Buckney, David

City Research Online
City, University of London Institutional Repository
Citation: Buckney, D. (2017). Clearance management in twin screw compressors.

(Unpublished Doctoral thesis, City, University of London)
This is the accepted version of the paper.
This version of the publication may differ from the final published
version.
Permanent repository link: https://openaccess.city.ac.uk/id/eprint/18235/
Link to published version:
Copyright: City Research Online aims to make research outputs of City,

University of London available to a wider audience. Copyright and Moral
Rights remain with the author(s) and/or copyright holders. URLs from
City Research Online may be freely distributed and linked to.
Reuse: Copies of full items can be used for personal research or study,
educational, or not-for-profit purposes without prior permission or
charge. Provided that the authors, title and full bibliographic details are
credited, a hyperlink and/or URL is given for the original metadata page
and the content is not changed in any way.
City Research Online: http://openaccess.city.ac.uk/ publications@city.ac.uk

CLEARANCE MANAGEMENT IN
TWIN SCREW COMPRESSORS
DAVID MALCOLM BUCKNEY
Thesis submitted for the degree of Doctor of Philosophy
in Mechanical Engineering
City, University of London
School of Mathematics, Computer Science and Engineering
August 2017
ii
Table of Contents
Table of Contents .......................................................................................................................... iii
List of Figures ................................................................................................................................. ix
List of Tables.................................................................................................................................. xv
Acknowledgements................................................................................................................... xvii
Declaration ................................................................................................................................. xviii
Abstract .......................................................................................................................................... xix
Notation ........................................................................................................................................... xx
Chapter 1 Introduction........................................................................................................... 1
1.1 Motivation ....................................................................................................................... 1
1.2 The Twin Screw Compressor................................................................................... 2
1.2.1 Basic Principles .................................................................................................... 2
1.2.2 Ports and Internal Compression.................................................................... 3
1.2.3 Capacity Control................................................................................................... 4
1.2.4 Oil Injection and Compressor Arrangement ............................................. 5
1.3 Design and Optimisation ........................................................................................... 6
1.3.1 Industrial Applications ...................................................................................... 6
1.3.2 Optimisation .......................................................................................................... 8
1.3.3 General Design...................................................................................................... 9
1.3.4 Rotor Profile Design ......................................................................................... 10
1.3.5 Clearance Design ............................................................................................... 12
1.3.6 Design Software ................................................................................................. 13
1.4 Preliminary Studies ................................................................................................... 13
1.4.1 Performance Sensitivity to Clearance ....................................................... 13
iii
1.4.2 Clearance Influences and Behaviour ......................................................... 14
1.5 Concluding Remarks ................................................................................................ 18
1.6 Thesis Overview ......................................................................................................... 19
Chapter 2 Background Theory and Literature Review ........................................... 21
2.1 Introduction ................................................................................................................. 21
2.2 Rotor Design ................................................................................................................ 21
2.2.1 Profile Generation............................................................................................. 21
2.2.2 Clearance Design ............................................................................................... 25
2.3 Geometric Characteristics ...................................................................................... 28
2.3.1 Identification of Geometric Characteristics ............................................ 28
2.3.2 Calculation of Geometric Characteristics ................................................ 33
2.4 Review of Twin Screw Compressor Models .................................................... 34
2.4.1 Chamber Models................................................................................................ 34
2.4.2 Alternative Models ........................................................................................... 39
2.4.3 Thermal Analysis .............................................................................................. 40
2.5 Concluding Remarks ................................................................................................ 43
Chapter 3 Research Objectives, Methods and Expected Contribution .............. 45
3.1 Statement of Research ............................................................................................. 45
3.2 Objectives ..................................................................................................................... 46
3.3 Methodology ................................................................................................................ 46
3.4 Expected Contribution............................................................................................. 48
3.4.1 Boundary Map .................................................................................................... 48
3.4.2 Model Integrating Operational Clearance Modification .................... 48
Chapter 4 Boundary Map for Rotor and Casing Surfaces ....................................... 49
4.1 Introduction ................................................................................................................. 49
4.2 Definition of Boundary Surfaces .......................................................................... 50
4.2.1 Coordinate System Conventions ................................................................. 50
iv
4.2.2 Casing Surfaces ................................................................................................... 54
4.2.3 Rotor Surfaces .................................................................................................... 55
4.3 Definition of Cycle Conventions ........................................................................... 58
4.4 Mapping of the Casing Surfaces ............................................................................ 62
4.5 Mapping of the Rotor Surfaces .............................................................................. 66
4.6 Definition of Rotor Surface Boundary Map ...................................................... 68
4.6.1 Example: Zero Blow-Hole Profile ................................................................ 73
4.6.2 Example: Symmetric Profile .......................................................................... 80
4.6.3 Example: Asymmetric Profile ....................................................................... 86
4.6.4 Example: N – Profile ......................................................................................... 90
4.7 Local Cycle Exposure over Full Rotors .............................................................. 92
4.8 Mapping Pressure and Temperature on Surfaces ......................................... 99
4.8.1 Temperature Distribution.............................................................................. 99
4.8.2 Pressure Distributions ................................................................................. 101
4.9 Concluding Remarks .............................................................................................. 102
Chapter 5 Operational Clearance Distortions .......................................................... 103
5.1 Introduction .............................................................................................................. 103
5.2 Approximating Temperature Distributions.................................................. 104
5.2.1 Casing Temperature ...................................................................................... 104
5.2.2 Rotor Temperature ........................................................................................ 105
5.3 Thermal Distortion of Interlobe Gap ............................................................... 111
5.3.1 Interlobe Sealing Line ................................................................................... 111
5.3.2 Local Rotor Distortion .................................................................................. 112
5.3.3 Local Casing Distortion ................................................................................ 113
5.3.4 Relative Gap Distortion ................................................................................ 114
5.4 Thermal Distortion of Radial Gap ..................................................................... 118
5.4.1 Radial Sealing Lines ....................................................................................... 118
v
5.4.2 Local Distortions ............................................................................................. 119
5.5 Modelling with Operational Clearance Corrections ................................... 121
5.5.1 Integration with Geometry Calculation Program ............................... 121
5.5.2 Integration with Chamber Model ............................................................. 123
Chapter 6 Case Studies ...................................................................................................... 127
6.1 Introduction ............................................................................................................... 127
6.2 Model Sensitivity to Rotor Parameters ........................................................... 129
6.2.1 Overview ............................................................................................................ 129
6.2.2 DRUM127 Compressor ................................................................................. 129
6.2.3 Mapped Boundary Conditions ................................................................... 131
6.2.4 Sensitivity to Lobe Combination ............................................................... 133
6.2.5 Sensitivity to Wrap Angle ............................................................................ 135
6.2.6 Sensitivity to Volume Index ........................................................................ 137
6.2.7 Discussion .......................................................................................................... 140
6.3 Model Sensitivity to Clearance Modifications .............................................. 141
6.3.1 Overview ............................................................................................................ 141
6.3.2 DRUM127 Compressor ................................................................................. 141
6.3.3 DRUM127 Test Measurement .................................................................... 141
6.3.4 DRUM127 Clearance Sensitivity ............................................................... 143
6.3.5 DRUM127 Performance Sensitivity ......................................................... 145
6.3.6 Discussion .......................................................................................................... 146
6.4 Investigating Interlobe Clearance Distortion ............................................... 147
6.4.1 Overview ............................................................................................................ 147
6.4.2 WCVTA510 Compressor .............................................................................. 148
6.4.3 WCVTA510 Test Measurement ................................................................. 150
6.4.4 WCVTA510 Thermal Analysis .................................................................... 151
vi
6.4.5 WCVTA510 Clearance Presentation........................................................ 152
6.4.6 WCVTA510 Clearance Distortion Results ............................................. 155
6.4.7 WCVTA510 Performance Results ............................................................ 156
6.4.8 WCVTA510 Teardown Results .................................................................. 157
6.4.9 Discussion ......................................................................................................... 158
6.5 Investigating Radial Clearance Distortion ..................................................... 159
6.5.1 Overview ............................................................................................................ 159
6.5.2 HS204 Compressor ........................................................................................ 159
6.5.3 HS204 Test Measurement ........................................................................... 161
6.5.4 HS204 Performance Results with Jacket Cooling .............................. 165
6.5.5 HS204 Performance Results Without Jacket Cooling ....................... 173
6.5.6 HS204 Radial Gap Distortion Results ..................................................... 176
6.5.7 Discussion ......................................................................................................... 179
Chapter 7 Conclusions and Future Work ................................................................... 184
7.1 Conclusions................................................................................................................ 184
7.2 Future Work .............................................................................................................. 186
References ................................................................................................................................... 188
Appendix A Geometry Calculations ............................................................................. 195
A.1 Basic Geometry and Terminology .................................................................... 195
A.2 Meshing Conditions for Conjugate Profile ..................................................... 199
A.3 Calculating Geometry Characteristics ............................................................. 203
A.3.1 Chamber Area and Volume ......................................................................... 203
A.3.2 Suction Port Areas.......................................................................................... 205
A.3.3 Blow Hole Area ................................................................................................ 206
A.3.4 Sealing Lines ..................................................................................................... 207
Appendix B Compressor Modelling ............................................................................. 210
vii
B.1 Chamber Model Details ......................................................................................... 210
B.1.1 Model Discharge Temperature .................................................................. 210
B.1.2 Adiabatic Fanno Flow.................................................................................... 211
B.2 Thermal Analysis Simplifying Assumptions ................................................. 212
B.2.1 Rotor Thermal Distortion ............................................................................ 212
B.2.2 Datum for Relative Clearance Distortion ............................................... 212
Appendix C Definition of Performance Characteristics ....................................... 217
C.1 Compressor Boundaries ....................................................................................... 217
C.2 Mass Flow ................................................................................................................... 218
C.3 Volumetric Efficiency ............................................................................................. 218
C.4 Adiabatic Efficiency ................................................................................................ 219
viii
List of Figures
Figure 1-1: Compressor operation.......................................................................................... 2
Figure 1-2: Meshing rotors highlighting single compression chamber.................... 4
Figure 1-3: Oil free compressor ............................................................................................... 5
Figure 1-4: Howden WRV range (courtesy of Howden Compressors Ltd.) ............ 7
Figure 1-5: Comparison of rotors with different basic geometry parameters ...... 9
Figure 1-6: SRM standard rotor geometry. Equal diameters; 4/6 lobes; L/D =
1.65; wrap angle = 300°. ........................................................................................................... 10
Figure 1-7: Rotor profile on transverse view ................................................................... 11
Figure 1-8: Rotor clearance gaps ........................................................................................... 12
Figure 1-9: Rotor retrofit test results with different clearances............................... 14
Figure 1-10: Location of proximity probe .......................................................................... 15
Figure 1-11: Measured relationship between operating temperature and
clearance reduction .................................................................................................................... 16
Figure 1-12: How rotor thermal growth and casing thermal growth affect
clearances ....................................................................................................................................... 17
Figure 2-1: Transverse profiles showing terminology ................................................. 22
Figure 2-2- Photo comparing symmetric and N-rotor profile rotors ...................... 23
Figure 2-3: ‘N’ rotor rack generation ................................................................................... 24
Figure 2-4- Clearance distribution on a rack projection .............................................. 27
Figure 2-5: Defining discrete leakage paths ...................................................................... 30
Figure 2-6: Leakage flow from chamber volume ............................................................ 31
Figure 2-7: Volume and area curves .................................................................................... 34
Figure 3-1: GEOM program integration with SCORPATH ............................................ 47
Figure 4-1: Compressor and rotor co-ordinate systems .............................................. 51
Figure 4-2: Parameters for surface definition .................................................................. 52
Figure 4-3: Exploded view of casing surfaces .................................................................. 54
Figure 4-4: Transverse profiles defined at LP plane ...................................................... 56
Figure 4-5: Full main rotor surface....................................................................................... 56
Figure 4-6: Rotor surface arrays............................................................................................ 57
Figure 4-7: Rotor home position ........................................................................................... 59
Figure 4-8: Rotor segments ..................................................................................................... 59
ix
Figure 4-9: Moving rotors to the start of the compression cycle ............................. 61
Figure 4-10: Single chamber projected onto casing surfaces .................................... 63
Figure 4-11: Cycle angle mapped onto casing bore surfaces ..................................... 66
Figure 4-12: Position of rotor chambers in the transverse plane ............................ 67
Figure 4-13: Notation used on boundary map ................................................................ 70
Figure 4-14: Local transverse rotor offset angle, λ, at given axial position, γ. .... 71
Figure 4-15: Zero blow-hole rotor profile ......................................................................... 73
Figure 4-16: Zero BH rotor boundary map for main rotor ......................................... 74
Figure 4-17: Zero BH rotor boundary map for gate rotor........................................... 75
Figure 4-18: Zero BH rotor boundary points (λ = 0) ..................................................... 76
Figure 4-19: Zero BH rotor boundary points on main rotor surface (λ = 0) ........ 77
Figure 4-20: Zero BH rotor boundary points on gate rotor surface (λ = 0) ......... 77
Figure 4-21: Zero BH rotor boundary points (λ = -φc1) ............................................... 78
Figure 4-22: Zero BH rotor boundary points on main rotor surface (λ = -φc1) .. 79
Figure 4-23 Zero BH rotor boundary points on gate rotor surface (λ = -φc1) ..... 79
Figure 4-24: Symmetric (circular) rotor profile ............................................................. 80
Figure 4-25: Symmetric rotor boundary points (λ = -φC1) ......................................... 81
Figure 4-26: Symmetric rotor boundary points (-φC1 < λ < 0)................................... 81
Figure 4-27: Symmetric rotor boundary points (λ = 0) ............................................... 82
Figure 4-28: Symmetric rotor boundary points on main rotor ................................ 83
Figure 4-29: Symmetric rotor boundary map for main rotor with new
boundaries ..................................................................................................................................... 83
Figure 4-30: Symmetric rotor boundary points on gate rotor .................................. 84
Figure 4-31: Symmetric boundary map for gate rotor with new boundaries ..... 84
Figure 4-32: Asymmetric rotor profile (hybrid of zero BH and symmetric) ....... 86
Figure 4-33: Asymmetric rotor boundary map for main rotor ................................. 87
Figure 4-34: Asymmetric rotor boundary map for gate rotor................................... 87
Figure 4-35: Asymmetric rotor boundary map for gate rotor with adjusted
radial SL .......................................................................................................................................... 89
Figure 4-36: N-profile rotors .................................................................................................. 90
Figure 4-37: N-profile boundary map for main rotor ................................................... 91
Figure 4-38: N-profile boundary map for gate rotor..................................................... 91
Figure 4-39: Full rotor surfaces (3D domain) .................................................................. 93
x
Figure 4-40: Main rotor chamber offsets ........................................................................... 95
Figure 4-41: Gate rotor chamber offsets ............................................................................ 96
Figure 4-42: Full rotor surfaces with surface contour plot of local cycle angle .. 98
Figure 4-43: Thermodynamic results from chamber model ...................................... 99
Figure 4-44: Instantaneous gas temperature at rotor surface................................ 100
Figure 4-45: Time averaged gas temperature at rotor surface .............................. 101
Figure 5-1: Average fluid temperature at casing surfaces ........................................ 104
Figure 5-2: Average fluid temperature at casing surfaces in 3D ............................ 105
Figure 5-3: Instantaneous boundary temperature at rotor surfaces ................... 106
Figure 5-4: Instantaneous temperature at main rotor surface at different cycle
angles............................................................................................................................................. 107
Figure 5-5: Time-averaged boundary temperature at rotor surfaces ................. 108
Figure 5-6: Time averaged boundary temperature on main rotor ....................... 109
Figure 5-7: Planar averaged boundary temperature on rotor surfaces .............. 110
Figure 5-8: Planar averaged temperature distribution on rotors ......................... 110
Figure 5-9: Side projection of interlobe sealing line ................................................... 111
Figure 5-10: Transverse planes at suction and discharge of casing ..................... 113
Figure 5-11: Transverse analysis of interlobe gap ...................................................... 114
Figure 5-12: Pressure angle intersecting with pitch point ....................................... 115
Figure 5-13: Angles on transverse section ..................................................................... 116
Figure 5-14: Radial sealing points on main and gate rotors .................................... 118
Figure 5-15: Radial sealing lines......................................................................................... 119
Figure 5-16: Flow chart for performance calculation with operational clearances
.......................................................................................................................................................... 123
Figure 6-1: DRUM127 air compressor ............................................................................. 129
Figure 6-2: Rotor boundary temperatures ..................................................................... 131
Figure 6-3: Averaged boundary temperatures used for thermal analysis ......... 131
Figure 6-4: Rotor temperature distribution .................................................................. 132
Figure 6-5: Comparison of rotor models with different lobe combinations ..... 133
Figure 6-6: Rotor temperature distributions with different lobe combinations
.......................................................................................................................................................... 134
Figure 6-7: Comparison of rotor models with different wrap angles .................. 135
Figure 6-8: Rotor temperature distributions with different wrap angles .......... 136
xi
Figure 6-9: Modelled temperature during cycle with different volume index . 137
Figure 6-10: Comparison of rotor models with different volume index ............. 137
Figure 6-11: Rotor temperature distributions with different volume index ..... 139
Figure 6-12: Comparison of casing temperatures with different volume index
.......................................................................................................................................................... 139
Figure 6-13: DRUM127 schematic ..................................................................................... 142
Figure 6-14: Interlobe clearance distribution corrected for thermal
deformations............................................................................................................................... 143
Figure 6-15: Variation of leakage areas throughout compression cycle ............. 145
Figure 6-16: Compressor performance from test and model .................................. 146
Figure 6-17: WCVTA510 compressor ............................................................................... 148
Figure 6-18: WRV range compressor vertical section ................................................ 148
Figure 6-19: WRVTA compressor on test stand............................................................ 150
Figure 6-20: Instantaneous and averaged fluid boundary temperature. Line AB
highlights the interlobe sealing line for a single compression chamber. ............ 151
Figure 6-21: Transverse cross section of rotors showing interlobe clearance
distribution of rotor surfaces ............................................................................................... 153
Figure 6-22: Comparison of example interlobe clearance distributions along
rack projection of rotors ........................................................................................................ 154
Figure 6-23: Original design clearance variations ....................................................... 155
Figure 6-24: Revised design clearance variations ....................................................... 156
Figure 6-25: Evidence of rotor rooting on main rotor of WCVTA510 .................. 157
Figure 6-26: Horizontal section of HS204 compressor .............................................. 159
Figure 6-27: Rotor profiles and design data ................................................................... 160
Figure 6-28: Interlobe clearance design data ................................................................ 161
Figure 6-29: HS204 on test stand ....................................................................................... 162
Figure 6-30: HS204 instrumentation schematic ........................................................... 162
Figure 6-31: Position of radial proximity probes in the test compressor........... 164
Figure 6-32: Proximity probe location ............................................................................. 164
Figure 6-33: HS204 volumetric efficiency with unmodified model clearances 167
Figure 6-34: HS204 discharge temperature with unmodified model clearances
.......................................................................................................................................................... 167
xii
Figure 6-35: HS204 volumetric efficiency with modelled rotor thermal
distortion ..................................................................................................................................... 169
Figure 6-36: HS204 discharge temperature with modelled rotor thermal
distortion ..................................................................................................................................... 169
Figure 6-37: HS204 volumetric efficiency with modelled rotor and casing
thermal distortion .................................................................................................................... 171
Figure 6-38: HS204 discharge temperature with rotor and casing thermal
distortion ..................................................................................................................................... 171
Figure 6-39: HS204 volumetric efficiency at 6000rpm with and without jacket
cooling ........................................................................................................................................... 174
Figure 6-40: HS204 discharge temperature at 6000rpm with and without jacket
cooling ........................................................................................................................................... 174
Figure 6-41: Measured and modelled radial gap on MAIN rotor bore ................ 177
Figure 6-42: Measured and modelled radial gap on GATE rotor bore................. 177
Figure 6-43: Local temperatures at MAIN rotor bore radial gap ........................... 178
Figure 6-44: Local temperatures at GATE rotor bore radial gap ........................... 178
Figure 6-45: Example of overall casing displacement due to thermal loading
(courtesy of Howden Compressors Limited.) with approximate rotor body and
proximity probe location superimposed ......................................................................... 182
Figure A-1: Basic rotor dimensions ................................................................................... 195
Figure A-2: Helix angle calculation .................................................................................... 196
Figure A-3: Angles to casing cusp ....................................................................................... 198
Figure A-4: Meshing condition ............................................................................................ 200
Figure A-5: Solving for meshing angle.............................................................................. 201
Figure A-6: Interlobe sealing line and normal projection......................................... 202
Figure A-7: Cross sectional area calculation using trapezoidal rule..................... 203
Figure A-8: Area integration limits for volume ............................................................. 204
Figure A-9: Designated port names ................................................................................... 205
Figure A-10: Blow-hole leakage area ................................................................................ 206
Figure A-11: Interlobe sealing line .................................................................................... 207
Figure A-12: Interlobe leakage area curves ................................................................... 208
Figure B-1: Chamber model setup ..................................................................................... 210
Figure B-2: Uniform 2D thermal distortion .................................................................... 212
xiii
Figure B-3: 2D rotor and casing distortion ..................................................................... 213
Figure B-4: DRUM127 casing temperature exposure ................................................. 214
Figure B-5: DRUM127 axial casing temperature variation....................................... 215
Figure C-1: Compressor boundaries for performance evaluation ......................... 217
Figure C-2: Comparison of model indicated PV curve and polytropic curve ..... 221
xiv
List of Tables
Table 1-1: Commonly grouped constraints ......................................................................... 8
Table 4-1: Compressor orientation ...................................................................................... 50
Table 4-2: Casing surface parameters ................................................................................. 55
Table 4-3: Local cycle angle corrections on main rotor ............................................... 95
Table 4-4: Local cycle angle corrections on gate rotor ................................................. 97
Table 6-1: Air ideal gas properties..................................................................................... 128
Table 6-2: Basic compressor details ................................................................................. 129
Table 6-3: Air duty ................................................................................................................... 130
Table 6-4: Model parameters for thermal analysis ..................................................... 130
Table 6-5: Temperatures at specific locations .............................................................. 132
Table 6-6: Temperatures with different lobe combinations ................................... 134
Table 6-7: Temperatures with different wrap angles ................................................ 136
Table 6-8: Temperatures with different volume index ............................................. 139
Table 6-9: DRUM127 instrumentation ............................................................................. 142
Table 6-10: Modelled clearance adjustments ................................................................ 143
Table 6-11: Basic compressor details ............................................................................... 149
Table 6-12: Air test for contract duty ............................................................................... 149
Table 6-13: Air test for high temperature ....................................................................... 149
Table 6-14: WRVTA510 instrumentation ....................................................................... 150
Table 6-15: Average fluid boundary temperature across outlet plane ............... 151
Table 6-16: Key Clearance Locations ................................................................................ 154
Table 6-17: Performance penalty with revised interlobe clearance .................... 156
Table 6-18: Basic compressor details ............................................................................... 160
Table 6-19: Air test duty ........................................................................................................ 161
Table 6-20: HS204 instrumentation ................................................................................. 163
Table 6-21: HS204 additional hardware ......................................................................... 163
Table 6-22: HS204 performance test results ................................................................. 166
Table 6-23: HS204 interpolated performance test results....................................... 166
Table 6-24: HS204 model performance results with unmodified clearances .. 168
Table 6-25: HS204 model performance results with rotor thermal distortion 170
xv
Table 6-26: HS204 model performance results with rotor and casing thermal
distortion ...................................................................................................................................... 172
Table 6-27: HS204 performance test results at 6000rpm with and without
jacket cooling .............................................................................................................................. 173
Table 6-28: HS204 interpolated performance test results at 6000rpm with and
without jacket cooling ............................................................................................................. 173
Table 6-29: HS204 ‘MODEL R’ performance results at 6000rpm with rotor
thermal expansion .................................................................................................................... 175
Table 6-30: HS204 ‘MODEL RC’ performance results at 6000rpm with rotor and
casing thermal expansion ...................................................................................................... 176
xvi
Acknowledgements
This work was made possible through collaboration between my employer -
Howden Compressors in Glasgow, and the Centre for Positive Displacement
Compressor Technology at City, University of London; where I enrolled as an
external student in 2010. Thanks must go to the management team in Howden
for the continuing support of this work in light of ongoing business pressures.
Thank you to Prof. Ahmed Kovacevic, Jim Fairbairn and Graeme Cook, among
others, for having the vision to initiate this work. I am indebted to many other
colleagues who have been supportive of this work over the years. Dr. Manoj
Heiyantuduwa and Dr. Maria Wilson have provided valuable advice about their
own postgraduate study experience and are great examples to me of how PhD
graduates can play a valuable leading role in a commercial research
environment.
Regularly visiting and working with the staff and students of the Compressor
Centre has been a fantastic opportunity - thank you to City, University of
London and all those involved in making my role as an external student
possible. It’s been a privilege to share in the vast knowledge and experience of
the Compressor Centre. My first supervisor – Prof. Ahmed Kovacevic - has been
instrumental in this work and has always generously dedicated time and energy
to helping me. I’m grateful to have received both the patient encouragement I
needed and the less patient encouragement when required. My meetings with
my second supervisor – Prof. Nikola Stosic – were less frequent but always
inspiring and have had a significant influence on the direction of this work.
Thanks must also go to Professor Ian Smith for kindly proof reading and
suggesting improvements.
Finally, thank you to my family who have helped get me to this point,
particularly my wife Kirsteen who has supported me throughout my studies
while we also embarked on other adventures together including finding a home,
planning a wedding, and most recently, bringing up our daughter, Erin.
David Buckney
Glasgow, August 2017
xvii
Declaration
I confirm that this work is my own except where indicated by reference in the
text.
I agree for this work to be submitted to the Institutional Repository of City,

University of London for use in accordance with the Thesis Deposit Agreement.
David Buckney
Glasgow, August 2017
xviii
Abstract
Although the performance of twin screw compressors is heavily dependent on
the rotor clearances within them, chamber models, used as design aids, allow
for the specification of their magnitude and distribution but do not account for
how these may vary during operation, as a result of internal temperature
changes caused by the compression process.
A validated procedure has therefore been developed to enable a chamber model

to predict compressor performance, while including the effects of rotor and
casing distortion resulting from dependant thermal effects. This has been
achieved by the use of surface boundary mapping to calculate the rotor and
casing temperature exposure within the compression chamber resulting from
initial performance estimates. These detailed temperature distributions are
processed analytically using appropriate assumptions that allow calculation of
component temperatures and thermal growth.
A program for calculation of leakage area curves has been adapted to support
locally calculated variations in clearances. These updated area curves can then
be fed back into the chamber model in an iterative procedure to simulate
performance with thermally distorted clearances.
The inclusion of thermal clearance corrections generally improved the accuracy

of the chamber model when predictions from it were compared with test results
over a wide range of operating pressures and temperatures.
Furthermore, this work was found to be applicable in the evaluation of the

interlobe clearance distribution between the rotors. Predicting clearance
distortions and likely areas of rotor to rotor contact at a particular operating
duty allows clearances to be optimised for the correct balance between
performance and reliability; the results thus obtained were supported by
findings from available test and tear down results.
xix
Notation
Symbol Units Description
A [m] distance between rotor axes
AI [m2] interlobe leakage area
AR [m2] radial leakage area
C [-] rotor chamber identifier
D [m] rotor diameter
GA [m] axial gap
GI [m] local interlobe gap
GI [-] interlobe boundary point
GR [m] local radial gap
GR [-] radial boundary point
H [m] rotor addendum
h [kJ kg-1] specific enthalpy
h [m] lead length
L [m] rotor length
l [m] distance from suction plane
lI [m] interlobe sealing line length
ṁ [kg s-1] mass flow rate
N [rpm] main rotor speed
n [-] integer / point number
p [Bara] pressure
P [-] point on rotor or casing surface
Q [kW] heat
r [m] radius / polar co-ordinate
rc [m] radius to casing bore
ro [m] max radius (at outer diameter)

rw [m] pitch radius
xx
S [-] co-ordinate system
s [-] surface parameter along transverse rotor curve
Ta [°C] ambient temperature
TB [°C] time averaged local temperature on rotor bore
TCP [°C] averaged temperature over full casing plane
TG [°C] gas temperature
T*R [°C] instantaneous local rotor boundary temperature
TR [°C] time averaged local rotor boundary temperature
TRP [°C] planar averaged rotor boundary temperature
U [kJ] internal energy
V1 [m3] chamber volume at closing of inlet port
V2 [m3] chamber volume at opening of outlet port
Vi [-] volume index – ratio V2 / V1
X, x [m] Cartesian co-ordinate
Y, y [m] Cartesian co-ordinate
Z, z [m] Cartesian co-ordinate
z [-] number of lobes (gear teeth) on rotor
αF [°] polar flank angle measured from radial line
β [°] local angle parameter
βsl [°] β at the start of cycle at the leading tip
βst [°] β at the start of cycle at the trailing tip
γ [°] main rotor wrap angle from suction plane
ε [-] normalized surface parameter along curve s
θ [°] cycle angle measured at main rotor
θl [°] minimum cycle angle for a given point on casing
θt [°] maximum cycle angle for a given point on casing
θM [°] meshing angle
θlocal [°] local cycle angle on rotor surface
λ [°] local transverse profile rotation
xxi
μC [°C-1] coefficient of thermal expansion for casing
μR [°C-1] coefficient of thermal expansion for rotors
φ [°] polar co-ordinate
φc [°] angle to casing cusp (intersection of casing bores)
φoff [°] main rotor lobe offset angle
φs [°] offset angle to start of cycle
φw [°] rotor wrap angle at length, L
φgate [°] gate tip angle at home position
ψ [°] transverse rotor surface normal measured on
rotor co-ordinate system
ψH [°] local helix angle
ψPA [°] pressure angle - transverse rotor surface normal
measured on global co-ordinate system
ω [rad s-1] rotational speed
Additional Subscripts
1 main rotor / co-ordinate system fixed to casing at main
rotor / inlet
2 gate rotor / co-ordinate system fixed to casing at gate
rotor / outlet
01, 02 co-ordinate systems fixed to rotors

T, x, y transverse, x, y components
in, out inflow, outflow
l, t leading, trailing
a, b, c misc. labels
Acronyms
1D 1-dimensional
xxii
2D 2-dimensional
3D 3-dimensional
BH blow hole
CAD computer aided design
CFD computational fluid dynamics
CMM co-ordinate measurement machine
FEA finite element analysis
HP high pressure
LP low pressure
OD outer diameter
SCORPATH screw compressor optimal rotor
profiling and thermodynamics
SL sealing line
SRM Svenska Rotor Maskiner
SYM symmetric profile
xxiii
Chapter 1: Introduction
Chapter 1
Introduction
1.1 MOTIVATION
Twin screw compressors are widely used in the refrigeration and gas
processing industries. They have increased in popularity over the decades and
have replaced reciprocating compressors in many applications (Ohama, Kurioka
et al. 2006). As a consequence of their widespread use, achieving modest gains
in performance can result in a significant reduction in global energy
consumption. One possible way to increase performance is to match the screw
compressor to the application, i.e. the properties of the working fluid and the
operating duty, by using uniquely optimised rotors as suggested by Singh et al.
(Singh, Onuschak 1984) and demonstrated by Stosic et al. (Stosic, Smith et al.
2003). This has not traditionally been attempted by screw compressor
manufacturers due to the specialised nature of designing the rotor profile
geometry and the expense and lead time associated with modifying or
procuring tooling to manufacture bespoke rotors. With the current availability
of modern flexible profile generation tools (Stosic, Hanjalic 1997) and advanced
manufacturing techniques (Holmes 2008), tailoring rotor designs for specific
applications is more feasible. However there are still considerable challenges
that need to be addressed in order to achieve this.
This work contributes to practical rotor optimisation for industrial applications

within the constraints of a commercial manufacturing environment where there
is a need to rationalise and limit the number of design variants engineered and
manufactured. To make optimisation of rotor profile viable, the procedure must
be reliable, fast and easy enough to implement for small batches of rotors. It
was found from a review of relevant literature and experience gained, while
working in industry, that existing rotor design and modelling tools are still
1
reliant on empirical observations, particularly in the prediction of absolute

performance results and the specification and analysis of reliable clearances.
Improving clearance analysis and design will allow an optimum balance

between performance and reliability to be achieved with minimum adaptation
of rotor tooling and manufacturing programs. While much work has been
published on the general design and optimisation of rotor profiles and twin
screw compressor configuration, more fundamental research in the area of
clearance analysis would be a useful contribution that would make the practical
application of rotor optimisation more achievable.
1.2 THE TWIN SCREW COMPRESSOR
1.2.1 BASIC PRINCIPLES

The twin screw compressor, though complex in design, essentially comprises of
only two principle moving parts. These are the two helical screw rotors which
are usually designed with integrated shafts for drive, bearings and seals. All
moving parts are in balanced rotating motion allowing high operational speeds
and power density.
Figure 1-1: Compressor operation
Figure 1-1 shows a representation of the rotors in mesh at various stages of the
compression cycle. These rotors are enclosed within a closely fitting casing,
resulting in the cavities that become the working chambers of the compressor.
As the rotors turn these cavities translate axially and change in volume due to
the helical form of the rotors. At the suction end new cavities are formed which
increase in volume. At the discharge end these cavities decrease in volume and
2
eventually disappear. The rotor on the right, with 4 lobes in this example, will
be referred to as the ‘main’ rotor and the rotor on the left, with 6 lobes, as the
‘gate’ rotor.
It is vital that clearances are maintained where close interaction between the
rotors and casing exists. As these clearances define part of the boundary for a
given chamber they are also referred to as the sealing lines. The clearances
along these sealing lines introduce leakage paths which need to be controlled
for efficient operation. The area of the leakage path depends on the length of the
sealing line and on the size of the clearance gap. Minimising the effect of leakage
paths is one of the main aims when designing twin screw compressors
(Fleming, Tang 1995).
1.2.2 PORTS AND INTERNAL COMPRESSION

A twin screw compressor does not require valve actuation for its operation
because the working chambers move past stationary ports, periodically
exposing each chamber to suction and discharge. In Figure 1-2 the axial and
radial parts of the discharge port are shown in blue and green respectively. The
chamber highlighted red has reached its smallest volume before it will be
exposed to the discharge ports. This represents the volume, V2, at the end of
compression. The suction port, situated on the opposite side from that shown,
will have closed when the maximum chamber volume is achieved to give the
volume, V1.
3
Radial Port
Axial Port Auxilliary Port
Figure 1-2: Meshing rotors highlighting single compression chamber
The volume index, or ‘Vi’, is the ratio of the inlet and outlet chamber volumes:
V1/V2. The compressor should be designed to match the internal and external
pressure increase so the target volume index is dependent on the compression
ratio and gas properties. Systems have been introduced that allow the size of
the ports to be altered, allowing a variable volume index. Ideally the port shapes
should be as large as possible to minimise flow velocity and consequent losses.
Depending on the casing design, the suction and discharge ports may only
consist of an axial or a radial opening.
1.2.3 CAPACITY CONTROL

