY2(I,J)
429 CONTINUE
430 CONTINUE
RETURN
END
310
C.4 Geometry
('GEOMETRY-WORKING VOLUMES-SEALING LINES- PEPRENDICULAR ');

('CIRCULAR ARCS OF SCROLL MACHINES-WITH RESPECT TO CRANK ANGLE');
fidxO=fopen('xO' ,'wt+');
fidyO=fopen('yO' ,' wt+');
fidxl=fopen('xl' ,'wt+');
fidyl=fopen('yl','wt+');
fidx2=fopen(' x2','wt+');
fidy2=fopen('y2' ,'wt+');
fidx3=fopefl('x3' ,'wt+');
fidy3=fopen('y3' ,'wt+');
fidx4=fopen(' x4' ,' wt+');
fidy4=fopen('y4' ,' wt+');
fidx5=fopen(' x5' ,' wt+');
fidy5=fopen(' y5' ,' wt+');
fidXk=fopen(Xk' ,'wt+');
fidYk=fopen('Yk' ,' wt+');
fidXk1zfopen('Xk1','wt+');
fidYkl=fopen('Ykl','wt+');
ZOOM ON
rb=O.75;
theta=27.O*piJ9;
a=theta;
c=O.026;
g=O.3;
1(pi_g).*rb;
1=1-c;
Pos=sqrt(((pi-g)."2./4)- 1);
Pis=pi+sqrt(((pig). A 2J4) 1 )-g;
M=5.5;
Pof=M*pi;
Pif=(M*pi)g;
if theta >2.O*pi
C SUCTION');
('starting and finishing angles for the working volumes');

fi fz40*pi+pii2.O;
f4f=f 1 fi-pi-g;
fi s=f 1 f+theta;
f4s=f4f+theta;
f4f=(M.*pi)g;
f4s=f4f+theta;
fi s=f4s-pi±g;
fi f=(M- I )*p;
('starling and finishing angles for the outer spirals');

Pos=fI s;
Pof=fl f,,
('starting and finishing angles for the inner spirals');
P1s=f45;
Pif=f4f;
311
('starting and finishing angles for the mid planes');

Posm=f1s;
Pofm=(4.5 .*p i+f 10/2;
end
if theta <_2.O*pi
('COMPRESSION-DISCHARGE');
Xos=rb. *(cos(Pos)+POS*sin(Pos));
Yos=rb.*(sin(Pos)Pos.*cos(Pos));
Xis=rb. *(cos( p is+g)+p is. *sjn(Pjs+g));
Yis=rb.*(sin(Pis+g)Pis. *cos(pjs+g));
Yoo 1=(((XosXis).A2)((2.*(XosXis).*tan(Pos).*Yis))+(Yos).A2.(Yis).A2);
Yoo2=2.*(YosYis(XosXis). *tan(po5));
Yoo=Yoo 1/Yoo2;
Xoo=Xis-tan(Pos). *(yooyis);
r=((Xis-Xoo).'2.+(Yis-Yoo)/2)!O.5;
Xfc=Xis:O.O1 :Xos;
Yoc=Yfc1.*sin(a);
XocXfc1.*cos(a);
end
('OUTER SPIRALS');
p=Pos:O. 1 :Pof;
('p=O:O.Ol:Pof');
xO=rb.*(cos(p)+p. *sin(p));
yO=rb.*(sin(p)p.*cos(p));
x1=rb.*(cos(p)^p.*sin(p))1.*cos(a);
y 1=rb.*(sin(p)p.*cos(p))1.*sjn(a);
('INNER SPIRALS');
p=Pis:O. 1 :Pif;
('p=O:O.Ol :Pif);
x2=rb.*(cos(p+g)+p.*sin(p+g));
y2=rb.*(sin(p+g)p.*cos(p+g));
x3_rb.*(cos(p+g)+p. *sin(p+g))(l. *cos(a));
y3=rb.*(sin(p+g)p. *cos(p^g))(1.*sin(a));
('MID PLANE 1 FIXED OUTER - ORBITING INNER');

p=Pos:O.1 :(M1).*pi;
('p=O.O:O.Ol :(M1).*pi;');
('THIS SECTION IS TO PLOT THE MID PLANE');

GPI=pi;
RGEN=rb;
GAMMA=g;
CLEARAN=c;
PHI=p;
THETA=a;
GPIGAMz(GPI-GAMMA)/2.O;
DCENTR(RGEN*(GPIGAMMA)CLEARAN);
COSPHI=cos(Pffl);
SINPHI=sin(PHI);
XMP=RGEN*(COSPHI+(PHI+QPIGAM). *SINP} DCENTR*cos(THETA)/2.O;
YMP=RGEN*(SJNPHI(PHI+GPIGAM).*COSPHI)DCENTR*sin(THETA)/2.O;
x4=XMP;
312
y4=YMP;
('THIS SECTION IS TO FIND THE ANGLE PHIFO');
('PEPPENDICULAR TO FIXED OUTER');
('CHOOSE A POINT ON THE MID PLANE');
p=7.O*piI9;
ppp=p;
PHI=p;
COSPHI=cos(PHI);
SINPHI=sin(PHI);
XMP 1 =RGEN*(COSPHI+(PHI+GPIGAM). *SINPHI) DCENTR*cos(THETA)/2.O;
YMP 1=RGEN*(SINPHI(PHI+GPIGAM). *COSPHI) DCENTR*sin(THETA)/2.0;
xmm=XMP1;
ymm=YMP1;
('!--->3');
RMP 1=sqrt(XMP 1 *XMP 1 +YMP 1 *YMP1);
COS_OMEGA=YMP1fRMP1;
if COS_OMEGA>0.999999
COS_OMEGA =1.0;
end
if COS_OMEGA<-0.999999
COS_OMEGA=- 1.0;
end
OMEGA=acos(COS_OMEGA);
if XMP1<0.0
OMEGA=2.0*GPIOMEGA;
end
('!--->4');
SINALPHA=RGENIRMP 1;
if SINALPHA>0.999999
SINALPHA =1.0;
end
if SINALPHA<-0.999999
SINALPHA=- 1.0;
end
ALPHA=asin(SINALPHA);
PHI_FO=GPI-(OMEGA+ALPHA);
while PHI_FO>0.0
PHI_FO=PHI_FO2.0*GPI;
end
while PHI_FO<0.0
PHI_FO=PHI_FO+2.0*GPI;
end
SINPHI_FO=sin(PHI_FO);
COSPHI_FO=cos(PHI_FO);
XFO=RGEN*(COSPHI_FO + PHI_FO*SINPHI_FO);
YFO=RGEN*(SINPHI_FO - PHI_FO*COSPHI_FO);
RFO=sqrt(XFO*XFO + YFO*YFO);
('!--->6');
while RFO<RMP1
PHI_FO=PHI_FO + 2.OD+00*GPI;
SINPHI_FOsin(PHIFO);
COSPHI_FOcos(PHI_FO);
313
App endix C Program Listing
YFO=:RGEN*(SINPHI_FO - PHI_FO*COSPHI_FO);
RFO=sqrt(XFO*XFO + YFO*YFO);
end
('IF(RFO NE. RMP1)i dont know how to do it in matlab');
PHI_FO=PHI_FO - 2.0*GPI;
phi=PHI_FO;
('!') ;
YFO=RGEN*(SINPHI_FO - PHLFO*COSPHI_FO);
xstep=(xmm-XFO);
ystep=(ymm-YFO);
xplot=XFO:xstep: xmm;
yplot=YFO:ystep:ymm;
('THIS SECTION IS TO FIND THE ANGLE PHI_Ol');

