See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/289788674

CFD grid generation and analysis of screw compressor with variable geometry
rotors

Article · December 2013

DOI: 10.1533/9781782421702.11.601

CITATIONS READS
4 175

4 authors:

Sham Rane Ahmed Kovacevic

University of Oxford City, University of London
42 PUBLICATIONS   109 CITATIONS    157 PUBLICATIONS   812 CITATIONS   

SEE PROFILE SEE PROFILE

Nikola Stosic Madhulika Kethidi

City, University of London City, University of London
172 PUBLICATIONS   1,112 CITATIONS    5 PUBLICATIONS   38 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

NextORC: Fundamental studies on organic Rankine cycle expanders View project

Grid Generation in positive displacement screw machines View project

All content following this page was uploaded by Ahmed Kovacevic on 27 March 2018.

The user has requested enhancement of the downloaded file.

 CFD Analysis of Screw Compressors with Variable
Geometry Rotors
Sham Rane, Ahmed Kovacevic, Nikola Stosic and Madhulika Kethidi

Centre for Positive Displacement Compressor Technology, SEMS

City University London, U.K. EC1V 0HB

Email: sham.rane.1@city.ac.uk

phenomena of interest for compressor design occur. In order

ABSTRACT to obtain boundary conformal representation of physical
Commonly screw compressor rotors are of the uniform pitch space, most grid generation techniques start from the
and profile along the rotor length. However rotors used in boundaries and proceed to the interior. There are three main
other twin screw machines, such as vacuum pumps more classes of mathematical techniques used for this process, a)
often use variable pitch to improve efficiency. The limited Algebraic methods, b) Differential Methods and c)
use of variable pitch rotors and/or section profile is mainly Variational methods. Authors such as [1], [9], [10], [11],
consequence of manufacturing constraints. With [12], [13] and [15] have described different grid generation
advancements in manufacturing and measurement techniques in detail.
technologies, the new generation of screw machines may
see rotors with variable pitch and profile which could bring
benefits to economy and environment. In order to analyze
such machines by use of Computational Fluid Dynamics
(CFD) it is necessary to produce suitable numerical mesh
capable of reliable calculation of 3D transient fluid flows
within domains with sliding and stretching interfaces.
An Algebraic grid generation algorithm applicable to Finite
Volume Method (FVM) in variable pitch screw machines is
described in this paper. It is based on the principles
developed for the uniform pitch rotors with constant cross-
section profiles and could be also used for rotors with
variable geometry.
CFD calculations are presented to evaluate performance of
an oil-free 4/5 lobed twin screw compressor with variable
pitch rotors and uniform ‘N’ profile. A comparison with the
performance of a compressor of same size and wrap angle,
but with uniform pitch rotors showed that variable pitch
rotors give better compression characteristics. This is
achieved by reduced throttling losses, reduced sealing line
length towards high pressure domains and a larger discharge
area for the same pressure ratio.

1. INTRODUCTION
Screw compressors are usually manufactured with helical
lobed rotors of uniform lead and a profile optimized for
efficient compression process. It was suggested in open
literature that rotors with variable lead may provide more Figure 1: Oil injected Twin Screw compressor and its
efficient process for high pressure applications. Figure 1 working chamber
shows an example of a CAD model of an oil injected twin Kovacevic et. al. in [5], [6], [7] and [8] have successfully
screw compressor in which the compression process, used algebraic grid generation method with boundary
leakage flows, heat transfer, oil injection and other