In principle the capacity control of a rotary positive displacement compressor
can be very straightforward as the speed of the drive is proportional to the
volume delivered. However, due to the prevalence of fixed speed motors,
compressors commonly use slide valves (O'Neill 1977, SRM 1966) or poppet
valves which allow recirculation from the chamber back to suction, prior to
compression. As a consequence of the internal leakage being virtually
unchanged, higher temperatures will occur within the compressor at reduced
capacity, making this an important operating condition for the analysis of
clearances and reliability.
4
1.2.4 OIL INJECTION AND COMPRESSOR ARRANGEMENT
Figure 1-3: Oil free compressor
Twin screw compressors can be ‘oil injected’ or ‘oil free’ (O'Neill 1966). Figure
1-3 shows a sectional view of an oil free compressor highlighting some of the
main components in different colours. The radial and axial bearings are
highlighted in red; the seals are highlighted in yellow; and the timing gears are
highlighted in green. With oil injected compressors the gate rotor is typically
driven directly from the main rotor along a contact band on the rotors,
eliminating the need for expensive timing gears. Another advantage of oil
injected machines is that oil can drain directly into the compression chamber
from the bearings eliminating the need for internal mechanical seals. Oil
injected twin-screw compressors are capable of achieving higher compression
ratios due to the presence of liquid in the clearance gaps and cooling of the
compression gas by the liquid. However the viscous losses associated with oil
injection limit the operational speed and therefore the capacity compared to an
oil free compressor. Oil injected machines are dominant in refrigeration and air
compression whereas oil free machines are common in gas processing
applications in which contamination of the gas stream is not acceptable.
Oil injected compressors will typically operate at discharge temperatures less

than 100°C while oil free compressors may operate with discharge
5
temperatures in excess of 200°C. So the clearance requirements are quite

different for these machines. In addition to oil free compressors having larger
overall clearances, extra measures are sometimes used such as tapering the
rotors to introduce larger clearances at the hot outlet end, or introducing liquid
cooling passages in the casings and/or rotor bodies. As oil injected compressors
feature direct drive via rotor to rotor contact the rotor to rotor clearances must
be designed in such a way that contact only occurs at a specific area where the
relative sliding motion between the rotor surfaces is small (Stosic, Smith et al.
2005).
1.3 DESIGN AND OPTIMISATION
1.3.1 INDUSTRIAL APPLICATIONS

Screw compressors have been adopted for a wide range of industrial
applications including:
 Refrigeration / Air-conditioning industries (cold stores, freezing)

 Extractive industries (mining, oil and gas)
 Process industries (chemical, petroleum refining, refrigeration, cement)
 Power generation (steam recovery and compression system, gas booster
for gas turbine)
 Manufacturing industries (pneumatic applications)
These applications cover a wide range of operating duties in terms of: capacity;
suction pressure and temperature; and pressure ratio. Typical process gases
handled (Howden Compressors Ltd. 2008) include: Carbon dioxide; Helium;
Hydrogen; Ammonia; Butane; Chlorine; Coke oven gas; Sour hydrocarbons;
Ethane; Propane; Town gas; Natural gas; Steam; and Refrigerants. Different
operating fluids can result in significantly different discharge temperatures and
internal leakages depending on their specific heat ratio and molecular weight.
6
Figure 1-4: Howden WRV range (courtesy of Howden Compressors Ltd.)
In order to meet the demand of different markets, a number of compressor

ranges have been developed over the years (O'Neill 1977). Figure 1-4 shows the
Howden WRV range of oil injected compressors which are popular in the
petrochemical and refrigeration industries. Generally, standard parts are used
as far as possible for each size; which includes standard rotor designs. Some
rotor options are available, for example in the Howden ‘HS’ oil-free range, the
standard temperature limitations can be extended by introducing rotor cooling
and rotor tapering. In this range, rotor cut-back is an option used to adjust the
displacement capacity of the rotors for a given speed. Similarly on the ‘WRV’
range, some rotors with the same diameter but different lengths use the same
rotor tooling, effectively resulting in a cut-back or extension of the same rotor
design. These measures help to limit the need for more tooling and parts.
Using a standard rotor profile for a range of compressors is a far cry from early
predictions (Singh, Onuschak 1984) and more recent examples in the literature
of rotor profile optimisation (Stosic, Smith et al. 2003). This work initially set
out to apply well documented rotor profile optimisation methods to small
batches of industrial compressors in a commercially viable way. Research based
on this remit identified some practical and design challenges without clear
solutions, namely, precisely how to design optimum clearances that safely
accommodate duty dependent distortions during operation and how to capture
the impact of clearance distortions during modelling. Tackling these challenges
7
led to a research niche that focuses specifically on clearance design and

evaluation rather than overall optimisation; this is explained in more detail in
the following sections.
1.3.2 OPTIMISATION
A thorough overview of screw compressor optimisation was presented by
Stosic (Stosic, Smith et al. 2003). This describes target functions; identification
of suitable optimisation parameters and constraints; and application of a
method to find local minima. Using a number of case studies, Stosic shows that
different rotor designs are optimal for different applications, this depends on
whether they are oil-injected or oil-free and on the working fluid and operating
conditions. Other optimisation methods have also been demonstrated to be
successful (Stosic 2005, Wu, Fong 2009, Hauser, Brummer et al. 2008). The
advent of accurate thread grinding (Holmes 2008) brings flexibility, as changes
can more readily be made to the form of the tool, making the possibility of
optimising smaller batches of rotors more realistic and economical.
The design and optimisation of twin screw compressors can be broken down
into three areas, namely: ‘General Design’, ‘Profile Design’ and ‘Clearance
Design’. The group of design variables that fall within each of these areas are
often constrained together such that different levels of optimisation can be
broadly described as shown in Table 1-1. These different areas will be
elaborated on in the following sections.
Table 1-1: Commonly grouped constraints
Variable Group Optimisation Level
General Design Constrained Constrained
Profile Design Constrained 3

2
Clearance Design 1
8
1.3.3 GENERAL DESIGN

‘General Design’ relates to the basic geometric proportions of a twin screw
compressor; whether it is oil injected or oil free; and other features such as the
type of bearings. Basic geometric parameters include the rotor diameter (D) for
a given rotor centre distance; the rotor lobe gear ratio (z1/z2); the wrap angle
describing the helix of the main rotor (φw); and the ratio of rotor length over
diameter (L/D).
a) 3/5 lobe combination, b) 5/7 lobe combination,

L/D = 1.65, L/D = 1.10,
wrap angle = 200° wrap angle = 300°
Figure 1-5: Comparison of rotors with different basic geometry parameters
A small lobe number combination, such as on the 3/5 lobe rotors shown in
Figure 1-5a, results in a compressor well suited to high flow applications. This
type of rotor geometry is often seen in applications such as engine
superchargers where the inlet air is ‘boosted’ to increase mass flow with only a
modest pressure increase.
The 5/7 lobe rotors shown in Figure 1-5b result in more working chambers but
the compressor will have a smaller net capacity for a given rotor diameter. This
larger number of lobes is well suited to higher pressure applications as the
differential pressure between working chambers will be smaller. The rotors in
Figure 1-5b also feature a smaller L/D ratio which when combined with the
larger root diameter for the 5/7 lobe combination, results in very rigid rotors
that can accommodate large bearings for high pressure operation.
9
Increasing the wrap angle of the rotors increases the ports areas and decreases
the rate of filling and discharge due to a slightly longer cycle time. These factors
both help reduce dynamic losses. However, the slower cycle time that reduces
the flow rate in the ports also has the negative consequence of increasing the
time over which leakage can occur.
Figure 1-6: SRM standard rotor geometry. Equal diameters; 4/6 lobes; L/D = 1.65;
wrap angle = 300°.
The need to optimise the general rotor arrangement is most applicable when
designing a new range of compressors. When retrofitting rotors or designing
rotors for a range of standard machines with finite casing geometry variations,
it is common for the general rotor arrangement to be mostly constrained.
Figure 1-6 shows the commonly used general rotor parameters used in the
most popular compressor ranges in Howden Compressors. This design was
based on experimental work and reports produced by SRM (SRM 1952) and
provides good performance over a wide range of duties.
1.3.4 ROTOR PROFILE DESIGN

‘Rotor Profile Design’ relates to the geometry of the actual rotor lobes; referred
to as the ‘rotor profile’. This profile is usually defined on the transverse cross
section for a single rotor lobe. An example profile is given in Figure 1-7.
10
Figure 1-7: Rotor profile on transverse view
In order to maintain sealing, the rotor pair must be conjugate, resulting in

continuous ‘line contact’ between the rotor surfaces (Litvin, Fuentes 2004). To
meet this condition, the relative velocity between two rotor surfaces at the
moment of ‘contact’ must be tangential to the rotor surfaces. Applying the
conjugate condition it is possible to calculate any of the transverse segments
from one of the others. The transverse rack segment is calculated as a rotor
with an infinite number of lobes, or infinite rolling ‘pitch’ radius. This is useful
as it is common for the main and gate rotor pair.
The rotor profile is defined using a rotor generation procedure that is usually
patented; if publically disclosed. Rotor generation procedures commonly
describe the profile on either the rotor or rack using various explicit curves
with known derivatives that allow analytical calculation of the profile
conjugates. Each rotor generation procedure has its own unique set of
parameters that define the component curves of the profile. Modern generation
procedures such as ‘N’ profile generation (Stosic, Hanjalic 1997, Stosic 2001),
used in this research, provide a high degree of flexibility to the profile design
while generating robust, reliable profiles. Rotor generation procedures will be
discussed in more detail in the next chapter.
11
1.3.5 CLEARANCE DESIGN

‘Clearance Design’ does not, typically, come under the umbrella of rotor
optimisation in the literature but if altering rotor designs for significantly
different operating duties then this is a critical part of the design process that
cannot be overlooked. The location and type of each clearance gap is
highlighted in Figure 1-8. These are the rotor to rotor ‘interlobe’ gap, the rotor
to casing ‘radial’ gap, and the rotor to casing ‘axial’ gap. The optimisation of
clearance gaps involves finding an acceptable balance between performance
and reliability for a given operating duty.
Rotor to casing – axial gap
Rotor to rotor – interlobe gap
Rotor to casing – radial gap
Figure 1-8: Rotor clearance gaps
A considerable challenge in this process is that it is difficult to determine what

the actual clearances are. Clearances specified on paper will not reflect the
clearances achieved once all component parts of the compressor are machined
and assembled. Furthermore, clearances measured after assembly will not
reflect the operational clearances which are subject to complex pressure and
temperature distributions. These issues and others related to clearance design
will be discussed in the following chapter. Later in this chapter the section
‘Preliminary Studies’ will introduce in more detail how clearances impact on the
performance.
12
1.3.6 DESIGN SOFTWARE

Specialised software called SCORPATH, developed at City University, was used
during this research. This software allows the user to generate new rotor
profiles with full control over all of the rotor design parameters previously
discussed. Critically, the software includes a simulation model allowing the
performance of the newly designed rotor to be evaluated. This simulation
model is detailed in the book: ‘Screw Compressors, Mathematical Modelling and
Performance Calculation’ (Stosic, Smith et al. 2005) and will be discussed in
more detail in the next chapter.
1.4 PRELIMINARY STUDIES
1.4.1 PERFORMANCE SENSITIVITY TO CLEARANCE

Reviewing clearance designs of previously tested compressors is a valuable
input to designing new clearances. Mechanical testing is a requirement for all
new compressors to ensure reliability therefore there is a growing bank of
empirical data. Testing is unfortunately an expensive and time consuming
process and, in an industrial manufacturing environment, reliability is
prioritised; – often the intended operating duty might be well below the
maximum design limit and may lead to non-optimal clearances.
13
Volume Flowrate Comparison

speed = 3500 rpm, p1 = 1.013 Bara
200
190
180
Suction Flow (m^3/min)
170
160 N_T1: original

clearances
150
140 N_T2: revised

clearances
130
Tested Extrapolated
120
110
100
1.0 2.0 3.0 4.0 5.0
Pressure Ratio
Figure 1-9: Rotor retrofit test results with different clearances
Figure 1-9 shows results from a Howden HS408165, oil free compressor. The
compressor was retrofitted with newly designed N profile rotors (N_T1); these
results are shown in red. Unfortunately, this rotor pair encountered serious
rotor contact issues and testing was stopped. After conservatively relaxing the
clearances, a second rotor pair was manufactured and tested (N_T2) with
results shown in green. This performed well in terms of reliability however the
flow was impaired as shown.
1.4.2 CLEARANCE INFLUENCES AND BEHAVIOUR

Due to the clearance sensitivities that relate to manufacturing and assembly
tolerance there is always likely to be some degree of error between a simulation
model based on ideal design clearances and actual test results. Stosic et al.
(STOSIC, SMITH et al. 2003) highlighted the relative importance of the
compressor housing and bearings on the interlobe clearances and presents a
mathematical analysis to quantify potential clearance reductions. Assembled
clearance deviations can be assessed with thorough examination of all
compressor components as shown in other work by Sauls et al. (Sauls 1996).
14
However, not all clearance sensitivities are unique to a particular compressor

build - it is possible to broadly distinguish clearance sensitivities as follows:
Duty INDEPENDENT clearance sensitivities:
• Manufacturing tolerance
• Assembly tolerance
• Bearing clearances
Duty DEPENDENT clearance sensitivities
• Pressure distortions
• Thermal distortions
Some operational clearance distortions can be considered to be dependent on

the operating duty and should in theory behave in a systematic and repeatable
manner for a given compressor design.
Figure 1-10: Location of proximity probe
This is supported by test results where the radial gap between the rotor and the
casing was directly measured using a proximity probe at the location shown in
Figure 1-10. These test results are later utilised in the last case study in chapter
15
6 where more detail will be provided. Figure 1-11 shows the measured radial
gap plotted against the ‘temperature increase’ across the compressor; i.e. the
difference between the discharge temperature and the suction temperature
where in this case the suction temperature is the same as ambient. Different
temperatures were achieved by testing the compressor over a range of
compression ratios. The nominal gap size in this case is 0.150mm. The radial
clearance can be seen to reduce at higher temperatures. This is only a limited
sample of data but it does show a fairly linear relationship between gap
distortion and operating temperature.
Measured Radial Gap

at Various Operating Temperatures
150
125
y = -0.6957x + 145.9
Radial Gap (μm)
R² = 0.9782
100
75
50
25
0
0 25 50 75 100 125
Temperature Increase (°C)
TEST Linear (TEST)
Figure 1-11: Measured relationship between operating temperature and

clearance reduction
Thermal distortions have been analysed in detail in the literature by Sauls et al.
(Sauls, Powell et al. 2007) and by Kovacevic et al. (Kovacevic, Stosic et al.
2002a) who showed that this is the dominant cause of operational deflections
for oil free compressors. Assuming a simplified 2D cross-section with uniform
rotor and casing temperatures, it is easy to visualise how clearances are
affected. The left hand side of Figure 1-12 represents the effect of rotor thermal
growth assuming no change in the casing geometry, including the rotor centre
distance, A. With uniform heating of the rotors, the measured gap ‘GapR’ will
16
decrease in a very predictable manner. The change in the rotor overlap from H
to H’ represents the reduction in the horizontal, transverse component of the
interlobe clearance.
A A’
A
dGapR
H
dGapR
H’
Figure 1-12: How rotor thermal growth and casing thermal growth affect
clearances
The right hand side of Figure 1-12 shows the scenario of the casing thermal
expansion without any deformations of the rotors. In this case the bearing
centres will move apart, due to thermal expansion of the casing that houses the
bearings. Due to this increase in centre distance from A to A’ the interlobe
clearance gap will increase. Additionally, the casing thermal growth will
increase the measured radial gap ‘GapR’.
In short, heating of the rotors acts to decrease clearances and heating of the
casing acts to increase clearances. With a uniform temperature increase over
the entire compressor where the rotors and casing are at the same temperature,
the change in clearance gaps would be negligible since all dimensions would
scale uniformly.
With this simplified model of clearance behaviour it is easy to imagine how

operational clearance adjustments could potentially be integrated into a
performance simulation model. The considerable difficulty lies in obtaining
realistic rotor and casing temperature distributions and distortions.
17
1.5 CONCLUDING REMARKS

Full optimisation of twin screw compressors for specific applications would
ideally involve a multivariable optimisation of every parameter that affects
performance, however, there needs to be some rationalisation of the number of
design variables. Modification of clearances requires minimum alteration to
manufacturing setup and tools so in a commercial manufacturing environment
this aspect of profile design and optimisation has the most potential to provide
significant benefits for the smallest cost.
In most commonly used compressor simulation models (Hanjalic, Stosic 1997,

Fleming, Tang et al. 1998b, Fujiwara, Kasuya et al. 1984, Sangfors 1984),
operating clearances are treated as an input parameter that must be assigned
by the design engineer. The effect of clearance distortions at elevated
temperature is shown in Figure 1-11. Changing clearances has a considerable
effect on compressor performance as observed in the results of Figure 1-9 and
as reported in the work of Fleming et al. (Fleming, Tang 1995). It follows that if
operational clearance deviations can realistically be represented in a
compressor simulation model, then the model will predict performance more
realistically over a wider range of operating conditions, without the need to
revise the clearance input parameters manually. Such a model would also be
invaluable in addressing the necessity to calculate optimal duty specific
clearances accurately.
Further research that contributes towards the goal of designing rotor

clearances for optimum balance between performance and reliability is
required and would be a valuable contribution to the field of twin screw
compressors. The sensitivity of the clearances to the operating duty and the
consequent effect on performance and reliability is a critical factor in any rotor
optimisation and is something that is not well addressed in available chamber
models. An important question that this research aims to address is how far
commonly used chamber models can reasonably be extended to predict and
correct for duty dependant thermal effects.
18
1.6 THESIS OVERVIEW

This chapter covers the motivation behind this research, introduces the reader
to some twin screw compressor basics and to the concept of optimising the
compressor rotors for specific applications. In summary this chapter discusses
the importance of clearance analysis and design to the rotor optimisation
process and highlights this as an area that would benefit from additional
research.
Chapter 2: Background Theory and Literature Review
This chapter draws on literature to set out details of twin screw compressor
rotor geometry generation and analysis that are important foundations of this
research. Some of the established geometry calculations used will be described
here or referred to in appendices. This chapter continues to discuss established
compressor models that will be used. Limitations in applying these current
tools to clearance analysis are discussed.
Chapter 3: Research Objectives, Methods and Expected Contribution
Within this chapter the research goals are set out along with specific details of
the expected contribution.
Chapter 4: Boundary Map for Rotor and Casing Surfaces
This chapter describes work to map the fluid properties from a chamber model
to the rotor and casing geometry.
Chapter 5: Operational Clearance Distortions
Building on Chapter 4, it is described how approximated rotor and casing

temperature distributions can be used to calculate possible operational
clearance distortions. Variations in the local clearances are fed into the
geometry calculation procedure in order to recalculate leakage areas used by
the chamber model. In this way it is possible to calculate the effect of
operational clearances on performance iteratively.
19
Chapter 6: Case Studies
This chapter presents case studies in which the methods outlined in Chapters 4
and 5 can be applied to the design and analysis of clearances for oil free and oil
injected compressors.
Chapter 7: Conclusions and Future Work
Finally, this chapter will conclude this thesis and identify any future work.
20
Chapter 2: Literature Review
Chapter 2
Background Theory and
Literature Review
2.1 INTRODUCTION
This chapter draws on available literature to: describe the work that the current
research is based on, introducing relevant theory as necessary; and clarify
exactly where new contributions are required. The generation of rotor profiles
is a necessary starting point for this research so this will be discussed first. This
is followed by a review of how the clearances are designed. Once the rotor
geometry has been generated the task of fully evaluating the geometry
characteristics of the assembled rotor and casing is considerable. So this is also
covered. Modelling of twin screw compressors is then reviewed; considering
both performance prediction and thermal analysis methods.
2.2 ROTOR DESIGN
2.2.1 PROFILE GENERATION

Figure 2-1 introduces some basic terminology used to describe rotor profiles.
The direction of rotation shown by the arrows is for compressor operation.
Assuming compressor operation the leading and trailing flanks of each
protruding lobe can be identified. The leading main flank meshes with the
trailing gate flank and vice versa so this terminology is not always ideal. The
flanks can also be referred to as the round flank or the straight flank as labelled.
The tangent circles projected on the main and gate rotors represent the pitch
circles; where rolling contact would occur.
21
Figure 2-1: Transverse profiles showing terminology
The development of rotor profile generation procedures for screw compressors

has been well documented (Stosic, Smith et al. 2005, Stosic, Smith et al. 2010).
The following is a summary of some of the key developments. Figure 2-2 shows
the end on view of two types of rotors which are significantly different in terms
of lobe width and shape. The rotors in the top part of Figure 2-2 are a
symmetric ‘circular’ profile which was used in the first manufactured screw
compressors; note that the profile lobes are symmetrical and both the leading
and trailing flanks of the main rotor are circular in shape. The rotors in the
lower part of Figure 2-2 have a modern asymmetric rotor profile and have
distinguishable round and straight flanks.
22
Figure 2-2- Photo comparing symmetric and N-rotor profile rotors
This asymmetry is one of the most important developments in screw

compressors and was first introduced by SRM in the ‘A’ Profile (Schibbye 1979).
In order to create the straight flanks, curves were generated by an undercutting
action so that single points on the main and gate rotors trace the curves on the
opposite rotors. It is this undercutting that allows the degree of asymmetry
required for the straight flank on the gate lobe to curve forward in the direction
of rotation and meet the tip of the main rotor closer to the cusp of the casing,
considerably reducing the blowhole area whilst ensuring continuous ‘contact’
between the rotors. The blowhole will be explained in more detail in the
following sections. Since the symmetric ‘circular’ and asymmetric ‘A’ profile,
many new designs have been patented by SRM and others; they all now use
asymmetric profiles.
The enveloping theory used to calculate conjugate rotors is explained by Stosic

et al. (Stosic 1998). Calculating the conjugate section of the generated profile
directly from the generating profile is most common. However, different
23
approaches have been proposed including the ‘transverse rack generation

method’, the ‘normal rack generation method’ (Wu 1995), and ‘meshing line
generation’ (Zaytsev, Infante Ferreira 2005). An early version of the transverse
rack generation method described by Rinder (Rinder 1987) highlighted the
advantages of rack generation. The use of straight lines on the transverse rack
in the pitch line area results in involute curves on the rotors with constant
pressure angle and excellent force transmission. The limitation of Rinder’s rack
was that the range of lines and circles used to define the rack resulted in a large
blow hole area.
Stosic’s N-profile improved the rack definition of the straight flank, and thus
reduced the blowhole area. This was achieved by generating the straight
portion of the rack from two small radii defined on the main and gate rotors
(Stosic 2001), similar to the SRM ‘D’ profile (Astberg 1984). The use of small
fillet radii rather than singular points is beneficial when designing a practical
profile that requires carefully controlled clearances. The N-profile rotors in
Figure 2-3 show a meshing rotor pair on the right and on the left the transverse
rack from which they were generated. The letters identify the extents of the
individual curve segments used to define the profile.
Figure 2-3: ‘N’ rotor rack generation
24
There are other interesting rack generation methods that have been proposed
such as using a number of discrete points connected using a spline (Sundt
1997) rather than the traditionally used set of explicit curves. Furthermore, it is
possible to derive meshing profiles using the Boolean operations built into
modern CAD packages (Stosic, Mujic et al. 2008) without the need to calculate
the meshing conditions. These types of approach are likely to gain more favour
among profile designers, due to the freedom they bring, providing numerical
errors are within acceptable limits.
It is important to understand that using the same profile generation method

will not always result in the same profile shape; this is because modern
methods allow several profile design parameters to be altered. For example, the
rotor segment BC defined in Figure 2-3 is defined by the N profile generation
procedure to be a straight line that results in involute curves on both rotors
however the angle of this line is a parameter that can be readily adjusted for a
specific N profile.
The patented ‘N’ profile procedure (Stosic 2001, Stosic, Hanjalic 1997) draws
together the strengths of previous profiles: such as the asymmetry of the A and
D profiles; and the involute curves at the pitch circle used in the Rinder rack;
while bringing new benefits, due to the novel rack generation method which
provides far more flexibility in the form of the remaining curves. This
generation method is accessible to rotor design engineers as part of a
commercially available design suite (Mujic, Kovacevic et al. 2010) and was used
for all new rotors designed during this research.
2.2.2 CLEARANCE DESIGN

Modern manufacturing methods allow rotor profiles to be machined to a
tolerance of 5 microns (Kovacevic, Stosic et al. 2002a) but consideration must
be given to manufacturing and assembly errors of all other compressor
elements (Stosic, Smith et al. 2003) and particularly to the operating
deformations which are generally larger than the manufacturing tolerances
(Sauls, Powell et al. 2007). As well as defining the size of the clearance gap it
must be specified how the clearances are to be applied between the
25
components. The radial and axial clearances that occur between the rotors and
the casing are fairly straightforward to implement; for example by designing the
rotors to a slightly smaller diameter than the casing bore diameter. The
interlobe clearance is introduced by removing material from one or both of the
generated profiles. Fleming (Tang, Fleming 1994) investigated suitable
practises to introduce interlobe clearances with tooling offsets however
modern rotor design tools such as for the ‘N’ profile generation allow a high
degree of control over the clearance distribution when designing the profiles.
Clearance distributions usually vary around the profile and can be significantly
different in local areas. Figure 2-4 shows a transverse rack view with clearances
mapped onto it. Clearances are represented by magnified vectors
perpendicular to the rotor rack curve. The point numbers, e.g. ‘#01’, on this
figure represent inspection points where the designed clearances may have
different values. The number that follows is the normal clearance gap
(measured perpendicular to the surface) in microns, e.g. ‘+90’. In this figure the
clearances are smaller on the round flank, on the left hand side, because this is
the side where contact normally occurs. The larger clearances on the straight
flank, on the right hand side, allow for rotor ‘backlash’ which is a necessary
design feature of any practical gear system.
26
Figure 2-4- Clearance distribution on a rack projection
In one sense the clearances in oil injected machines are easier to manage than
in oil free machines, because the temperatures are usually kept within much
lower limits. However the lack of timing gears and the direct rotor to rotor
drive adds more complexity to clearance analysis. Rotor contact is necessary for
one rotor to drive the other; the objective is to control precisely where contact
is to occur. Best practice for direct drive clearance design ensures only rolling
contact at the pitch radius of each rotor as presented by Stosic et al (Stosic,
Smith et al. 2005). Deviations from nominal design clearances, whether due to
manufacturing and assembly variations or due to operational distortions, can
result in a shift in the relative rotation between the main and gate rotors (Sauls,
Powell et al. 2007): this further distorts the interlobe clearance distribution and
must be considered in clearance evaluation for direct drive, oil injected
compressors.
The discontinuities at each side, on the top and at the bottom of the clearance
distribution in Figure 2-4, are introduced by ‘tip seals’, or ‘wear strips’ located
at the outer diameter of each rotor. These narrow bands of protruding metal are
a compromise that slightly increases the total interlobe leakage area but they
27
allow the radial clearance gap to be kept to a minimum without compromising

reliability. It was claimed in Fleming et al (Fleming, You et al. 1994) that, if
correctly designed, sealing strips can even reduce the viscous drag between the
rotors and casing in oil injected compressors.
On the topic of clearance management, Stosic (Stosic, Smith et al. 2005)

describes a sensible approach wherein the designed clearance distribution
during operation should be such that rotor contact is not possible in any
location that would lead to damage. Therefore the design clearance distribution
is the target operating clearance distribution with additional allowance built in
for thermal expansion. The optimal clearance design is therefore dependent on
the operating temperature. More specifically it is dependent on the thermal
distribution and distortion of the compressor components which are
considerably more difficult to predict. Designing clearance allowance for
operational distortions requires the temperature distribution within the
compressor to be calculated. Examples of using numerical methods to calculate
temperature distribution on the rotors and casing and thus the change in
clearances have been given by Kovacevic et al (Kovacevic, Stosic et al. 2002a)
and by Sauls et al (Sauls, Powell et al. 2006b). The former uses Computational
Continuum Mechanics to allow the solution of both the solid and fluid domains
on a 3D numerical mesh of the actual compressor geometry while the latter
uses temperature results from a chamber model that are mapped to a 3D finite
element grid. Both approaches showed good correlation when compared with
experimental data.
2.3 GEOMETRIC CHARACTERISTICS
2.3.1 IDENTIFICATION OF GEOMETRIC CHARACTERISTICS

Compressor models used for the calculation of screw compressor
thermodynamic and fluid flow processes, from non-dimensional
thermodynamic chamber models (Singh, Onuschak 1984, Hanjalic, Stosic 1997,
Seshaiah, Ghosh et al. 2007, Stosic, Hanjalic et al. 1986, Fujiwara, Kasuya et al.
1984, Sauls 1996) to three-dimensional computational fluid dynamics (CFD)
28
procedures (Kovacevic, Stosic et al. 2006, Kovacevic, Stosic et al. 2002b),

require accurate representation of the rotor geometry. With 3D modelling
methods, that represent the full fluid domain, the effect of the geometry on the
thermodynamic result is inherent however when working with non-
dimensional models, identification and quantification of geometric parameters
such as the chamber volume, port areas and leakage areas is needed. Such
parameters vary throughout the cycle and it is necessary to know the
instantaneous value and gradient of each one.
As the helical rotors rotate within the compressor casing the actual volume of
trapped gas within any one of the compressor chambers is continuously
changing in shape, size and position. Each compression chamber is comprised
of one main rotor and one gate rotor flute. These are the individual channels
threading their way around the rotors. As the compression volume moves
through the compressor it interacts with fixed ports (refer back to Figure 1-2)
in the casing, resulting in port areas that continually vary and exist for only a
certain portion of the cycle. Due to the necessity for clearances, leakage paths
also exist at the sealing boundaries. Again these are continuously changing in
shape, size and position throughout the compression cycle. Calculating these
geometric characteristics of the twin screw machine, that are relevant to its
performance, can become very complex depending on the level of realism and
accuracy required.
29
Figure 2-5: Defining discrete leakage paths
Accurate presentation of screw compressor geometry was given in the

comprehensive screw compressor textbook by Sakun (Sakun 1960) and the
screw compressor handbook by Amosov (Amosov 1977), both published in
Russian. Rinder (Rinder 1979) presented a similar work in German. SRM, who
were the first to license twin screw compressor technology have published
numerous reports detailing how to analyse twin screw compressor geometry
(SRM 1954, SRM 1953) however this material is still only available to licensees.
Singh and Onuschak (Singh, Onuschak 1984) provide an overview of the main
screw compressor geometric characteristics. Further work was published by
Singh et al describing in more detail the methods used to calculate the
geometric parameters (Singh 1990, Singh, Bowman 1990). Fleming and Tang
(Fleming, Tang 1995) presented a model which described 6 distinct leakage
paths – the interlobe gap; the radial gap; the discharge face axial gap; the
suction face axial gap; the high pressure (HP) blowhole; and the low pressure
(LP) blowhole; these discrete leakage paths are illustrated in Figure 2-5. Other
geometric factors that can be considered include injection ports and part load
re-circulation passages. Detailed breakdown of the flow areas can vary but the
principle of describing volumes and areas as a function of the cycle is the same.
Hauser et al (Hauser, Brummer 2010) demonstrated a procedure whereby
30
abstracted geometric parameters could be directly used for efficient

optimisation of profile thermodynamic performance.
Of the leakage paths shown in Figure 2-5, the radial gap and interlobe gap are
directly related to the design clearances. The discharge end face gap - also called
the axial gap, is an assembly feature while the blow-hole area is an inherent
feature of the rotor profile geometry. Not all leakages paths are present for the
entire duration of a compression cycle and the size and shape of the leakage
path will vary. Defining each of the leakage areas as a function of the cycle angle
sufficiently captures the most important leakage characteristics in a very
efficient manner. This efficiency comes at the cost of losing details such as how
the area evolves along the path of the leakage flow. When describing leakage
areas with non-dimensional area curves the simple objective is to calculate the
minimum instantaneous area along each flow path as accurately as possible.
Radial Leakage Blow-hole Leakage
Interlobe Leakage
Figure 2-6: Leakage flow from chamber volume
The radial leakage path for outflow can be visualised as the flow along the
curved arrows shown in Figure 2-6. For a given rotor position the length of this
sealing line is easily calculated as it follows two helical paths that terminate at
31
the casing cusp where the main and the gate rotors meet, or at one of the axial
end faces. If the radial clearance gap is constant, then the leakage area is the
product of the sealing line length and the clearance gap. This simple
representation of the clearance gap is common and has shown to be sufficiently
accurate for many models in predicting the general performance metrics of
compressors.
Adding a little more sophistication by allowing the local clearance gap to be

varied along the length of the sealing line opens up the potential for
investigation such as by Wu (Hsiao, Wu et al. 2012) who assessed how different
clearance distributions along the interlobe sealing line affected performance.
This approach also allows for investigation of how local clearances are affected
by relative movements of the rotors and casing, for example, due to the relative
rotation of the rotors. This was discussed by Stosic (Stosic, Smith et al. 2005)
who highlighted that when contact occurs on the trailing flank of the main rotor
(straight side) the net leakage area is reduced due to a longer sealing line on
this side of the rotor profile. In these examples of non-uniform clearance
analysis, the interlobe gap varies around the rotor profile but was considered
constant along the length of the rotor and throughout the cycle. Additional
sophistication is needed if the effect of tapered rotors or thermal distortions in
local rotor hotspots is to be investigated.
32
2.3.2 CALCULATION OF GEOMETRIC CHARACTERISTICS

A screw compressor geometry calculation program name ‘GEOM’, developed at
City University, was made available during this research. This featured accurate
calculation of the actual port area, as described by Mujic (Mujic 2009). This
program allows the detailed calculation of geometric characteristics and
provided a basis for the novel work described in Chapters 4 and 5; therefore it
is worth providing a brief overview.
The objective of ‘GEOM’ is to produce all necessary geometric characteristics

required for the assessment and thermodynamic simulation of any given pair of
twin-screw rotors for a positive displacement compressor or expander. The
requirements are defined:
 to accurately calculate the required geometric characteristics of a screw

compressor independently, prior to thermodynamic calculations
 to represent them as a function of the main rotor angle, θ
 to give the output of all the rotor geometric parameters in a single

matrix
 to accept rotor co-ordinates, of any type and from any source, for
calculation, allowing flexibility and independent rotor comparisons
 to support the use of arbitrary ports
 to support non-uniform clearance distributions along sealing lines and

to allow local clearance modifications to be applied
The last point that is underlined was introduced during the course of this
research to allow specific analyses of the clearances. An example of some of the
typical results obtained is shown in Figure 2-7.
33
IN-FLOW Tip OUT-FLOW Tip