('PEPPENDICULAR TO ORBITING INNER');
('!--->7');
XMP 1 =XMP 1 +DCENTR*cos(THETA);

YMP 1 =YMP 1 +DCENTR*sin(THETA);
('--->3');
RMP1=sqrt(XMP1 *XMp1+yMpl *YMP1);
COS_OMEGA=YMP1/RMPI;
if COS_OMEGA>0.999999
COS_OMEGA =1.0;
end
if COS_OMEGA<-0.999999
COS_OMEGA=- 1.0;
end
OMEGA=acos(COS_OMEGA);
if XMP1<0.0
OMEGA=2.0*GPIOMEGA;
end
('!--->4');
SINALPHA=RGEN/RMP 1;
if SINALPHA>0.999999
SINALPHA =1.0;
end
if SINALPHA<-0.999999
SINALPHA=- 1.0;
end
ALPHA=asin(SINALPHA);
('!--->8');
PHI_OI=-(OMEGA+ALPFIA+GAMMA);
while PHI_OI>0.0
PHI_OI=PHI_OI2.0*GPI;
end
while PHI_OI<0.0
PHI_OI=PHI_OI+2.0*GPI;
end
SINPHI_OIG=sin(PHI_OI+GAMMA);
CO5PHI_OIGcos(PHI_OI+GAMMA);
314
Appendi)' C Program Listing
XOI=RGEN*(CO5PHJ_OJG + PHI_OI*SINPHI_OIG);
YOI=RGEN*(SINPHI_OJG - PHI_OI*COSPHI_OIG);
ROI=sqrt(XOI*XOI+YOI*YOI);
while ROI< RMP1

PHI_OI=PHI_OI + 2.0*GPI;
SINFHI_OIG=sin(PHI_OJ+GAMMA);
COSPHI_OIG=cos(PHJ_OJ+GAMMA);
XOI=RGEN*(COS pHI_OJG+pHJ OI*SINPHI 010);
YOI=RGEN*(SINPIJJ_OJGpHJ_OJ*CO5pHJ_OJG);
ROI=sqrt(XOI*XOI+YOI*YOI);
end
PHI_OI=PHI_OI;
phi 1=PHI_OI;
XOI=RGEN*(COSpHJ_OJG+PHIOJ*SINpHIOJG)DCENTR*cos(THETA);
YOI=RGEN*(5INPHI_OJGpHI_OJ*COSpFJJ_OJG)DCENTR*sjn(THETA);
xstep 1=(xmm-XOI);
ystep 1 =(ymm-YOI);
xplotl=XOI:xstepl :xmm;
yplotl=YOI:ystepl :ymm;
('SUBROUTINE THREE_POINT_ARC');
xl 1=XOI;
Yl 1=YOI;
X22=xmm;
Y22=ymm;
X33=XFO;
Y33=YFO;
IVERTIC12_YESP=1;
IVERTIC 1 2_NOP=2;
IVERTIC 1 3_YESP=1O;
IVERTIC 1 3_NOP=20;
('!--->3');
X12=O.5D+00*(X1 1+X22);
Y12=O.5D+0O*(Y1 1+Y22);
X13=0.5D+O0*(X1 1+X33);
Y13=O.5D+OO*(Y1 1+Y33);
('--->4');
IAUX=IVERTIC 1 2_YESP;
AUX=abs(Y1 1-Y22);
if AUX>0.0
IAUX=IVERTIC 1 2_NOP;
end
IAUX1=IVERTIC 1 3_YESP;
AUX=abs(Y1 1-Y33);
if AUX> 0.0
IAUX1=IVERTIC 1 3_NOP;
end
IAUX=IAUX+IAUX 1;
('!--->5');
AUX=IVERTIC 1 3_NOP+IVERTIC 12_NOP;
gnk 1=IVERTIC 1 3_NOP-f-IVERTIC 1 2_YESP;
gnk2=IVERTIC I 3_YES P+IVERTIC 1 2_NOP;
if IAUX == AUX;
GRAD 12 =-(X22-X1 1)/(Y22-Y1 1);
315
A ppendiA ' C Program Listing
C12 =Y12GRAD12*X12;
GRAD 13 =-(X33-X1 1)/(Y33-Y1 1);
C13 =Y13GRAD13*X13;
AUX=abs(GRAD 12-GRAD 13);
if AUX<=0.0
ANGLE=GPI;
('RETURN')
end
X0=-(C12-C1 3)/(GRAD 12-GRAD 13);
Y0=-(C 12*GRAD 13-C 13*GRAD1 2)/(GRAD 12-GRAD 13);
elseif IAUX==gnkl
GRAD13 -(X33-X11)/(Y33-Y11);
C13 =Yl3GRAD13*X13;
X0=X12;
YO=GRAD13*X0+C13;
elseif IAUX=gnk2
GRAD12 =-(X22-X1 l)/(Y22-Y1 1);
C12 =Y12GRAD12*X12;
X0=X13;
YO=GRAD12*X0+C12;
else
('All points on straight AND horizontal line')
ANGLEGPI;
end
X10=X1 l-X0;
Y1OYI 1-YO;
X30=X33-X0;
Y30=Y33-Y0;
RADIUS2X1O*XlO+YlO*Y1O;
RADIUS=sqrt(RADIUS2);
ANGLE=acos((X1 O*X30+Y1 O*Y30)/RADIUS2);
X12=Xl1-X22;
Y12=Y1 1-Y22;
X32=X33-X22;
Y32=Y33-Y22;
gnk4=X 1 2*Y32X32*Y 12;
if gnk4> 0.0
ANGLE=-ANGLE;
end
GONIA_A=asin(abs(Y 10)/RADIUS);
XIOY1O=X10*Y10;
if X1OYI0<0.0
GONIA_A = -GONIAA;
end
if X1O<0.0
GONIA_A =GPI+GONIA_A;
end
if GONTA_A<0.0
GOMA_A =2.OD+00*GPI+GONIA_A;
end
gpi2=GPIJ2.0;
if GONTA_A==gpi2
if Y10<0.0
316
AppendA' C Program Listing
GONJA_AGPJ+GONIAA;
end
end
gnkangle=ANGLE;
gnkgonia_a=GONIAA;
aktina=RADIUS;
step=ANGLE/ 10.0;
ANGLECGQNIAA+ANGLE;
de1taphi=GONIAA•step:ANGLEC
Xk=RADIUS*cos(deltaphj)+XO;
Yk=RADIUS *sin(deltaphi)+YO;
Xkl =XkDCENTR*cos(THETA);
Yki =YkDCENTR*sin(THETA);
('MID PLANE 2 ORBITING OUYFER FIXED INNER');

p=Pos:0.O1:(M1).*pi;
('p=O.O:O.Ol :(M_1).*pi;');
('x5=(b.*(cos(p+pi)+(p+pig). *sin(p+pi)))+(.rb. *(cos(p)+(p).*sin(p))..(1 *cos(a))))/2;');
x5=x4DCENTR*cos(THETA);
y5=y4DCENTR*sin(THETA);
p=ppp;
xmm=((rb. *(cos(p+pi)+(p+pi..g). *sin(p^pi)))+(..rb. *(cos(p)+(p) *sjn(p)) (I *cos(a))))/2;
('PEPPENDICULAR TO ORBITING OUTER');

p=phi;
xm55=rb . *(cos(p)+p . *sin(p))1*cos(a);
ym55=rh.*(sin(p)p.*cos(p))1.*sjn(a);
xstep5=(xmm-xm55);
ystep5=(ymm-ym55);
xplot5=xm55 :xstep5 :xmm;
yplot5=ym55 :ystep5 :ymm;
('PEPPENDICULAR TO FIXED INNER');

p=phil;
xm66=rb.*(cos(p+g)+p.*sin(p+g));
ym66=rb. *(sin(p+g)p. *cos(p+g));
xstep6=(xmm-xm66);
ystep6=(ymm-ym66);
xplot6=xm66 :xstep6:xmm;
yplot6=ym66:ystep6:ymm;
Yoo=0;
Xoo0;
('FIXED CIRCLE');
Xfcl 1=-rb:0.01:rb;
YfcI 1=((rb!'2.-(Xfcl 1-Xoo)."2)/'O.5)+Yoo;
Yfc22=((rb A2 . (Xfc 11 -Xoo).'2)/'O.5)+ Yoo;
('ORBITING CIRCLE');
317
Xocl l=-Xfcl ll.*cos(a);