1
adaptation to generate numerical mesh for twin screw
machines with constant pitch rotors. This has been
implemented in the custom made program for the
calculation of a screw compressor numerical mesh called
Screw Compressor Rotor Grid Generator – SCORG©. The
motivation for the present work was to enhance the
functionality of SCORG© to allow grid generation for twin
screw rotors with variable pitch and variable section profiles
which in turn would allow CFD simulations of these new
types of machines.
Numerical treatment of the CFD models for screw
compressors with uniform lead and variable lead rotors
differs mainly at the grid generation stage. In the grid
generation technique for constant pitch rotors the axial
distance between the grid points is constant. The challenge
in generating a grid for variable lead rotors is in establishing
the relationship of the axial distance between consecutive
cross sections and the angular rotation of nodes which
changes continuously along the rotor axis. To accommodate
this change, the grid needs variable spacing in the axial
direction which will still provide a conformal mesh. Figure
1 shows the numerical grid in one of the cross-sections. The
topology of the mesh remains same over the entire length of
the rotor but the rotors change their relative position
governed by the variable lead. Still, the grid generator needs
to accommodate the large difference in length scales of the
main domain and the clearances.
Figure 2: Meshing of Uniform Pitch Twin Screw Rotors
Figure 3: Meshing of Variable Pitch Twin Screw Rotors
Rotors with variable lead were patented by Gardner [3] in
1969, but such machines are still rarely used due to the lack
of efficient and economical manufacturing techniques to
produce such rotors. Figure 2 shows twin screw rotors with
uniform pitch while Figure 3 shows twin screw rotors of the
same size but with variable pitch. It has been shown in
2
literature that for the same rotor lengths, diameter, wrap
angles and lobe profiles, variable pitch rotors can be
designed to provide higher pressure ratios and larger
discharge port opening areas with reduced throttling losses
when compared to constant pitch rotors [3].
As confirmed by many authors, screw compressor
efficiency depends upon the profile of the rotors, number of
lobes on the male and female rotors, the length and
diameter, the wrap angle and the rotor clearances [2], [4]
and [14]. Based on these and the original work of Gardner
[3] it is suggested that the effects of variable lead rotor
designs are as follows;
a) If all other variables are unchanged for a constant lead
rotor pair and a variable lead rotor pair, the length of the
sealing line will be less for the variable rotor pair as the
rotor lead reduces from suction to discharge. Since the
leakage loss is directly proportional to the length of the
sealing line and the sealing line is shorter for the
variable lead rotors in the high pressure regions, the
leakage loss will be reduced. This may lead to higher
efficiencies with variable lead rotors.
Figure 4: Volume-Angle diagram
Figure 5: Pressure-Angle Diagram
b) For variable lead rotors, the reduction of volume during
the compression process will be faster than for constant
lead rotors, as shown in Figure 4. Consequently the
pressure will rise more rapidly for the variable lead
rotors as shown in Figure 5.
c) The built-in volume index (Vi) is the ratio of the suction
to the discharge volumes. The suction volume is the
maximum volume at which the suction port is usually
closed and where the compression process begins. The
 discharge volume is the size of the compression chamber
at the moment of opening of the discharge port. As
shown in Figure 4, to achieve the same volume index for
variable pitch rotors the discharge port should be opened
earlier which allows it to be bigger than in the constant
lead case. Hence it is possible to have a greater
discharge area at a similar pressure ratio and this will
reduce the throttling losses.
d) If the same size of discharge port is retained for both the
variable and uniform lead rotors, the discharge pressure
of the variable rotors will be higher. This indicates that
variable lead rotors can achieve a higher Vi index.
The advantages and disadvantages identified on the basis of
previous research have not been tested so far on physical Figure 6: Axial spacing difference between uniform pitch
prototypes due to difficulties in producing such rotors with and variable pitch rotor grids
existing manufacturing methods. However, these can be In order to achieve this, the existing structure of the
investigated further in detail by use of CFD analysis if an procedure used in SCORG© has been reformulated to be
appropriate numerical mesh can be generated. adaptable with the variable pitch and variable profile screw
The advantages of the variable lead rotors could be rotors.
enhanced even further by introducing variable rotor profiles The rotor lead variation could be constant, linear or stepped.
which could enhance rate of change of volume in the The former is used in screw vacuum pump technology while
machine and allow for even higher compression ratios and the latter is applied in some car superchargers. The
larger discharge port. Analysis of such machines requires expressions can be specified in the form of equations 1 – 4.
further step in adapting the existing grid generation
technique which requires introduction of a new rotor profile
for each axial cross section. However, if such modifications
are implemented, other positive displacement machines such
as single screw compressor with variable rotor profile along
the axis could also be analyzed.