Radial Suction Radial Discharge
IN-FLOW Blow-hole OUT-FLOW Blow-hole
Axial Suction Axial Discharge
OUT-FLOW Suction face OUT-FLOW Discharge face
CHAMBER VOLUME
OUT-FLOW Interlobe CHAMBER VOLUME
120000 3.5E+6 120 3.5E+6
100000
3.0E+6 3.0E+6
100
2.5E+6
Volume (mm^3)
2.5E+6
Volume (mm^3)
Area (mm^2)
80000
Area (mm^2)
80
2.0E+6 2.0E+6
60000 60
1.5E+6 1.5E+6
40000 40
1.0E+6 1.0E+6
20000 20
5.0E+5 5.0E+5
0 0.0E+0 0 0.0E+0
0 200 400 600 800 0 200 400 600 800
Main Rotor Angle (degrees) Main Rotor Angle (degrees)
a. port areas b. leakage areas
Figure 2-7: Volume and area curves
The port areas are shown in Figure 2-7a while the smaller leakage path areas
are shown in Figure 2-7b, from which it can be seen that different leakage paths
exist during different phases of the compressor cycle and for a particular
working chamber there can potentially be two of each type of leakage path: one
for in-flow from the previous chamber or from discharge; and one for out-flow
to the next chamber or to suction. Details on the calculation of the volume curve
and selected areas are included in Appendix A.
2.4 REVIEW OF TWIN SCREW COMPRESSOR MODELS
2.4.1 CHAMBER MODELS

Screw compressors are positive displacement machines which when modelled
at a basic level can be represented by the compression cycle of a single control
volume which is analogous to a piston cylinder arrangement. Calculated
geometric characteristics and discrete leakage paths described as functions of
the cycle can be utilised in a thermodynamic simulation to predict the expected
performance of the new rotors without the need to fully represent the complex
3D geometry. By the 1980’s there were a number of examples in literature of
relatively mature computer programs to calculate screw compressor
performance using such non-dimensional models (Singh, Onuschak 1984,
34
Hanjalic, Stosic 1997, Seshaiah, Ghosh et al. 2007, Stosic, Hanjalic et al. 1986,
Fujiwara, Kasuya et al. 1984, Sauls 1996).
Sangfors (Sangfors 1984) published the principles of differential modelling of

the screw compressor thermodynamic process. Stosic et al. (Stosic, Hanjalic et
al. 1986) published a comprehensive application of such a differential model,
previously widely used in the modelling of internal combustion engines and
reciprocating compressors, to calculate screw compressor performance.
Fujiwara et al (Fujiwara, Kasuya et al. 1984) presented a model where the
volume curve was obtained using the principle of virtual work. Singh et al.
(Singh, Bowman 1990, Singh, Onuschak 1984, Singh 1990) presented papers
dealing with screw compressor geometry which was then utilised in a
performance prediction model. Fleming and Tang introduced a model
particularly suited to refrigeration applications (Fleming, Tang et al. 1998b,
Tang 1995, Tang, Fleming 1992). The ability to estimate the performance of
screw compressors quickly and accurately, as confirmed by many other
authors, for example in Fujiwara and Osada (Fujiwara, Osada 1995) has
revolutionised the field of the screw compressor design and optimisation and
has been identified as a 'success story of the twentieth century', as stated by
Fleming et al.(Fleming, Tang et al. 1998a, Fleming, Tang et al. 1998b).
By applying initial conditions at the compressor inlet, including pressure,

temperature, and the resulting thermodynamic properties for a given working
fluid, differential equations of conservation of energy and continuity are solved
to describe the compression process. The changing volume of the compression
chamber is factored into the energy conservation and the mass balance allows
for mass inflow and outflow to simulate flow through the port and leakage
paths. The thermodynamic properties throughout the compressor cycle can
then be calculated, as a function of time, or shaft angle.
The model used in this research is that described by Hanjalic and Stosic
(Hanjalic, Stosic 1997). Here the compression process is described as a classic
open thermodynamic system using the non-steady flow energy equation with
internal energy as the derived function from which all other properties such as
35
instantaneous pressure are found. The main thermodynamic assumptions of

this model are:
 The fluid flow in the model is assumed to be quasi 1D i.e. only 1D flow
through ports and leakage paths is considered while the chamber in non-
dimensional
 This assumes that kinetic energy changes of the working fluid within the
working chamber are negligible compared to internal energy changes
 Gas or gas-liquid inflow to and outflow from the compressor ports is
assumed to be isentropic
 Leakage flow of the fluid through the clearances is assumed to be
adiabatic.
Thus the non-steady flow energy equation can be written as:
(1)
Each term in equation (1) describes the rate of energy change within the
working chamber. On the right hand side the enthalpy of suction, discharge,
leakage to and leakage from the chamber are all described with a mass flow
balance (other fluids such as oil are included in this total enthalpy). The final
terms are the rate of heat transfer and the compression work. Once the
compressor has reached a steady operating temperature, the heat transfer, Q,
describing the heat flux between the working fluid and the compressor and
surroundings, is relatively small compared to the overall compression power
and can therefore be neglected in terms of its effect of the thermodynamic
process (Stosic, 2015). Therefore, in this work, Q has been set to zero so that the
compressor surfaces are adiabatic. Note that this does not mean the
compression cycle is adiabatic as heat is transported via leakage paths and to
cooling liquid, if present, which have a far more significant impact on the
thermodynamic process.
Each mass flow is calculated using orifice or nozzle flow theory and continuity.
The areas used are those calculated from the rotor geometry which varies as a
function of the compressor angle, θ. Similarly, the last term ωp dV/dθ which
36
describes the rate of compression work done on the system is dependent on

how the volume varies as a function of the compressor angle. From this, all
other properties are calculated (Stosic, Smith et al. 2005).
In the current model, the flow through the ports is assumed to be isentropic,
corrected using an orifice flow coefficient. In other models the flow through the
leakage paths is also often calculated as isentropic (Fujiwara, Mori et al. 1974,
Sauls 1996, Seshaiah, Ghosh et al. 2007), if applicable, by using the critical
pressure ratio for choked flow. Sachs (Sachs 2002) has confirmed
experimentally that supersonic leakage flow can occur. These models neglect
the effect of friction resulting in the need for correction factors. Bell (Bell, Groll
et al. 2012) presents a numerical leakage model that uses a Fanning friction
factor. Results from this model are compared against an isentropic nozzle flow,
highlighting the large difference. In the currently used leakage model (Stosic,
Smith et al. 2005) the leaking gas velocity is derived from the momentum
equation, which accounts for the fluid-wall friction. The mass flow is found by
integrating along a gap from the high pressure to the low pressure side so that
the flow can be defined as adiabatic Fanno flow. This flow has been described in
detail elsewhere (Stosic, Smith et al. 2005); a brief description has been
included in B.1.2 Adiabatic Fanno Flow.
It has been demonstrated in a number of models that, for the purposes of

performance calculation, simplified approximations of the leakage gap are
adequate. In the adopted model the leakage areas between the sealing lines, i.e.
through the interlobe, radial and axial gaps (GI, GR, and GA), are assumed to be
directly proportional to the length of the sealing line. This takes the clearance
gap to be a constant, uniform value, denoted G*. This is shown in equation (2)
which calculates the leakage area AI through the interlobe gap:
(2)
What value this average gap size should take is unclear. Assuming nominal
values for all manufacturing and assembly tolerances it is possible to calculate
what is sometimes described as the ‘cold clearance’, which is what the gap
would be when the compressor is at ambient temperature and stationary. It is
37
known that the operational clearances can vary significantly when the
compressor is running (Sauls, Powell et al. 2007). Fleming dealt with this by
using empirically derived flow coefficients to correct the flows predicted using
the cold clearances (Fleming, Tang et al. 1998b); this approach is less suitable
for a model which must be accurate over a wide range of operation duties and
working fluids. If the temperature distribution of the rotors and casing is
known, operational clearances can be predicted by considering thermal
expansion. In a series of papers (Sauls, Powell et al. 2006a, Weathers, Sauls et al.
2006, Powell, Weathers et al. 2006) Sauls et al accurately predicted the rotor
and casing temperature distribution of an oil injected compressor by applying
boundary conditions from a chamber model to a finite element model. This
analysis deals with duty dependent clearance sensitivities and provided
excellent insight into the rotor and casing temperature distributions which can
be fed back into a chamber model. However, it would be difficult to run such a
detailed simulation for numerous rotors, requiring geometry updates; and for
different applications, requiring re-calculation of boundary conditions.
Sauls et al have reported on a model which can make thermodynamic

predictions based on 3D sealing geometry (Sauls, Powell ); in this they tested
the assumption that using an average clearance gap will provide accurate
results. When comparing results using an average interlobe clearance against
the actual interlobe clearance distribution for a number of scenarios, the results
showed little difference in the net performance characteristics such as flow and
power. However, if the primary interest is in clearance design and reliability it
will be useful to fully represent this clearance distribution. The first step
towards a full 3D clearance analysis is to calculate the leakage area by
integrating the local gap value as shown in equation (3); in the case when the
local gap can vary along the length of the sealing line. Hence it has been
described as a function GI(lI):
(3)
This describes the approach taken by Hsiao et al (Hsiao, Wu et al. 2012) to show
how certain non-uniform clearance distributions can indeed have an effect on
38
compressor performance. However, this approach only really describes the

change in the sealing line along the sealing line path with the assumption that
the clearance distribution around the rotor profile is the same along the length
of the rotor profile. To allow variation in the clearance distribution along the
length of the rotors, either for rotor taper or for thermal distortion, that
increase towards the hot end of the rotor, the local gap also depends on the
cycle angle, θ, which describes where the sealing line is situated along the
length of the rotors. It is not detailed by Sauls et al precisely how the local
clearance gap is handled but this can generally be described in equation (4).
There is a need for detailed procedures to be developed in this work in order to
define unique local clearance gap values at any location in the compressor, at
any rotor position.
(4)
Other developments in chamber models in the literature include analysing

assembly data acquired for each component using a co-ordinate measurement
machine (CMM) and feeding this into the model (Sauls 1996); this deals with
duty independent clearance sensitivities. Another novel model determines flow
coefficients as variable functions by using findings of CFD analysis (Sauls 2011).
The philosophy here is different from that of using empirical flow coefficients
from test results in that the emphasis is on developing a deeper understanding
of the compressor behaviour so that there is less reliance on empirical
coefficients. This approach is important so that a design tool can be used
reliably over a wide range of conditions with minimum ‘calibration’. Such a
powerful yet flexible model is a necessity for optimising a rotor profile for
specific applications and significant contributions can still be made, particularly
in the analyses of duty dependent clearance sensitivities.
2.4.2 ALTERNATIVE MODELS

Singh and Schwartz (Singh 1990) proposed that for highly 3D problems such as
the calculation of the blowhole area, full 3D surface representation would be
more appropriate for calculation of leakage areas. Standard 3D CAD modelling
packages can be used, but developing a robust procedure for extracting the
39
minimum cross sectional area on a plane perpendicular to the actual flow

direction is a challenge. Computational fluid dynamics has been used to
describe the full 3D flow field of a twin screw compressor by Kovacevic et al
(Kovacevic, Stosic et al. 2003) giving insight into the internal behaviour of the
compression volume. It can be argued that the actual flow through a highly 3D
blow-hole can only be accurately modelled by using a 3D representation of the
flow domain. However, the high level of expertise, computing power, and run
time can make the use of CFD prohibitive in some instances.
To find a compromise between the accuracy of the expensive multidimensional

calculations and fast, but less detailed chamber models, which is adequate for
evaluation and general performance prediction and optimisation, a hybrid
model was proposed by Mujic et al. (Mujic 2009) which used chamber models
for analysis of flow within the rotor domains and a 3D approach at the
compressor ports.
2.4.3 THERMAL ANALYSIS

The prediction of thermal distribution and distortion is beyond the scope of
typical chamber models used for performance prediction and requires
additional calculation procedures. With the application of reasonable
assumptions the temperature distribution of the rotors can be approximated
analytically based on the temperature of the operating fluid at the compressor
inlet and outlet (SRM 1950). This procedure essentially performs a lumped
analysis for each 2D cross section of the rotor based on an empirically observed
axial temperature distribution.
If the assumptions are to be minimised, the complexity of the analysis can

quickly increase. Generally some assumptions are required to prevent the
analysis becoming prohibitively complex, this complexity is clear when
considering the physical processes involved: Heat is generated due to
compression; and kinetic losses in bearings, seals and oil. The convective heat
transfer coefficient can vary locally on surfaces due to the complex flow and
changing fluid properties within the compressor. The problem is compounded
40
by the interdependence between the compression process and component

temperature and geometry.
Kovacevic et al (Kovacevic, Stosic et al. 2002a) developed a 3D numerical

method to calculate flow and deformation by using a numerical grid for both
the solid and fluid domains. This work showed how the ‘operational clearances’
for a given compressor and compression application are influenced by
movement due to pressure and, more significantly in oil free machines, by
thermal distortions. This work suggested a more or less uniform temperature
across the transverse section of each rotor however because this work focused
on analysis of the rotor bodies it is not apparent what impact heat conduction
along the shafts, to or from the bearings, would have on the thermal
distribution.
A commonly used approach to facilitate analysis is to de-couple the calculation

of the thermodynamic boundary conditions from thermal analysis of the solids.
The thermodynamics are calculated independently then used as an input for the
thermal analysis. Hsieh et al (Hsieh, Shih et al. 2011) use the thermodynamic
boundary conditions from a lumped parameter, chamber model. These
boundary conditions were used with a finite element model of the rotors to
calculate the temperature distribution in oil injected screws. This work
discusses the numerous boundary conditions that the full rotor shafts are
exposed to; including the bearing surfaces. The main challenge in this work was
in establishing empirical constants to calibrate the unknown heat transfer
coefficients. This is achieved by adjusting multiple coefficients to try and match
temperatures at 3 points on each rotor, which were measured on test. The
reported results state that there is a heat flux from the discharge end bearings,
along the rotor bodies and outwards toward the compression fluid. For this to
be the case the implication is that the main rotor outlet bearing temperature is
greater than the bulk temperature of the working fluid. It is not clear from this
work how the boundary conditions from the chamber model are applied to
rotor surfaces though it is implied that the discharge end of the rotor is exposed
to higher gas temperatures. Interestingly, the temperature variation from the
41
hot to the cold end of the rotors is actually said to be due to the larger heat flux
from the discharge bearing, rather than the gas temperature distribution.
The last two examples resulted in quite different rotor temperature

distributions over the transverse cross section of the rotors, demonstrating how
sensitive these kinds of analysis are to any difference in the way boundary
conditions are defined. In the thermal analysis done by Sauls et al, referred to
previously (Sauls, Powell et al. 2006b, Sauls, Powell et al. 2006c), using a
chamber model and FEA, the first of these papers goes into detail about how
surface boundary conditions are extracted from the chamber model results and
mapped onto the surface of the compressor casing and rotors. The significance
of the interlobe sealing line boundary is discussed although the way in which
the full chamber boundary is analysed, is not disclosed. These results show the
round, or high pressure, side of the main rotor to be hotter than the opposite
flank with the explanation that it is exposed to a high pressure chamber for
longer. This result conflicts with Hsieh whose results, discussed previously,
show the rotor to be slightly cooler on the round side.
Analyses from closely related fields are interesting to compare. Nilolov et al

(Nikolov, Brummer 2012) uses an iterative coupling of a thermodynamic
chamber model and FEA thermal simulations. This highlights the
interdependence between the thermal distortion, leakage gap size and the
thermodynamic performance for a twin screw expander. Gao et al (Gao, Yang et
al. 2011) performs a temperature and thermodynamic analysis for a twin screw
multi-phase pump in which temperature and pressure transducers are
embedded into the rotor bodies. The results from this experiment supports the
assumption that the temperature at a point on the rotor surface can be
considered to be steady state at it rotates due to the high rotation speed and
good thermal conductivity of the rotors. The presented temperature
distribution on the rotors shows the surface temperature to increase from inlet
to outlet due to the gas temperature increase as would be expected. In the core
of the rotor the temperature gradient along the length of the rotors is smaller
due to conduction. As a result the transverse cross-section of the rotors at the
discharge end of the rotors is hotter outside and cooler inside; at the inlet end
42
the opposite is true. This paper also considers deformation due to the pressure
distribution but notes that this is negligible compared to the temperature
deformation. Kovacevic et al apply a solid-fluid interaction analysis to a novel
screw compressor-expander (Kovacevic, Stosic et al. 2006a). This analysis is
built on in (Kovacevic, Stosic et al. 2006b) which estimates clearance gap
distortions using the gas discharge temperature by mapping numerical and
experimental results. Both Kovacevic and Gao discuss the need to consider the
rotor and casing material properties and how this influences gap distortion
during operation.

Some relevant background has been given on profile generation and the way in
which clearances are applied to the rotors. The concept of abstracting the
compressors geometric characteristics for the purposes of modelling and
clearance analysis is a balance between efficient modelling and accurately
capturing enough detail for the type of investigation required. To investigate
the behaviour of clearances in detail, the conventional clearance approximation
that assumes uniform gaps is not suitable. Additional sophistication is needed if
the effect of thermal distortions in local rotor hotspots is to be investigated.
The need to understand the thermal distribution in the compressor is vital to

understanding operational clearance distortions. However, the thermal analysis
procedures that should provide the most detail and realism are not necessarily
the most appropriate for a robust clearance design and optimisation procedure.
This review has identified some variations in rotor temperature distribution
results from different analyses; this doesn’t necessarily mean that one analysis
is wrong but it highlights how these sophisticated tools are still highly subject to
the boundary conditions assumed, whether correctly or incorrectly.
Chamber models are well established tools that have been shown to be accurate
for general performance prediction, with this in mind it can sometimes be
counter-productive to introduce additional complexity since the model
efficiency may be reduced without an overall improvement in accuracy due to
43
other inherent assumptions such as the way in which leakage flow is calculated.
However in certain specific analyses, adapting existing chamber models can
provide valuable new insights with minimum decrease in model efficiency. For
example, when Mujic (Mujic 2009) replaced approximated sine curves with port
area curves derived from actual rotor and casing geometry, he enabled the
investigation of pressure pulsations due to subtly different port area curves.
An objective of this research was to introduce new modifications that allow

compressor analyses, focusing on the estimation of operational deformations in
clearance gaps and quantifying how performance and reliability is affected. It
has already been shown that the thermodynamic results from a chamber model
can provide valuable and detailed boundary conditions for FEA thermal analysis
(Sauls, Powell et al. 2006b), thus giving very detailed results, albeit with
accuracy still heavily dependent on the assumptions used. On the other hand it
has been shown that simple empirical adjustments to clearances, as a function
of the discharge temperature, can improve the correlation between a chamber
model and test results (Buckney, Kovacevic et al. 2011, Kovacevic, Stosic et al.
2006). There is significant scope to process and utilise the detailed
thermodynamic results from a chamber model in a simplified analytical analysis
using appropriate assumptions. With minimum additional complexity an
adapted chamber model and geometry calculation tool could efficiently provide
useful results for the design and optimisation of rotor clearances for specific
applications. This effect of duty dependent thermal clearance sensitivities is not
directly fed back into any of the existing chamber models reviewed.
44
Chapter 3
Research Objectives, Methods
and Expected Contribution
3.1 STATEMENT OF RESEARCH

Efficient estimation of the size of operational clearances is a critical
requirement for reliable calculation of a twin screw compressor performance
and reliability; these are necessary inputs for design and optimisation of screw
compressors for specific applications.
The aim of this research is to develop and validate procedures which will enable
analysis, design and optimisation of the operational clearances in industrial twin
screw compressors in order to improve their performance and reliability.
Thermodynamic results obtained from chamber models are readily available

but underutilised. Such results could be efficiently processed and, with the
application of suitable assumptions, used for the analysis of distortions of
compressor elements due to thermal loading. The distortions of the compressor
elements, i.e. rotors and casing, cause clearances to change from ‘cold’ to
‘operational’ clearances. The operational clearances can be effectively estimated
using proposed mapping of two dimensional computational surfaces of rotors
and casing and then these can be utilised in a chamber model to predict the
performance of screw machines with operational clearances more accurately.
This approach sacrifices some of the accuracy and realism of the 3D simulations
of the solid and fluid domains in order to maintain the efficiency and speed of
calculation with chamber models.
A key output of this research will be validated tools and procedures that can be
readily applied to the design and optimisation of twin screw compressors for
specific applications.
45
3.2 OBJECTIVES
Develop a boundary mapping procedure to fully describe exposure of
compression chambers to rotor and casing surfaces, thus allowing
temperatures and pressures calculated by thermodynamic model to be mapped.
Develop a procedure to estimate change in radial and interlobe clearances due

to thermal loading on the rotors and casing.
Integrate boundary mapping and clearance calculation procedures with a

thermodynamic model to calculate operational clearances and allow calculation
of performance with updated clearances.
Validate the developed procedure for different oil free and oil injected
compressors by the use of experimental results.
Study the geometric characteristics of oil free and oil injected rotor profiles.
3.3 METHODOLOGY
Generation of Rotor Profiles:
 For the purposes of this research rotor profiles will be generated using
existing methods such as those used for the N-Profile
Calculation of Geometric Characteristics:
 A geometry calculation program (GEOM) will be further developed based

on previous work at City University. The input to this program will be
the x and y co-ordinates of the rotor profiles, thus allowing profile input
data from other sources
 All rotor geometry including swept areas and volumes will be calculated
as a function of the compressor cycle angle
 Particular detail will be given to accurately representing leakage paths
individually and incorporating functions to vary the leakage areas as a
result of operational effects
 Support calculation of areas affected by complex clearance distributions
46
Calculate Performance:
 The geometry calculation program will provide the input to the chamber
model that will calculate the performance of the compressor. The basic
integration of the new program is shown in Figure 3-1
 Verify performance predictions with available test data
 Investigate how operational clearances vary at operational temperatures
and pressures and quantify the impact this has on predicted
performance
 The clearance areas and thermodynamics will be solved iteratively to
include operational effects where possible. This should improve the
accuracy of performance prediction and form a basis for optimum
clearance design for specific applications
 Study geometry characteristics of rotor profile using a new program suite
 Validate predicted results and compressor reliability with test results
 Utilise test results to refine models / procedures
Figure 3-1: GEOM program integration with SCORPATH
47
3.4 EXPECTED CONTRIBUTION
3.4.1 BOUNDARY MAP

The procedure of mapping properties to the surface of the rotors will utilise a
novel ‘Rotor Boundary Map’. The rotor boundary map will facilitate input of the
distribution of temperatures and pressures on the rotor and casing boundaries
i.e. it will define instantaneous boundary conditions for screw compressor
rotors and casing. There will be a unique boundary map for the main and the
gate rotors; the form of which will depend only on the transverse rotor profiles.
This presentation of the rotor boundaries will provide a unique way of
visualising and comparing the key geometrical properties of different profiles
such as the length of the sealing lines and how the sealing lines are related to
the formation of the blow hole leakage area. This will be a direct contribution to
compressor technology, since it will define a procedure which has not been
used before.
3.4.2 MODEL INTEGRATING OPERATIONAL CLEARANCE MODIFICATION

Results from chamber models are currently underutilised and it is possible to
use the temperature and pressure of the fluid within the chamber, calculated
throughout the compression cycle, to assess local exposure to rotor and casing
surfaces. Knowing local fluid boundary temperatures allows approximate
estimates to be made of metal temperature and distortion during operation by
using analytical procedures and appropriate assumptions. This allows the
clearance reliability to be evaluated at an early design stage with minimum
additional complexity. In addition, iterating the thermodynamic performance
calculation with modified operational leakage areas is expected to improve the
model realism. In particular, this newly developed model supports temperature
management of operational clearances.
48
Chapter 4
Boundary Map for Rotor and
Casing Surfaces
4.1 INTRODUCTION
Chamber models simulate compressor performance efficiently by assuming a
non-dimensional control volume. This control volume varies in size throughout
the compression cycle as a function of the main rotor angle, or cycle angle. At
discrete cycle angles the fluid properties within the control volume are
calculated thus the fluid properties are described throughout the compression
process. From this data, critical performance characteristics such as indicated
power are easily derived without the need to represent actual 3D flow domains.
However, in order to calculate some performance characteristics such as rotor
torque it is necessary to perform additional analysis of the compressor
geometry (Stosic, Smith et al. 2005). Taking this analysis further it is possible to
establish the instantaneous fluid properties at any local point on the rotor or
casing compression surfaces. This allows for more detailed investigation of the
inner workings of the compressor.
The objective of this chapter is to define a method by which instantaneous or

averaged fluid properties from a non-dimensional chamber model can be
mapped onto surfaces that represent the rotor and casing geometry. Fluid
properties from the chamber model are known at a given cycle angle therefore
the challenge is to establish what portion of the cycle any given point, fixed to
the surface of the casing or rotor, is exposed to. This will make it possible to
establish the boundary temperature or pressure at any location on the
compression surfaces of the rotors and casing.
49
4.2 DEFINITION OF BOUNDARY SURFACES
4.2.1 COORDINATE SYSTEM CONVENTIONS

The following work refers to the ‘compressor’ for simplicity, however all
calculations are equally applicable to expanders with the only differences being
in the direction of rotation and the inlet and outlet directions. In order to
generalise the compressor inlet plane, which is the expander outlet, it can also
be been referred to as the low pressure (LP) plane.
As twin screw compressors can be installed in various orientations a ‘global’ co-

ordinate system will be fixed to the casing. Before placement of the global co-
ordinate system, the compressor orientation needs to be defined. For twin-
screw compressors the only significant variable that affects the relative position
of the compressor features is whether the main rotor has a right-hand or a left-
hand helix, the latter type resulting in a compressor that is the mirror image of
the former. To prevent confusion all geometric analysis will be performed
assuming a main rotor with a left-hand helix regardless of the actual rotor
geometry. To clarify the helix convention – the main rotor shown further on in
Figure 4-5 has a left-hand helix. If necessary the rotor helix can easily be
corrected at a post-processing stage. Constraining the compressor to have a
main rotor with left-hand helix and orientating this compressor to be viewed
from the inlet end with the high pressure (HP) cusp at the top, the following
convention will always apply:
Table 4-1: Compressor orientation
Main rotor helix: LEFT-HAND

Compressor orientation: HP cusp at top (top out discharge port)
Main rotor position (viewed from inlet): LEFT
Main rotor rotation (viewed from inlet): CLOCKWISE
50
Figure 4-1: Compressor and rotor co-ordinate systems
Any location on the compressor rotors or casing can be described using the
global Cartesian co-ordinate system S(X,Y,Z) defined in Figure 4-1. This is
located on the high pressure (HP) plane with the z-axis aligned with the main
rotor axis. The x-axis is co-incident with the centre line formed between each
rotor axis on the HP plane. Additional co-ordinate systems have been defined on
the low pressure (LP) plane: S1(x1,y1,z1) and S2(x2,y2,z2). These co-ordinate
systems are also fixed to the compressor casing and are simply offset from the
global origin, S. Any point on these local co-ordinate systems can be related
back to the global system, S(X,Y,Z), using the equations (5) to (7). ‘L’ is the
nominal rotor length between the LP and HP plane, and ‘A’ is the nominal rotor
centre distance between the rotor axes:
(5)
(6)
51
(7)
With the cylindrical form of the rotor bores and the rotational movement of the
rotors it will be more convenient to define points using a cylindrical co-ordinate
systems located at the origins S1 and S2. In the standard convention for a polar
co-ordinate system, (r,θ,z), the angle θ is measured counter-clockwise from the
x-axis and z is in the direction already defined. To simplify later equations it is
preferable to define less conventional angle parameters that are aligned with
the direction of rotation of each respective rotor. In addition, it is easier to use
an axial length parameter that is aligned from the compressor inlet to outlet.
Figure 4-2: Parameters for surface definition
Parameters ‘r’, ‘β’ and ‘l’ have been defined on each co-ordinate system S1 and S2
as shown in Figure 4-2. These parameters effectively describe a non standard
cylindrical co-ordinate systems where r is the local radius, β is the angle
(measured in the same direction as the rotor rotates) and l is a measure of the
axial position from inlet.
The rotors on twin screw compressors feature a helical twist along the axis
therefore the position along the axis that has been defined by parameter, l, can
alternatively be defined by using a new angular parameter, γ; this describes the
52
local wrap angle measured from the inlet (LP) plane. For the case of rotors with
constant pitch helix the relationship is as shown in equation (8) where φw1 is
the main rotor wrap angle over the rotor length, L. The parameter γ will always
be defined using the main rotor wrap angle, whether measured on the main
rotor side or on the gate rotor side. It is convenient to describe the axial
position in terms of the wrap angle on the main rotor for later calculations.
(8)
Equations (9) to (14) describe transformations from the parameters r, β and γ,

back to the global Cartesian compressor system located at S. The subscripts ‘1’
and ‘2’ for the rotor parameters distinguish whether they refer to the main or
gate rotor sides which are measured relative to the origins S1 or S2 respectively:
Main rotor side:
(9)
(10)
(11)
Gate rotor side:
(12)
(13)
(14)
The internal geometry of the compressor will be broken down into a number of
discrete surfaces which are exposed to various compression chambers. These
surfaces are: casing inlet plane, casing outlet plane, casing bore and rotor
section (ridge or flute). Each of these surfaces is repeated for the main rotor and
gate rotor sides of the compressor resulting in a total of eight surfaces. A 2D
53
computational array can then be defined for each surface in order to efficiently
describe any point on each respective surface. The transformation between the
3D spatial domain and the 2D computational domain will be done by utilising
the co-ordinate systems and parameters that have been defined. These surfaces
will now be defined in more detail.
4.2.2 CASING SURFACES
Figure 4-3: Exploded view of casing surfaces
The static casing will be considered first. The six casing surfaces are labelled in
Figure 4-3 which shows an exploded view of the inner surfaces of the
compressor casing. The vertices A, B, C and D are also shown on Figure 4-1 for
reference. This view is formed by cutting along the line AB on the low pressure
cusp. To explain Figure 4-3 in reference to Figure 4-1 – imagine lying inside the
compressor looking upwards with your head towards the HP plane – this is why
the main rotor bore now shown on the right hand side.
This figure shows how the parameters r, β and γ align to these ‘unwrapped’ 2D
surfaces. Table 4-2 summarises how the location on each of the surfaces can be
defined using only two of the three parameters.
54
Table 4-2: Casing surface parameters
Surface Co-ordinates for

Surface Description Constant
Label surface array
Main rotor side, inlet plane IP1 (r1,β1) γ=0
Main rotor side, bore B1 (β1,γ1) r1 = ro1
Main rotor side, outlet plane OP1 (r1,β1) γ = φw1
Gate rotor side, inlet plane IP2 (r2,β2) γ=0
Gate rotor side, bore B2 (β2,γ2) r2 = ro2
Gate rotor side, outlet plane OP2 (α2,β2) γ = φw1
It is worth clarifying that these casing surfaces, described using parameters ‘r’,
‘β’ and ‘γ’, are representations of idealised casing surfaces that would exist with
perfect rotor alignment and zero clearances. This would be inappropriate for
other full 3D analysis using FEA or CFD where the clearances would be derived
from actual component surface dimensions. The planned analysis will of course
use local clearance data however the local clearances will be handled as scalar
values only; this vastly simplifies the problem while still providing insight into
the 3D clearance distributions.
4.2.3 ROTOR SURFACES

Transverse 2D rotor co-ordinates are defined on S01(x01,y01) and S02(x02,y02) for
the main and gate rotors respectively; as shown in Figure 4-4. These rotor co-
ordinate systems are only partially constrained so that S01 freely rotates about
the z1-axis of S1 and S02 freely rotates about the z2-axis of S2.
55
Figure 4-4: Transverse profiles defined at LP plane
The full 3D surface of the rotors (see Figure 4-5) can be derived using the
transverse rotor co-ordinates that are extruded along a helical path.
Figure 4-5: Full main rotor surface
An instantaneous point on this rotor surface could be represented globally

using (X,Y,Z); locally on origin S1 using (x1,y1,z1); or with parameters (r1,β1,l1).
56
The length parameter, l1(γ), a function of γ, is conveniently aligned along the

rotor axis from the inlet to outlet and is unaffected by rotor rotation so this is a
suitable parameter to describe the axial position on the rotor surface. The local
radius, r1, is also a constant for a given point on the surface however the angle,
β1, is affected by the rotational position of the rotor defined by the cycle angle,
θ.
Each rotor lobe is geometrically equivalent and therefore the surface of each
lobe can be generated from another by indexing the cycle angle, θ, through
2π/z1, where z1 is always the number of lobes on the main rotor - rotating the
rotor pair by cycle angle θ = 2π/z1 will index the gate rotor by an angle
corrected for the gear ratio: (z1/z2)2π/z1, in other words by: 2π/z2
A 2D computational array has been defined to simply identify a surface location

on a single main rotor ridge; or in the case of the gate rotor, the corresponding
flute. This array can be thought of as the surface produced by straightening out
the rotor helix and flattening the resulting surface to form a rectangle.
Main rotor surface, R1
Gate rotor surface, R2
Figure 4-6: Rotor surface arrays

57
In Figure 4-6 the length of a curve, s1, between points A and B is:
(15)
The parameter, ε, is defined as the relative position along the line, s, of the
transverse rotor profile curve such that 0 ≤ ε ≤ 1. This is defined separately for
the main and gate rotor segments as shown in Figure 4-6. This allows the rotor
surface to be presented two dimensionally and independently of the rotor
rotation.
The rotor segments are in this case a main rotor ridge (or lobe) and a gate rotor
flute (or interlobe) section. It is later assumed that the limits of ε2 on the gate
rotor form the boundary between adjacent rotor chambers on that rotor so it is
important that this point is located approximately central on the gate rotor OD
or at the tip seal protrusion, if applicable.
In practise, the parameter ε is numerically approximated by linearly

interpolating between discrete profile points. As this parameter is primarily
used for lookup purposes in order to relate the computational array to the
actual geometry, any approximation errors are not important. In fact, this
dimension of the rotor surface array could easily be replaced by identifying the
discrete profile point number, n, as was done when developing a program to
deploy these procedures. The value of presenting the rotor surface arrays using
the ε parameter is that the scale along line s is preserved in the surface array
plots and the generalised procedure can be presented in an efficient manner.
4.3 DEFINITION OF CYCLE CONVENTIONS