Yoci l=-Yfcl ll.*sin(a);
Yoc22=-Yfc22-l. *sin(a);
('GRANK CIRCLE');
r=1;
Xgc=-r:O.O1 :r+O.0054;
Ygc=((r.A2.(XgcXoo).A2).AO.5)+Yoo;
('saveabcde xO yO xl yl');
('save abcde xO yO xl yl -ascii -double -tabs');
('IF YOU WANT TO PRINT INTO FILES THE COORDINATES OF THE');

('PERPENDICULAR STRAIGHT LINES PUT PRINTLINES=1 OTHERWISE 0');
('NOTE ALSO THAT THE COORIDINATES WILL BE STORED IN THE FILES');
('xO,yO and xl,yl RESPECTIVELY');
PRINTLINES=2;
if PRINTLII'TES==1
fprintf(fidxO,' % 10.4f\n' ,xplot);
fprintf(fidyo,' % 10.4f\n',yplot);
fprintf(fidx0,' % 1 0.4ñn' ,xplotl);
fprintf(fidy0,' % 1 0.4f\n' ,yplotl);
fprintf(fidxl ,'% 1O.4ñn' ,xplot5);
fprintf(fidy 1,' % 1O.4f\n' ,yplot5);
fprintf(fidx I,' % 1 0.4f\n' ,xplot6);
fprintf(fidy 1,' % 1 O.4f\n' ,yplot6);
end
fprintf(fidxO,' % l0.4f\n' ,xO);

fprintf(fidy0,' % 1 0.4f\n' ,yO);
fprintf(fidx 1,' % 1 0.4f\n' ,x 1);
fprintf(fidy 1,' % 1 0.4ñn' ,yl);
fprintf(fidx2,' % 1 0.4f\n' ,x2);

fprintf(fidy2,' % 1 O.4f\n' ,y2);
fprintf(fidx2,' % 1 O.41\n' ,Xfc);
fprintf(fidy2,' % 1 O.4ftn' ,Yfc);
fprintf(fidx3,' % 1 0.4ñn' ,x3);

fprintf(fidy3,' % 10.4f\n' ,y3);
fprintf(fidx3,' % 1 O.4f\n' ,Xoc);
fprintf(fidy3,' % 1 O.4ñn' ,Yoc);
fprintf(fidx4,' % 10.4f\n' ,x4);

fprintf(fidy4,' % 1 O.41\n' ,y4);
fprintf(fidx5,' %1O.4f\n',x5);
fprintf(fidy5,' % I O.4f\n' ,y5);
fprintf(fidXk,' % 10.4ñn' ,Xk);
fprintf(fidYk,' % 10.4f\n' ,Yk);
fprintf(fidXkl ,' % 1O.4f\n' ,Xkl);
fprintf(fidYkl ,' % 1O.4f\n' ,Ykl);
fclose(fidxo);
fclose(fidyo);
fclose(fidxl);
fclose(fidy 1);
fclose(fidx2);
318
fclose(fidy2);
fclose(fidx3);
fclose(fidy3);
fclose(fidx4);
fclose(fidy4);
fclose(fidx5);
fclose(fidy5);
fclose(fidXk);
fclose(fidYk);
fclose(fidXkl);
fclose(fidYkl);
s= 1;
if s>2
('WORKING VOLUME SPIRALS, GENERATING CIRCLES, CRANK CIRCLE');
plot(xO,yO,x l,yl ,x2,y2,x3,y3 ,Xfc,Yfc,Xoc,Yoc,Xfc 11 ,Yfc 11 ,Xfc 11 ,Yfc22,Xoc 11 ,Yoc 11 ,Xoc 11 ,Y
oc22,Xgc,Ygc,Xgc,Ygc 1)
axis([-13,13,-13,13])
axis(' equal')
end
s=3;
if s>2
('WORKING VOLUME SPIRALS, MIDPLANES PERPENDICULAR STRAIGHT LINES');
plot(xO,yO,x l,yl ,x2,y2,x3,y3,x4,y4,x5,y5,Xfc,Yfc,Xoc,Yoc,xplot,yplot,xplotl ,yplotl ,xplot5,yplot5,
xplot6,yplot6)
axis([-13,13,-13,13])
axis(' equal')
end
s=3;
if s>2
('WORKING VOLUME SPIRALS, MIDPLANES PERPENDICULAR CIRCULAR ARCS');
plot(xO,yO,x I,yl ,x2,y2,x3,y3 ,x4,y4,x5,y5,Xfc,Yfc,Xoc,Yoc,Xk,Yk,Xk I ,Yk 1)
axis([-13,13,-1 3,13])
axis('equal')
end
s= 1;
if s>2
('WORKING VOLUME SPIRALS, GENERATING CIRCLES,');
('MIDPLANES PERPENDICULAR STRAIGHT LINES');
plot(xO,yO,x3,y3,x l,yl ,x2,y2,x4,y4,x5 ,y5 ,Xfc 11 ,Yfc II ,Xfc 11 ,Yfc22,Xoc Ii ,Yoc 11 ,Xoc II ,Yoc22,x
plot,yplot,xplotl ,yplot 1 ,xplot5 ,yplot5 ,xplot6,yplot6)
axis([-13,13,-13, 13])
axis('equal')
end
if theta >2.O*pi
('COMPRESSION - DISCHARGE');
('SPIRALS, GENERATING CIRCLES,');
('MIDPLANES PERPENDICULAR STRAIGHT LINES');
plot(xO,yO,x3,y3,xl ,yl ,x2,y2,x4,y4,x5,y5,Xfc,Yfc,Xoc,Yoc,Xfc 11 ,Yfc 11 ,Xfcl 1 ,Yfc22,Xoc 11 ,Yoc

11 ,Xoc 11 ,Yoc22,xplot,yplot,xplotl ,yplotl ,xplot5 ,yplot5,xplot6,yplot6)
axis([- 14,14,- 14,14])
axis(' equal')
end
319
A ppendA ' C Program Listing
C5 Command Language (CFX4)
/* SCROLL EXPANDRER *1
/* TRANSIENT GRID COMPRESSIBLE, TURBULENT FLOW */
/**********************************************************/
>>CFXF3D
#CALC
NII=60;
NJ1=20;
NK 1=6;
N12=20;
NJ2=20;
NK2=6;
N13=60;
NJ3=20;
NK3=6;
N14=20;
NJ4=20;
NK4=6;
NI5=10;
NJ5=10;
NK5=20;
#ENDCALC
>>SET LIMITS
TOTAL INTEGER WORK SPACE 5000000
TOTAL REAL WORK SPACE 15000000
MAXIMUM NUMBER OF BLOCKS 5
MAXIMUM NUMBER OF PATCHES 50
MAXIMUM NUMBER OF INTER BLOCK BOUNDARIES 30
END
>>OPTIONS
THREE DIMENSIONS
TURBULENT FLOW
HEAT TRANSFER
COMPRESSIBLE FLOW
TRANSIENT FLOW
TRANSIENT GRID
END
>>USER FORTRAN
USRGRD
USRTRN
END
>>MODEL TOPOLOGY
/*****************/
/* CREATE BLOCKS */
/*****************I
>>CREATE BLOCK
BLOCK NAME 'BLOCK-i'
NUMBER OF I CELLS #NI1
NUMBER OF J CELLS #NJi
NUMBER OF K CELLS #NKi
END
>>CREATE BLOCK
BLOCK NAME 'BLOCK-2'
NUMBER OF I CELLS #NI2
320
A ppendi)c C Program Listing
NUMBER OF J CELLS #NJ2

NUMBER OF K CELLS #NK2
END
>>CREATE BLOCK
NUMBER OF I CELLS #N13
END
>>CREATE BLOCK
END
>>CREATE BLOCK
END
/*****************/
1* CREATE PATCH */
/*****************/
>>CREATE PATCH
PATCH NAME 'PATCH-i. 1'
PATCH TYPE 'INTER BLOCK BOUNDARY'
LOW I
END
>>CREATE PATCH
PATCH NAME 'PATCH-1.2'
PATCH TYPE 'WALL'
HIGH J
END
>>CREATE PATCH
PATCH NAME 'PATCH-i.3'
PATCH TYPE 'WALL'
HIGH I
END
>>CREATE PATCH
BLOCK NAME 'BLOCK-I'
PATCH TYPE 'WALL'
LOW J
END
>>CREATE PATCH
PATCH NAME 'PATCH-i.5'
PATCH TYPE 'WALL'
LOWK
END
>>CREATE PATCH
321
A ppendbc' C Program Listing
PATCH TYPE 'WALL'

HIGH K
END
>>CREATE PATCH
PATCH NAME 'PATCH-2.l'
PATCH TYPE 'WALL'
LOW J
END
>>CREATE PATCH
PATCH TYPE 'WALL'
HIGH I
END
>>CREATE PATCH
HIGH J
END
>>CREATE PATCH
LOW I
END
>>CREATE PATCH
PA TCH NA ME 'PA TCH-2.5'
PATCH TYPE 'WALL'
LOW K
END
>>CREATE PATCH
PATCH TYPE 'WALL'
HIGH K
END
>>CREATE PATCH
LOW I
END
>>CREATE PATCH
PATCH TYPE 'WALL'
HIGH J
END
>>CREATE PATCH
PATCH TYPE 'WALL'
322
Appendi) C Program Listing
HIGH I
END
>>CREATE PATCH
PATCH TYPE 'WALL'
LOW J
END
>>CREATE PATCH
PATCH TYPE 'WALLs
LOWK
END
>>CREATE PATCH
PATCH TYPE 'WALL'
HIGH K
END
>>CREATE PATCH
PATCH NAME 'PATCH-4.I'
PATCH TYPE 'WALL'
LOW J
END
>>CREATE PATCH
PATCH TYPE 'WALL'
HIGH I
END
>>CREATE PATCH
HIGH J
END
>>CREATE PATCH
LOW I
END
>>CREATE PATCH
PATCH NAME 'PATCH-4.5 1'
PATCH TYPE 'WALL'
LOWK
PATCH LOCATION 1 5 1 20
END
>>CREATE PATCH
PATCH TYPE 'WALL'
LOWK
323
PATCH LOCATION 6 15 11 20 11
END
>>CREATE PATCH
PATCH TYPE 'WALL'
LOW K
END
>>CREATE PATCH
PATCH TYPE' INTER BLOCK BOUNDARY'
LOW K
END
>>CREATE PATCH
PATCH TYPE 'WALL'
HIGH K
END
>>CREATE PATCH
PATCH TYPE 'WALL'
LOW J
END
>>CREATE PATCH
PATCH TYPE 'WALL'
HIGH I
END
>>CREATE PATCH
PATCH TYPE 'WALL'
HIGH J
END
>>CREATE PATCH
PATCH TYPE 'WALL'
LOW I
END
>>CREATE PATCH
HIGH K
END
>>CREATE PATCH
PATCH NAME 'INLET'
PATCH TYPE 'PRESSURE BOUNDARY'
LOW K
324
END
I*****************/
1* GLUE PATCHES */
/*****************/
>>GLUE PATCHES
FIRST PATCH NAME 'PATCH-li'
SECOND PATCH NAME 'PATCH-2.3'
ORIENTATION CHANGE 'HIGH J' 'HIGH I' 'HIGH K'
END
>>GLUE PATCHES
FIRST PATCH NAME 'PATCH-2.4'
ORIENTATION CHANGE 'LOW I' 'LOW J' 'HIGH K'
END
>>GLUE PATCHES
ORIENTATION CHANGE 'HIGH J' 'HIGH I' 'HIGH K'
END
>>GLUE PATCHES
END
>>MODEL DATA
>>TITLE
PROBLEM TITLE 'SCROLL EXPANDER R134a'
END
>>PHYSICAL PROPERTIES
>>FLUID PARAMETERS
VISCOSITY 0.206E-4
END
>>COMPRESSIBILITY PARAMETERS
WEAKLY COMPRESSIBLE
UNIVERSAL GAS CONSTANT 8314.0
FLUID MOLECULAR WEIGHT 102.03
REFERENCE PRESSURE 9.489E+05
END
>>HEAT TRANSFER PARAMETERS
FLUID SPECIFIC HEAT 1020.0
THERMAL CONDUCTIVITY 0.0211
ENThALPY REFERENCE TEMPERATURE 288.0
END
>>TRANSIENT PARAMETERS
TIME STEPS 200*8.333E5
END
>>SOLVER DATA
>>PROGRAM CONTROL
MAXIMUM NUMBER OF ITERATIONS 100
OUTPUT MONITOR BLOCK 'BLOCK-4'
OUTPUT MONITOR POINT 1053
MASS SOURCE TOLERANCE 1.OE-5
END
>>DEFERRED CORRECTION
EPSILON START 101
325
EPSILON END 101

K START 101
KEND 101
END
>>CREATE GRID
>>MODEL BOUNDARY CONDITIONS

>>SET VARIABLES
#CALC
UIINL=40.0;
TEINL=0.002*UINL*UINL;
CH=0.02;
EPSINL=TEINL** 1 .5/(0.3*CH);
#ENDCALC
PATCH NAME 'INLET'
PRESSURE 9.489E+05
TEMPERATURE 310.0
END
>>OUTPUT OPTIONS
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 10
END
>>DTJMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 20
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 30
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 40
END
>>DJJMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
326
A ppendi)c C Program Listing
PRESSURE
GEOMETRY DATA
TIME STEP 50
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 60
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 70
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 80
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 90
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 100
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 110
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
327
A ppendix C Program L/stiig
GEOMETRY DATA
TIME STEP 120
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 130
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 140
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 150
END
>>DTJMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 160
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 170
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 180
END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
328
Appendbc' C Program Listing
TIME STEP 190

END
>>DUMP FILE OPTIONS
U VELOCITY
V VELOCITY
W VELOCITY
PRESSURE
GEOMETRY DATA
TIME STEP 200
END
>>DUMP FILE FORMAT

UNFORMATTED
END
>>STOP
329
References And Bibliography
Alderton J.H., Wilkes N.S. (1988): Some applications of New Finite Difference Schemes
for Fluid Flow Problems. Harwell Report AERE-R 13234
Anderson D.A., Tannehill J.C., Pletcher R.H. (1986): Computational Fluid Mechanics and
Heat Transfer. McGraw-Hill Book Company
Anonymous (1990a): STAR-CD Manual, Computational Dynamics, London, 1990