In order to relate the actual rotor positions to the compression cycle it is
necessary to define a datum from which rotations can be measured. Figure 4-7
shows a transverse rotor profile pair in what will be referred to as the ‘home
position’. The main rotor on the left is rotated so that the tip of one lobe is
coincident with the centre line formed between the centre points of the main
and gate rotors. The gate rotor is positioned to be in mesh with the main rotor.
This datum for rotor rotation is measured on the LP plane; the inlet plane for a
58
compressor. This convention has been adopted from the N-Profile generated
rotors, example coordinates of which are shown in Figure 4-8.
Any rotor position can now be defined by the offset angle of the main rotor tip
from this home position. To relate this 2D rotor position at the LP plane to an
instant during the compressor cycle it must be explicitly stated whether the
referenced tip is on the leading or trailing lobe of the control volume.
undercut area
Figure 4-7: Rotor home position
The transverse profile co-ordinates for a single main rotor ridge and its
corresponding gate rotor flute are plotted with respect to their local co-ordinate
systems in Figure 4-8.
y01, y02
Main
Gate
x01, x02
Figure 4-8: Rotor segments
59
Before defining the rotor position at the start of the compression cycle (when θ
= 0°) it is necessary to explain a little about the cycle duration and the various
steps that must be considered in its calculation:
1. Full rotation of main rotor – this is the obvious minimum travel before
the cycle would repeat
2. Transfer from leading main lobe to trailing main lobe – the interlobe area
which defines the volume is initially created by the leading main lobe but
will not disappear until the trailing main rotor lobe has advanced to the
position of the leading lobe, thus adding 360/z1° to the cycle duration.
3. Wrap angle lag – the wrap angle of the main rotor introduces a lag
during which the chamber advances axially along the rotors, thus adding
the main rotor wrap angle, γ1, to the cycle duration.
4. Undercut area – all asymmetric rotors will have non-zero area on the
trailing flank of the main rotor when in the home position as shown in
Figure 4-7. Depending on the rotor profile this requires the rotors to be
reversed by approximately 30° before this area, and consequently the
chamber volume, is zero. This adds to the cycle duration.
5. Idling gate rotor lobe offset – if z2 – z1 is greater than or equal to 2 then
the area formed within a particular main rotor interlobe area and gate
rotor interlobe area will not finish with the same gate rotor interlobe.
The area formed in the gate rotor interlobe will idle for one additional
lobe pass when z2 – z1 = 2, adding 360/z1° to the total cycle duration.
The last item listed is often omitted in chamber models. Since the offset of the
idling gate interlobe volume would generally by added at the filling stage it is
not important for the chamber model accuracy. However in order to represent
the net volume accurately for all chambers, at a given instant, this item is
important. Likewise, it is necessary for mapping results from a chamber model
onto the actual rotor geometry where the full duration of all gate rotor lobes
must be described.
60
Figure 4-9 breaks down how the rotor start position (when θ = 0°) is defined in
order to preserve the correct cycle duration. Steps A) to D) of Figure 4-9 show
the transverse rotors at the LP plane:
Figure 4-9: Moving rotors to the start of the compression cycle
A) The ‘rotor home’ position before any adjustments

B) This simply clarifies the interlobe area of interest – the tip of the main
rotor in the home position will be the leading tip
C) φs1 is defined as the ‘start angle’ through which the rotors must be
reversed to close the ‘Undercut area’
D) φoff is the ‘offset angle’ to correct for ‘Idling gate rotor lobe offset’. Note
that this rotation is only applied to the main rotor. This angle is
calculated as:
(16)
61
This rotor offset is not an obvious concept to understand and might be slightly
better illustrated later in Figure 4-12: Position of rotor chambers in the
transverse plane.
The start angle, φs1, has been defined graphically in Figure 4-9. This angle can
be derived from the sealing line because the rotor position in Figure 4-9C is also
the maximum rotor reversal before the contact point at the tip of the main rotor
is broken off. φs1 equals the meshing angle, θM, at the upper limit of the sealing
line ySL_MAX. The calculation of the meshing angle and sealing line co-ordinates is
explained in Appendix A.
Once both rotors are adjusted for φs1 and the main rotor is adjusted by φoff, if
applicable, the rotors will be in the start position as shown in Figure 4-9D. With
reference to this start condition the rotor positions for any cycle angle, θ, can
now be described.
With the assumption that the compressor is operating at steady state conditions
the problem is cyclic in nature and therefore only one chamber has to be
considered. For a point on the casing the cycle repeats for every passing lobe of
the compressor rotors – therefore only seeing a limited range of the full cycle
duration. For a point on the rotors, the cycle repeats for every full rotation of
the particular rotor on which the point is located – again only being exposed to
a limited range of the overall cycle. The overall compression cycle is distinct
from these local exposure cycles because it follows a single control volume as it
passes through the compressor.
4.4 MAPPING OF THE CASING SURFACES

The casing surfaces previously presented in Figure 4-3 have been presented
again in Figure 4-10 to show the projection of the nearby rotors. At the instant
shown, any points on any of the surfaces that are within the hatched areas are
in the same chamber.
62
Figure 4-10: Single chamber projected onto casing surfaces
The following paragraphs detail how to calculate the cycle limits that are ‘seen’
by a fixed point on the static rotor casing. A similar procedure was presented by
Sauls et al. (Sauls, Powell et al. 2006b) and describes the cycle limits for a static
point on the casing bore by considering the cycle angle at the moment the
leading and trailing tip of the rotor pass by.
Consider the rotors in the start position defined in Figure 4-9D. At this position
the rotor cycle angle θ = 0. The position of the leading tip on the main rotor can
be described using the parameter, β, as described in section 4.2.1 to be positive
in the forward direction of rotation for a compressor. Moving the rotors
through the angles φs1 and φoff, as defined in Figure 4-9, would result in a
backward rotation. At the start of the cycle at the leading tip on the main rotor
(subscript sl1) the parameter β can be calculated as:
(17)
63
At the trailing tip (subscript st1); where z1 is the number lobes, or gear teeth,
on the main rotor:
(18)
Referring to the rotor ‘home position’ defined in Figure 4-9-A: at the start of the
cycle at the leading tip on the gate rotor (subscript sl2) the rotor is already
offset from the rotor centreline by the angle φgate. The only further adjustment
to the gate rotor rotation corresponds to the reverse rotation of the main rotor
by the angle φs1 which is adjusted by the gear ratio z1/z2:
(19)
On the gate rotor at the trailing tip (subscript st2):
(20)
Consider a fixed point on the main rotor bore, B1, described by the parameters
β1 and γ. When the leading tip of the main rotor intercepts this point
corresponds to one rotor position and cycle angle. When the trailing tip of the
main rotor intercepts the same point this corresponds to a different rotor
position and thus a different cycle angle. For clarity the corresponding cycle
angles will be labelled θl and θt respectively.
The distance that the main rotor must travel through, i.e. the cycle angle, θ, until
the leading tip of the active chamber meets the point on the casing is the
difference between β1 and βsl1. When moving away from the inlet plane the
effect of the wrap angle along the length introduces a lag which is accounted for
by the parameter γ:
(21)
64
The cycle angle at which the trailing tip of the main rotor passes the same point
is:
(22)
θl and θt represent the limits of the compression cycle that the point on the
casing (β1,γ) is exposed to.
The calculation of the cycle limits for a point on the gate bore surface is similar;
differences are due to the fact that the cycle angle is always referenced from the
main rotor position.
(23)
(24)
At every location on any of the static casing surfaces the limits of cycle exposure
defined by θl and θt can be calculated. From these values a single ‘average’ cycle
angle can be calculated at each location:
(25)
(26)
Figure 4-11 was generated to show the average cycle angle using a surface
colour contour plot. The colour represents the local average cycle angle
calculated using the appropriate values of either β1 and γ or β2 and γ. These
same surface parameters are then fed into equation (8) to (14) in order to
65
calculate the position in the global co-ordinates (X,Y,Z) at each location (these
co-ordinates are as defined in Figure 4-1).
Figure 4-11: Cycle angle mapped onto casing bore surfaces
This surface plot has been smoothed and only serves to show visually how
different parts of the casing are exposed to different parts of the cycle. In
practice, surfaces B1 and B2 of each rotor bore are discretized using a finite
number of points defined by β and γ. The explicit calculations are repeated at
each point or node and the number of points has no bearing on the accuracy at a
given node.
4.5 MAPPING OF THE ROTOR SURFACES

In the same way that the cycle angle has been mapped onto the rotor bore
surfaces B1 and B2 in the previous section, this will be done for the rotor
surfaces R1 and R2 (surfaces as defined in Figure 4-6). In the previous case of
the rotor bore surfaces, the instantaneous boundaries that divided the surfaces
into different chambers were simply defined by the rotor to casing sealing lines.
In the case of the rotor surfaces, the chamber boundaries are defined by the
rotor to casing, radial sealing lines and the rotor to rotor, interlobe sealing line.
With this added complexity it was found that the previous method (identifying
the cycle limits) was not suitable, furthermore it is useful to be able to calculate
66
the instantaneous boundary conditions on the rotor surface at any given rotor
angle.
Figure 4-12 shows a transverse cross section of the compressor at the inlet
plane. This shows the rotor to casing and rotor to rotor interactions that result
in separate compression chambers as highlighted with different colours. The
key on the right hand side gives the cycle angle that defines each compression
chamber. For most chamber cross-sections the entire rotor flute (or interlobe
area) is exposed to the same chamber which is separated from its adjacent
chambers by the radial rotor to casing sealing points. During rotor to rotor
‘contact’, typically up to three sealing points will be formed between the rotors.
When this transverse section is extended to form the full 3D rotors these sealing
points will all fall on the same 3D sealing line.
Considering one meshing rotor segment on the main rotor, identified by the
white lines forming a ‘slice’ of the rotor, it can be seen that the surface of this
particular rotor segment is simultaneously exposed to four different
compression chambers labelled: blue, green, red and yellow. A procedure is
needed to identify the location of the boundaries and the chamber exposure for
any given rotor angle.
Figure 4-12: Position of rotor chambers in the transverse plane
67
It can be observed from Figure 4-12 that the adjacent blue and red chambers at
the low pressure cusp are not separated – this is not due to any conventionally
defined leakage path such as the low pressure blow-hole area and this flow path
between chambers is in fact just a consequence of the 4/6 lobe combination. If
the gate rotor instead had 5 lobes then the matching chamber sections
highlighted in blue would be formed together at suction. This is the rationale
behind the rotor offset term previously described in equation (16). This
necessary offset is applied at the low pressure side where both chambers will
be exposed to the low pressure port anyway. Hence, the interaction between
these chambers is not significant. However, neglecting this offset term will
result in incorrect cycle duration on the gate rotor side and it must be included
in order to relate the compression cycle to the rotor and casing geometry
accurately.
Due to the complexity of the chamber boundaries on the rotor surface, work to
investigate the cycle exposure led to the development of a rotor surface
boundary map which will now be discussed in more depth.
4.6 DEFINITION OF ROTOR SURFACE BOUNDARY MAP

The following procedure analyses the rotor boundary and chamber interaction
along the line of a single transverse rotor profile section from ε = 0 to 1. This
abstracted procedure can later be used to map boundaries and fluid properties
along the full rotor surface, R1, then the full rotor body for all lobes.
For a given rotor profile the radius, r, varies only as a function of the rotor
surface parameter, ε. In practise the value of the corresponding radius can be
saved in an array that can be looked up.
(27)
(28)
The original profile co-ordinates were defined on the systems S01(x01,y01) and
S02(x02,y02) as shown in Figure 4-6 and Figure 4-8. Each of these points can be
described using the corresponding polar co-ordinates (r01,φ01) and (r02,φ02) for
68
the main and the gate rotor respectively. As was the case for the local radius, the
local profile angle is a profile constant that can be described as a function of the
rotor surface parameter ε. Again, in practice, the value of the corresponding
local angle can be saved in a look up array.
(29)
(30)
Recall that the actual instantaneous geometry of the rotors at any point during
the cycle can be represented by the parameters r,β,γ, set out in Figure 4-2. The
angle parameters, β1 and β2, for a point on the main or gate rotors are
calculated:
(31)
(32)
The term λ has been defined below by subtracting the terms for the local profile
angle from β; this new parameter more generally describes the movement of
the rotor pair rather than individual points on either rotor:
(33)
(34)
These equations will be used to relate the local transverse profile rotation, λ, to
the cycle angle, θ, for the main and gate rotors respectively; as the cycle angle is
referenced against the main rotor rotation, λ is also referenced against the main
rotor. The subscripts define whether λ refers to the main or gate rotor
boundary map. The only difference between these λ values is due to the main
rotor offset term, φoff; for rotors with a 5/6 lobe combination it would be the
case that λ1 = λ2. In the special case of the point located at the tip of the main
rotor, φ01 = 0 and therefore λ 1 = β1. At the corresponding point at the root of the
gate rotor, φ02 = 180° and therefore λ2 = β2(z2/z1). Going forward it will be
69
clarified whether the boundary map is for the main rotor or the gate rotor so
these subscripts can be dropped.
Figure 4-13: Notation used on boundary map
The parameter λ has been shown in Figure 4-13 – note that it is only referenced
to the rotation of the main rotor but describes the motion of both rotors. Also
note the convention where a negative rotation of the rotors from the ‘home
position’ results in a negative value for λ. The angles to the casing cusp, φc1 and
φc1, are also defined in this figure, which will be used to define the limit of
interaction between the rotor and the casing.
The bottom part of Figure 4-14 shows the rotor surface, R1, on which a single
transverse section (γ = constant) has been highlighted. The top part of the
figure shows the ‘Boundary Map’ with the rotor surface parameter, ε1,
represented on the horizontal axis. The vertical axis represents the local
transverse rotor offset angle, λ1.
70
Boundary Map
λ1
R1
Figure 4-14: Local transverse rotor offset angle, λ, at given axial position, γ.
Because λ is adjusted for the parameter, γ, the Boundary Map is common all
along the length of rotor, whatever the value of γ.
This ‘Boundary Map’ will be used to plot the transverse rotor boundary
interactions. A key advantage of this generalised boundary map is that both
interlobe and radial boundaries can be plotted. This will allow the analysis of
different boundary interactions with the aim to fully define distinct
compression chambers on the rotor surface.
The boundaries on this map are created by plotting the radial and interlobe
sealing lines with respect to ε and λ. ε represents the local position on the
transverse rotor profile while λ defines the relative rotation of the profile. This
shows at precisely what rotor position the different boundaries come into play.
There will be a unique boundary map for the main and the gate rotors.
The interlobe sealing line boundary is plotted onto the boundary map by setting
λ to equal the local meshing angle, θM:
71
(35)
The meshing angle is calculated at each location along the rotor profile curve
from ε = 0 to 1. The parameter λ is equal to the interlobe meshing angle, θM,
with the specific condition that λ = θM = 0 at the tip of the main rotor (as is the
case when the meshing angle is calculated when the rotors originate in the
‘home position’). The theory for calculation of the meshing angle is provided in
Appendix ‘A.2 Meshing Conditions for Conjugate Profile’.
The rotor to casing boundary will always occur at the same position on the
profile, i.e. at the tip of the rotor. The limits of this boundary depend on the
casing geometry. This will be explained in more detail in the following
examples. These examples will look at the boundary maps produced for
different profiles and how these boundary maps relate to sealing points
identified on transverse rotors for various rotor positions. The examples
illustrate how the full boundary interactions, which would otherwise need to be
presented in 3D or on numerous transverse rotor plots, can be captured in a
single 2D boundary map efficiently. The section that follows on from these
examples will then pick up on the theory of how to apply the boundary map to
analysis of a full 3D rotor.
72
4.6.1 EXAMPLE: ZERO BLOW-HOLE PROFILE

The concept of the boundary map will be discussed by using idealised,
theoretical profile examples. Figure 4-15 shows a simple profile concept in
which the root of the main rotor is at the pitch circle – in other words there is a
zero dedendum on the main rotor and a zero addendum on the gate rotor. The
flanks of the main rotor lobes are solely generated by a point on the tip of the
gate rotor to form part of an epicycloid curve; a special type of epitrochoid
where the generating point lies on the rolling radius (Weisstein ). The gate rotor
flute is similarly generated by the tip of the main rotor to form another
epitrochoid curve, this time with the generating point offset from the rolling
radius. In this profile the sealing line extends to the casing cusp resulting in zero
blow-hole.
Figure 4-15: Zero blow-hole rotor profile
Because of its limited number of component curves and the fact the interlobe
sealing line connects to the radial sealing line it is well suited to analyse the
interaction of these various sealing lines and how they form distinct
compression chambers. In Figure 4-16, the interlobe SL (sealing line) and radial
SL have been plotted onto the main rotor boundary map.
73
Main Rotor Surface - Chamber Boundaries

60
C3 C4
Local transverse rotor offset, λ (deg)
45
30
15
0
Interlobe SL
-15 Radial SL
-30
-45
C1 C2
-60
0 1
Rotor surface parameter, ε
Figure 4-16: Zero BH rotor boundary map for main rotor
The Radial SL occurs at a fixed region on the rotor profile. In the case of this
profile the main rotor radial SL is a single point at the tip of the main profile that
occurs when parameter ε = 0.5; hence the radial SL will fall along a vertical line.
The radial SL on the main rotor does not come into play until the rotor has
rotated through the main rotor cusp angle, φc1. Calculation of the cusp angles is
explained in Appendix A.
The Interlobe sealing line is calculated at each location along the rotor profile
curve from ε = 0 to 1. The values of the parameter λ for the interlobe SL cover a
range equal to 360°/z1. All examples presented in this chapter have 4 lobes on
the main rotor so the range is 360°/4 or 90°, in this case from -45° to +45° since
the profile is symmetrical. The scale on the y-axis from -60° to +60° degrees
allows the important details to be captured for asymmetric rotors. The pattern
of the SL boundaries will repeat when λ is offset by 360° therefore it is not
necessary to extend the scale any further as simple corrections can be applied
for larger values of λ. In this region of interest 4 distinct chambers have been
created by the interaction of the interlobe SL and radial SL boundaries. These
74
chambers have been identified as chambers C1, C2, C3 and C4 in Figure 4-16. At
this stage the objective is to clearly identify the boundaries that distinguish
these chambers and not what each ‘local’ chamber cycle angle is.
Gate Rotor Surface - Chamber Boundaries

60
C4
45
30
C2
15
-15
C3 Interlobe SL
Radial SL
-30
-45
C1
-60
0 1
Figure 4-17: Zero BH rotor boundary map for gate rotor
In Figure 4-17 a similar map has been produced for the gate rotor. Recall from
Figure 4-6, in which the rotor surface arrays are defined, that the limits of
parameter ε2 correspond to the tip of the gate rotor. Since the gate rotor tip is
formed by a cylindrical line of constant outer diameter (OD) for this rotor the
parameter ε has been set to zero at the mid-point of this OD line.
On the gate rotor the radial SL can occur for any location on the profile that
comes into ‘contact’ with the casing. The radial SL will come into effect for a
given value, ε, at a value of λ which can be derived using the interlobe SL: The
lower and upper parallel lines that bound chamber C3 are the interlobe and
radial SLs respectively. These lines are offset by a constant related to the gate
rotor cusp angle, φc2. The interlobe contact occurs along the centre line between
rotors for a cylindrical curve so that point must then rotate by φc2 to reach the
cusp.
75
It is important to remember that the parameter, λ, used on the gate rotor

boundary map references the offset of the main rotor. As a consequence, this
offset between the interlobe SL and the radial SL for the gate rotor boundary
map is actually φc2(z2/z1).
In Figure 4-18 a section of the profile has been highlighted from points A to B on
both the main and gate rotors. ‘A’ corresponds to ε = 0 and ‘B’ corresponds to
ε=1. For this given rotation (λ=0), the boundary points, G, have been
highlighted. The first subscript identifies the type of boundary which is either
interlobe - I, or radial – R. The lower case letter in the subscript is a unique
identifier to distinguish between each point.
The points identified in Figure 4-18 are then plotted on Figure 4-19 and Figure
4-20 in order to highlight the line AB which occurs when λ = 0. This illustrates
how for any given rotation angle, λ, the rotor surface boundary map can be used
to identify what chamber a given point on the surface is exposed to. For
example: a point within the region, GIa < ε < GIb, will be exposed to the chamber,
C1, when λ = 0.
C2 B2
B1 C4
GIc
GIb
GIa
A1 C1
C3 A2
Figure 4-18: Zero BH rotor boundary points (λ = 0)
76

60

45 C3 C4
30
15
0 Interlobe SL
-15
A1 GIa GIb GIc B1 Radial SL
Line AB
-30
-45
C1 C2
-60
0 1
Figure 4-19: Zero BH rotor boundary points on main rotor surface (λ = 0)

60
45
30
15
0 Interlobe SL
A2 GIa GIb GIc B2 Radial SL
-15
Line AB
-30
-45
-60
0 1
Figure 4-20: Zero BH rotor boundary points on gate rotor surface (λ = 0)
77
A second rotor position is presented in Figure 4-21, Figure 4-22 and Figure
4-23. The profiles have been rotated backwards until the tip of the main rotor
meets the HP cusp. In these figures line AB has now moved down to the
appropriate value of λ = -φC1.
At the tip of the main rotor the interlobe boundary, GIb, has been identified. At
this point the main rotor and gate rotor also both meet the cusp of the casing
therefore this point could equally have been identified by the radial boundaries
since: GIb = GR1 = GR2. This is confirmed on the rotor surface maps that show GIb
is located at the intersection of the interlobe SL and radial SL for both main and
gate rotor surfaces.
Between point GIb and B2 on the gate rotor it can be observed that the rotor
surface is not exposed to any rotor chambers for this rotor angle.
B1 B2
GIb
C2
-λ C1
GIa
A1 A2
C3
Figure 4-21: Zero BH rotor boundary points (λ = -φc1)
78

60

45
30
15
0 Interlobe SL
Radial SL
-15
A1
Line AB
-30 B1
-45 GIa GIb
-60
0 1
Figure 4-22: Zero BH rotor boundary points on main rotor surface (λ = -φc1)

60
45
30
15
0 Interlobe SL
Radial SL
-15
A2 Line AB
-30
B2
-45 GIa GIb
-60
0 1
Figure 4-23 Zero BH rotor boundary points on gate rotor surface (λ = -φc1)
79
4.6.2 EXAMPLE: SYMMETRIC PROFILE

For various reasons the zero BH profile is not practical for use in twin screw
compressors. Figure 4-24 shows a representation of the first manufactured
profile - the SRM symmetric profile. The root of the main rotor is now situated
on a circle below the pitch circle resulting in a dedendum on the main rotor and
an addendum on the gate rotor which is a feature that is present to some degree
on all modern profiles. All curves on the gate rotor are circles with centres
located on either the axis of the rotor or on the pitch circle. This results in
similar circular curves on the main rotor. Only the lower portion of the main
flank is non-circular – this is generated by the point on the gate rotor where the
small tip radius intersects with the large radius of the flute.
Figure 4-24: Symmetric (circular) rotor profile
This profile presents a complete contrast to the previous example as the

interlobe sealing line is very short and does not extend anywhere close to the
cusps on the casing. This poses some difficulty when distinguishing clear
chamber boundaries and is therefore a useful case study. The chambers and
boundaries have been identified on the rotors in Figure 4-25, Figure 4-26 and
Figure 4-27 which each show a different rotor position.
80
GR2
C2
GR1
C1
GIa
C3
Figure 4-25: Symmetric rotor boundary points (λ = -φC1)
C2 GR2’
C1
GIa’
C3
Figure 4-26: Symmetric rotor boundary points (-φC1 < λ < 0)
81
C2 GIb’’ C2
C3 GIa’’
Figure 4-27: Symmetric rotor boundary points (λ = 0)
At rotor position λ = -φc1 (as in Figure 4-25), chambers C1 and C2 are separated
by the boundary point GR1. As the rotor angle λ increases this point disappears
and a connection is formed between the chambers, known as the BH leakage
path. This connection does not actually disappear until the boundary point G Ib’’
is formed when λ = 0 (as in Figure 4-27). At this point the chamber C1 has
reduced to zero cross-sectional area.
The points identified in the previous rotor figures have been plotted on the
newly generated symmetric rotor boundary map for the main rotor in Figure
4-28. In this map the four chambers are not fully separated. Figure 4-25 shows
that when λ = -φc1, (the line through GIaGR1 on Figure 4-28 ) the range of the
rotor surface when ε > GR1 is certainly only exposed to chamber C2. By assessing
Figure 4-27, when λ = 0, it is reasonable to state that just before this, when λ < 0,
the range of the rotor surface GIa’’ < ε < GIb’’ will be exposed to chamber C1.
Between these two values of λ no boundary exists on the surface between

chambers C1 and C2. In order to map chamber properties to the surface, the
chamber exposure must be explicitly defined. Therefore in Figure 4-29 new
boundaries have been defined by linearly interpolating between the known
points.
82

60

45
30
15
GIa’’ GIb’’
0
Interlobe SL
-15 Radial SL
-30
GIa’
GR1
-45
GIa
-60
0 1
Figure 4-28: Symmetric rotor boundary points on main rotor

60
C3 C4
45
30
15
0 Interlobe SL
Radial SL
-15
New Boundary
-30
-45
C1 C2
-60
0 1
Figure 4-29: Symmetric rotor boundary map for main rotor with new boundaries
83

60
45
30
15
0
GIa’’ GIb’’
Interlobe SL
-15 Radial SL
-30
GIa’ GR2’
GR2
-45 GIa
-60
0 1
Figure 4-30: Symmetric rotor boundary points on gate rotor

60
C4
45
30
15 C2
0 Interlobe SL
Radial SL
-15 C3 New Boundary
-30
-45
C1
-60
0 1
Figure 4-31: Symmetric boundary map for gate rotor with new boundaries
84
The boundary points have also been plotted on the symmetric rotor boundary
map for the gate rotor in Figure 4-30. Following a similar procedure as for the
main rotor, new boundaries have been generated and plotted in Figure 4-31.
From Figure 4-25 it can be stated that between the points GIa and GR2 the gate
rotor surface is exposed to C1 so it follows that this is true along this horizontal
line between these points on Figure 4-30. Similarly from Figure 4-27 it can be
stated that when ε > G’’Ib this part of the surface is exposed to C2. These facts can
be stated with some degree of certainty based on this reasoning.
Between these two states no boundary exists and the surfaces between the
rotors and casing combine to form a complex 3D flow channel. This area on the
boundary map has been highlighted by the blue box. A considerable amount of
time was spent trying to define the most appropriate curve to define this
boundary on the gate rotor boundary map – the orange curve on Figure 4-30
actually approximates a transformation of the curve formed by the blow hole
definition in Appendix ‘A.3.3 Blow Hole Area’. Ultimately, any curve relies on an
assumed boundary surface in the region of what is a non homogeneous flow in
terms of pressure and temperature distributions. It was finally decided to avoid
unnecessary, unjustifiable complexity and to stick to linearly interpolating
between the two explicit states resulting in the boundaries added to Figure
4-31.
These new boundaries can be referred to as the blowhole boundaries. The new
boundary on the left hand side of the boundary map, where the values of ε are
smaller, is on the leading flank of the main rotor and forms the LP blowhole.
Conversely, on the right hand side which is the trailing flank of the main rotor,
the new boundary forms the more important HP blowhole.
85
4.6.3 EXAMPLE: ASYMMETRIC PROFILE
Figure 4-32: Asymmetric rotor profile (hybrid of zero BH and symmetric)
By combining the simple features of the zero BH profile and the symmetric
(circular) profile it is possible to approximate a more typical asymmetric profile
as shown in Figure 4-32. This example shows how the boundary map
generation procedure and the rules defined for creating new boundaries can be
applied to any typical profile. Due to the asymmetry of the profile, the tip of the
main rotor is now located at a position ε > 0.5, rather than at the mid-point.
86

60

45 C3 C4
30
15
0 Interlobe SL
Radial SL
-15
New Boundary
-30
-45
C1 C2
-60
0 1
Figure 4-33: Asymmetric rotor boundary map for main rotor

60
C4
45
30
15
C2
0 Interlobe SL
Radial SL
-15
C3 New Boundary
-30
-45
C1
-60
0 1
Figure 4-34: Asymmetric rotor boundary map for gate rotor
87
It can now be observed that with this asymmetric profile the length of the new
boundary line required to bridge the gap that forms the blow hole leakage path
is considerably smaller on the HP blowhole on the right hand side, particularly
on the main rotor boundary map.
When considering the boundary map generated by the gate rotor in all of these
examples the most notable difference from the main rotor maps is the greyed
out area bounded by the radial SL, within which the rotor surface is not exposed
to any chambers. This is certainly true for a theoretical compressor with zero
clearances however in practise fluid will exist between the rotor and the casing
due to the radial clearance gap. Assessing the fluid properties along the length
of this leakage path is somewhat beyond the capabilities of the thermodynamic
chamber model currently used. A reasonable compromise for this procedure is
to define an arbitrary boundary at the mid-point of the arc that lies on the gate
rotor OD. This approach is certainly appropriate if a tip sealing strip is located
at this point. This modified boundary map for the gate rotor presented in Figure
4-35 allows boundary conditions to be mapped onto the full surface of the gate
rotor.
88

60

45 C4
30
15
C2
0 Interlobe SL
Radial SL
-15 C3 New Boundary
-30
-45
C1
-60
0 1
Figure 4-35: Asymmetric rotor boundary map for gate rotor with adjusted radial
SL
89
4.6.4 EXAMPLE: N – PROFILE
Figure 4-36: N-profile rotors
Figure 4-36 shows a pair of N-profile rotors that are representative of the type
of profile used in modern screw compressors. This shares many of the basic
features of the asymmetric profile shown in Figure 4-32 but there is much more
control over a range of additional design parameters. For example: the flank
angles can be adjusted in the vicinity of the pitch circle where involute curves
ensure good torque transmission when required; the radius of all curves can be
adjusted; and undercutting generation from one rotor to another uses a small
radius on the generating profile instead of a single point.
The corresponding rotor boundary maps for this profile are shown in Figure
4-37 and Figure 4-38. These profiles will be used in exemplary figures in the
next section which addresses how the four chambers, C1 to C4, relate to the
more numerous chambers that actually exist within a twin screw compressor at
any instant.
90

60

45 C3 C4
30
15
0 Interlobe SL
Radial SL
-15
New Boundary
-30
-45
C1 C2
-60
0 1
Figure 4-37: N-profile boundary map for main rotor

60
C4
45
30
15
0 C3 C2 Interlobe SL
Radial SL
-15
New Boundary
-30
-45
C1
-60
0 1
Figure 4-38: N-profile boundary map for gate rotor
91
4.7 LOCAL CYCLE EXPOSURE OVER FULL ROTORS

The full 3-dimensional rotor surfaces are plotted in Figure 4-39 at a reference
compression cycle angle, θ. All rotor sections have been generated from the
surface arrays R1(ε1,γ1) and R2(ε2,γ2) as shown in Figure 4-6. The parameters r
and β can be found for a point on the rotor surface at any cycle angle using the
equations set out previously. The parameters r, β and γ were then transformed
to the global X,Y,Z co-ordinates and plotted using ‘Gnuplot’ (Williams, Kelley et
al. 2013).
In order to build up the full 3D surface, the surface at cycle angle θ is written to
memory then the cycle angle is advanced by 2π/z1 and the next surface is
written to memory until all lobes have been generated.
Due to the way the respective rotor parameters, r, β and γ, have been defined
for R1 and R2 with respect to a reference cycle angle , θ, the rotors would be
displayed in mesh once related back to the global Cartesian compressor co-
ordinate system, S0. However, in order to show the detail along the meshing
line the rotors have been plotted separately and are viewed as if from the other
rotor.
The current objective is to set out the procedure by which the instantaneous
cycle angle for the local compression chamber at any point on the surfaces
displayed in Figure 4-39 can be retrieved. The required parameter is the ‘local
cycle angle’, θlocal; this is based on θ which is effectively a ‘reference cycle angle’
because it only defines the cycle angle for a single ‘reference compression
chamber’ that is used to define the rotor position.
92
Figure 4-39: Full rotor surfaces (3D domain)
The first step to finding the local cycle angle, θlocal, is to identify where the
surface point lies on the rotor boundary map in order to establish what chamber
offset is required. This requires the rotor surface parameter ε; and the local
transverse rotor offset λ. The latter is a function of the axial position parameter,
γ; and the reference cycle angle, θ, as set out previously and summarised below:
(36)
(37)
In order to use the rotor chamber boundary map the offset should lie in the
range: -180 < λ1 < 180. In the case of the gate rotor: -180(z2/z1) < λ1 <
180(z2/z1). To ensure the parameter λ is within this range a correction must
be applied:
93
(38)
(39)
Where n is an appropriate positive or negative integer. The maximum value of

this integer is limited by the fact that the reference cycle angle used to define
the cycle angle in the reference chamber cannot exceed the maximum cycle
angle for the rotor pair.
Once known, λ is used along with the rotor surface parameter, ε1 or ε2, for the
main and gate, to identify the relevant rotor chamber as C1, C2, C3 or C4 by using
the rotor boundary maps of Figure 4-37 and Figure 4-38 for the main and gate
rotors respectively.
Figure 4-40 has been created based on the same boundary map of Figure 4-37
in order to illustrate how the local cycle angle needs to be adjusted for each of
the four chambers. Chamber C4 has been set as the reference chamber. By
assuming that -180 < λ < 180, no adjustment of λ is required and the local cycle
angle in chamber C4 will equal the reference angle θ. Crossing the radial SL from
C4 to C3 advances the local cycle angle by 360/z1. Crossing the interlobe SL from
C4 to C2 advances the local cycle angle by 360. Crossing the radial SL again from
C2 to C1 results in a total offset of 360 + 360/z1.
94

60
θ + 360/z1 θ
45
30
C3 C4
15
0 Interlobe SL
Radial SL
-15
New Boundary
-30
-45
C1 C2
-60 θ + 360 + 360/z1 θ + 360
0 1
Figure 4-40: Main rotor chamber offsets
If any adjustment is made to λ using equation (38) it is important to include that

term were calculating the local cycle angle. The general equations for
calculation of the local cycle angles for each chamber on the main rotor are
given in Table 4-3.
Table 4-3: Local cycle angle corrections on main rotor
Chamber Reference Calculation of local theta angle , θlocal
C1
C2
C3
C4
95
Figure 4-41 shows the chamber offsets for the gate rotor boundary map. The
boundaries and chambers on this map are the same as those on the main rotor
boundary map, albeit with different topography, so it may come as a surprise to
see that the equations describing the chamber offsets are slightly different
across the interlobe sealing line. The new term highlighted in blue is required to
account for any rotor lobe offset that occurs if the difference between z2 and z1
is greater than 1.