Anonymous (1990b): The PHOENICS Reference Manual. Report CHAM/TRIIOO.
CHAM Ltd, London 1990.
Anonymous (1990c): Harwell-FLOW3D, Release 2.3: User Manual, Harwell Laboratory,

1990
Anonymous (1991a): User guide to FEAT. Engineering Analysis Centre, Nuclear Electric
Plc, UK
Anonymous (1991b): FLOW3D, Release 2.4: User Manual. AEA md-Tech, Harwell-
Laboratory, Oxfordshire U.K., 1991
Anonymous (1992): FLOW3D, Release 3.1.2: User Manual. AEA md-Tech, Harwell-
Anonymous (1993): FLOW3D, Release 3.2: User Manual. AEA md-Tech, Harwell-
Anonymous (1994a): FLOW3D, Release 3.3: User Manual. AEA md-Tech, Harwell-
Anonymous (1994b): ENVIROMENT User Guide Release 1.5: User Manual. AEA md-
Tech, Harwell-Laboratory, Oxfordshire U.K., 1994
Babuska I., Aziz A.K. (1972): Survey lectures on the mathematical foundation of the
finite element method. In: Mathematical Formulation of the Finite Element Method. Ed.
A.K. Aziz. Academic Press, New York
Baker A.J. (1983): Finite Element Computational Fluid Mechanics. McGraw-Hill
Baker A.J. (1984): Finite Element Penalty Algorithm for the Parabolic Navier-Stokes
Equations for Turbulent Three Dimensional Flows. Computer Methods in Applied
Mechanics and Engineering Vol. 46 pp. 277-93
Baliga B.R., Patankar S.V. (1980): A New Finite Element Formulation for Covection-
References and Bibliography
Diffusion Problems. Num. Heat Transfer Vol.3 pp. 393-409
Baliga B.R., Patankar S.V. (1993): A Control Volume Finite Element Method for Two
Dimensional Fluid Flow and Heat Transfer. Num. Heat Transfer Vol.6 pp. 245-61
Brooks A.N., Hughes T.S.R. (1982): Streamline UpwindlPetrov-Galerkin Formulations

for Convection Dominated Flows with Particular Emphasis on the Incompressible Navier
Stokes Equations. Comp. Mech. Appi. Mech. Eng. Vol.32 pp. 199-259
Burns A.D., Wilkes N.S. (1987): A Finite Difference Method for the Computation of
Fluid Flows in Complex Three-Dimensional Geometries. UKAEA Report AERE-R 12342
Burns A.D., Wilkes N.S., Jones I.P., Kightley J.R. (1996): FLOW3D: Body-Fitted Co-
ordinates. AERE-R 12262
O Bush W.B., Cailat J.L., Seibel M.S. (1986): Dimensional Optimization of Scroll
Compressors. mt. Compressor Eng. Conf. Pardue Univ.
Canuto C., Hussaini M.Y., Quarteroni A., Zang T.A. (1987): Spectral Methods in Fluid-
Dynamics. Berlin. Springer-Verlag
Cebeci T., Hirsh R.S., Keller H.B., Williams P.G (1981): Studies of Numerical Methods
for the Plane Navier-Stokes Equations. Computer Methods in Applied Mechanics and
Engineering Vol.27 pp. 13-44
Chen C.J., Chen H.0 (1984): Finite analytic numerical method for unsteady two-
dimensional Navier-Stokes Equations. J. Comp. Physics Vol.53 pp. 209-226
Chen C.J., Naseri-Neshat H., Ho K.S. (1981): Finite Analytical Numerical Solution of
Heat Transfer in Two-Dimensional Cavity Flow. Numerical Heat Transfer Vol.4 pp. 179-
97
Chorin A.J. (1967): A Numerical Method for Solving Incompressible Viscous Flow
Problems. J. Comput. Phys. Vol.2 pp. 12-26
Ciofalo M. (1992): Large-Eddy Simulation of Turbulent Flows with Heat Transfer in

Simple and Complex Geometries. Ph.D. Thesis, Dept of Mechanical Engineering and
Aeronautics, City University, London, U.K.
Ciofalo M., Collins M.W. (1989): ic-c Predictions of Heat Tranfer in Turbulent
Recirculating Flows using an Improved Wall Treatment. Numerical Heat Transfer, Part B
Vol.15 pp. 21-47
Courant R., Isaacson E., Rees M. (1952): On the solution of Non-Linear Hyperbolic
Differential Equations by Finite differences. Comm. Pure Appl. Math. Vol.5 p. 243
331
DeBlois R.L., Cutts D.G., Drost R.T. (1990): Lubrication System for a Low-Side
Horizontal Scroll Compressor. 10th Tnt. Compressor Eng. Conf. Pardue Univ.
Demirdzic I., Peric M. (1988): Space conservation law in finite volume calculations of
fluid flow. Intern. J. Numerical Methods in Fluids Vol.88 pp. 1037-58
Demirdzic I., Peric M. (1990): Finite Volume Method prediction of fluid in arbitrarily
shaped domains with moving boundaries. Intern. J. Numerical Methods in Fluids Vol.88
pp. 77 1-790
Douglas R.W., Ramshaw J.D (1994): Future Research Directions in Computational Fluid-
Dynamics. J. Fluids Eng. ASME Vol.116 pp. 2 12-5
Eiseman P.R., Eriebacher G. (1982): Grid Generation for the Solution of Partial
Differential Equations. NASA Contract Report 178365
Eisman P. (1979): A Multi -Surface Method of Coordinate Generation. J. Comp. Phys.

Vol.33 pp. 118-150
Etemad S., Nieter J. (1988): Computational Parametric Study of Scroll Compressor

Efficiency, Design, and Manufacture. 9 mt. Compressor Eng. Conf. Pardue Univ.
Fearn R.M., Mullin T., Cliffe K.A. (1990): Nonlinear flow phenomena in a symmetric
sudden expansion. J. Fluid Mech. Vol.2 11 pp.595-608
Finlayson B.A. (1972): The Method of Weighted Residuals and Variational Principles.
Academic Press, New York
Fletcher D.F., McCaughey m., Hall R.W. (1993): Numerical Simulation of A Laminar Jet
Flow- A Comparison of 3 CFD Models. Computer Physics Communications Vol.78 n.l-2
pp.1 13-20
Folkow B., Neil E (1971): Circulation. Oxford University Press, New York
Forsey C.R., Billing C.M. (1988): Some Experiences with Grid Generation on Gurved
Surface using Variational and Optimmization Techniques. In: Numerical Grid Generation
Methods for Fluid Dynamics (3). Ed. By K.W. Morton and M.J. Baines, Oxford
University Press.
Frankel S.P. (1950): Convergence Rates of Iterative Treatments of Partial Differential