60
C4
45
30
θ
15
C3 C2
0 Interlobe SL
θ + 360((z2-1)/z1)
θ + 360/z1
Radial SL
-15
-30
C1 New Boundary
θ + 360((z2-1)/z1) + 360/z1
-45
-60
0 1
Figure 4-41: Gate rotor chamber offsets
Equations (31), (33) and (36) all include the rotor lobe offset angle, φoff, which
was defined in equation (16). When used in conjunction with the offsets defined
above, these equations ensure that the cycle angle calculated for surfaces on
each rotor, when exposed to the same chamber near the end of the cycle, is the
same. Inevitably this means that the cycle angle will be different for surfaces on
each rotor, when exposed to a chamber near the start of the cycle; when the
difference between z2 and z1 is greater than 1. The general equations for
calculation of the local cycle angles for each chamber on the gate rotor are given
96
in Table 4-4. The other additional terms highlighted in yellow are due to the
difference between equations (38) and (39).
Table 4-4: Local cycle angle corrections on gate rotor
Chamber Reference Calculation of local theta angle , θlocal
C1
C2
C3
C4
In one sense it is a fairly trivial task to correct chamber offset across the main
interlobe and radial boundaries. However only by analysing the rotor boundary
maps can the subtle complexity of the problem be appreciated. All of the rotor
profile examples show that along at least some portion of the interlobe SL, the
chamber C1 is adjacent to C4 due to the discontinuity of the radial SL so the
offset across this portion of the interlobe SL is 2π + 2π/z1. Referring back to
Figure 4-12 in which the sealing boundaries are shown as boundary points, the
adjacent chambers in green and red around meshing lobes of the main rotor are
equivalent to C1 and C4.
The rotor surfaces have been re-plotted in Figure 4-42 with the local cycle
angle, θlocal, represented by the colour contours. Areas that are the same colour
can be easily identified as individual compression chambers. In order to build
up the full rotor the procedure was repeated after advancing the reference cycle
angle θ for each lobe:
(40)
97
On the gate rotor on the right hand side of Figure 4-42 two white dashed lines
have been added. These are the newly defined boundaries that were introduced
in Figure 4-38 to form a boundary across the blow-hole leakage paths – the
more important HP blowhole is the smaller dashed line on the left. The fact that
these seem to separate the chambers in a sensible manner provides some
degree of validation regarding the boundary definition.
Figure 4-42: Full rotor surfaces with surface contour plot of local cycle angle
In the region of the boundaries some of the tiles used to produce this plot
suggest a transition region across the boundary however this is an artefact
introduced by the averaging across nodes that fall in different chambers. All
computations are done at the nodal positions therefore the calculated values
will only fall within the specified local cycle angles.
98
4.8 MAPPING PRESSURE AND TEMPERATURE ON SURFACES
4.8.1 TEMPERATURE DISTRIBUTION

The homogenous fluid properties within a single compression chamber are
calculated for a given operating fluid and duty point using the established
chamber model detailed in Chapter 2. The fluid properties are output with
respect to the compression cycle angle (in this case normalised to be zero at
maximum chamber volume) as presented in Figure 4-43.
Figure 4-43: Thermodynamic results from chamber model
With these results, and knowing the local cycle angle at any particular point on
the rotor and casing surfaces it is a matter of referencing the appropriate fluid
properties from the thermodynamic results.
Figure 4-44 shows the instantaneous, homogeneous gas temperature from the
chamber model results mapped onto the main and gate rotor surfaces. This
output will vary depending on the reference cycle angle θ, which controls the
rotor positions, however once the cycle has gone through 360°/z1 the result
presented would repeat.
99
Figure 4-44: Instantaneous gas temperature at rotor surface.
In Figure 4-45 the time-varying fluid properties have been averaged over the
full rotor cycle (360°) for each location on the rotor surface in order to
approximate steady state boundary conditions. Averaged boundary conditions
would greatly simplify any thermal analysis.
100
Figure 4-45: Time averaged gas temperature at rotor surface
The temperature results mapped onto the rotors are mapped onto the casing
surfaces in the same way using the calculated local casing cycle exposure. The
averaging of casing and rotor temperatures is discussed in more detail in the
next chapter.
4.8.2 PRESSURE DISTRIBUTIONS

Pressure distributions are found in exactly the same way as the temperature
distributions as shown in Figure 4-43 and Figure 4-44. With pressure, the time
varying forces on the rotor are of interest therefore it is not appropriate to
average the instantaneous pressure distributions. The pressure results would
need to be repeated over the range 360°/z1, with appropriate time resolution
to capture the full time history for the pressure boundary conditions.
In order for pressure results to be utilised in a useful way it is necessary to

resolve the net pressure forces on the rotors due to the pressure distribution on
all rotor surfaces i.e. radial loads, axial loads and torque. Calculation of these
forces was not included in the scope of this thesis and will be addressed in
further work.
101

Accurate surface boundary mapping and calculation of chamber exposure is
critical for any kind of analysis that relates fluid properties within the chamber
to the actual rotor geometry. This includes determination of thermal boundary
conditions which will be utilised in the next chapter, and in future work, the
calculation of net forces of the rotor body due to pressure loading.
These procedures will allow existing results from a non-dimensional chamber

model to be utilised in new ways such as estimating local operational clearance
distortions. In addition it provides a better understanding of the complexities of
the chamber interaction for various rotor profiles including: interlobe and
radial sealing line interaction with the blow-hole leakage path; and the degree
of cycle offset between various adjacent chambers.
This will be used as the basis for an initial estimate of the actual temperature of
the rotor metal in the next chapter, without using FEA. This will allow fast,
efficient estimation of thermal distortions during rotor design for a specific
operating duty.
Other future work that has been identified is the use of the pressure and
temperature results for 3D FEA analyses. Another potential use of this kind of
result would be to use the instantaneous, homogeneous fluid properties at a
fixed rotor position as the initial boundary conditions for CFD simulation.
102
Chapter 5
Operational Clearance
Distortions
5.1 INTRODUCTION
The boundary conditions calculated in the previous chapter can be used for
estimation of the surface temperature distribution on the rotors and casing. It is
then possible to estimate thermal displacements of individual components and
ultimately how this affects the operational clearances.
A procedure is presented here to estimate local clearance distortions in a twin

screw compressor using boundary conditions derived from a chamber model.
Time varying boundary conditions from the non-dimensional model are
mapped onto rotor and casing surface arrays. The fluid boundary temperatures
are time-averaged and then used to estimate the local rotor and casing
temperatures. Heat transfer assumptions that represent the extreme case for
component temperature distributions are presented. This approach aims to
meet the objective to provide a fast, efficient, estimation of thermal distortions
for any rotor design or operating duty early in the design process. Estimated
rotor and casing temperatures will be used to analyse how thermal distortions
will change the compressor clearances, and consequently how this affects the
compressor performance and reliability. With this procedure, relative local
change in clearances between rotors and casing can be estimated without the
need for FEA.
103
5.2 APPROXIMATING TEMPERATURE DISTRIBUTIONS
5.2.1 CASING TEMPERATURE

In the previous chapter a procedure was described to find the local fluid
boundary temperature on the internal surfaces of the casing bores, B1 and B2.
The local temperatures TB1 and TB2 are time averaged values which are unique
for every position of the rotor bore surfaces; the position is defined by the
parameters β and γ so that the local temperature on the main and gate rotor
bores are defined respectively: TB1(β1,γ) and TB2(β2,γ)
In Figure 5-1 the average local temperature has been plotted onto the surfaces
B1 and B2 as a colour contour which shows the temperature distribution with
respect to the surface parameters β and γ.
Outlet End
B2 - Gate bore B1 - Main bore
γ γ
β2 Inlet End β1
Figure 5-1: Average fluid temperature at casing surfaces
Converting the surface location to be defined on the global co-ordinate system

for the compressor, this temperature distribution can be plotted in 3D as shown
in Figure 5-2; which shows more intuitively how these temperatures vary
around the compressor. This shows the two intersecting cylinders that form the
compressor casing bores and is a very limited representation of only the
internal casing geometry. Externally, the casing will have some thickness and
various other features such as port, ribs, flanges and feet which will all effect the
actual temperature distribution in the casing body.
104
Gate rotor bore
Location of
Main rotor bore
discharge port
Figure 5-2: Average fluid temperature at casing surfaces in 3D
With such a simplified representation of the actual casing geometry there is

limited scope for thermal analysis. By considering a hypothetical scenario; with
no conduction across the surface of the casing; and no external heat transfer
from the casing, the steady state result is that the casing surface temperature of
the metal will be the same as the fluid boundary temperature. This hypothetical
case is of interest as it results in the peak local temperatures and temperature
gradients that are possible.
As will be discussed later - it is useful to average the temperature over certain

regions of the casing – namely, over a transverse cross-section of the casing. For
a given transverse plane on the compressor, when the parameter γ, which
describes axial position, is fixed, the average temperature over that plane can be
found by averaging the temperature around both rotor bores:
(41)
5.2.2 ROTOR TEMPERATURE

The instantaneous local fluid boundary temperature on the rotor surfaces, R1
and R2 and defined: T*R1(ε1,γ,θ) and T*R2(ε2,γ,θ)
105
The parameters ε and γ describe the location on the surface, while θ describes
the cycle angle. An example of the instantaneous boundary temperature
distribution is plotted on the surfaces R1 and R2 as shown in Figure 5-3
Outlet End Outlet End °C
R1 R2
γ γ
Inlet End ε1 ε2 Inlet End
Figure 5-3: Instantaneous boundary temperature at rotor surfaces
Figure 5-3 shows similar chamber boundaries to those that can be observed in
the rotor chamber boundary maps of Figure 4-40 and Figure 4-41 in the
previous chapter (albeit with different orientation). The subtle difference is that
rotor boundary maps were defined using a parameter λ that meant they were
independent of the cycle angle. In this case the vertical axis is now represented
by the axial parameter γ; as the cycle advances, the chambers appear to
translate as shown in Figure 5-4.
106
Outlet End
°C
R1 R1
At cycle angle θ At cycle angle θ + 60°
γ γ
Inlet End ε1 ε1
Figure 5-4: Instantaneous temperature at main rotor surface at different cycle

angles
The local temperature will cycle every time θ is incremented by 360°. This
means that for even a large slow compressor running at 1500rpm, the
temperature will cycle every 0.04 seconds. ‘Steady state’ conditions can be
approximated by taking time averaged values that will now only vary as a
function of the surface parameters:
(42)
(43)
107
R1 R2
γ γ
Figure 5-5: Time-averaged boundary temperature at rotor surfaces
Surfaces R1 and R2 have been re-plotted in Figure 5-5 with the colour contour
representing the time averaged surface boundary temperature. The colour
contours representing temperature have been plotted on the same scale as used
in Figure 5-3 which highlights how much lower the peak temperature is after
averaging.
In Figure 5-6, the time averaged boundary temperatures have been plotted on
the 3D main rotor surface. This is an accurate representation of the main body
of the rotor however it is still a simplification of the actual geometry as the
details of the rotor shafts that extend from this rotor body are not known. These
shafts would of course have an effect of the thermal behaviour of the rotors.
108
°C
Figure 5-6: Time averaged boundary temperature on main rotor
As was the case for the casing representation, there is a limit to the scope for
thermal analysis that can be done without defining the full 3D rotor bodies
including all shaft dimensions. To facilitate analysis the following assumption is
made: there is no heat conduction in the axial direction along the length of the
rotors. This assumes no heat flux from the hot to the cold end of the rotors and
no heat flux from the rotor body to the rotor shafts and bearings. Neglecting
axial conduction results in the peak transverse temperatures and temperature
gradients along the axis of the rotors. This allows the problem to be reduced to
a 2D problem for each transverse cross section of the rotors. However, this
would still require a numerical solution to determine the 2D temperature
distribution due to surface heat transfer and conduction within the rotor.
Averaging the boundary temperature over the transverse cross section of each
respective rotor using equations (44) and (45) yields the rotor ‘planar averaged’
temperatures; these are plotted in Figure 5-7:
(44)
(45)
109
R1 R2
γ γ
Figure 5-7: Planar averaged boundary temperature on rotor surfaces
As this temperature only varies with axial location the axial temperature
distribution for both rotors can be directly compared on a single plot as in
Figure 5-8. The reason the discharge end temperature is hotter on the main
rotor is because it is exposed to the 4 hottest chambers over a full rotation
(since z1 = 4) while the gate rotor is exposed to these same 4 chambers for only
4/6 (z1/z2) of a rotation and the remaining time it is exposed to 2 cooler
chambers.
120
100
Temperature (°C)
80
60
Main Rotor
40 Gate Rotor
20
0
0 50 100 150 200 250 300
Axial parameter, γ (deg)
Figure 5-8: Planar averaged temperature distribution on rotors
110
5.3 THERMAL DISTORTION OF INTERLOBE GAP
5.3.1 INTERLOBE SEALING LINE

The objective here is to approximate the relative change in the local interlobe
sealing line gap, ΔGI, due to thermal distortion of the rotors and casing. This
local distortion will be unique for each position along the full length of the
sealing line. In addition to this the position of the sealing line changes so that a
given point on the sealing line will be formed between rotor surfaces that are
exposed to different temperatures resulting in different thermal distortions as
the cycle progresses.
GI
Figure 5-9: Side projection of interlobe sealing line
Figure 5-9 shows the entire length of the interlobe sealing line from a side
projection (viewed along the negative X direction of Figure 4-1). The sealing
line between the rotors exists between the limits of the low pressure (LP) and
high pressure (HP) cusps. As the rotors turn, this sealing line translates from
the suction face (LP plane), at the left hand side, to the discharge face (HP
plane), at the right hand side.
Over the course of the compression cycle, every location on surfaces R1 and R2
will be in mesh at some point. During steady state operation, the local interlobe
111
gap, GI, and, just as importantly, the change in the gap, ΔGI, will remain constant
when defined with respect to a fixed point on each of the rotor surfaces. For this
reason it is appropriate to define the local interlobe gap with respect to its
corresponding location on the main rotor surface, R1: GI(ε1,γ). Note that this is
an equivalent alternative to equation (4) in Chapter 2.4.1, which described the
local interlobe gap as a function of the position along the interlobe sealing line
and the cycle angle GI(l1,θ).
5.3.2 LOCAL ROTOR DISTORTION

In order to approximate thermal distortion of each rotor a 2D transverse cross
section of the rotors was considered. By using the planar averaged
temperatures defined previously, a uniform temperature can be assumed across
the entire transverse cross section of each rotor. With this assumption uniform
thermal expansion occurs without any thermal stresses. This neglects shear
stresses between adjacent transverse sections. Taking the axis centre point as a
datum, thermal expansion can be calculated for each rotor:
(46)
(47)
The local radius, r, is known at any location for a given rotor profile and can be
identified with respect to the surface parameter, ε: r1(ε1)
μR, is the coefficient of thermal expansion for the rotor material which is
assumed to be common for the main and gate rotors.
The ambient temperature, Ta, is the temperature at which the design clearances
are defined. The planar averaged temperature of the rotor varies as a function
of the axial surface parameter, γ (see Figure 5-8): TRP1(γ)
Therefore, for a given rotor pair running at a steady state duty, the local thermal
distortion on each rotor can be approximated in such a way that it can be
described in terms of its position of the rotor surface: Δr1(ε1,γ); Δr2(ε2,γ).
112
5.3.3 LOCAL CASING DISTORTION

The bearings that locate the rotors are situated in the casing so the interlobe
gap can also be affected by casing thermal distortions. In reality the bearings
will be located on the rotor shafts some distance from the meshing region of the
rotor bodies. Even if the actual geometry were defined it is not a trivial matter
to approximate the casing temperature remote from the compression process.
This would require detailed analysis of a specific case which is not the aim of
this exercise. For this more general investigation of thermal behaviour, two
transverse planes were defined as shown in Figure 5-10, one at the suction end
and the other at the discharge end.
A, suction
φW1
A, discharge
Figure 5-10: Transverse planes at suction and discharge of casing
Where the centre line of each rotor bore intersects with one of these planes the
distance, A, has been identified. The averaged temperature on each respective
plane will be used to estimate ΔA at each end, resulting in non-parallel axes.
The 2D thermal analysis is similar to what was done for the rotors. A uniform
casing temperature is estimated on each plane using the casing planar averaged
113
temperature equation defined previously (41); where these temperatures are

defined as follows:
Suction plane temperature: TCP1 = TCP(γ = 0)
Discharge plane temperature: TCP2 = TCP(γ = φw1)
Assuming that each rotor axis intersects with the appropriate casing bore axis
at these planes; and that rotor deflection due to rotor bending is negligible; the
rotor centre distance can be defined at any axial position, γ, by interpolating
between the end planes:
(48)
5.3.4 RELATIVE GAP DISTORTION

Procedures have now been detailed to find the local distortions: Δr1, Δr2 and ΔA
for a given cross section of the rotors. This section details how these component
distortions are used to calculate the resulting relative distortion at the interlobe
gap. The first step of this analysis will be to determine the transverse
component of the interlobe gap, GIT.
ψPA
Figure 5-11: Transverse analysis of interlobe gap
The dimensions r1, r2, and A have been plotted on Figure 5-11. In order to relate
these dimensions to the transverse interlobe gap, GIT, the horizontal and vertical
components have been defined. The relationship between the horizontal
components is:
114
(49)
Similarly:
(50)
For the interlobe gap highlighted in Figure 5-11 the relationship between the
vertical components is:
(51)
However, the above is only true when |y2| > |y1| AND when y1 and y2 are
positive. In the magnified gap in Figure 5-11, the body of the gate rotor is
‘above’ the body of the main rotor; this is related to the surface normal. For
conjugate rotors, the line of the surface normal at any location, currently in
mesh, must intersect with the rotor pitch point. This behaviour can be used to
set out general rules as illustrated in Figure 5-12. An axial projection of an
arbitrary sealing line is shown in red; this is representative of the sealing line
path for any profile with an addendum and dedendum. The line of the surface
normal will always radiate from the pitch point as shown. Where this line
intersects the meshing point between rotors, the rotor closer to the pitch point
will be ‘smaller’ in the y direction. This behaviour was used to construct
equation (52) and (53).
y2
y1
+y y1
y2
y1 y2
r1w
|y1| > |y2| |y2| > |y1|
Figure 5-12: Pressure angle intersecting with pitch point
115
(52)
(53)
The change in the transverse interlobe gap can now be calculated as:
(54)
The pressure angle, ψPA, is defined in Figure 5-13 and is the angle between the x
axis and the surface normal line at a meshing point. This angle is measured on
the global compressor co-ordinate system and is common for both rotors.
Figure 5-13: Angles on transverse section
(55)
θM is the meshing angle that describes the rotation of the main rotor. ψ01 can be
described as the ‘home pressure angle’ (Holmes 1990); this is the normal to the
profile curve measured relative to the profile co-ordinate system S01. Details for
calculation of θM and ψ01 and provided in appendix ‘A.2 Meshing Conditions for
Conjugate Profile’.
The minimum gap between the surfaces of the main and gate rotors is the
distance along a line normal to the rotor surfaces – the normal gap. In the case
of rotors with no helix angle this gap would be the same as the transverse gap,
GIT. When a helix angle is introduced the normal gap will be smaller than the
transverse gap. The degree to which the normal gap is reduced due to the local
116
rotor helix also depends on the local steepness of the rotor flank. Holmes
(Holmes 1990) provides a derivation of the following equation used to convert
from a transverse to a normal gap:
(56)
ψH1 is the local helix angle on the main rotor, this increases with an increase in
the local radius according to the equation below, where h1 is the lead length
that describes the main rotor helix:
(57)
αF1 is described by Holmes as the polar flank angle, which is measured from a
radial line that intersects the meshing point, to the line of pressure. This is
included in Figure 5-13 and can be defined as below; where φ01 is the local
profile angle measured from the profile co-ordinate system:
(58)
Since GI will typically be many orders of magnitude smaller than the profile
dimensions, the following approximations would result in negligible errors. In
practical terms this allows the calculations to be performed using only the main
rotor profile co-ordinates.
(59)
117
5.4 THERMAL DISTORTION OF RADIAL GAP
5.4.1 RADIAL SEALING LINES

The radial sealing points between the rotor and casing have been highlighted
for the main and gate rotors in Figure 5-14. These occur at the maximum rotor
radii, ro1 and ro2. In the event that the gate rotor features a region where the OD
has the same max value, the radial sealing point has been set to be on the
trailing side on the rotor lobe such that this point defines the port opening,
when timing is most critical. This is as opposed to choosing a point on the
leading side of the lobe that would control the port closing.
Figure 5-14: Radial sealing points on main and gate rotors
A single compression chamber will be enclosed by one radial sealing line on the
leading tip of each rotor and one on the trailing tip of each rotor. For the
purposes of thermal analysis of the radial gap only one of these sealing lines
needs to be considered since one is just an offset of the other. Each of these
sealing lines is made up of one helical curve on the main rotor and another on
the gate rotor. These sealing lines have been projected onto the casing bore
surfaces in Figure 5-15.
118
Figure 5-15: Radial sealing lines
On the main rotor on the right hand side of Figure 5-15, the angular position of
the radial sealing lines have been identified with β1l and β1t for the leading and
trailing tip respectively. These angles have been measured at γ = 0, on the inlet
(LP) plane. Due to the overlap of the rotor bores the surfaces B1 and B2 only
extend from one cusp to the other therefore the radial sealing line can only exist
in this region where: φc1 < β1 < (360 – φc1)
During steady state operation, the local radial gap, GR, and the change in the gap,
ΔGR, will remain constant when defined with respect to a fixed point on the
casing surface. For this reason it is appropriate to identify the local radial gap
with respect to its corresponding location on the casing surfaces surface, B1 or
B2 respectively: GR1(β1,γ) and GR2(β2,γ)
5.4.2 LOCAL DISTORTIONS

The local rotor distortions calculated previously in equations (46) and (47) are
used to calculate the thermal distortion at the tip of the main and gate rotors
using the outer radii ro1 and ro2 respectively:
(60)
(61)
The outer radius is a constant along the full length of the radial sealing line for
each rotor. The planar averaged temperature, TRP, will vary with respect to the
119
axial parameter, γ. As previously stated, these equations assume a datum along

the axis of the rotors for the rotor thermal expansion.
The intuitive datum for analysing displacements in the full casing due to
thermal distortion would generally be where the casing is secured to its base. In
this instance the only distortion of interest is along the lines that radiate from
the common axis of the rotor and casing bore. For this reason, two datum
locations have been defined on the casing - one on the main rotor axis and one
on the gate rotor axis; this assumption is discussed in more detail in ‘Appendix
B.2.2 Datum for Relative Clearance Distortion’. Axial distortions will have a
negligible effect on the radial gap therefore these will be neglected. This
essentially allows the analysis of the radial gap to be treated as a 1D problem.
The rotor distortion has already been discussed so the remaining unknown is
the radial distortion on the casing bore, relative to the centre axis of that bore:
Δrc1(β1,γ) and Δrc2(β1,γ)
As for the rotors, the realistic thermal distortion of the casing simply cannot be
calculated without taking into account geometrical features and how these
interact with temperature gradients to cause thermal stresses. If the casing
temperature was approximately uniform over a transverse cross section of the
rotors it would be possible to perform a similar analysis as for the rotors since
there are zero thermal stresses over a section with uniform temperature
(neglecting stresses transferred from adjacent planar cross sections).
Unfortunately, referring back to Figure 5-1, it is clear that the high

temperatures on the casing are highly localised to the region of the discharge
port. While using a planar average temperature is reasonable for distortions in
centre distance, A, this kind of averaging is not appropriate for analysis of the
radial gap. The only option, short of creating a finite element model, is to treat
each point on the casing in isolation, and deem the rest of the casing to be at the
same temperature, circumventing the issue of thermal stresses, for the purpose
of estimating the local distortion at that point.
(62)
120
(63)
Providing casing thermal gradients are not very severe; and that the
compressor casing has been sensibly designed in such a way that the thermal
expansion of the bore will not be restricted; the expectation is that an overall
approximation of casing thermal distortion can be achieved; this will be an
improvement over completely neglecting casing thermal expansion and in no
way attempts to provide a fully realistic thermal analysis of the casing.
The analysis of the local radial gap, normal to the surface of the casing bore, is
considerably easier to calculate than for the interlobe gap:
(64)
(65)
5.5 MODELLING WITH OPERATIONAL CLEARANCE

CORRECTIONS
5.5.1 INTEGRATION WITH GEOMETRY CALCULATION PROGRAM

The described calculations were implemented in the program GEOM used for
the calculation of all geometric parameters. This program outputs only the main
geometric characteristics required; namely, the chamber volume and all flow
areas through the ports and leakage paths, and how these vary throughout the
compression cycle.
The minimum requirement for calculation of the leakage area through a sealing
line gap is the instantaneous length of that sealing line and the average gap
through the sealing line. With non-uniform clearance distributions around the
surface of the rotor profile it is necessary to integrate the local gap along the
path of the sealing line:
121
(66)
For a designed clearance distribution the local gap, GI, varies along the length of
the sealing line. Assuming there is no rotor taper or axis offset towards one end
of the rotors, a point fixed on this sealing line, as shown previously in Figure
5-9, would have a constant local gap by design such that the gap could be
described as a function of the position on the rotor surface: GI(ε1). In this
situation only the limits of integration for the active sealing line are required.
When considering distortions due to local rotor and casing temperature, the
position of the sealing line within the compressor must also be calculated since
the local distortion depends on the surface location: ΔGI(ε1,γ). The local gap
corrected for operational distortions is therefore a function that varies with
respect to: GI*(ε1,γ).
In practise, equation (66) is solved numerically by calculating GI* at discrete

points along the sealing line. For a given point in the cycle, θ, running at a steady
state operating condition, the corrected leakage area through the interlobe gap,
AI*, has a constant value. This calculation is repeated over a finite number of
steps during the compression cycle in order to describe the leakage area curve,
AI*(θ).
The procedure is similar when calculating the corrected radial gap so that:
GR1*(β1,γ) and GR2*(β2,γ).
The leakage area for the radial gap is the sum of the area on the main and gate
rotors; where ‘s’ is the length along the relevant curve:
(67)
Examples of operational clearance distortions and the resulting impact of

clearance distributions and area curves will be provided for the case studies in
Chapter 6.
122
5.5.2 INTEGRATION WITH CHAMBER MODEL
Fluid
Map boundary
boundary
conditions
conditions
Calculate operational Clearance

deflections corrections
Compressor Calculate geometry Geometry

input data characteristics output data
No
Application Solve Performance

input data thermodynamics output data
Converged?
Yes
New / Updated Procedures

End
Existing Procedures
Figure 5-16: Flow chart for performance calculation with operational clearances
The boxes in Figure 5-16 represent the operations that are performed while the
parallelograms represent the data that is input and output from each program.
More specifically, this data includes:
Fluid boundary conditions:
 Main rotor temperature distribution, TR1(ε1,γ)

 Gate rotor temperature distribution, TR2(ε2,γ)
 Casing surfaces temperature distribution, TB1(β1,γ) and TB1(β1,γ)
Clearance corrections:
 Local change in interlobe gap, ΔGI(ε1,γ)
123
 Local change in radial gap, ΔGR1(β1,γ) and ΔGR2(β2,γ)
Compressor input data:
 Rotor profile co-ordinates

 Rotor parameters (such as wrap angle and Vi)
 Clearance parameters including designed clearance distribution
Geometry output data:
 Chamber volume curve, V(θ)

 Flow area curves, A1(θ), A2(θ)…
Application input data:
 Fluid properties (ideal gas properties were used)

 Compression duty, p1, T1, pressure ratio
 Compressor parameters (such as bearing type / oil injection)
Performance output data:
 Fluid properties during compression cycle, p(θ), T(θ)…

 Net flow / volumetric efficiency
 Net power / adiabatic efficiency
The procedures: ‘Calculate geometry characteristics’ and ‘Solve

thermodynamics’ were introduced in Chapter 2. The procedure to ‘Map
boundary conditions’ was described in the Chapter 4. The procedure to
‘Calculate operational deflections’ is as detailed in this chapter. All procedures
were programmed using FORTRAN. The program to ‘Calculate geometry
characteristics’ was based on existing routines in a program GEOM developed at
City University which were modified to support the more complex clearance
distributions during calculation of the leakage areas.
‘Geometry characteristics’ are calculated using the appropriate ‘Compressor

input data’ and ‘Clearance corrections’ (initially set to zero). These geometry
characteristics are used along with the ‘Application input data’ to solve the
thermodynamics, giving the ‘Performance output data’. The performance output
124
data includes the fluid temperature as it varies throughout the compression

process, T(θ). Once available, the temperature curve is used to estimate ‘Fluid
boundary conditions’ and ‘Clearance corrections’. If clearance distortions are
significant these will alter the flow area through the leakage paths within the
compressor therefore the geometry characteristics and thermodynamics should
be re-calculated. If the results are significantly different compared to the
previous solution, then the process should be iterated until the solution
converges.
125

For the purposes of assessing how thermal distortions affect local clearances
and performance at a conceptual design stage, significant simplifying
assumptions must be made. This approach does not attempt to provide
maximum realism for a particular case but rather provides a general
approximation of rotor thermal behaviour. An objective in this work to quantify
local clearance variations is to assess reliability therefore the assumptions that
have been made aim to represent extreme cases for thermal distortions.
Ultimately, temperature distribution and thermal distortion need to be

validated as far as possible which will be the subject of the Chapter 6. The effect
of the revised clearances on the performance will be quantified and compared
with experimental results.
126
Chapter 6
Case Studies
6.1 INTRODUCTION
Procedures for the analysis of the thermal deformation of clearances have been
described in the Chapters 4 and 5. These have been integrated with a well
established twin screw compressor modelling procedure, detailed in Chapter 2.
The resulting model enables investigation of localised clearance distortions and
quantification of how these changes affect the compressor performance. In this
chapter, case studies are presented in order to show the capabilities of this
model and to assess how accurate and applicable the model is when used in the
practical context of designing screw compressors for industrial applications.
This research was performed in a commercial engineering environment. This

means that there was a good resource of test data available. However the
downside is that there is less flexibility in modifying designs, measuring specific
parameters or in running at extended operating duties. The solution to this
issue was to utilise test data from various different test cases which each
allowed a specific detail or application of the model to be assessed.
The first case covers Model Sensitivity to Rotor Parameters. The objectives of
this case study were to validate that surface boundary conditions were
produced as expected, with sensibly bounded chambers that are consistent with
known compressor behaviour. The results were used to assess how these
boundary conditions are affected by changes to the rotor geometry to ensure
the calculation procedures are robust. This initial validation was only
concerned with the distribution of temperature.
In section ‘6.3 Model Sensitivity to Clearance Modifications’, the procedures

detailed in Chapter 5 to predict clearance distortions were employed. The effect
of modified operational clearances on the simulated thermodynamic
127
performance is presented and compared against available performance tests to

give a general assessment of the model behaviour and sensitivity.
Section ‘6.4 Investigating Interlobe Clearance Distortion’ particularly focuses on

the impact of thermal distortions on the interlobe clearance distribution for an
oil injected machine with direct rotor to rotor drive.
‘6.5 Investigating Radial Clearance’ draws on extensive test results obtained

from an oil free HS204 compressor, over a wide range of temperatures and
speeds. This case study features the added complexity of the compressor casing
having a cooling jacket. The results obtained with the HS204 compressor
include measurement of the radial clearance gap which was used to provide
some verification of the predicted radial clearance deformation.
Table 6-1: Air ideal gas properties
Specific heat ratio, CP/CV 1.4

Gas Constant, RG (Jkg-1K) 287
Compressibility factor, Z 1
All tests were performed with atmospheric air. This fluid has been represented
in all models as an ideal gas with the properties provided in Table 6-1. All test
results presented in this chapter have flows calculated in accordance with the
standard ISO 5167-2:2003 (International Organization for Standardization b)
and have been corrected to a standard suction pressure of 1 Bar absolute using
the corrections described in ISO 1217:2009 (International Organization for
Standardization a). The measured flow was also corrected to account for
deviations in the actual speed achieved on test; within the acceptable deviations
set out in this standard.
128
6.2 MODEL SENSITIVITY TO ROTOR PARAMETERS
6.2.1 OVERVIEW
Assessment of the model sensitivity to rotor parameters was based on an oil
free machine compressing air. The modelled temperature throughout the
compression process was mapped onto the rotor and casing surfaces. The
resulting instantaneous and averaged surface temperature distributions are
presented. These are compared with the temperature of the compression fluid
at various locations within the compressor. The compressor model was based
on the ‘DRUM127’ test compressor but different rotor geometry was generated
and modelled to evaluate how this changed the temperature distribution when
running at a similar duty.
6.2.2 DRUM127 COMPRESSOR
Figure 6-1: DRUM127 air compressor
Table 6-2: Basic compressor details
Main rotor diameter, D (mm) 127

Length over diameter ratio, L/D 1.6
Main rotor wrap angle, φw1 (deg) 285°
Lobe combination, z1/z2 3/5
Profile type N
Volume index, Vi 1.5
Oil injection No
129
The DRUM127, shown in Figure 6-1, is a twin screw compressor which features
timing gears but has no oil or water cooling within the chamber. The analysis
was based on the following operating conditions and assumed machine
parameters.
Table 6-3: Air duty
Inlet pressure, p1 (Bara) 1

Outlet pressure, p2 (Bara) 2 - 2.5
Inlet temperature, T1 (°C) 25
Speed, N (rpm) 7000
Table 6-4: Model parameters for thermal analysis
Ambient temperature, Ta (°C) 20

Rotor thermal expansion coef., μr (°C-1) 1e-5
Casing thermal expansion coef., μr (°C-1) 1e-5
Nominal interlobe gap, GI, (μm) 100
Nominal radial gap, GR, (μm) 100
Nominal axial gap, GA, (μm) 100
The model was set up with a uniform interlobe clearance distribution – this is a
reasonable assumption for this kind of oil free, N-profile rotor, which mainly
features a uniform clearance distribution that only reduces slightly in the pitch
areas. Thermal expansion coefficients aren’t used at this stage but have been
included here as this DRUM127 test case is used later to investigate thermal
distortions.
130
6.2.3 MAPPED BOUNDARY CONDITIONS
a) instantaneous temperature b) time averaged temperature

Figure 6-2: Rotor boundary temperatures
a) planar averaged temperature b) time averaged temperature

Figure 6-3: Averaged boundary temperatures used for thermal analysis
Figure 6-2 and Figure 6-3 show various fluid temperature distributions over the
rotor and casing surfaces. The planar averaged temperature on the rotors from
Figure 6-3 a) is used to generate a 2D plot in Figure 6-4, which shows how the
main rotor is exposed to a higher average temperature than the gate rotor
towards the outlet end of the rotors. These rotor results are also tabulated in
131
Table 6-5. The averaged rotor temperature exposure is not as extreme as the
minimum and maximum fluid temperatures. However, on the static casing, the
time averaged temperature exposure limits are very similar to the minimum
and maximum fluid temperatures.
Table 6-5: Temperatures at specific locations
Temperature (°C)
Fluid Main Rotor Gate Rotor Casing
Inlet Outlet Inlet Outlet Inlet Outlet Min Max
25 161 37 122 35 98 24 160
Main rotor Gate rotor

Average temperature, T (°C)
120
100
80
60
40
20
0 50 100 150 200
Distance from inlet end (mm)
Figure 6-4: Rotor temperature distribution
132
6.2.4 SENSITIVITY TO LOBE COMBINATION
a) tested 3/5 lobe rotors b) 3/4 lobe model comparison

Figure 6-5: Comparison of rotor models with different lobe combinations
New rotors were generated with a 3/4 lobe combination. The new rotors are
shown next to the original 3/5 rotor geometry in Figure 6-5. The length to
diameter ratio and wrap angle on the main rotor is the same for each rotor pair.
The results, given in Table 6-6 and Figure 6-6 show that reducing the number of
lobes on the gate rotor from 5 to 4 has increased the average temperature at the
outlet so that the main and gate rotor outlet temperatures are similar. An
additional case with 4/5 rotor geometry was run and added to Table 6-6; this
supports the theory that the larger lobe difference between the 3/5 rotors is
what causes the temperature difference.
133
Table 6-6: Temperatures with different lobe combinations
Temperature (°C)
Fluid Main Rotor Gate Rotor
Lobe
Comb. Inlet Outlet Inlet Outlet Inlet Outlet
3/5 25 161 37 122 35 98
3/4 25 156 35 118 36 114
4/5 25 172 37 130 38 124
Main rotor, 3/5 lobes Gate rotor, 3/5 lobes