Equations. Mathematical Tables and other Aids to Computation Vol.4 pp. 65-75
Galiat J.L., Ni S., Daniels M. (1988): A Computer Model For Scroll Compressors. 9tI mt.
Compressor Eng. Conf. Pardue Univ.
332
References and B/biography
Gentry R.A., Martin R.E., Daly B.J. (1996): An Eulerian Differencing Method for Un-
steady Compressible Flow Problems. J. Comput. Phys. Vol.1 pp.87
Glass J., Rodi W. (1982): A Higher Order Numerical Scheme for Scalar Transport.
Comput. Mech. App!. Mech. Engin. Vol.3 pp. 37-58
Glowinski R., Pironneau (1992): Finite Element Methods for Navier-Stokes Equations.
Ann. Rev. of Fluid Mechanics Vol.24 pp.
Gosman A.D., Peric M. (1985): A Flux Bendin Sceme for Redusing Numerical Diffuesion
in Convective Transport, Mech. Eng. Dept. Report, Imperial College, London
Gosman A.D., Pun W.M. (1974): Lecture Notes for a Course Entitled "Computation of
Recirculating Flows". Heat Transfer Sec. Report HTS/74/2, Imperial College, London,
1974
Gosman A.D., Pun W.M., Runchal A.K., Spalding D.B., Wolfshtein M. (1969): Heat and
Mass Transfer in Recirculating Flows. Academic, New York
Graham D.I., Jawes P.W., Jones T.E.R., Davies J.M., Delo E.A (1992): Measurement and
Prediction of Surface Shear-Stress in Annular Flume. J. of Hydraulic Enginnering ASCE
VOL.118 n.9 pp. 1270-86
Greox Leon (1905): U.S. Patent 801,182
Gresho P.A. (1984): A Modified Finite Element Method for Solving the Time-Dependent
Incompressible Navier-Stokes Equations. Part I: Theory. Tnt. J. Mathematics in fluids
Vol.4 pp. 557-98
Gresho P.M., Lee R.L., Sani R.L. (1980): On the Time Dependent Solution for the
Incompressible Navier-Stokes Equations in two and three Dimensions In: Recent
Advances in Num. Methods in Fluids. Eds. C. Taylor and K. Morgan. Pineridge Press,
Swansea Vol.1
o Grunwald S., Beagle W. (1990): Check Valve Movement in a Scroll Compressor. 10th mt.
Hall M.G. (1981): Computational Fluid Dynamics - A Revolutionary Force in

Aerodynamics. AIAA Paper 81-1014, Palo Alto, California
Hassan Y.A., Rice J.G., Kim J.H. (1983): A Stable Mass-Flow-Weighted two-dimensional
Skew Upwind Sceme. Num. Heat Transfer Vol.6 pp. 395-408
Hawkins I.R., Wilkes N.S. (1991): Moving Grids in Harwell FLOW3D. UKAEA Report
AEA-Im Tech-0608
333
Hayano M., Sakata H., Nagamoto S., Murasaki H. (1988): An Analysis of Loses in Scroll
Compressors. . 9 mt. Compressor Eng. Conf. Pardue Univ.
Hayano Makoto et at. (1986): Performance Analysis of Scroll Compressor for Air
Conditioners. 8 Tnt. Compressor Eng. Conf. Pardue Univ.
Hedberg P.K.M. (1989): NONDIF: A Method to Avoid Numerical Diffusion and over-
and under-shoots. Proc. 6th mt. Conf. Num. Meth. Lam. Turb. Flow. Swansea UK, 11-15
July 1989 Ed. By C. Taylor, P. Gresho, P.L. Sani and J. Hauser. Pineridge Press, Swansea
pp. 193-202
Henry F.S., Pearcey H.H (1994): Numerical Model of Boundary-Layer Control using Air-
Jet Generated Vortices. AIAA J. Vol.32 n.12 pp. 2415-25
Hindman R.G., Spencer J (1983): A New Approach to Truly Adaptive Grid Generation.
ATAA Paper 83-0450, Reno Nevada
Hirano T., Matsumura N. (1988): Development of High Efficiency Scroll Compressors for
Air Conditioners. 91b mt. Compressor Eng. Conf. Pardue Univ.
Huiqing L., Disheng W., Huanran W., Penggao C. (1992): Research of oil-injected Scroll
Compressor Working Process. 11± Tnt. Compressor Eng. Conf. Pardue Univ.
Hussaini M.Y., Zang T.A. (1987): Spectral Methods in Fluid-Dynamics. Ann. Review
Fluid Mech. Vol.19 pp. 339-67
Hutton A.G., Smith R.M., Hickmott S. (1987): The Computattion of Turbulent Flows of
Industrial Complexity by the Finite Element Method- Progress and Prospects. mt. J. for
Numerical Methods in Fluids Vol.& pp. 1277-98
Ishi N., Yamamura M., Muramatsu S., Yamamoto S., Sakai M. (1990): Mechanical
Efficiency of a Variable Speed Scroll Compressor. 10th Tnt. Compressor Eng. Conf.
Pardue Univ.
Ishi Noriaki et al. (1986): A Study on Dynamic Behavior of a Scroll Compressor. 8th Tnt.
Issa R.I. (1985): Solution of Implicitly Discretised Fluid Flow Equations by Operator
Splitting. J. Computational Physics Vol.6 1 pp.40
Jeff J.N., Douglas P.O. (1992): Analytical modeling of Discharge flow dynamics in scroll
compressors. 1 1th mt. Compressor Eng. Conf. Pardue Univ.
Jones I.P.(1991): CFDS Software- Vlidation Report: Part T: AEA-Tn Tech Report Harwell
Laboratory, Nov 1991
334
Ref erences and Bibiography
Jones I.P., Kightley J.R., Thompson C.P., Wilkes N.S. (1985): FLOW3D, a Computer
Code for the Prediction of Laminar and Turbulent Flow and Heat Transfer Release 1.
Harwell Report AERE -R 11825, 1985
Jones I.P., Kightley J.R., Thompson C.P., Wilkes N.S. (1986): FLOW3D Release 1: User
Manual. Harwell Report AERE -R 11893, 1986
( Kazutaka S., Masao S., Kenji T. (1992): Performance Analysis of Hermetic Scroll
Compressors. 1 1th mt. Compressor Eng. Conf. Pardue Univ.
C) Kenji Tojo et al. (1984): A Scroll Compressor For Air Conditioners. 7th Tnt. Compressor
Eng. Conf. Pardue Univ.
O Kenji Tojo et a!. (1986): Computer modeling of scroll compressor with self adjusting
back-pressure mechanism. 8th mt. Compressor Eng. Conf. Pardue Univ.
Kobayashi M.H., Pereira J.C.F. (1994): A computational method for solving arbitary two-
dimensional physiological flows. ASME J. Biomechanical Engineering Vol.116 pp. 3 15-7
Latimer B.R., Pollard A. (1989): Comparison of Pressure-Velocity Coupling Algorithms.
Num. Heat Transfer Vol.8 pp. 635-52
Leonard B.P. (1979): A Stable and Accurate Convective Modeling Procedure based on
Quadratic Upstream Interpolation. Comp. Meth. AppI. Mech. Eng. Vol.19 pp.59-98
Levine R.D. (1982): Supercomputers. Sc. Am. Vol.246 pp.1 18-135
C) Liansheng L., Yongzhang Y. (1990): The Theoretical Analysis and Experimental

Research for A Refrigerating Scroll Compressor. 10th mt. Compressor Eng. Conf. Pardue
Univ.
Lillington J.N. (1981): A Vector Upstream Differencing Scheme for Problems in Fluid
Flow involving Significant Source Term in Steady-Linear Systems. Tnt. J. Num. Meth.
Fluids Vol.1 pp. 3-6
Lin A. (1985): The Parameterized Strongly Implicit Method for Solving Elliptic
Difference Equations. mt. J. Num. Meth. In Fluids Vol.1 pp. 3-6
Lohner R. (1987): Finite Elements in CFD: What lies Ahead?. mt. J. Numer. Meth.
Engng. Vol.24 pp.1741-56
Lysholm A. (1967): Screw Rotor Machine. U.S. Patent No. 3,314,598
McDonough J.M., Catton I. (1982): A mixed Finite difference-Galerkin Procedure for

Two Dimensional Convection in A square Box. mt. J. Heat and Mass Transfer Vol.25 pp.
1137-46
335
Mom P., Kim J. (1982): Numerical Investigation of Turbulent Channel Flow. J. Fluid
Mech. Vol.118 PP. 341-77
Morishita E., Kitora Y., Nishida M. (1992): Basic Study of Engine with Scroll
1th
Compressor and Expander. 1 mt. Compressor Eng. Conf. Pardue Univ.
Morishita Etsuo et al. (1984): Scroll Compressor Analytical Model. 7th hit. Compressor
Eng. Conf. Pardue Univ.
Nieter J. (1988): Dynamics of Scroll Suction Process. 9th Tnt. Compressor Eng. Conf.
Pardue Univ.
Patankar S.V. (1980): Numerical Heat Transfer and Fluid Flow. McGraw-Hill Book
Company
Patankar S.V. (1981): A Calculation Procedure for Two Dimensional Elliptic Situations.
Numerical Heat Transfer Vol.4 pp.409-25
Patankar S.V. (1988): Recent Developments in Computational Heat Transfer. ASME J.