Main rotor, 3/4 lobes Gate rotor, 3/4 lobes
120
100
80
60
40
20
0 50 100 150 200
Figure 6-6: Rotor temperature distributions with different lobe combinations
134
6.2.5 SENSITIVITY TO WRAP ANGLE
a) model with 255° wrap angle b) model with 315° wrap angle
Figure 6-7: Comparison of rotor models with different wrap angles
Figure 6-7 shows an example of rotors with different wrap angles. In this case
the length to diameter ratio and lobe combinations are preserved. It is shown in
Figure 6-8 and Table 6-7 that, with increased wrap angle, there is a modest
decrease in the rotor temperature at the inlet and an increase in the rotor
temperature at the outlet. This is as expected, since if the rotors had a wrap
angle of zero, as is the case for roots blowers, then the temperature exposure
would be the same along the full length of the rotors.
135
Table 6-7: Temperatures with different wrap angles
Temperature (°C)
Wrap
Angle Inlet Outlet Inlet Outlet Inlet Outlet
255 25 159 38 117 37 94
285 25 161 37 122 35 98
315 25 163 35 129 34 104
Main rotor, wrap = 255 Gate rotor, wrap = 255

Main rotor, wrap = 315 Gate rotor, wrap = 315
120
100
80
60
40
20
0 50 100 150 200
Figure 6-8: Rotor temperature distributions with different wrap angles
136
6.2.6 SENSITIVITY TO VOLUME INDEX
Vi = 1.5 Vi = 2.0
200
180
Fluid Temperature (°C) 160
140
120
100
80
60
40
20
0
0 50 100 150 200 250
Cycle angle - from begining of compression (degrees)
Figure 6-9: Modelled temperature during cycle with different volume index
a) model with Vi = 1.5 b) model with Vi = 2.0

Figure 6-10: Comparison of rotor models with different volume index
In this case the rotor geometry in unchanged but the casing has been modified.
In fact this ‘casing geometry’ modification was simply achieved by changing the
timing of the port opening in the thermodynamic model, resulting in a different
temperature curve as shown in Figure 6-9. In Figure 6-10 the chambers
highlighted with the red ellipse show the instantaneous fluid temperature
immediately prior to opening of the discharge port. With Vi = 1.5 the
temperature is less than discharge in this chamber but once open to discharge
will be exposed to a higher temperature earlier in the cycle. With Vi = 2.0 the
137
temperature is greater than the final discharge temperature because it will

decrease once expanded to the lower discharge pressure. Both scenarios result
in higher than ideal temperatures at different points in the cycle however it is
shown in Table 6-8 and Figure 6-11 that the overall effect on the rotor
temperature exposure is similar.
The most notable difference in the surface temperature distribution was on the
casing bores, as shown in Figure 6-12. The minimum and maximum time
averaged temperatures are similar. However, with Vi = 1.5, a larger area of the
casing bores is exposed to the higher temperatures.
138
Table 6-8: Temperatures with different volume index
Temperature (°C)
Volume
Index Inlet Outlet Inlet Outlet Inlet Outlet
1.5 25 161 37 122 35 98
2 25 159 36 122 35 98
Main rotor, Vi = 1.5 Gate rotor, Vi = 1.5

Main rotor, Vi = 2.0 Gate rotor, Vi = 2.0
120
100
80
60
40
20
0 50 100 150 200
Figure 6-11: Rotor temperature distributions with different volume index
a) model with Vi = 1.5 b) model with Vi = 2.0

Figure 6-12: Comparison of casing temperatures with different volume index
139
6.2.7 DISCUSSION
These results confirm that surface boundary conditions have been mapped as
expected. Testing with different rotor geometry has confirmed that the
procedures outlined in Chapter 4 are sufficiently robust and furthermore have
been correctly implemented.
This exercise demonstrates how the procedures developed in this work can be
readily applied to the evaluation of rotor designs at a very early stage in the
design process. Further case studies, in this thesis, utilise these temperature
distributions in order to investigate operational clearances due to thermal
distortions.
140
6.3 MODEL SENSITIVITY TO CLEARANCE MODIFICATIONS
6.3.1 OVERVIEW
The estimated thermodynamic performance is compared with experimental
results. The aim of this exercise was to evaluate the sensitivity of the model
performance to predicted operational clearance variations.
6.3.2 DRUM127 COMPRESSOR

This case is based on the same setup as detailed in 6.2.2 ‘DRUM127
Compressor’.
6.3.3 DRUM127 TEST MEASUREMENT

Air test data was provided by City, University of London, for the range of
operating duties detailed previously. Due to the lack of compressor cooling and
the high adiabatic exponent of air, the achievable pressure ratio was restricted
by the operating temperature; however this is well suited to investigate thermal
distortions.
A torque meter was installed on the motor shaft while the digital encoder for
the speed measurement was mounted on the male rotor shaft. The pressure and
temperature of the gas were measured at the inlet, the discharge, and upstream
of the orifice plate. The air flow through the compressor was measured by use
of an orifice plate installed in the discharge line of the system. The discharge
line contained a control valve for regulation of the discharge pressure. A
simplified schematic of the test setup is shown in Figure 6-13 and a list of
instrumentation used in provided in Table 6-9.
141
T1
Air Inlet
p1
Motor Drive
Screw
Compressor
Speed
T2 Throttle
Torque Valve
Orifice
p2
Air Outlet
Tu pu
Δp
Figure 6-13: DRUM127 schematic
Table 6-9: DRUM127 instrumentation
Measured
Instrument Description
Parameter
shaft speed Shaft Encoder. (BHG 16.25W.3600- 360 TTL pulses per revolution.
B2-5)
TRP-500 torque meter. ( strain max capacity: 500Nm, Calibration level:

shaft torque
gauge transducer) 335Nm Range = 0 - 6000 rpm, Supply
volt=10v dc
inlet pressure, PDCR 110/w –pressure transducer Operating range = 3.5bar(abs). Excite
p1 voltage=10V dc
inlet temp. , T1 Platinum Resistance Thermometer Range= -750C to 3500C
outlet pressure, PDCR 922-pressure transducer Operating range =15 bar (abs). Excite
p2 voltage=10V dc, Output voltage= 100 mV
outlet temp., T2 Platinum Resistance Thermometer Range= -750C to 3500C
orifice inlet PDCR 922-pressure transducer Operating range =15 bar (abs). Excite
pres., pu voltage=10V dc, Output voltage= 100 mV
orifice inlet Platinum Resistance Thermometer Range= -750C to 3500C
temp., Tu
orifice diff. PDCR 2120- pressure transducer Pressure diff= 0.35 bar, Excitation
pres., Δp Voltage=10V dc
Data acquisition and logging was done using a National Instrument Compact-
RIO (CRIO-9022) Real-Time with an 8 slots chassis CRIO-9114. Programming
142
was done using LabVIEW FPGA (Field Programmable Gate Array) which is
suitable for high frequency data acquisition.
6.3.4 DRUM127 CLEARANCE SENSITIVITY

Three different model cases have been described in Table 6-10. Case A uses the
unmodified design clearances, set up in the model for the radial and interlobe
leakage paths, in order to provide a benchmark. In Case B the interlobe and
radial gaps have both been modified for rotor thermal distortion only, while in
Case C the estimated thermal distortion of the casing has been included.
Table 6-10: Modelled clearance adjustments
Clearance Adjustments Case A Case B Case C

ROTOR thermal distortion No Yes Yes
CASING thermal distortion No No Yes
In Figure 6-14, the local interlobe clearance gap is plotted against the relative
position along the interlobe sealing line for a single compression chamber. In
Figure 6-14 the location of the rotor root and tip is annotated and the local rotor
radius is plotted on the secondary axis, to provide an indication of the local
clearance positions.
Main Rotor Main Rotor

0.12 Root Tip 120
Local Radius on Main Rotor (mm)
Local Clearance Gap (mm)
110
0.10 100
0.08 90
80
0.06 70
60
0.04 50
0.02 40
30
0.00 20
0 0.2 0.4 0.6 0.8 1
s, relative position along sealing line
Case A Case B Case C Radius
Figure 6-14: Interlobe clearance distribution corrected for thermal deformations
143
In Case A the local clearance gap has been taken to be constant with a value of
0.1mm. With the introduction of the rotor thermal expansion in Case B,
significant clearance reduction will occur with some local areas more affected
than others. The location of the biggest clearance reduction is at the root of the
main rotor. This indicates the location where rotor contact is most likely to
occur. The thermal expansion of the casing in Case C mitigates the rotor
expansion, as would be expected. However, there is still a slight net decrease in
the interlobe clearances for this case.
The results shown in Figure 6-14 are for a specific instant in the compression
cycle when the interlobe sealing line is fully developed i.e. the full length of the
sealing line is an active leakage path. The instantaneous interlobe leakage area
is taken as the area under the relevant curve (corrected for the actual sealing
line length). As the cycle progresses, the length and position of the sealing lines
change and the local clearance deformations need to be re-evaluated based on
the new location. The changing history of two leakage areas is shown in Figure
6-15, beginning at the start of compression (in this case when θ = 0°). The
interlobe leakage area curve is shown on the left graph of Figure 6-15. This is
related to the clearance distribution of Figure 6-14. The combined leakage area
curve for all out-flow radial clearances is shown in the right hand graph. It is
interesting to note that for Case C, with both rotor and casing expansion
included, a net increase in the radial gap is predicted in the latter stages of the
compression cycle due to the local hot spot on the compressor casing. The
radial leakage area is greater than the interlobe leakage area. Note the different
scales used on each y-axis. However, the cycle offset and consequently the
pressure difference across the radial leakage area is significantly smaller than
across the interlobe gap which determines the relative significance of how these
flow areas affect the thermodynamic performance. Note that a similar radial in-
flow leakage area would feed some mass flow back into the rotor chamber from
the previous chamber, though this is not shown.
144
2.5E-05 7.0E-05
Radial Out-flow Leakage Area (m2 )

(m2)
6.0E-05
2.0E-05
Area
2.5E-05 5.0E-05
Area (m
Leakage 2) 2.0E-05
1.5E-05 4.0E-05
1.5E-05
3.0E-05
Flow
1.0E-05
1.0E-05
Interlobe
5.0E-06 2.0E-05
Interlobe Leakage
5.0E-06
0.0E+00 1.0E-05
0 50 100 150 200 250 300 350
0.0E+00 Cycle Angle, θ (degrees)
0.0E+00
0 100 200 300 0 100 200 300
Cycle Angle, θ (degrees) Cycle Angle, θ (degrees)
Case A Case B Case C
Figure 6-15: Variation of leakage areas throughout compression cycle
6.3.5 DRUM127 PERFORMANCE SENSITIVITY

As shown in Figure 6-16, measured tests results of compressor flow and
discharge temperature have been used as a benchmark to compare model
predictions. Initial temperature boundary conditions were always calculated,
based on results from a chamber model with no clearance modifications (as in
Case A). For case B and C these initial temperatures were used to calculate new
operational clearances so the thermodynamic model had to be recalculated
(following the flowchart of Figure 5-16), to include the temperature boundary
conditions. Further iteration was required until the clearance corrections had
no significant effect on the recalculated thermodynamic performance. After 3
iterations, the discharge temperature varied by less than 0.3% compared to the
previous iteration. This was taken to be satisfactory therefore 3 iterations were
used in all models where clearance modification was included.
145
12.0 190
11.5 180
11.0 12.0 170
Temperature ( C)
Flow (m3/min) 11.5
Flow (m3/min)
11.0 160
10.5
10.5 150Measured
10.0 Measured
10.0 9.5 140Model - Case A Model - Case A
9.0
9.5 8.5 130Model - Case B Model - Case B
8.0
9.0 1.9 2.0 2.1 2.2 120 Model - Case
2.3 2.4 C 2.5 2.6 Model - Case C
8.5 pressure ratio
110
8.0 100
1.9 2.0 2.1 2.2 2.3 2.4 2.5 2.6 1.9 2.0 2.1 2.2 2.3 2.4 2.5 2.6
pressure ratio pressure ratio
Measured Model - Case A Model - Case B Model - Case C
Figure 6-16: Compressor performance from test and model
The results of the model running Case B - with interlobe and radial clearances
corrected for rotor thermal growth only - show a significant increase in flow
compared to the results of Case A. Case B also has a smaller drop off in flow with
increased pressure ratios (flatter gradient) that is closer to the measured
gradient than Case A. The results of the model running Case C, with interlobe
and radial clearances corrected for rotor and casing thermal growth, show an
improvement in the absolute flow values. However there is a deviation from the
measured gradient. The modelled temperature, shown in the right hand
diagram in Figure 6-16 is closer to the measured results in terms of absolute
values and gradient (with respect to pressure ratio) for both cases with
clearance corrections. For temperature comparison the measured discharge
temperature (T2 in Figure 6-13) directly downstream from the compressor is
used; the equivalent model temperature is used as discussed in Appendix B.1.1.
6.3.6 DISCUSSION
Results show that the modelled flow is highly sensitive to changes in the
operational clearances. The largest deviation in predicted performance is
between Case A – with no clearance modification and Case B – with clearances
adjusted due to rotor distortion only. These are the two extreme cases for
clearance values and it’s a reasonable assumption that actual operational
clearances will be somewhere in between.
146
In the case presented, the model accuracy improved with the introduction of
local clearance corrections used to calculate leakage areas iteratively. However
care must be taken when drawing conclusions, based on absolute test values,
because it is difficult to know, with absolute certainty, that the unmodified
model clearances are representative of the actual assembled clearances; which
are subject to a complex array of manufacturing and assembly tolerances.
The predictions shown in Case C are closest to the test results, in terms of flow
prediction. However the predicted temperature shown in the results of Case B
was closer. Put another way, the modelled discharge temperature is slightly
high for a given flow compared to test results. It is difficult to say if this is the
result of something inherent to the chamber model used; or if the relative size
of the various clearances during set-up is not representative of the actual
assembled clearances; or if the relative adjustment for operational distortions is
inaccurate.
6.4 INVESTIGATING INTERLOBE CLEARANCE DISTORTION
6.4.1 OVERVIEW
This case focuses on the impact of thermal distortions on the interlobe
clearance distribution. As before, the predicted operational interlobe clearances
are presented for different combinations of rotor and casing distortion in order
to assess how sensitive the calculated clearances are to each.
This case study is on an oil injected WCVTA510 compressor. In contrast to the

previous DRUM127 case, this adds further complexity to the clearance analysis
due to the lack of timing gears and direct rotor to rotor drive. More emphasis is
put on representing the actual interlobe clearance distribution around the rotor
profile accurately. The test results are from two different compressors, before
and after a design revision of the interlobe clearances was made in order to
address reliability concerns when running at higher than normal temperatures.
147
6.4.2 WCVTA510 COMPRESSOR

The WCVTA510 is an oil injected compressor which at the time of testing was
the largest manufactured by Howden Compressors, Figure 6-17 gives a sense of
the scale and Figure 6-18 shows how the rotors are situated.
Figure 6-17: WCVTA510 compressor
Figure 6-18: WRV range compressor vertical section
148

Profile type N
Volume index, Vi 2.6 (fixed)
Oil injection Yes
Table 6-12: Air test for contract duty

Outlet pressure, p2 (Bara) 5
Outlet temperature, T2 (°C) 90 (targeted)
Speed, N (rpm) 1400
Table 6-13: Air test for high temperature

Outlet pressure, p2 (Bara) 11
Outlet temperature, T2 (°C) 120 (targeted)
Speed, N (rpm) 750*
*Due to power restrictions on the testing of this large compressor, the speed
was reduced from the contract speed of 1400rpm to 750rpm during the
overload testing. It is this overload testing at higher discharge temperature
which was of interest for clearance analysis.
149
6.4.3 WCVTA510 TEST MEASUREMENT

Test data was provided from contract testing in Howden Compressors. A typical
test setup is shown in Figure 6-19. Though not visible in the photo, an orifice
plate for flow measurement is situated in the inlet ducting. Some details of test
instrumentation are provided in Table 6-14.
Figure 6-19: WRVTA compressor on test stand
Table 6-14: WRVTA510 instrumentation
Manufacturer Description Range
Budenberg Discharge Pressure 0-400psig
Budenberg Discharge Pressure 0-100psig
Kosmos Discharge Temperature
G&W Tachometer 1500rpm
Elliott Electrical power to motor 0-1200W
150
6.4.4 WCVTA510 THERMAL ANALYSIS

The overload test condition, with a discharge temperature of 120°C, was
modelled to determine how the temperature varied throughout the
compression cycle for this duty.
As before, the fluid temperature from the compression cycle was mapped onto
the surfaces of the rotors and the casing bores. The temperature of the fluid
boundary on the main and gate rotor surfaces is shown in Figure 6-20 . The left
image shows the instantaneous fluid temperature for a particular rotor
position. In the right hand image these instantaneous temperatures has been
averaged over one full rotation and over each transverse plane of the rotors.
Figure 6-20: Instantaneous and averaged fluid boundary temperature. Line AB

highlights the interlobe sealing line for a single compression chamber.
Table 6-15: Average fluid boundary temperature across outlet plane
Component Description Temperature (°C)

Main rotor 86.8
Gate rotor 73.2
Casing 69.4
The peak temperatures occur at the outlet plane of the rotors. Therefore the
analysis presented in this work was based on the average fluid boundary
151
temperatures presented in Table 6-15. It is clear that the averaged fluid

boundary temperatures at critical locations on the compressor are actually
much lower than the peak fluid discharge temperature of 120°C. Note that while
the peak temperature that the casing is exposed to is close to the discharge
temperature the planar average used in this case is the average of the
temperature all the way around the rotor bores in the discharge plane.
6.4.5 WCVTA510 CLEARANCE PRESENTATION

Due to the 3D nature of the interlobe sealing line it is difficult to present
clearance information clearly. Figure 6-21 shows an example of clearances
mapped onto each of the rotors as vectors normal to the transverse rotor curve.
The magnitude of the vectors actually shows the magnitude of clearances
normal to the 3D rotor surface. This approach is very intuitive and clearly
shows how clearances relate to difference areas on the rotor profiles. The
difficulty with this clearance presentation is comparing small differences in
similar clearance distributions. Magnifying the vectors quickly distorts the
apparent clearance gap due to the curvature of the profiles. In the previous case,
of the DRUM127, local clearances were plotted against the ‘relative position on
the sealing line’ but this makes it difficult to determine where the clearances
actually occur on the rotors. For this example, where the design clearance is
non-uniform, a different approach has been taken in order to show local
clearances more clearly.
152
Figure 6-21: Transverse cross section of rotors showing interlobe clearance

distribution of rotor surfaces
The compromise that has been taken is to plot the local clearances on the axis of
a rack projection along which the rack would translate. Thus, the magnitude of
the local clearances can be simply graphed and superimposed against the
somewhat familiar shape of the rack as presented in Figure 6-22. Rather than
putting numerical values on the horizontal axis, vertical lines have been added
to identify key points along the length of this sealing line projection, details of
which are provided in Table 6-16.
Line AB represents the conjugate rack, common to both rotors. The limits of this
rack segment, points A and B, are shown in Figure 6-20 where it can be
observed that these points occur on the boundary between different
compression chambers. This point is important as the distortions over this
specific section of the interlobe sealing line are used when calculating the
interlobe leakage area for that chamber at that moment.
The original and revised clearance distributions are both presented in Figure
6-22. In all cases the relative rotation between the rotors has been adjusted in
order to provide zero clearance at location 3: the pitch radius on the round side
of the rotors. This is where the driving force is transferred from the main to the
gate rotor. In the revised clearance design the main difference is that the
clearance in the main rotor root has been increased and to a lesser extent also
153
at the main rotor tip. Notice that the clearance on the round, or drive, flank of
the rotors (between locations 3 and 4) has actually been decreased in the
revised clearance design – this clearance will not significantly reduce during
operation because the gap is maintained due to the nearby contact at the pitch
point (location 3, where the main rotor drives the gate rotor), therefore this
clearance was deemed to be unnecessarily large. Conversely, on the opposite
flank (from 4 to B, and from A to 1) the gap is adversely affected by thermal
distortions at location 3 which are transferred over to the non-drive flank.
Therefore it was necessary to increase the clearances on the non-drive flank.
To quantify how these clearance changes impact the leakage area through the
interlobe gap, the local clearances were integrated over the length of the 3D
sealing line. The total leakage area is 108.5mm2 for the original design and
142.7mm2 for the revised design, an increase of 31.5%.
Interlobe Clearance Distribution Along Rack Projection of Rotors

A 1 2 3 4 B
2.0 1.0
1.8
Relative Local Clearance
Original Clearance
1.6 0.9
Design
Relative 'Radius'
1.4
1.2 0.8
Revised Clearance
1.0
Design
0.8 0.7
0.6
Rack Projection
0.4 0.6
0.2
0.0 0.5
Relative Position on Rack Projection
Figure 6-22: Comparison of example interlobe clearance distributions along rack

projection of rotors
Table 6-16: Key Clearance Locations
Location on rack Description

A Limit of sealing line for single chamber
1 Pitch radius on straight (undercutting) side of rotors
2 Root of main rotor / tip of gate rotor
3 Pitch radius on round (non-undercutting) side of rotors
4 Tip of main rotor / root of gate rotor
B Limit of sealing line for single chamber
154
6.4.6 WCVTA510 CLEARANCE DISTORTION RESULTS

The following results show the predicted operational clearance distributions for
the original and revised designs respectively. As previously discussed, a similar
temperature increase, in both the rotors and casing, results in little net change
in the operational clearances. This case has been presented with the longer
dashed lines on each figure and the trend is not far off the original clearance.
A second case has also been presented where the rotors do not move apart due
to casing thermal expansion. This case is quite possible in the event that sudden
temperature changes cause the rotors to heat up faster than the casing, or,
alternatively the outlet end bearings might be situated remotely from the rotor
outlet plane in a location where the casing temperature is much lower than
estimated in Table 6-15. Unfortunately the distance between the two dashed
curves for the different operational scenarios results in a large band of
uncertainty but it does serve to highlight 1) where rotor contact is most likely to
occur for a given clearance design, and 2) where clearances can be further
reduced without compromising reliability.
Possible Variation in Interlobe Clearance Distribution at Outlet End

(Original Design)
2.0
1.8
1.6 Cold Clearance (Design)

1.4
1.2
1.0
0.8
0.6 Clearance adjusted for
0.4 potential rotor distortion
0.2
0.0
-0.2 Clearance adjusted for
-0.4
-0.6 potential rotor and casing
-0.8 distortion
-1.0
Figure 6-23: Original design clearance variations
155
Possible Variation in Interlobe Clearance Distribution at Outlet End

(Revised Design)
2.0
1.8
1.6 Cold Clearance (Design)

1.4
1.2
1.0
0.8
0.6 Clearance adjusted for
0.4 potential rotor distortion
0.2
0.0
-0.2 Clearance adjusted for
-0.4
-0.6 potential rotor and casing
-0.8 distortion
-1.0
Figure 6-24: Revised design clearance variations
This analysis supports the suitability of the revised clearance design. The
results show that at the root of the main rotor, contact cannot now occur due to
thermal distortion alone. The chance of contact at the tip of the main rotor has
been significantly reduced but the analysis suggests contact is still possible
there under extreme circumstances.
The most problematic clearance area is on the non-drive (undercutting) flank of

the rotors. Since distortions are transferred over from the contacting drive
flank, these non-drive clearances vary more than in other areas. What the
figures do not show is that this represents a long part of the sealing line path
due to the effect of undercutting. So this has a large effect on the overall leakage
area. This area also experiences faster relative sliding motion between the rotor
surfaces. So contact in that region should certainly be avoided.
6.4.7 WCVTA510 PERFORMANCE RESULTS

The models were also run at the nominal air test duty which is closest to the
designed operating conditions: with a speed of 1400rpm, a pressure ratio of 5
and a discharge temperature maintained at approximately 90°C. The modelled
deviation in the volume flow due to the revised design is compared with the
deviation measured on test in Table 6-17.
Table 6-17: Performance penalty with revised interlobe clearance
change in volumetric flow

Modelled -0.5%
Measured on Test -1.7%
156
The difference between the model result and the test result suggests that the
model is slightly underestimating the effect of leakages through the interlobe
gap. Unfortunately as the test results were obtained using two completely
different compressors the effect of other manufacturing and assembly
tolerances cannot be ruled out for the test results.
Importantly, the resulting change in flow on test was small enough for the
performance of both compressors to still be within normal operating
tolerances. With the reduced flow there was a similar reduction in shaft power
so there was very little change in overall efficiency.
6.4.8 WCVTA510 TEARDOWN RESULTS

In the absence of direct measurement of the operational interlobe clearances
tear down of the compressors often provides useful insight. On the compressor
with the original clearance design, tear down inspection revealed that rotor to
rotor contact had occurred at the root of the main rotor, resulting in a polished
surface finish as highlighted in Figure 6-25. In other words the clearance had
reduced to zero locally.
Figure 6-25: Evidence of rotor rooting on main rotor of WCVTA510
The type of contact observed and the fact that this standard compressor had not
presented any problems, while operating within normal temperature limits,
points to thermal distortion of the rotors being the most likely cause of this
157
rotor contact. This tear down also indicated that some rotor to casing contact
had occurred as well as some localised interlobe contact around the tip of the
main rotor.
The revised clearance design was implemented on a different compressor at a

later date and though teardown inspection revealed some minor localised rotor
contact, related to the tip sealing strip, there was no contact over the main part
of the root.
6.4.9 DISCUSSION
Local variations in the interlobe clearance distribution, due to thermal
distortions, have been approximated. This provides valuable data at an early
design stage about where and when, in the compression process, clearances are
not adequate for a given compression application. Integrating this information
into the rotor design process will allow a more optimised balance between
reliability and compressor performance.
For validation purposes, the analysis was repeated for two interlobe clearance
designs, when operating at an elevated rotor discharge temperature of 120°C in
an oil injected compressor, with direct rotor drive. While there is a fairly large
uncertainty on the predicted clearance deviations, the modelled results are in
agreement with the available experimental findings because they predicted a
high likelihood of rotor contact at the main rotor root on the original clearance
design.
158
6.5 INVESTIGATING RADIAL CLEARANCE DISTORTION
6.5.1 OVERVIEW
This case study draws on extensive test results obtained from an oil free HS204
compressor, operating over a wide range of temperatures and speeds. This
allows a more comprehensive comparison of modelled and measured
compressor behaviour than provided in the previous case studies.
In addition to performance results from the tests, these results include real time
measurement of the radial clearance gap which was used for more direct
verification of the predicted clearance deformations.
6.5.2 HS204 COMPRESSOR

The HS204 is an oil free twin screw compressor which features timing gears at
the suction end and mechanical seal arrangements on the shafts between the
bearings and main rotor bodies. Hence there are two seals per rotor. A sectional
arrangement of this compressor is shown in Figure 6-26. The rotors used in
these tests were newly designed with an N rotor profile as shown in detail in
Figure 6-27.
Figure 6-26: Horizontal section of HS204 compressor
159
INTERLOBE CLEARANCE DISTRIBUTION: ROTOR DATA:
Point Normal Clearance [microns] MALE ROTOR

Min Max NUMBER OF LOBES =4
1 143 173 HELIX = RIGHT HAND
2 139 169 WRAP ANGLE = 300°
3 106 135 ROTOR LENGTH = 336.600
4 105 135 LEAD = 405.000
5 105 135 Pitch HELIX ANGLE AT PCD = 44.795°
CLEARANCE DISTRIBUTION: 6 116 146
ROTOR DATA: LEAD ANGLE AT PCD = 45.204°
7 126 156
Normal Clearance [microns] 8 143MALE ROTOR
173 FEMALE ROTOR
Min Max 9 116NUMBER 146
OF LOBES =4 NUMBER OF LOBES =6
143 173 10 107HELIX 137 = RIGHT HAND HELIX = LEFT HAND
139 169 11 96WRAP ANGLE
126 Pitch = 300° WRAP ANGLE = 200°
106 135 12 96ROTOR LENGTH
126 = 336.600 ROTOR LENGTH = 336.600
105 135 13 143LEAD 173 = 405.000 LEAD = 607.500
105 135 Pitch 14 143 173 AT PCD = 44.795°
HELIX ANGLE HELIX ANGLE AT PCD = 44.795°
116 146 15 145LEAD ANGLE
174 AT PCD = 45.204°
LEAD ANGLE AT PCD = 45.204°
126 156
143 173 FEMALE ROTOR
116 146 NUMBER OF LOBES =6
107 137 HELIX = LEFT HAND
96
96
126
126
Pitch
Figure 6-27: Rotor profiles and design data
WRAP ANGLE = 200°
ROTOR LENGTH = 336.600
143 173
LEAD = 607.500
143 173
HELIX ANGLE AT PCD = 44.795°
145 174 All of the basic compressor parameters are given in Table 6-18 and more details
LEAD ANGLE AT PCD = 45.204°
of the interlobe clearance distribution are shown in Figure 6-28. The radial
clearance gap for this compressor is nominally 0.150mm – this was the initial
value used in the models.

Profile type N
Volume index, Vi 1.98
Oil injection No
Jacket Cooling Yes
160
INTERLOBECLEARANCEDISTRIBUTION: ROTORDATA:
Point Normal Clearance [microns] MALEROTOR
Min Max NUMBEROF LOBES =4
1 143 173 HELIX = RI
2 139 169 WRAPANGLE = 30
3 106 135 ROTORLENGTH = 33
4 105 135 LEAD = 40
5 105 135 Pitch HELIXANGLEAT PCD = 44
6 116 146 LEADANGLEAT PCD = 45
7 126 156
8 143 173 FEMALEROTOR
9 116 146 NUMBEROF LOBES =6
10 107 137 HELIX = LE
11 96 126 Pitch WRAPANGLE = 20
12 96 126 ROTORLENGTH = 33
13 143 173 LEAD = 60
14 143 173 HELIXANGLEAT PCD = 44
15 145 174
LEADANGLEAT PCD = 45
Figure 6-28: Interlobe clearance design data
The compressor was tested over a range of speeds and pressures as detailed
below:
Table 6-19: Air test duty

Outlet pressure, p2 (Bara) 1.5 – 2.5
Speed, N (rpm) 4000 – 7000
6.5.3 HS204 TEST MEASUREMENT

Extensive test data for this oil free compressor was available from previous
R&D work at Howden Compressors. The compressor is shown on the test stand
in Figure 6-19. Real time data monitoring and logging equipment was used to
continuously record the parameters shown in Figure 6-30; details of the
instrumentation used are provided in Table 6-20 and Table 6-21. By using the
continuously monitored results, sufficient dwell time was allowed between
different operating points to ensure that the measurements had stabilised.
Furthermore, each stabilised results were averaged over a 10 second period.
161
Figure 6-29: HS204 on test stand
Throttle
Valve
Orifice
T1
Air Inlet Air Outlet
Tu pu
p1 T2
Δp
p2
Motor Drive
Screw
Compressor
Speed
Torque GapR
Figure 6-30: HS204 instrumentation schematic
162
Table 6-20: HS204 instrumentation
Parameters Supplier Part No Description
Omni 0 to 2 BarA, 2mV/V, pressure

pu, p1 Pi610C3-2BarA
Instruments transducer
Omni 0 to 6 BarA, 2mV/V, pressure

p2 Pi610C3-6BarA
Instruments transducer
RS Honeywell Comp SCX press sensor, 0-

orifice Δp 455-5071
Components 5psi, differential
Mineral insulated type thermocouple

Tu, T1, T2 TC Direct 405-987 with IP67 standard weatherproof head
(Type T, 6 mm x 150 mm)
Proximity Probe: Straight mount, 5mm

Omni tip, 2.0 m cable, armoured, female SMC
ΔGapR PRS02/2.0A01
Instruments connector, 30 mm M6 x 0.75 thread, 0 to
2.5 mm range.
02-2350-00,
speed, torque Torquemeters ET2350 Torque meter and readout unit
613/M
Table 6-21: HS204 additional hardware
Use Supplier Part No Description
data aq. National 779508-06 NI cDAQ-9172 8-slot USB 2.0 Chassis

Instruments for CompactDAQ
data aq. National 779014-01 NI 9221 8-channel, +/- 60V, 12-bit

Instruments analogue input module
data aq. National 779519-01 NI 9205 32-channel, +/- 10V, 16-bit

Instruments analogue input module
data aq. National 780493-01 NI 9213 16-channel, 24-bit,

Instruments thermocouple module
driver unit Omni ECD02/20 Probe Driver: 2.0m cable, 0 to 2.5 mm

Instruments range (8mV/um, API 670)
Proximity probes were used to measure the radial gap, GapR, on both the main
and gate rotors. The probes situated in the bores of the casing at the bottom,
30mm from the discharge end face, as shown in Figure 6-31. Note that the
probes are situated opposite the discharge port which is located on the top of
the casing.
163
Eddy current probes
Figure 6-31: Position of radial proximity probes in the test compressor
Rotor lobe
a
b
Casing
Eddy current probe
Figure 6-32: Proximity probe location
The proximity probe is slightly recessed into the casing bore as shown in Figure
6-32. The measured gap ‘b’ is therefore slightly bigger than the radial gap ‘a’.
While the compressor was running the gap obviously varies due to passing of
the rotor lobe and the objective here was to record the minimum value.
164
Due to the nature of this set-up, the static gap could not be measured and
instead the value of ‘b’ was recorded immediately after start-up to provide a
reference un-deformed measurement. Once thermally stable at each operating
point the change in the value of ‘b’ was used to assess gap distortion.
6.5.4 HS204 PERFORMANCE RESULTS WITH JACKET COOLING

At each test speed, the discharge pressure was adjusted for each measurement
point and the compressor was given time to stabilise before measurements
were taken. Model results were produced for the same range of operating
points. Table 6-22 shows the measured volumetric efficiency and discharge
temperature at each test point; with the values corrected for suction pressure
and speed. The results at new pressure points have been linearly interpolated
from the actual pressure points in Table 6-23. This has been done in order to
provide test reference points so that the model error can be calculated at
specific pressure ratios. In all the graphs presented, the originally tested
pressure ratios will be used.
Graphs that compare the model results against the test results are shown in
Figure 6-33 and Figure 6-34. In this case the model is based on nominal design
clearances that are unmodified. These model results are denoted ‘MODEL’. The
relative error of the model across the range of operating points is tabulated in
Table 6-24.
Similar results for a model that predicts clearance distortion due to rotor
thermal expansion only, denoted by ‘MODEL R’, are given in Figure 6-35 and
Figure 6-36 and Table 6-25.
The same set of results, produced by a model that predicts clearance distortion
due to both rotor AND casing thermal expansion, denoted ‘MODEL RC, are
shown in Figure 6-37 Figure 6-38 and Table 6-26 .
165
Table 6-22: HS204 performance test results
Speed Pressure
(rpm) Ratio Volumetric efficiency T2 (K)
1.63 78.3% 349.2
1.79 77.3% 361.3
4000
1.97 76.3% 373.2
2.07 75.8% 379.7
1.63 81.8% 354.3
1.81 80.9% 364.3
5000
2.00 80.1% 375.5
2.10 79.5% 381.5
1.65 84.2% 355.6
1.84 83.4% 367.5
6000
2.01 82.9% 376.9
2.11 82.5% 381.9
1.69 85.7% 363.1
1.88 85.1% 371.3
7000
2.06 84.5% 380.6
2.16 84.3% 386.7
Table 6-23: HS204 interpolated performance test results

Speed Pressure
(rpm) Ratio Volumetric efficiency T2 (K)
1.50 79.0% 340.7
1.75 77.5% 357.9
4000 2.00 76.1% 375.1
2.25 74.7% 392.3
2.50 73.3% 409.5
1.50 82.4% 346.7
1.75 81.2% 361.2
5000 2.00 80.0% 375.6
2.25 78.8% 390.1
2.50 77.6% 404.6
1.50 84.7% 347.7
1.75 83.8% 361.8
6000 2.00 82.9% 376.0
2.25 82.0% 390.2
2.50 81.1% 404.4
1.50 86.3% 352.9
1.75 85.5% 365.4
7000 2.00 84.7% 377.9
2.25 84.0% 390.3
2.50 83.2% 402.8
166
100%
95% TEST, 4000rpm
90% TEST, 5000rpm
85%
Volumetric Efficiency
TEST, 6000rpm
80%
75% TEST, 7000rpm
70%
MODEL, 4000rpm
65%
MODEL, 5000rpm
60%
55% MODEL, 6000rpm
50%
1.4 1.6 1.8 2.0 2.2 2.4 2.6 MODEL, 7000rpm
Pressure Ratio
Figure 6-33: HS204 volumetric efficiency with unmodified model clearances
450
TEST, 4000rpm
425
TEST, 5000rpm
Discharge Temperature, T2 (K)
400
TEST, 6000rpm
375 TEST, 7000rpm
MODEL, 4000rpm
350
MODEL, 5000rpm
325
MODEL, 6000rpm
300
1.4 1.6 1.8 2.0 2.2 2.4 2.6 MODEL, 7000rpm
Pressure Ratio
Figure 6-34: HS204 discharge temperature with unmodified model clearances
167
Table 6-24: HS204 model performance results with unmodified clearances
Speed Pressure Relative Relative

(rpm) Ratio Volumetric efficiency Error T2 (K) Error
1.50 76.3% -3.4% 340.6 0.0%
1.75 69.9% -9.9% 369.5 3.3%
4000 2.00 65.2% -14.4% 396.4 5.7%
2.25 62.4% -16.5% 420.6 7.2%
2.50 60.4% -17.6% 443.1 8.2%
1.50 81.1% -1.6% 337.1 -2.8%
1.75 75.9% -6.6% 362.4 0.4%
5000 2.00 72.2% -9.8% 385.6 2.6%
2.25 70.1% -11.1% 406.4 4.2%
2.50 68.6% -11.6% 425.8 5.2%
1.50 84.1% -0.7% 335.3 -3.6%
1.75 79.8% -4.8% 358.4 -1.0%
6000 2.00 76.7% -7.5% 379.3 0.9%
2.25 75.0% -8.5% 398.1 2.0%
2.50 73.9% -8.9% 415.7 2.8%
1.50 86.2% -0.1% 334.2 -5.3%
1.75 82.5% -3.5% 355.8 -2.6%
7000 2.00 79.8% -5.8% 375.3 -0.7%
2.25 78.5% -6.5% 392.8 0.6%
2.50 77.6% -6.7% 409.3 1.6%
average relative error: -7.8% 1.4%
maximum relative
error: -17.6% 8.2%
168
100%
TEST, 4000rpm
95%
90% TEST, 5000rpm
85%
TEST, 6000rpm
80%
TEST, 7000rpm
75%
70% MODEL R, 4000rpm
65%
MODEL R, 5000rpm
60%
55% MODEL R, 6000rpm
50%
MODEL R, 7000rpm
1.4 1.6 1.8 2.0 2.2 2.4 2.6
Pressure Ratio
Figure 6-35: HS204 volumetric efficiency with modelled rotor thermal distortion
450
TEST, 4000rpm
425
TEST, 5000rpm
400
TEST, 6000rpm
375 TEST, 7000rpm
MODEL R, 4000rpm
350
MODEL R, 5000rpm
325
MODEL R, 6000rpm
300
1.4 1.6 1.8 2.0 2.2 2.4 2.6 MODEL R, 7000rpm
Pressure Ratio
Figure 6-36: HS204 discharge temperature with modelled rotor thermal

distortion
169
Table 6-25: HS204 model performance results with rotor thermal distortion

1.50 79.7% 0.9% 337.2 -1.0%
1.75 75.5% -2.6% 362.3 1.2%
4000 2.00 73.1% -4.0% 384.5 2.5%
2.25 72.3% -3.3% 404.1 3.0%
2.50 71.9% -2.0% 422.0 3.1%
1.50 83.5% 1.3% 335.0 -3.4%
1.75 79.9% -1.6% 357.8 -0.9%
5000 2.00 77.8% -2.8% 377.9 0.6%
2.25 77.0% -2.3% 395.9 1.5%
2.50 76.7% -1.2% 412.5 1.9%
1.50 86.0% 1.5% 333.7 -4.0%
1.75 82.8% -1.2% 355.0 -1.9%
6000 2.00 81.0% -2.3% 373.9 -0.6%
2.25 80.3% -2.1% 390.7 0.1%
2.50 80.0% -1.4% 406.3 0.5%
1.50 87.6% 1.5% 333.0 -5.6%
1.75 85.0% -0.6% 353.2 -3.3%
7000 2.00 83.3% -1.7% 371.1 -1.8%
2.25 82.8% -1.4% 387.1 -0.8%
2.50 82.5% -0.9% 402.2 -0.2%
average relative error: -1.3% -0.5%
maximum relative
error: -4.0% -5.6%
170
100%
95% TEST, 4000rpm
90% TEST, 5000rpm
85%
TEST, 6000rpm
80%
75% TEST, 7000rpm
70%
MODEL RC, 4000rpm
65%
MODEL RC, 5000rpm
60%
55% MODEL RC, 6000rpm
50%
1.4 1.6 1.8 2.0 2.2 2.4 2.6 MODEL RC, 7000rpm
Pressure Ratio
Figure 6-37: HS204 volumetric efficiency with modelled rotor and casing thermal
distortion
450
TEST, 4000rpm
425
TEST, 5000rpm
400 TEST, 6000rpm
375 TEST, 7000rpm
MODEL RC, 4000rpm

350
MODEL RC, 5000rpm

325
MODEL RC, 6000rpm
300
1.4 1.6 1.8 2.0 2.2 2.4 2.6 MODEL RC, 7000rpm
Pressure Ratio
Figure 6-38: HS204 discharge temperature with rotor and casing thermal
distortion
171
Table 6-26: HS204 model performance results with rotor and casing thermal
distortion

1.50 75.5% -4.4% 341.6 0.3%
1.75 68.4% -11.8% 371.8 3.9%
4000 2.00 62.6% -17.8% 400.8 6.8%
2.25 58.3% -22.0% 427.3 8.9%
2.50 55.2% -24.7% 452.2 10.4%
1.50 80.7% -2.1% 337.7 -2.6%
1.75 75.0% -7.7% 363.7 0.7%
5000 2.00 70.5% -11.9% 387.9 3.3%
2.25 67.4% -14.5% 409.8 5.0%
2.50 65.3% -15.9% 430.2 6.3%
1.50 83.9% -1.0% 335.6 -3.5%
1.75 79.2% -5.5% 359.1 -0.7%
6000 2.00 75.5% -8.9% 380.8 1.3%
2.25 73.1% -10.9% 400.1 2.5%
2.50 71.6% -11.7% 418.2 3.4%
1.50 86.0% -0.3% 334.4 -5.2%
1.75 82.0% -4.1% 356.3 -2.5%
7000 2.00 78.9% -6.9% 376.3 -0.4%
2.25 77.0% -8.3% 394.1 1.0%
2.50 75.8% -8.9% 410.8 2.0%
maximum relative
error: -24.7% 10.4%
172
6.5.5 HS204 PERFORMANCE RESULTS WITHOUT JACKET COOLING

Additional tests were performed to investigate the effect of the casing water
cooling jacket on the compressor performance. This repeat testing did not cover
all of the original test points and therefore only results at 6000rpm are
presented.
In Table 6-27 the test results with and without jacket cooling are summarised
and as before these have been adjusted, through linear interpolation, to
produce the results in Table 6-28 at specific pressure ratios.
Table 6-27: HS204 performance test results at 6000rpm with and without jacket
cooling
Pressure
Test Ratio Volumetric efficiency T2 (K)
1.65 84.2% 355.6
1.84 83.4% 367.5
jacket cooling
2.01 82.9% 376.9
2.11 82.5% 381.9
1.57 78.4% 355.6
1.80 77.5% 365.3
no jacket cooling
2.09 76.2% 385.1
2.60 73.8% 412.1
Table 6-28: HS204 interpolated performance test results at 6000rpm with and
without jacket cooling
Pressure
Test Ratio Volumetric efficiency T2 (K)
1.50 84.7% 347.7
1.75 83.8% 361.8
jacket cooling 2.00 82.9% 376.0
2.25 82.0% 390.2
2.50 81.1% 404.4
1.50 78.8% 350.5
1.75 77.7% 364.5
no jacket cooling 2.00 76.6% 378.5
2.25 75.4% 392.5
2.50 74.3% 406.5
173
100%
95%
TEST: jacket cooling
90%
85%
80% TEST: no jacket

cooling
75%
70%
65% MODEL R
60%
55%
MODEL RC
50%
1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8
Pressure Ratio
Figure 6-39: HS204 volumetric efficiency at 6000rpm with and without jacket
cooling
450
TEST: jacket cooling

425
400
TEST: no jacket
cooling
375
350 MODEL R
325
MODEL RC
300
1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8
Pressure Ratio
Figure 6-40: HS204 discharge temperature at 6000rpm with and without jacket
cooling
174
The results from MODEL R and MODEL RC are given in Figure 6-39 and Figure
6-40 for testing with and without jacket cooling. MODEL R is more closely
aligned to the test results with jacket cooling and MODEL RC is more closely
aligned to the test results without jacket cooling. These results are presented in
Table 6-29 which compares MODEL R against each test case. Then in Table
6-30, MODEL RC is compared with the same test cases.
Table 6-29: HS204 ‘MODEL R’ performance results at 6000rpm with rotor thermal
expansion
Test Pressure Relative Relative

reference Ratio Volumetric efficiency Error T2 (K) Error
1.50 86.0% 1.5% 333.7 -4.0%
1.75 82.8% -1.2% 355.0 -1.9%
jacket
2.00 81.0% -2.3% 373.9 -0.6%
cooling
2.25 80.3% -2.1% 390.7 0.1%
2.50 80.0% -1.4% 406.3 0.5%
average relative error: -1.1% -1.2%
maximum relative error: -2.3% -4.0%

1.50 86.0% 9.2% 333.7 -4.8%
1.75 82.8% 6.6% 355.0 -2.6%
no jacket
2.00 81.0% 5.8% 373.9 -1.2%
cooling
2.25 80.3% 6.4% 390.7 -0.5%
2.50 80.0% 7.6% 406.3 0.0%
average relative error: 7.1% -1.8%
maximum relative error: 9.2% -4.8%
175
Table 6-30: HS204 ‘MODEL RC’ performance results at 6000rpm with rotor and
casing thermal expansion

1.50 83.9% -1.0% 335.6 -3.5%
1.75 79.2% -5.5% 359.1 -0.7%
jacket
2.00 75.5% -8.9% 380.8 1.3%
cooling
2.25 73.1% -10.9% 400.1 2.5%
2.50 71.6% -11.7% 418.2 3.4%
maximum relative error: -11.7% -3.5%

1.50 83.9% 6.5% 335.6 -4.2%
1.75 79.2% 2.0% 359.1 -1.5%
no jacket
2.00 75.5% -1.4% 380.8 0.6%
cooling
2.25 73.1% -3.1% 400.1 1.9%
2.50 71.6% -3.7% 418.2 2.9%
average relative error: 0.1% -0.1%
maximum relative error: 6.5% -4.2%
6.5.6 HS204 RADIAL GAP DISTORTION RESULTS

The changes in the radial gap at the location of the proximity probe; as
measured on test, and as predicted by the model, are shown in Figure 6-41 and
Figure 6-42. For clarity, these have been plotted as the absolute value of the
local radial gap. Where there is some uncertainty about the absolute clearance
values, as is the case for the test results, this has been set to a nominal gap size
of 0.150 mm and then adjusted by the actual measured clearance distortion. It is
the degree of operational distortion from this nominal gap at a given operating
temperature that is the focus of interest.
The corresponding temperatures, used to calculate the modelled distortions,

are given in Figure 6-43 and Figure 6-44. These give some additional insight
into the model behaviour. The differences in temperature between the rotor
and casing, ΔR and ΔRC for each respective model, are related to the gap
distortion as set out in equations (60) to (65) in Chapter 5.
176
Radial Gap on MAIN Rotor Bore at 6000rpm

150
125
Local Radial Gap (μm)
100
75
50
25
0
0 25 50 75 100 125 150
T2 - Tamb (K)
TEST: jacket cooling TEST: no jacket cooling MODEL R MODEL RC
Figure 6-41: Measured and modelled radial gap on MAIN rotor bore
Radial Gap on GATE Rotor Bore at 6000rpm

150
125
Local Radial Gap (μm)
100
75
50
25
0
0 25 50 75 100 125 150
T2 - Tamb (K)
TEST: jacket cooling TEST: no jacket cooling MODEL R MODEL RC
Figure 6-42: Measured and modelled radial gap on GATE rotor bore
177
Local Temperature at MAIN Rotor Bore at 6000rpm

150
125
Temperature (deg C)
100
75 ΔRC
50
ΔR
25
0
0 25 50 75 100 125 150
T2 - Tamb (K)
MODEL R: local casing temp, T_B1 MODEL RC: local casing temp, T_B1
MODEL R: local rotor temp, T_RP1 MODEL RC: local rotor temp, T_RP1
Figure 6-43: Local temperatures at MAIN rotor bore radial gap
Local Temperature at GATE Rotor Bore at 6000rpm

150
125
Temperature (deg C)
100
75
50 ΔRC
ΔR
25
0
0 25 50 75 100 125 150
T2 - Tamb (K)
MODEL R: local casing temp, T_B2 MODEL RC: local casing temp, T_B2
MODEL R: local rotor temp, T_RP2 MODEL RC: local rotor temp, T_RP2
Figure 6-44: Local temperatures at GATE rotor bore radial gap
178
6.5.7 DISCUSSION
All test results indicate that the volumetric efficiency and temperature vary
more or less linearly with the pressure ratio and suggest that there is a limiting
volumetric efficiency that will not be exceeded regardless of how low the
external pressure ratio is; this is due to the internal compression that occurs
within the compressor. In contrast, the model results do show some
nonlinearity, particularly at lower pressure ratios where the rate of change in
volumetric efficiency is higher. When operating at higher, more realistic
pressure ratios for the designed volume index, the model shows a more linear
relationship between volumetric efficiency and pressure ratio. The discrepancy
at lower pressures merits further investigation and possible refinement to the
setup of the chamber model used. However it should not impede the current
investigations significantly.
Assuming no heat transfer from the gas to the compressor and surroundings
prior to measurement of the outlet temperature, T2, on test, the adiabatic
efficiencies were calculated to range from 70.6% (at the lowest speed and
highest pressure ratio) to 83.2% (at the highest speed and lowest pressure
ratio). These efficiencies are in the expected range which supports the
assumption that heat transfer from the gas can be neglected up to the point of
measurement. The same assumption is made in the compressor model to
ensure that the temperature comparisons presented are appropriate. More
discussion on the calculation of T2 is included in appendix ‘B.1.1 Model
Discharge Temperature’. More detail on the measurement of T2 and how this is
used to calculate efficiency is provided in appendix ‘C.1 Compressor
Boundaries’ and ‘C.4 Adiabatic Efficiency’.
Initial results with the first model that uses unmodified nominal design
clearances (labelled MODEL), underestimate flow and overestimate the
discharge temperature. In addition, the range of modelled volumetric efficiency
and temperature over both speed and pressure ratio is greater than that
measured on test.
179
The second set of modelled results, with clearances corrected for rotor thermal
distortion (MODEL R) show a considerably smaller error against the test results
over the full range of operating points.
The third set of modelled results, with clearances corrected for rotor and casing
thermal distortion (MODEL RC) show the largest deviations from the test
results.
For the results shown in Figure 6-33 to Figure 6-38, where the test compressor
casing jacket is water cooled, the most accurate performance predictions were
achieved using Model R, based on nominal design clearances that are modified
to estimate rotor thermal distortion only. The fact that this model, which
resulted in the closest match to the measured flow, shown in Table 6-25, also
resulted in the closest match to the measured temperature, provides some
validation of the model assumptions.
A comparison of the model predicted results at 6000rpm, with additional test

results, with no jacket cooling and the covers for the cooling jacket removed, is
given in Figure 6-39 and Figure 6-40. The model predictions that agreed with
the test results, with no jacket cooling, most closely, was MODEL RC, which
estimated rotor AND casing thermal distortion. This is an important result that
further confirms that operational temperature distributions and clearances are
indeed changing compressor performance and need to be accounted for during
modelling.
Figure 6-41 shows good agreement between the measured radial gap on the
MAIN rotor bore when tested with jacket cooling and the MODEL R results. On
the GATE rotor bore (Figure 6-42) these results are not as good and the
measured radial gap is smaller than predicted by MODEL R. Figure 6-43 and
Figure 6-44 help illustrate how the modelled gaps are calculated - the modelled
change in radial gap (from the nominal 0.150mm) is proportional to the
difference between the rotor and casing temperatures for each respective
model. In MODEL R the casing is constant at 20 deg C and the bigger reduction
in the main rotor is simply due to that rotor being hotter in the model (based on
the transverse plane average temperature). The measured clearances have
180
actually reduced more on the GATE rotor bore with jacket cooling; it is not
possible to say whether this is due to the gate rotor actually being hotter
because the actual temperature distribution and distortion of the casing are
unknown. Considering the simple nature of the thermal assumptions used in
each model, the overall agreement between the jacket cooled radial gap
measurements and the MODEL R predictions are fairly reasonable.
For MODEL RC the modelled clearance reduction is less than for MODEL R
because the local casing expansion mitigates the local rotor expansion to some
extent. This ties in with the modelled performance results as the bigger radial
clearances reduce volumetric efficiency. Based on this theory, the hypothesis
was that the measured radial gap without jacket cooling would be bigger than
with jacket cooling however the results from the proximity probes on the MAIN
and GATE rotor bores show the opposite case to be true. This seems to
contradict the performance results without jacket cooling which suggest a net
clearance increase had occurred. The logical conclusion is that the measured
radial clearance reduction is a localised phenomenon.
The simplified thermal analysis, implemented in the model, is useful for

capturing net changes in the compressor clearance in order to adjust the
predicted performance but it is limited in its ability to predict localised thermal
distortions accurately, particularly as a result of casing thermal distributions
and distortions. The complexity of this problem is illustrated in Figure 6-45
which was produced using FEA, in an investigation performed by Howden
Compressors Ltd (Howden Compressors Ltd. 2013). This was not created
within the scope of the current research and is only included to illustrate
possible distortion. The casing temperature distribution was predicted, based
on boundary temperatures and assumed heat transfer coefficients. The
distortions due to thermal expansion and resulting stress distributions were
then calculated. The colour contours show the total magnitude of the
displacement in any direction, where zero displacement occurs at the
compressor feet. The suction side of the casing on the right hand side of the
image has actually moved downward slightly. The approximate locations of the
rotor body and the proximity probe have been superimposed on to the figure.
181
With the opposing movement of the bearings at each end and the position of the
constraints at the feet of the compressor, it is easier to understand how it could
be possible for the local radial gap to be reduced with a hotter casing
temperature.
Figure 6-45: Example of overall casing displacement due to thermal loading

(courtesy of Howden Compressors Limited.) with approximate rotor body and
proximity probe location superimposed
Effectively, the compressor is hotter on top than it is on the bottom due to the
location of the high pressure discharge port on top. The temperature difference
causes the entire compressor to bow downwards at each end so that the
relative position of the proximity probe is higher, resulting in reduced clearance
to the rotor. This would be accompanied by a larger than anticipated increase of
the radial clearance at the top of the compressor and it is likely that the net
radial leakage area does increase, as predicted.
182

The applied routines for mapping boundary conditions from a chamber model
onto the compressor rotor and casing surfaces work as expected, with sensibly
bounded chambers. When used to update operational clearances and
thermodynamic performance, iteratively, a significant change in the net
performance was predicted for the DRUM127 compressor. This varied,
depending on whether just the rotors, or both the rotors and the casing were
distorted. At a minimum, this provides useful details of any modelled
uncertainty due to thermal effects. With some additional analyses and
understanding of the thermal behaviour of a given compressor this can be used
to improve the accuracy of performance predictions without the need to correct
operating clearance values manually.
Measuring the change in performance due to predicted change in operational

clearances, as was done for the DRUM127 case, is not the ideal way to validate
the modelled thermal distortions themselves. Ideally, direct measurements of
the operating interlobe gap should be obtained. The fact is that it is very difficult
to measure the rotor temperatures directly, let alone the distorted clearance
gap, and it was not possible to obtain these results during this study. In the
absence of direct operational measurements, the tear down results for the
WCVTA510 oil injected case are particularly useful as they show evidence of
rotor contact that was predicted by the model. Results suggest that the
developed model is a useful tool in the evaluation and design of interlobe
clearance distributions for a given operating duty.
In the case of the HS204 oil free case study, in addition to performance results, a
direct measurement of the radial gap between the rotors and casing was
utilised to assess the model. This showed that while an appropriate thermal
distortion model could improve performance results for a specific operating
duty and cooling regime, it was not an appropriate tool for detailed thermal
analysis of localised rotor to casing clearances which are too dependent on the
full casing geometry and this is not represented by this model.
183
Chapter 7
Conclusions and Future Work
7.1 CONCLUSIONS
This research was carried out while working full time in the R&D department at
Howden Compressors Ltd; working on various projects but primarily in the
design of new retrofit rotors for an existing range of compressors. From this it
soon became clear that, from an industrial viewpoint a major problem in
optimising performance was to be able to evaluate how internal clearances
could be minimised, without the risk of failure as a result of distortion due to
thermal effects during operation. A detailed literature survey showed very little
published in this area and clearance design was found to be mostly dependent
on empirical data and historical experience. It was clear that this was the single
most critical area of rotor design and manufacture in terms of performance and
reliability. This is particularly true for Howden whose products serve a wide
range of applications and operating duties.
A validated procedure has therefore been developed whereby a chamber model

has been adapted to predict and correct for duty dependant thermal effects.
This can readily be applied to the design and optimisation of twin screw
compressors for specific applications.
In the case studies reviewed, the inclusion of thermal clearance corrections

generally improved the accuracy of the model when compared with test results.
However, it has been shown from the oil free case studies that the most
applicable type of clearance corrections, i.e. whether only rotor thermal
distortions or rotor and casing distortions are used, is dependent on the
compressor configuration such as whether the casing is cooled. This shows that
while this approach can improve model performance there is not a universal
setup that works for all compressor types.
Case studies show that this work is useful and applicable in the design and
review of interlobe clearance distributions for specific applications; in terms of
184
predicting the likelihood of rotor to rotor contact; and predicting where contact
is most likely to occur. Mapping thermodynamic results from the chamber
model on to the rotor geometry has provided useful insight into rotor
temperature exposure, how this varies between the main and gate rotors, and
how it’s affected by other rotor parameters. Wrap angle and lobe combination
are two examples of parameters that have been shown to change rotor
temperature exposure. Calculated local rotor temperature exposure was used in
an analytical clearance correction procedure which was shown to predict the
same interlobe clearance behaviour evident from tear down inspection of an oil
injected compressor.
Investigation of localised radial clearances did however highlight the limited

applicability of these procedures to detailed analysis of rotor to casing
clearances. Without a cooling jacket, the predicted change in the radial gap due
to thermal distortion was different to the measured change in the gap. While the
thermal model used can predict general trends in the overall clearance change,
as supported by the modelled performance results, it can’t adequately capture
local distortions on the compressor casing which are dependent on the full
casing geometry and temperature distribution.
A useful outcome of developing this mapping procedure was the novel Rotor
Boundary Map which provides a unique way of visualising and comparing the
key geometrical properties of different profiles such as the length of the sealing
lines and how the sealing lines are related to the formation of the blow hole
leakage area.
185
7.2 FUTURE WORK

One of the biggest challenges in this work was obtaining results that would
provide direct, simple, verification of the basic constituent parts of the model.
For example, as well as measuring the radial gap on test and comparing it with
the modelled gap, it would ideally be possible to make direct comparisons of the
modelled and tested rotor and casing temperatures at this location. As
mentioned, there were limitations to this research in the test data that could be
obtained. Furthermore, it is a considerable challenge measuring the
temperature of the rotor while running and direct measurement is sometimes
not possible as in the case of the interlobe gap. Improving test instrumentation
for clearance and thermal analysis would be very useful.
In the absence of more extensive test results it would be useful to compare the
thermal distortions predicted by this model against FEA results for the same
case. Similarly it would be useful to compare the performance sensitivity to
clearances against CFD results for the same case.
The developed model would be a useful tool in a more general and

comprehensive rotor optimisation and design framework. This would also
include rotor profile optimisation. During profile optimisation there is a trade
off between features such as interlobe sealing line length and blow-hole area.
Better understanding of how the clearance gap along the interlobe sealing line
is affected by thermal distortion at a particular operating temperature would be
a valuable input to this. Some performance indicators of a given profile such as
specific displacement for a given rotor size can contribute to rotor optimisation
prior to thermodynamic simulation: similarly there may be potential to evaluate
different rotor designs by using the Rotor Boundary Map to review the
interaction between the various sealing lines and the blow-hole.
In very high pressure applications rotor movement and deflection will become
more significant and may be worth including in duty dependent operational
clearance distortions, in the same way that temperature distribution on the
rotor and casing surfaces has been utilised in this work. It follows that there is
186
clear potential to use pressure distribution for a wide range of purposes

including rotor optimisation, bearing design, and rotor deflection analysis.
187
References
References
AMOSOV, P.E., 1977. Vintovie kompresornie mashinii – Spravochnik (Screw
Compression Machines-Handbook). Mashinstroienie, Leningrad.
ASTBERG, A., 1984. US Patent 4435139 'Screw Rotor Machine and Rotor Profile
Therefor'.
BELL, I.H., GROLL, E.A., BRAUN, J.E. and HORTON, W.T., 2012. A computationally
efficient hybrid leakage model for modelling leakage in positive displacement
compressors, International Compressor Engineering Confernce at Purdue 2012,
Paper 2038.
BUCKNEY, D., KOVACEVIC, A. and STOSIC, N., 2011. Consideration of Clearances

in the Design of Screw Compressor Rotors, 7th International Conference on
Compressors and their Systems, London, 5th - 6th September 2011.
FLEMING, J.S. and TANG, Y., 1995. Analysis of leakage in a twin screw
compressor and its application to performance improvement. Proceedings of the
Institution of Mechanical Engineers, Part E: Journal of Process Mechanical
Engineering, 209, pp. 125-136.
FLEMING, J.S., TANG, Y. and COOK, G., 1998a. The twin helical screw
compressor. Part 1: development, applications and competitive position.
Proceedings of the Institution of Mechanical Engineers, Part C: Journal of
Mechanical Engineering, 212(5), pp. 355-367.
FLEMING, J.S., TANG, Y. and COOK, G., 1998b. The twin helical screw
compressor. Part 2: a mathematical model of the working process. Proceedings
of the Institution of Mechanical Engineers, Part C: Journal of Mechanical
Engineering, 212(5), pp. 369-380.
FLEMING, J.S., YOU, C.X. and TANG, Y., 1994. Rotor tip design in oil injected
helical twin screw compressors with respect to viscous friction loss,
INSTITUTION OF MECHANICAL ENGINEERS CONFERENCE PUBLICATIONS 1994,
pp. 115-115.
FUJIWARA, M., MORI, H. and SUWAMA, T., 1974. Prediction of the Oil Free
Screw Compressor Performance Using Digital Computer, International
Compressor Engineering Conference at Purdue 1974, Paper 119.
FUJIWARA, M. and OSADA, Y., 1995. Performance analysis of an oil-injected

screw compressor and its application. International Journal of Refrigeration,
18(4), pp. 220-227.
FUJIWARA, M., KASUYA, K., MATSUNAGA, T. and WATANABE, M., 1984.

Computer Modeling for Performance Analysis of Rotary Screw Compressor.
188
References
Proceedings of the 1984 International Compressor Engineering Conference - At

Purdue. 1984, Paper 503, pp. 536-543.
GAO, T.Y., YANG, D.F., CAO, F. and JIAO, J.C., 2011. Temperature and
thermodynamic deformation analysis of the rotors on a twin screw multiphase
pump with high gas volume fractions. Applied Physics and Engineering, 12(9),
pp. 720.
HANJALIC, K. and STOSIC, N., 1997. Development and optimization of screw

machines with a simulation model - Part II: thermodynamic performance
simulation and design optimization. Journal of Fluids Engineering, Transactions
of the ASME, 119(3), pp. 664-670.
HAUSER, J. and BRUMMER, A., 2010. Geometrical abstraction of screw

compressors for thermodynamic optimization. Proceedings of the Institution of
Mechanical Engineers, Part C: Journal of Mechanical Engineering Science, (225),
pp. 1399.
HAUSER, J., BRUMMER, A. and KAUDER, K., 2008. Rotor Profile Generation and
Optimization of Screw Machines Using NURBS, International Compressor
Engineering Conference at Purdue 2008, Paper 1861.
HOLMES, C., 1990. A study of screw compressor rotor geometry leading to a

method for inter-lobe clearance measurement, Huddersfield Polytechnic.
HOLMES, C.S., 2008. Form grinding of precision helical rotors and gears using
closed-loop profile control, 10th International Power Transmission and Gearing
Conference, presented at - 2007 ASME International Design Engineering Technical
Conferences and Computers and Information in Engineering Conference,
IDETC/CIE2007, September 4, 2007 - September 7, American Society of
Mechanical Engineers, pp. 775-781.
HOWDEN COMPRESSORS LTD., 2013. HCLRD0111. Glasgow, UK: Howden

Compressors.
HOWDEN COMPRESSORS LTD., 2008. Howden Academy: Screw Compressor

Selection Aspects.
HSIAO, H.H., WU, Y.R. and HSIEH, H.C., 2012. Non-Uniform Clearance between
Rotor Surfaces and Its Effect on Machine Performance in Twin-Screw
Compressors, International Compressor Engineering Conference at Purdue 2012,
Paper 2193.
HSIEH, S.H., SHIH, Y.C., HSIEH, W., LIN, F.Y. and TSAI, M.J., 2011. Calculation of
temperature distributions in the rotors of oil-injected screw compressors.
International Journal of Thermal Sciences, 50(7), pp. 1271-1284.
INTERNATIONAL ORGANIZATION FOR STANDARDIZATION, a. ISO 1217:2009

Displacement compressors -- Acceptance tests.
189
References
INTERNATIONAL ORGANIZATION FOR STANDARDIZATION, b. ISO 5167-2:2003

Measurement of fluid flow by means of pressure differential devices inserted in
circular cross-section conduits running full -- Part 2: Orifice plates. International
Organization for Standardization;.
KOVACEVIC, A., STOSIC, N. and SMITH, I.K., 2002a. The Influence of Rotor
Deflection upon Screw Compressor Performance. VDI BERICHTE, 1715, pp. 17-
28.
KOVACEVIC, A., STOSIC, N. and SMITH, I.K., 2002b. Numerical simulation of fluid
flow and solid structure in screw compressors, ASME 2002 International
Mechanical Engineering Congress and Exposition, IMECE2002, November 17,
2002 - November 22, 2002 2002b, American Society of Mechanical Engineers,
pp. 409-416.
KOVACEVIC, A., STOSIC, N. and SMITH, I.K., 2003. Three Dimensional Numerical
Analysis of Screw Compressor Performance. Journal of Computer Methods in
Applied Mechanics and Engineering, 3(2), pp. 259.
KOVACEVIC, A., STOSIC, N., MUJIC, E. and SMITH, I.K., 2006a. Influence of
thermal dilation upon design of screw machines. International Design
Conference - Design 2006, Dubrovnik - Croatia, .
KOVACEVIC, A., STOSIC, N. and SMITH, I.K., 2006b. Numerical simulation of

combined screw compressor-expander machines for use in high pressure
refrigeration systems. Simulation Modelling Practice and Theory, 14(8), pp.
1143-1154.
LITVIN, F.L. and FUENTES, A., 2004. Gear Geometry and Applied Theory. 2nd edn.
Cambridge.
MUJIC, E., KOVACEVIC, A., STOSIC, N. and SMITH, I.K., 2010. Advanced Design
Environment for Screw Machines, International Compressor Engineering
Conference at Purdue 2010, Paper 1971.
MUJIC, E., 2009. A numerical and experimental investigation of pulsation induced

noise in screw compressors, City University London.
NIKOLOV, A. and BRUMMER, A., 2012. Influence of thermal deformation on the

characteristic diagram of a screw expander in the automotive application of
exhaust heat recovery, International compressor engineering conference at
Purdue 2012, Paper 2128.
OHAMA, T., KURIOKA, Y. and TANAKA, H., 2006. Process gas applications where
API 619 screw compressors replaced reciprocating and centrifugal
compressors, Proceedings of the Thirty-Fifth Turbomachinery Symposium, 2006
2006.
190
References
O'NEILL, P., 1977. The Screw Compressor: A Short History of its Development
and its Application to the Fields of Air Conditioning and Mine Cooling, A paper
presented to the symposium 'Latest Concepts in Cooling of Deep Mines'. South
Africa.
O'NEILL, P.A., 1966. Mechanical Design and Efficiency of Screw Compressors.