Heat Transfer Vol.110 pp. 1037-45
Patankar S.V., Spalding D.B. (1972): A Calculation Procedure for Heat, Mass and
Momentum Transfer in Three Dimensional Parabolic Flows. Tnt. J. Mass Transfer Vol.15
PP . 1787-1806
Patera A.T. (1984): A Spectral Method for Fluid Dynamics: Laminar Flow in a Channel
Expansions. J. Comp. Phys. Vol.54 Pp.468-88
Pedley T.J. (1985): Flow along a channel with a time-dependent indentation in one wll:
the generation of vorticity waves. J. Fluid Mech. Vol.160 pp. 337-367
Peraire J., Morgan K., Peiro J.(1990): Unstructured mesh methods for CFD. Von Karman
Institute for Fluid Dynamics. "Numerical Grid Generation" Lecture Notes
Peric M. (1987): Efficient Semi-Implicit Solving Algorithm for Nine-Diagonal Coefficient

Matrix. Numerical Heat Transfer Vol.11 pp.251-79
Proceedings of the First CFDS International User Conference 6-7 May 1993, Eynsham
Hall. Oxford
Puttre'M. (1991): FEA programs band together. ASME Mech. Engng Vol.113 pp.77-80
Rai M.M., Anderson D.A. (1982): Application of Adaptive Grids to Fluid Flow Problems
with Asymptotic Solutions. AIAA J. Vol.20 pp.496-502
Raithby G.D. (1976): Skew Upwind Differencing Schemes for Problems involving Fluid
336
Flow. Comp. Meth. Appi. Mech Eng. Vo19 pp.l53-65
Reddy J.N. (1982): On Penalty Function Methods in the Finite Element Analysis of Flow
Problems. mt. J. Num. Metho. In Fluids Vol.2 pp.151-71
Rhie C.M. (1981): A Numerical Study of the Flow Past an Isolated Airfoil with
Separation. Ph.D. Thesis Dept. Mech. And Industrial Engineering University of Illinois at
Urbana-Champaign
Rhie C.M., Chow W.L. (1983): Numerical Study of the Turbulrnt Flow Past an Airfoil
with Trailing Edge Separation. AIAA J. Vol.2 1 pp. 1527-32
Rizzi A. Erikson L.E. (1981): Transfinite Mesh Generation and Damped Eulerian
Equation Algorithm for Transonic Flow around Wing-Body Configurations. Proc. AIAA
5th Computational Fluid Dynamics Conference, Palo Alto, Califomia pp. 43-69
Roache P.J. (1972): Computational Fluid Dynamics. Hermosa Publishers
Rogers R.J., Wagner T.C. (1990): Scroll Compressors Flow Modeling: Experimental and
Computational Investigation. 10th Tnt. Compressor Eng. Conf. Pardue Univ.
O Ronald T.D., Quesada J.F. (1992): Analytical and Experimental Investigation of a Scroll
Compressor Lubrication System. 11th Tnt. Compressor Eng. Conf. Pardue Univ.
Runchal A.K. (1987): CONDIF: A Modified Central-Difference Scheme for Convective

Flows. mt. J. Num. Meth. Eng. Vol.24 pp.1593-608
Smith I.K (1985): "PROJECT SPHERE: An Assessment of the Practicability of the

Trilateral Wet Vapour Cycle for Recovery of Power from Low Grade Heat Sources",
Thermo-Fluids Engineering Research Centre Report, The City University
Smith I.K. (1986): "A Comparison of Three Surface Plant Options for the Conversion of
Geothermal Energy From HDR", IRD 86/17
Smith I.K. (1993): Development of the Trilateral Flash Cycle System: 1 Fundamental
Considerations. J. of Power and Energy, Proc IMechE Vol.207 pp. 179-194
Smith I.K., Aldis C.A (1990): Lysholm Screw Expanders in place of throttle valves in
large heat pump and refrigeration systems. Proc. Meeting of Commission E2, Stockholm,
Aug 29-3 1
Smith I.K., Pitanga Marques da Silva R. (1994): Part 2: Increasing power output with
working fluid mixtures. J. of Power and Energy, Proc. IMechE Vol.208 pp. 135-144
0 Smith I.K., Stosic N., Aldis C.A. (1994): Lyshoim twin screw machines as two-phase
337
expanders. 12th mt. Compressor Engineering Conf. Pardue Univ. July
Smith I.K., Stosic N., Aldis C.A. (1995): Development of the Trilateral Flash Cycle
System: Part 3: The Design of High Efficiency Two-phase Screw Expanders. J. of Power
and Energy Proc. IMechE
Spalding D.B. (1970): A Novel Finite Difference Formulation for Differential

Expressions Involving both First and Second Derivatives. mt. J. Num. Meth. Eng. Vol.4
pp.55 1-9
Spalding D.B. (1972): Mathematical Modeling of Fluid Mechanics, Heat Transfer and
Mass Transfer Processes. Imperial College, London, Mech. Eng. Dept. Rep. N. HTS/80/1
Sprankle R.S. (1973): Electrical Power Generating Systems. U.S. Patent 3,751,673,
August 1973
Sprankle R.S. (1976): Throttling Means for Geothermal Stream. U.S. Patent 3,977,818
Starius G. (1977): Construction of Orthogonal Curvilinear Meshes by solving Initial

Values Problems. Num. Math. Vol.28 pp.25-48
Steidel R.F., Weiss H., Flower J.E. (1982): Characteristics of the Lysholm Engine as
Tested for Geothermal Applications in the Imperial Valey. J. Eng. For Power Vol.104 pp.
231-240
Stone H.L. (1968): Iterative Solution of Implicit Approximations of Multi-Dimensional

Partial Differential Equations. SIAM J. Num. Anal. Vol.5 pp. 530-58
Stosic N., Hanjalic K. (1994): Development and optimization of screw engine rotor pairs
on the basis of computer modeling. 12th mt. Compressor Eng. Conf. Pardue Univ. July
Stosic N., Milutinovjc L., Hanjalic K., Kovacevic A.(1992): Investigation of oil injection
influence upon the screw compressor working process. Intj. Refrig. Vol.15 pp. 206-220
Syed S., Chiappeta L. (1985): Finite Difference Methods For Reducing Numerical
Diffusion in TEACH-type Calculations. AIAA Paper 85-0056
Taniguchi H. (1988): Analytical and Experimental Investigation of Two-Phase Flow

Screw Expanders for Power Generation. Tran. Os ASME, J. of Eng. For Gas Turbines and
Power, Vol.110
Thompson J.F. (1984): A Survey of Dynamically-Adaptive Grids in the Numerical

Solution of Partial Differential Equations. J. of Numerical Mathematics
Thompson J.F. (1984): Grid Generation Techniques in Computational Fluid Dynamics.
338
AIAA Journal Vol.22 n.h pp. 1505-23
Thompson J.F. (1988): A Composite Grid Generation Code for General 3_D Regions. The
Eagle Code. AIAA J. Vol.26 pp. 27 1-2
Thompson J.F., Thames F.C., Mastin C.W. (1974): Automatic Numerical Generation of
Body-Fitted Curvilinear Coordinate System for Field Containing Any Number Arbitrary
Two Dimensional Bodies. J. Computational Physics Vol.15 pp. 299-3 19
Thompson J.F., Warsi Z.u., Mastin C.W. (1982): Boundary Fitted Coordinate Systems for
Numerical Solution of Partial Differential Equations - A Review. J. Comp. Phys. Vol.47
pp. 1-108
Thompson J.F., Warsi Z.U., Mastin C.W. (1985): Numerical Grid Generation. Elsevier
Science Publishing
Thompson J.F., Warsi Z. J.A. (1983): Three Dimensional Grid Generation from Elliptic
Systems. AIAA -83-1905 AIAA 6th Computational Fluid Dynamics Conference, Danvers,
Massachusetts
Thompson J.F., Wilkes N.S. (1982): Experiments with Higher-Order Finite Difference
Formulae. AERE 10493
Tsutomu Inaba et al. (1986): A scroll compressor with sealing means and low side shell.
8Lh Tnt. Compressor Eng. Conf. Pardue Univ.
Van Doormal J.P., Rairhby G.D. (1984): Enhancements of the SIMPLE Method for
predicting Incopressible Fluid Flow. Heat Transfer Vol.7 pp. 147-63
Wang Zongyan (1992) A New Type of Curve Used in the Wrap Design of the Scroll
Compressor. 1 1th Tnt. Compressor Eng. Conf. Pardue Univ.
Weatherill N.C., Forsey C.R. (1984): Grid Generation and Flow Calculations for Complex
Aircrafts Geometrics using a Multiblock Scheme. AIAA -84-1665. AIAA l7 Fluid
Dynamics, Plasma Dynamics, and Lasers Conference, Snowmass, Co
Wong H.H., Raithby G.D. (1979): Improved Finite Difference Methods Based on a
Critical Evaluation of the Approximation Errors. Num. Heat. Transfer Vol.2 pp. 139-63
( Yamamura M., et al. (1990): Compact Type Scroll Compressor for Air Conditioners.
Tnt. Compressor Eng. Conf. Pardue Univ.
C Yanagisawa T. et a!. (1990): Optimum Operating Pressure Ratio For Scroll Compressors.
lO Tnt. Compressor Eng. Conf. Pardue Univ.
339
Yogzhang Y., Yuhua X., Lianheng L. (1992): The Mechanical Analysis of a Scroll
Compressor. 1 1th mt. Compressor Eng. Conf. Pardue Univ.
Yong H. (1994): Leakage Calculation Through Clearances. 12th mt. Compressor Eng.
Conf. Pardue Univ.
Zaneli R., Favrat D. (1994): Experimental Investigation of a Hermetic Scroll Expander-

Generator. l2 mt. Compressor Eng. Conf. Pardue Univ.
Zhang H., Reggio M., Trepanier J.Y., Camarero R. (1993): Discrete Form of the CGL for
Moving Meshes and its Implementation in CFD schemes. Computers Fluids Vol.22 N. I
pp. 9-23
Zhenquan L., Guirong D., Shicai Y., Mingzhi W. (1992): The Grapjics Method Of
Modified Wrap Of Scroll Compressor. 1 1th Tnt. Compressor Eng. Conf. Pardue Univ.
Zhu J., Wang D. (1990): A research of scroll compressor working process computer
simulation and testing. 10th Tnt. Compressor Eng. Conf. Pardue Univ.
Zienkiewicz O.C. (1977): The Finite Element Method. Third Edition, McGraw-Hill
Zienkiewicz O.C., Cheung Y.K. (1965): Finite Elements in the Solution of Field
Problems. The Engineer pp. 507-10
340

Flow Modelling of Scroll Compressors and Expanders

Uploaded by

Copyright:

Available Formats

Flow Modelling of Scroll Compressors and Expanders

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Flow Modelling of Scroll Compressors and Expanders

Uploaded by

Copyright:

Available Formats

Karagiorgis, G. (1998). Flow modelling of scroll compressors and expanders.

City Research Online

Permanent City Research Online URL: http://openaccess.city.ac.uk/7573/

Copyright & reuse

COMPRESSORS AND EXPANDERS

Department of Mechanical Engineering and Aeronautics

CHA PTER 1: INTRODUCTION .

1.1 Introductory Remarks .................................................................

1.2 Present Contribution .................................................................. 2

1.3 Thesis Layout .......................................................................... 3

CHA PTER 2: LITERATURE SURVEY .............................................. 5

2.1 Introduction ............................................................................ 5

2.2 Compressors ........................................................................... 5

2.2.1 Aerodynamic Compressors ..................................................... 6

2.2.1.1 Centrifugal Compressors .................................................... 7

2.2.2 Positive Displacement Compressors ........................................... 7

2.2.2.1 Reciprocating Compressor .................................................. 7

2.2.2.2 Screw Compressor ............................................................ 8

2.2.2.3 Scroll Compressor ............................................................ 9

2.2.2.3.1 Working Process ......................................................... 10

2.2.2.3.2 Scroll Members .......................................................... 12

2.2.2.3.3 Principle Of Operation ................................................... 12

2.2.2.3.4 Analytical Model ......................................................... 13

2.3 Two Phase Expanders ................................................................ 22

2.3.1 Scroll Expanders .................................................................. 24

2.3.2 Lyshoim Total Flow Expander .................................................. 25

2.3.2.1 City University And Lyshoim Expander .................................. 27

2.3.3 Two Phase Turbines .............................................................. 27

2.4 Conclusions ........................................................................... 29

CHAPTER 3: OUASI ONE DIMENSIONAL FLOW MODELING ............ 31

3.1 Introduction ............................................................................ 31

3.2 Theory ................................................................................... 32

3.2.1 Assumptions ....................................................................... 32

3.2.2 Conservation Equations .......................................................... 32

EnergyGain Due To The Gas Inflow ............................................... 33

EnergyGain Due To The Gas Outflow ............................................. 34

3.2.3 Gas Leakage ........................................................................ 35

3.2.4 Oil Injection ........................................................................ 37

3.2.5 Computation Of Fluid Properties ............................................... 38

3.3 Results Of The 1-D Flow Modelling ................................................ 39

3.3.1 Scroll Compressor ................................................................. 39

3.3.2 Scroll Expander .................................................................... 53

3.4 Applications Of The Scroll Compressor - Expander .............................. 58

3.4.1 Cycle Proposed .................................................................... 58

3.4.2 The Proposed Model ............................................................... 59

3.4.3 Expander Characteristics ......................................................... 60

3.4.4 Compressor Characteristics ...................................................... 60

3.4.5 Combined Characteristics ........................................................ 61

CHAPTER 4: GEOMETRY AND GRID GENERATION ........................ 64

4.1 Introduction ............................................................................ 64

4.2 Geometry ............................................................................... 64

4.3 Theory of Grid Generation ........................................................... 68

4.3.1 Algebraic Methods ............................................................... 69

4.3.2 Partial Differential Equation Methods ......................................... 69

4.3.3 Adaptive Grids .................................................................... 71

4.3.4 Moving Boundary Grids ......................................................... 72

4.3.5 Computational Grids For Complex Geometry Domains ..................... 73

4.3.5.1 Multi Block Grids ............................................................ 74

4.3.5.2 Unstructured Grids ........................................................... 74