Paper presented to the Institute of Mechanical Engineers: .
POWELL, G., WEATHERS, B. and SAULS, J., 2006. Transient Thermal Analysis of
Screw Compressors, Part III: Transient Thermal Analysis of a Screw
Compressor to Determine Rotor-to-Rotor Clearances, Proceedings of
International compressor Engineering Conference at Purdue 2006, Paper 1813.
RINDER, L., 1987. Screw rotor profile and method for generating. US4643654 A
edn. US: .
RINDER, L., 1979. Schraubenverdichter (Screw Compressors). New York:

Springer Verlag.
SACHS, R., 2002. Experimental Investigations of Gas Flows in Screw Machines,

University of Dortmund.
SAKUN, I.A., 1960. Vintovie kompresorii (Screw Compressors). Mashinostroenie

Leningrad: .
SANGFORS, B., 1984. Computer Simulation of the Oil Injected Twin Screw
Compressor, International Compressor Engineering Conference at Purdue 1984,
Paper 502.
SAULS, J., POWELL, G. and WEATHERS, B., 2006a. Transient Thermal Analysis of
Screw Compressors, Part I: Use of Thermodynamic Simulation to Determine
Boundary Conditions for Finite Element Analyses, Proceedings of International
Compressor Engineering Conference at Purdue 2006a, Paper 1811.
SAULS, J., 2011. Use of finite element and computational fluid dynamics analysis
in the development of positive displacement compressor simulations, 7th
International Conference on Compressors and Their Systems 2011, September 5,
2011 - September 6, 2011 2011, Woodhead Publishing Limited, pp. 5-13.
SAULS, J., 1996. Development of a Comprehensive Thermodynamic Modelling

System for Refrigerant Screw Compressors, Proceedings of International
Compressor Engineering Conference at Purdue 1996, 1098.
SAULS, J. and POWELL, G., Influence of Manufacturing and Operational Effects

on Screw Compressor Rotor Pair Clearances, Tagung Schraubenmaschinen,
October 2010.
SAULS, J., POWELL, G. and WEATHERS, B., 2007. Thermal deformation effects on
screw compressor rotor design, International Conference on Compressors and
191
References
their Systems, September 10, 2007 - September 12, 2007 2007, Chandos
Publishing, pp. 159-168.
SAULS, J., POWELL, G. and WEATHERS, B., 2006b. Transient thermal analysis of
screw compressors part I - Development of thermal model. VDI Berichte, (1932),
pp. 19-29.
SAULS, J., POWELL, G. and WEATHERS, B., 2006c. Transient thermal analysis of
screw compressors part II - Application. VDI Berichte, (1932), pp. 195-204.
SCHIBBYE, L.B., 1979. US Patent 4140445 'Screw-Rotor Machine with Straight

Flank Sections'.
SESHAIAH, N., GHOSH, S.K., SAHOO, R.K. and SARANGI, S.K., 2007. Mathematical
modeling of the working cycle of oil injected rotary twin screw compressor.
Applied Thermal Engineering, 27(1), pp. 145-155.
SINGH, P.J., 1990. Exact analytical representation of screw compressor rotor

geometry, International Compressor Engineering Conference at Purdue 1990,
Paper 785.
SINGH, P.J. and BOWMAN, J.L., 1990. 'Calculation of blow-hole area for screw
compressors'. Proceedings of the 1990 International Compressor Engineering
Conference, , pp. 938-938.
SINGH, P.J. and ONUSCHAK, A.D., 1984. Comprehensive, Computerized Method

for Twin-Screw Rotor Profile Generation and Analysis. Proceedings of the 1984
International Compressor Engineering Conference - At Purdue. 1984, Purdue
Univ, Ray W. Herrick Lab, pp. 519-527.
SRM, 1966. Oil Injected Slide Valve Controlled Refrigeration Compressor K-211
(Howden WRVD 204). K371129. SRM.
SRM, 1954. Calculation of the Characteristics of the Screw Compressor . 839/K-

74. SRM.
SRM, 1953. Calculation of the volume curves of screw compressors. 892/K-109.

SRM.
SRM, 1952. Standardisation of Rotor Dimensions of 1-Stage Screw Compressor

with 4+6 Lobe Combination and Profile According to K.364942. 887/K-99. SRM.
SRM, 1950. The temperature of noncooled screw compressor rotors. 838/K73.

SRM.
STOSIC, N., 2001. US Patent 6296461 B1 'Plural Screw Positive Displacement

Machines'.
192
References
STOSIC, N., HANJALIC, K. and KOPRIVICA, J., 1986. A Contribution to the

Mathematical Modelling of Screw Compressor Working Processes, Strojarstvo
Journal, Zagreb, 28(2), pp. 95-100.
STOSIC, N., MUJIC, E., SMITH, I.K. and KOVACEVIC, A., 2008. Profiling of Screw
Compressor Rotors by Use of Direct Digital Simulation, International
STOSIC, N., SMITH, I.K. and KOVACEVIC, A., 2005. Screw Compressors:
Mathematical Modelling and Performance Calculation. UK: Springer.
STOSIC, N., SMITH, I.K. and KOVACEVIC, A., 2003. Optimisation of screw
compressors. Applied Thermal Engineering, 23(10), pp. 1177-1195.
STOSIC, N., SMITH, I.K., KOVACEVIC, A. and MUJIC, E., 2010. Three decades of
modern practise in screw compressors, International Engineering Compresssor
Conference 2010, Paper 1942.
STOSIC, N., 1998. On gearing of helical screw compressor rotors. Proceedings of

the Institution of Mechanical Engineers, Part C: Journal of Mechanical
Engineering, 212(7), pp. 587-594.
STOSIC, N. and HANJALIC, K., 1997. Development and optimization of screw

machines with a simulation model - Part 1: profile generation. Journal of Fluids
Engineering, Transactions of the ASME, 119(3), pp. 659-663.
STOSIC, N., SMITH, I. and KOVACEVIC, A., 2003. Rotor interference as a criterion
for screw compressor design. Journal of Engineering Design, 14(2), pp. 209.
STOSIC, N., 2005. Identification of constraints in the optimal generation of

screw compressor rotors by the pressure angle method, , September 4, 2005 -
September 7, 2005 2005, John Wiley and Sons Inc, pp. 33-42.
STOSIC, N., 2015. On heat transfer in screw compressors, International Journal

of Heat and Fluid Flow, 51, pp. 285-297.
SUNDT, C.V., 1997. Method and apparatus for generating profiles of cooperating
screw rotors. US5608648 A edn. US: .
TANG, Y., 1995. Computer aided design of twin screw compressors, University of
Strathclyde.
TANG, Y. and FLEMING, J.S., 1992. Obtaining the optimum geometrical

parameters of a refrigeration helical screw compressor, Proceedings of the
International Compressor Engineering Conference, Purdue, USA 1992, pp. 221-
227.
TANG, Y. and FLEMING, J.S., 1994. Clearances between the rotors of helical
screw compressors: their determination, optimization and thermodynamic
193
References
consequences. Proceedings of the Institution of Mechanical Engineers, Part E:

Journal of Process Mechanical Engineering, 208, pp. 155-170.
WEATHERS, B., SAULS, J. and POWELL, G., 2006. Transient Thermal Analysis of
Screw Compressors, Part II: Transient Thermal Analysis of a Screw Compressor
to Determine Rotor-to-Housing Clearances, Proceedings of International
WEISSTEIN, E.W., , "Epitrochoid." From MathWorld--A Wolfram Web Resource.

Available: http://mathworld.wolfram.com/Epitrochoid.html2015].
WILLIAMS, T., KELLEY, C. and 'MANY OTHERS', 2013-last update, Gnuplot 4.6:
an interactive plotting program. Available: www.gnuplot.info/.
WU, Y., 1995. Novel Design Methods of Rotor and Cutter Profiles for the Twin-
Screw Compressor, National Chung-Cheng University, Chia-Yi, Taiwan.
WU, Y. and FONG, Z., 2009. Optimization design of an explicitly defined rack for
the generation of rotors for twin-screw compressors. Mechanism and Machine
Theory, 44(1), pp. 66-82.
ZAYTSEV, D. and INFANTE FERREIRA, C.A., 2005. Profile generation method for
twin screw compressor rotors based on the meshing line. International Journal
of Refrigeration, 28(5), pp. 744-755.
194
Appendix A: Geometry Calculations
Appendix A
Geometry Calculations
A.1 BASIC GEOMETRY AND TERMINOLOGY
There are a number of general rotor profile features that are common for all
twin screw compressor rotors. The rotor pair always comprises of a main and a
gate rotor. The main rotor has a smaller number of lobes than the gate resulting
in a gear ratio that reduces the speed of rotation of the gate. Despite having
fewer teeth the main rotor is often close or equal to the gate rotor in diameter,
this is because the main rotor has protruding rounded lobes which mesh with
scalloped recesses in the gate rotor known as flutes. This arrangement results in
rotors with an interlobe sealing line that extends most of the way between the
two cusps on the casing thus reducing the leakage area between the rotors and
the casing.
H1 H2
D1 D2
rw1
rw2
Figure A-1: Basic rotor dimensions
Figure A-1: Basic rotor dimensions, shows the cross section of a 4-6 rotor
combination in the transverse plane (perpendicular to the rotor axis).
195
2π(rw1)
φw1(rw1)
λw1
ψw1
h1
Figure A-2: Helix angle calculation
The relationship between the main pitch radius, rw1, and the helix angle at this
particular radius is described geometrically in Figure A-2 which is a triangle of
the unwrapped helix where the horizontal line is parallel with the rotor axis. In
this figure the larger of the similar triangles represents the helix for a full 360°,
the parameter h is known as the ‘lead length’. The smaller triangle is truncated
to the actual rotor length, L. The parameter φw1 is the wrap angle and is
proportional to the rotor length for a given helix angle, ψ. The lead angle, λ, is
defined as the complement of the helix angle. The wrap angle is typically about
300° for the main rotor but can vary. The helix angle defined at the pitch point
is common for the main and gate rotors so that ψw1 = ψw2. The actual rotor
length is usually defined by the ratio L/D and similarly the main addendum and
dedendum are sometimes defined as a proportion of the rotor diameter, D, this
makes for easier scaling of rotors and provides a more intuitive feel for how the
parameters affect the profile geometry. The following equations are commonly
used in rotor geometry calculations:
Pitch circles:
(1)
196
(2)
The gate rotor wrap angle is:
(3)
The L/D is defined for the main rotor therefore the main rotor outer diameter is
used to find the length:
(4)
Rotor lead lengths:
(5)
(6)
The lead per unit angle (usually degrees) for the main and gate rotors are:
(7)
(8)
Rotor helix angle at pitch radius (common for both rotors):
(9)
Lead angle at pitch radius (common for both rotors):
( 10 )
The lobe offset angles are useful for adjusting the rotor profiles therefore these
are defined:
( 11 )
197
( 12 )
r1o r2o
φc1
α1cp α2cp
φc2
C
A
Figure A-3: Angles to casing cusp
The position of the casing cusp (neglecting clearances) is calculated using the
outer radii of each rotor and the rotor centre distance by applying the Cosine
Rule (see Figure A-3: Angles to casing cusp):
( 13 )
( 14 )
The Cartesian cusp co-ordinates can thus be calculated:
( 15 )
( 16 )
These cusp angles define the limiting values for the parameter β that describes
the angular position on the casing bore surfaces:
( 17 )
198
( 18 )
A.2 MESHING CONDITIONS FOR CONJUGATE PROFILE

For an effective positive displacement machine the main and gate rotors must
mesh to form a continuous seal along the full length of the rotors at all times.
This requires that the gate rotor is conjugate to the main rotor so that the rotors
mesh with continuous line contact (if zero clearance) between the rotors. This
conjugate condition can be used to calculate the sealing line co-ordinates.
Note that when there is interlobe clearance between the rotors and other
geometrical features such as sealing strips are present the actual rotors are not
necessarily an exact conjugate pair and calculating the sealing line in this way
can potentially introduce errors.
The meshing angle, θm, for a given point, n, is defined as the angle through
which the main rotor must be rotated so that meshing line contact occurs at
that point. The follow steps describe how this meshing angle is calculated.
Pressure angle, ψ01
The pressure angle defined on the local rotor co-ordinate system is calculated
as the inverse tan of the main rotor surface normal:
( 19 )
As the gradient of the surface normal does not give the direction of the
transverse surface vector the angle, ψ01, must be corrected based on the
quadrant in which the point lies and the known orientation of the main rotor
co-ordinates.
Profile angle, φ01
The local profile angle is the polar angle, φ01, calculated from the Cartesian rotor
co-ordinates, x01 and y01.
199
Meshing angle, θm
The meshing angle, θm, is defined in Figure A-4 as the angle of rotation of the
local profile co-ordinate system, x01y01, measured relative to a static co-ordinate
system, X1Y1. The angles φ01 and ψ01 are defined relative to the local profile
system, x01y01. For conjugate meshing, the line of pressure (normal to the rotor
surface) at the meshing point, M, will always intersect the pitch point, P. Thus,
the meshing angle, θm, can be defined geometrically. This meshing angle is the
same whether the main rotor profile shown is meshing with a rack or any gate
rotor defined on this transverse plane.
Y1
y01
1 M
xx01
1
θ101
ψ
φ
ψ01
1
θm
X1
P
Figure A-4: Meshing condition
In Figure A-5, two right angle triangles are defined, the angles identified can be
related to the common opposite length, a, as given in equation ( 20 ).
200
Y1
y101
xx01
1
(ψ
(θ011 – φ
ψ101) )
r1
r1w X1
(ψ 1 ++θθmm))
(θ01
a
Figure A-5: Solving for meshing angle
( 20 )
After rearranging this is an explicit equation that is used to calculate the

meshing angle at each point.
( 21 )
Interlobe sealing line co-ordinates
The x and y sealing line co-ordinates are found by rotating each co-ordinate on
the main rotor by its respective meshing angle:
( 22 )
( 23 )
The axial co-ordinate, z, is directly related to the meshing angle using the unit
lead, p:
( 24 )
201
The sealing line co-ordinates xSL, ySL, zSL can be calculated for a
number of discrete points around the rotors in order to generate
the full interlobe sealing line as presented in Figure A-6; the
interlobe sealing has been superimposed onto the main rotor
body.
Figure A-6: Interlobe sealing line and normal projection
The interlobe sealing line follows a complex 3D path and it is commonly

presented using axial or ‘side-on’ 2D projections. Figure A-6 shows that when
projected onto a ‘normal’ plane (offset from the transverse plane by ψw) the
‘normal’ rack is generated. When viewing a meshed rotor pair from this
orientation the entire sealing line, and consequently the clearance gap between
the rotors, can be viewed by shining a light between the rotors.
202
A.3 CALCULATING GEOMETRY CHARACTERISTICS
A.3.1 CHAMBER AREA AND VOLUME

Chamber cross-sectional areas are initially formed between the two rotors and
then between each rotor and the casing before finally reducing between the
rotors again. The biggest challenge to develop an area calculation procedure is
therefore in identifying the correct area boundaries for each rotor position. This
problem is exacerbated by the fact that the rotors and casing never actually
come into contact. A numerical approach using a trapezoidal rule has been
adopted for robust area calculation with all profile types (Mujic 2009). The
cross section is divided into a number of divisions along the radius for each
respective rotor as shown in Figure A-7.
Figure A-7: Cross sectional area calculation using trapezoidal rule
The maximum theoretical chamber volume is achieved when the both the main
and gate cross sectional areas are fully formed. In this case the volume is the
sum of the cross sectional area multiplied by the rotor length. In practise, the
chamber cross section is not always fully formed along the full length of the
rotor and the maximum volume is therefore reduced. For this reason it is far
more reliable to determine the volume curve by integrating the area curve
between limits that are defined by the cycle angle. This is described by equation
(25) with the limits of integration shown in Figure A-8. This figure includes a
203
representation of the rotors in an attempt to show, in what is hopefully a more

intuitive way, how the ‘volume’ in the rotor chambers corresponds to the
‘volume’ under the area curve.
( 25 )
Discharge Face Suction Face
(a) θ = 100 degrees

z = 140.25 mm
-500 -250 0 250 500 750 1,000 1,250 1,500

z - axis
(b) θ = 488 degrees

z = 684.42 mm
-500 -250 0 250 500 750 1,000 1,250 1,500

z - axis
(c) θ = 800 degrees

z = 1122.0 mm
-500 -250 0 250 500 750 1,000 1,250 1,500

z - axis
Figure A-8: Area integration limits for volume
204
A.3.2 SUCTION PORT AREAS

In Figure A-9 the suction port is abbreviated to SP and the discharge to DP. Each
port comprises of the area on the main rotor and the area on the gate rotor;
identified by subscripts 1 and 2 respectively. In addition the ports can be
situated axially or radially or be made of both parts; these identified by
subscripts ‘a’ and ‘r’.
SP1r SP2r
SP1a SP2a
Suction Port
DP2a DP1a
r DP1r
DP2
Discharge Port
Figure A-9: Designated port names
The shape of the suction and discharge ports is defined by the rotor profile
shape and helix at the rotor tips. The size of the discharge port is defined by the
rotor position at which the required volume index, V1/V2, has been achieved.
The ports highlighted in Figure A-9 are the full port windows; the effective port
areas, for a given compression chamber, depends how the rotors interact with
these port windows. The areas through the main suction and discharge ports as
205
the rotors interact with the casing are calculated using a similar procedure to
that used with the chamber cross sectional areas.
A.3.3 BLOW HOLE AREA
Figure A-10: Blow-hole leakage area
The blowhole area is formed between the tip of the main, the tip of the gate and
the cusp of the casing where the rotors meet. For most practical rotor designs
the main rotor, gate rotor and casing cusp will not coincide at the same point
resulting in the leakage area highlighted in Figure A-10. The area is solely
defined by the transverse rotor profiles (highlighted in green) and the rotor
helix. This leakage area is handled with its own subroutine that calculates the
area between lines 1-2 and 1-3 in Figure A-10b.
206
A.3.4 SEALING LINES
Figure A-11: Interlobe sealing line
Figure A-11 shows the interlobe sealing line - the blue line shows where the
main rotor is in closest proximity to the gate rotor when the rotors are in mesh.
The part of the line highlighted in red represents the interlobe sealing line for a
single compression volume therefore it is the leakage area through this segment
that is of interest. Calculation of the sealing lines co-ordinates was explained
previously. The length of the sealing line through a discrete number of sealing
lines points can be numerically approximated:
( 26 )
In order to calculate the sealing line length for a given compression chamber, it
is necessary to check that the sealing line is within the limits of the rotor length.
At the rotor position shown in Figure A-11 the interlobe sealing line highlighted
in red is fully formed within the limits of the rotor so the leakage area is
maximum. As point A moves beyond the discharge face on the LHS of the figure
207
the sealing line length and consequently the leakage area will reduce until point
C reaches the same discharge face and the leakage area has reduced to zero.
The interlobe area curve shown in blue in Figure A-12 assumes a constant,
uniform interlobe clearance distribution. By introducing a relative rotation
between the rotors the clearance distribution was altered and the resulting area
curve was plotted in green. This serves to demonstrate how changing the
clearance distribution, whether by design, due to rotation, reduced centres or
some other manufacturing or assembly variation can change both the
magnitude and the shape of the interlobe leakage area.
4.50E-06
Interlobe leakage area
4.00E-06
3.50E-06
3.00E-06
Area (m^2)
2.50E-06
2.00E-06
1.50E-06
1.00E-06
5.00E-07
0.00E+00
0 200 400 600 800 1000
-5.00E-07
Cycle Angle (degrees)
uniform gap flat side contact
Figure A-12: Interlobe leakage area curves
It is evident in Figure A-12 that there is a region, when the sealing line is fully
formed, where the leakage area is constant. This would not be the case if
clearances vary along the length of the rotors due to non-parallel rotors or
operational distortions caused by high pressures or temperatures. In order to
investigate these scenarios the GEOM program was modified to allow the
clearance distribution to vary along 2 dimensions: around the transverse rotor
profile; and along the length of the rotors. In other words – the surface of the
rotor lobe under investigation is divided up using a 2 dimensional grid where
the local clearance at each node is unique. The same approach was applied for
208
the radial sealing line – the local radial clearance at any position on the casing
bore can be uniquely defined.
209
Appendix B: Compressor Modelling
Appendix B
Compressor Modelling
B.1 CHAMBER MODEL DETAILS
B.1.1 MODEL DISCHARGE TEMPERATURE

Boundary properties
Chamber properties
pc-1, Tc-1
ṁ3h3
ṁ1h1 ṁ2h2
pin, Tin p c, T c pout, Tout
ṁ4h4
pc+1, Tc+1
Figure B-1: Chamber model setup
A high level explanation of how the single chamber thermodynamic model

works is shown in Figure B-1. In practise, each of the paths along which mass,
and therefore enthalpy, can be transported comprise of various ports and
leakage paths but these have been grouped to show the overall chamber
interaction. The enthalpy can transfer to or from other ‘chambers’ which are
offsets of the single chamber or to or from inlet and outlet boundaries which
can be thought of as very large chambers with constant pressure and
temperature. The parameters pin, Tin, and pout are constant input boundary
properties; Tout is constant but unknown.
Prior to running the cycle convergence loop, Tout is initially calculated as the
adiabatic discharge temperature based on the pressure ratio. This initial Tout is
210
used to calculate the initial specific enthalpy at discharge which is in turn used
to calculate the isentropic flow through the discharge port. Knowing the flow
rate to the outlet and the specific enthalpies in the chamber or outlet
(depending on flow direction) at any given time, the total enthalpy change to
outlet can be integrated over the cycle duration. The average specific enthalpy
of the gas to outlet is calculated as the total enthalpy to outlet divided by the
total mass to outlet – this specific enthalpy is then used as the outlet boundary
enthalpy in the next iteration for flow calculation. For an ideal gas, the average
temperature, Tout, at the outlet boundary can be calculated as the average
specific enthalpy divided by the specific heat, Cp.
In the version of the chamber model used in this research, the outlet
temperature, T2, was taken to be the value of the temperature in the
compression chamber, TC, at the end of the compression process. In this single
chamber model, the instantaneous TC value at the end of compression is similar
to the averaged boundary temperature, Tout, providing there isn’t significant
over or under compression.
B.1.2 ADIABATIC FANNO FLOW

An available description of the leakage flow calculation is repeated here for
reference, further detail can be found in (Stosic et al, 2005) and (Stosic, 2015).
The leakage velocity through the clearances is considered to be adiabatic
Fanno-flow through an idealised clearance gap of rectangular shape and the
mass flow of leaking fluid is derived from the continuity equation. The effect of
fluid-wall friction is accounted for by the momentum equation with friction and
drag coefficients expressed in terms of the Reynolds and Mach numbers for
each type of clearance.
( 27 )
where c is the speed of sound, = FL/D + characterises the leakage flow

resistance, L the clearance length in the leaking flow direction, f the friction
factor and the local resistance coefficient.
211
B.2 THERMAL ANALYSIS SIMPLIFYING ASSUMPTIONS
B.2.1 ROTOR THERMAL DISTORTION

In isolation, equation (46) in ‘5.3.2 Local Rotor Distortion’ describes 1D thermal
distortion but when applied to the radius, for which the angular position is
known, and with the stated assumption of a uniform planar temperature, this
equation can be used for calculation of 2D thermal distortions.
B’
ΔrB
Δx
B ΔrB
x A’
ΔrA
A
αB-αA
rB
AB = x
αB rA
A’B’ = x + Δx
αA
O
Figure B-2: Uniform 2D thermal distortion
With uniform 2D thermal distortion all relative proportions are preserved.

Providing the direction of the radial distortions, ΔrA and ΔrB, of points, A and B,
are applied in the correct direction using the appropriate angles, αA and αB
respectively, then the relative distortion, Δx, between points A and B will be
accounted for. This has been demonstrated with the use of similar triangles in
Figure B-2.
B.2.2 DATUM FOR RELATIVE CLEARANCE DISTORTION

While the interlobe clearance gap was analysed by investigating the thermal
distortion of each rotor and centre distance using a 2D approach, as explained
in ‘5.3 Thermal Distortion of Interlobe Gap’, the analysis of the radial gap was, in
contrast, reduced to a 1D analysis, as explained in ‘5.4 Thermal Distortion of
Radial Gap’. The main assumption that allows this simplification is the use of
212
each respective rotor axes, on the main and gate rotor bores respectively, as a
common datum so that only radial distortions in the rotors and casing are
considered.
The validity of this assumption will be assessed in more detail by considering

transverse cross sections of the rotors and casing with uniform temperatures
obtained by averaging the surface temperature exposure across that plane.
Simplified FEA analysis was considered however as shown in the previous
section, if the 2D expansion of each component is uniform it can simply be
analysed by considering a representative dimension; in this case the centre
distance, A.
Figure B-3: 2D rotor and casing distortion
In Figure 1-1 two casing distortion scenarios have been presented. The outer
dashed line shows the distorted casing bores where the centres of the bores are
also marked as dashed lines. In the first scenario the solid circles that represent
the rotors are positioned in the undistorted centre distance, A. This kind of
scenario could occur in a compressor where the casing is relatively cooler in the
vicinity of the bearings than the cross section being analysed, thus resulting in
rotors that are not concentric to the casing bore. As a consequence of this the
radial clearance will be non uniform so that near the cusp Gap1 would be
reduced while Gap2 would be increased.
213
In the second scenario the rotor positions are represented by the circles with
dashed lines that have been adjusted to the same distorted centre distance as
the casing at that particular plane. This results in rotors that are concentric to
the casing bore and in this case with uniform rotor and casing temperatures
across the plane the radial gap is constant so that Gap1’ = Gap2’.
Obviously, if the distorted centre distance of the casing is different from the
centre distance of the rotors this should ideally be accounted for by using a
single common datum for both rotors and casing and calculating the resulting
non-uniformity of the radial gap due to the non-concentric rotors. In the current
procedure which simplifies the calculations by defining a datum on each axis,
the rotors are assumed to be concentric to the casing.
Figure B-4: DRUM127 casing temperature exposure
In the current model the distortion of the rotor axis is based on the distortion of
the casing centre distance, A, at the inlet and outlet end planes, as described in
‘5.3.3 Local Casing Distortion’. So by definition each rotor will be concentric to
the casing bore at the inlet and outlet plane. Between these end planes the rotor
centre distance will vary linearly since the rotor axes are taken to be straight.
However, the axial variation in the average temperature of each casing cross
section is not necessarily linear and there is likely to be some deviation between
214
the rotor and casing centre distances. In order to quantify this effect, results
from the DRUM127 compressor model are used; the casing exposure is shown
in Figure B-4 where the temperature ranges from 25°C to 160°C. The
temperature exposure was averaged at each transverse plane in order to
estimate the average 2D casing temperature at that plane. The variation in the
casing temperature, which describes distortion of the casing centre distortion,
is compared against the interpolated temperature, which describes the rotor
centre distortion, in Figure B-5. Note that the temperature shown in this figure
is actually the ΔT, which is the different between the absolute temperature and
the reference ambient temperature which has been taken as 20°C.
100
90
Average Planar ΔT (deg C)
80
70
60
50 casing ΔT
40
casing ΔT linear
30
interpolation
20
10
0
0 100 200 300
Axial Plane Position / γ (deg)
Figure B-5: DRUM127 axial casing temperature variation
The maximum variance between the casing ΔT and the interpolated ΔT is 8.4°C.
When applied to the centre distance for this compressor, where A = 0.093m, for
cast iron, this equates to an error of less than 10μm. This results in a
concentricity error of less than 5μm on each rotor. The nominal radial clearance
for this compressor is 100 μm therefore the maximum relative error due to
non-concentricity (along a limited region of the axial length) is +/- 5%. Recall
from the example presented in Figure B-3 that the non-concentric rotor results
in opposing changes between Gap1 and Gap2 which further mitigates this error
in terms of the net radial leakage area used for performance calculation. In
215
conclusion the assumption of using each rotor axis as a datum in justified for
the current model.
216
Appendix C: Definition of Performance Characteristics
Appendix C
Definition of Performance
Characteristics
C.1 COMPRESSOR BOUNDARIES
In order to evaluate the compressor performance the metrics used must be
clearly defined. The compressor can be considered to be a ‘black box’ that can
be dropped into any system and described by only its inputs and outputs. The
inputs will be defined at the inlet flange, the outputs at the outlet flange and the
work in for a compressor is defined at the drive shaft – this has been shown in
Figure C-1.
1 - INLET
FLANGE
p1
T1
V 1 2 - OUTLET
FLANGE
DRIVE
SHAFT
p2
T2
N
τ V 2
Figure C-1: Compressor boundaries for performance evaluation
For a full energy balance of the compressor there would be additional inputs
and outputs to describe heat rejection to surroundings or oil. However, for an
evaluation of the compressor performance it is usually sufficient to measure the
useful work done on the gas by considering these described inputs and outputs.
For a known fluid, all thermodynamic properties can be derived if both

pressure, p, and the temperature, T, are known. For example, knowing p1 and
T1 at the inlet, the density, ρ1, can be derived. Similarly the density at the outlet,
ρ2, can be calculated using the pressure at temperature at that flange.
217
C.2 MASS FLOW

The other crucial measurement for performance evaluation is the mass flow, ṁ.
This is usually measured remotely from the compressor, usually with an orifice
situated upstream of the compressor on the low pressure ducting. There is an
assumption here that no leakage occurs to the surroundings between the
position of flow measurement and the discharge of the compressor. This is a
reasonable assumption since the only external seal is on the drive shaft which
under normal operating conditions will leak only a small volume of oil that is
negligible relative to the gas flow. From mass continuity, the actual volume flow
at a specific location with a known density can be found. In this work, the
volume flow is always defined at the inlet using, , which is calculated:
( 28 )
The shaft power, PSHAFT, is simply calculated as the product of the measured
speed and torque. This shaft power is often referred to as the ‘total power’ as it
is the sum of the power supplied to the compression fluid and the power lost to
all other inefficiencies including mechanical losses incurred in the bearings,
shaft seals and timing gears, if applicable. The focus of this research was on the
thermodynamic model of the compression process therefore the shaft power
was not used for model evaluation because of the uncertainty in the proportion
of work acting on the gas.
C.3 VOLUMETRIC EFFICIENCY

Generally, absolute flow results can be used to compare the modelled
performance against test measurements, as was done in the DRUM127 case
study. In the HS204 case study the volumetric efficiency was compared instead
of absolute flow – this provides essentially the same comparison while giving a
little more insight into how well the compressor is performing. In this work the
volumetric efficiency is defined as the ratio of the actual inlet volume flow rate
to the theoretical volume flow rate for that compressor at the same operating
speed:
218
( 29 )
The theoretical volume flow rate is calculated using the volume displacement
constant; this is expressed as the displaced volume per revolution of the drive
shaft and is a geometric feature of the compressor.
C.4 ADIABATIC EFFICIENCY

Typically, the adiabatic efficiency is described as the ‘total efficiency’ using the
shaft power:
( 30 )
However, as discussed previously the shaft power was not investigated in this
research - instead the adiabatic efficiency can be described using the indicated
power from the thermodynamic compression process – this includes the effect
of any heat transfer during compression but neglects any mechanical losses:
( 31 )
This is the most suitable efficiency for evaluation of the thermodynamic

performance at it captures the influence of all heat transfer, leakages, and under
or over compression. Indicated power is not normally available from
compressor test results so a useful way to approximate the adiabatic efficiency
of the thermodynamic process is to use the actual measurements p1, p2, T1 and
T2, as defined in Figure C-1. The gas properties at the inlet and outlet of the
compressor are used to define a polytropic process path. This is an
approximation of the indicated process path that doesn’t account for effects
such as over or under compression and as such the efficiency calculated in this
way can be slightly higher than the value from the indicated power:
( 32 )
219
Where the specific enthalpy for the adiabatic process is defined:
( 33 )
The specific enthalpy for the polytropic process is defined:
( 34 )
And the polytropic exponent is calculated using:
( 35 )
In practise for an oil free compressor, the measured outlet temperature, T2, is
higher than the adiabatic outlet temperature, T2_ADI, resulting in a polytropic
exponent: n > k. This implies that there is heat transfer to the compression gas
which reduces efficiency. In reality, this heat is primarily supplied due to a net
enthalpy increase introduced through leakage flows into the control volume.
Both specific enthalpies changes in equations (33) and (34) are calculated
assuming a system with constant mass. This is ok for the adiabatic process
curve as that represents the ideal compression process however the polytropic
process curve is intended to approximate an open system with net leakage
losses therefore the specific enthalpy change calculated in equation (34) is not
representative of the actual process. To compensate for the difference in
specific enthalpy change caused by the theoretical and the actual delivered
mass flow the volumetric efficiency was included in equation (32).
An example case was modelled in order to evaluate the approximate polytropic

curve, as shown in Figure C-2. The adiabatic efficiency based on the indicated
PV curve was 71.0% while the efficiency based on the temperatures and
polytropic path was 72.7%. The indicated curve captures a slight over-
compression hence that efficiency is a little lower.
220
3.0
2.5
2.0
Pressure (Bar)
1.5
model indicated
polytropic
1.0
0.5
0.0
0.0000 0.0005 0.0010 0.0015 0.0020
Volume (m3)
Figure C-2: Comparison of model indicated PV curve and polytropic curve
221

Buckney, David

Uploaded by

Copyright:

Available Formats

Buckney, David

Uploaded by

Document Information

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

Buckney, David

Uploaded by

Copyright:

Available Formats

City Research Online

City, University of London Institutional Repository

Citation: Buckney, D. (2017). Clearance management in twin screw compressors.

This is the accepted version of the paper.

Permanent repository link: https://openaccess.city.ac.uk/id/eprint/18235/

Link to published version:

Copyright: City Research Online aims to make research outputs of City,

City Research Online: http://openaccess.city.ac.uk/ publications@city.ac.uk

DAVID MALCOLM BUCKNEY

Thesis submitted for the degree of Doctor of Philosophy

City, University of London

School of Mathematics, Computer Science and Engineering

Table of Contents .......................................................................................................................... iii

List of Figures ................................................................................................................................. ix

Declaration ................................................................................................................................. xviii

Abstract .......................................................................................................................................... xix

1.1 Motivation ....................................................................................................................... 1

1.2 The Twin Screw Compressor................................................................................... 2

1.2.1 Basic Principles .................................................................................................... 2

1.2.2 Ports and Internal Compression.................................................................... 3

1.2.3 Capacity Control................................................................................................... 4

1.2.4 Oil Injection and Compressor Arrangement ............................................. 5

1.3 Design and Optimisation ........................................................................................... 6

1.3.1 Industrial Applications ...................................................................................... 6

1.3.2 Optimisation .......................................................................................................... 8

1.3.3 General Design...................................................................................................... 9

1.3.4 Rotor Profile Design ......................................................................................... 10

1.3.5 Clearance Design ............................................................................................... 12

1.3.6 Design Software ................................................................................................. 13

1.4 Preliminary Studies ................................................................................................... 13

1.4.1 Performance Sensitivity to Clearance ....................................................... 13

1.5 Concluding Remarks ................................................................................................ 18

1.6 Thesis Overview ......................................................................................................... 19

Chapter 2 Background Theory and Literature Review ........................................... 21

2.1 Introduction ................................................................................................................. 21

2.2 Rotor Design ................................................................................................................ 21

2.2.1 Profile Generation............................................................................................. 21

2.2.2 Clearance Design ............................................................................................... 25

2.3 Geometric Characteristics ...................................................................................... 28

2.3.1 Identification of Geometric Characteristics ............................................ 28

2.3.2 Calculation of Geometric Characteristics ................................................ 33

2.4 Review of Twin Screw Compressor Models .................................................... 34

2.4.1 Chamber Models................................................................................................ 34

2.4.2 Alternative Models ........................................................................................... 39

2.4.3 Thermal Analysis .............................................................................................. 40

2.5 Concluding Remarks ................................................................................................ 43

Chapter 3 Research Objectives, Methods and Expected Contribution .............. 45

3.1 Statement of Research ............................................................................................. 45

3.2 Objectives ..................................................................................................................... 46

3.3 Methodology ................................................................................................................ 46

3.4 Expected Contribution............................................................................................. 48

3.4.1 Boundary Map .................................................................................................... 48

3.4.2 Model Integrating Operational Clearance Modification .................... 48

Chapter 4 Boundary Map for Rotor and Casing Surfaces ....................................... 49

4.1 Introduction ................................................................................................................. 49

4.2 Definition of Boundary Surfaces .......................................................................... 50

4.2.1 Coordinate System Conventions ................................................................. 50

4.2.3 Rotor Surfaces .................................................................................................... 55

4.3 Definition of Cycle Conventions ........................................................................... 58

4.4 Mapping of the Casing Surfaces ............................................................................ 62