2. GRID GENERATION FOR VARIABLE PITCH

AND VARIABLE PROFILE SCREW ROTORS
For the Uniform Pitch rotors, there is a fixed relation for the
axial distance between the cross sections and the unit
rotation angle over the entire length of the rotor. This is
convenient because the profile is constant and the grid
generated for one interlobe space can be reused in
consecutive interlobes. However, for rotors of variable
pitch, this relation varies along the length of the rotor.
Therefore it is impossible to use the same method with a
constant axial distance between sections for grid generation
of such rotors. At the same time the rotors rotate at a
constant angular speed similar to the rotors of constant
pitch. In such a case the angular and axial intervals for grid
definition do not relate directly to the angular rotation.
Figure 6 shows the grid difference between the constant and
variable pitch rotors.
Two approaches are proposed here to the solution of grid
generation for the CFD analysis of such machines.
Approach 1 is easier to implement by modifying the
existing procedure and is suitable for variable pitch
machines with a uniform rotor profile. Approach 2 is more
complex in nature but is generally applicable for any cross
section, including those of conical rotors.
2.1 Approach 1
In this approach, the pitch function as in Equation (5) is
used to derive a relationship between the fixed angular
increments Δα, from one section to the other and the
required variable axial displacements ΔZ1, ΔZ2, ΔZ3…. ΔZn
for each cross section of the rotor. By this means, the grid
vertex data generated for one interlobe angle are reused and
positioned in the axial direction with variable ΔZ which
conforms to the pitch variation function.

3
 Figure 7: Example of variable pitch grid with uniform
profile: 5/6 ‘N’ rotors

A linear variation as in Equation 5 has been used to find out

the axial position of each of the sections over the rotor
length as shown in Figure 7.

Additional computational effort is required compared to the

uniform pitch rotors to calculate this axial position of cross
sections. The assembly of a grid from 2D to a 3D structure
remains unchanged. However, this approach cannot be used
if there is any variation in the section profiles of the rotor.

2.2 Approach 2
This approach addresses the more generic requirement for
the rotor pitch variation along with a variable cross section
profile over the length of the rotor. Hence in addition to the
axial variation of cross section in Approach 1, the algorithm
needs to capture the variation of the profile in cross sections.
The grid generation algorithm is represented in a block
diagram in Figure 8 and Figure 9. The foundation of the
approach is that each cross section of the screw compressor
rotor is a conjugate profile pair in which case they can be
calculated independently and the grid generation process
from the splitting of the rotor domain by a rack to the
allocation of 2D vertex coordinate data can be repeated.
2.2.1 Procedure
 The process starts with the division of the rotor length
into ‘n’ number of cross sections and proceeds with the Figure 8: Variable Pitch and Variable Profile Grid
generation of 2D vertex data (Nodal x and y coordinates) Generation in 2D
in the first cross section.

4
 This involves boundary discretization and adaptation.
Interior nodes at this section are calculated using
Transfinite Interpolation and the vertex data are recorded Discharge End
after grid orthogonalisation and smoothing operations.
This step is labeled as ‘Subroutine Run – 1’ in the block
diagram Figure 8.
 The process is repeated over the second cross section
and additionally this section will receive its axial
position (nodal z-coordinate) from Pitch variation
function. Suction End
 The same 2D grid generation process is repeated over all
‘n’ cross sections and then the data are used together to
assemble a set of grid files representing the rotor domain
at every time step.
Discharge End

Suction End

Figure 10: Example of variable geometry rotor grid with

uniform pitch: 3/5 ‘Rotor generated demonstrator’ profile

Figure 9: Variable Pitch and Variable Profile Grid
Generation in 3D
 From each cross section; a block of vertices which
represents the first rotor angular position is collected to
construct the 3D assembly with correct z coordinate
variation of the Mesh-0. The process is repeated for all
sets of data until the Mesh-n is generated. The process is
graphically shown in Figure 9.
2.2.2 Example of application of approach 2
An example of variable geometry rotor grid generated using
Approach 2 has been shown in Figure 10. The cross sections
changes continuously from suction end to the discharge end
of the rotors. As the center distance is fixed, the rotors are of
parallel axis and a tapered shape. The outer diameter on the
main rotor is tapered while the inner diameter is constant
while on the gate rotor, the inner diameter is tapered and the
outer is constant.
Figure 11 shows the 2D cross sections grids for variable
rotor profile at the suction end, middle of the rotors and the
discharge end.
3. CFD ANALYSIS
The numerical analysis was carried using ANSYS CFX
solver with the objective of validating the new grid
generation procedures and to study the flow behavior in a
variable lead screw compressor with uniform profile. The
following aspects are considered:
a. Compression characteristics.
b. Increase in discharge port area for a given pressure
ratio.
c. Reduction in sealing line length towards the high
pressure end of the rotors.
Three test cases are defined to analyze the above.
Case 1. Uniform Pitch rotors and a discharge port opening
area to give a built in volume index Vi of 1.8.
Case 2. Uniform Pitch rotors and a reduced discharge port
opening area to give built in volume index Vi > 1.8.
In this case, the compression chamber is exposed to
the discharge pressure relatively late in the cycle as
shown in Figure 15. This corresponds to an
additional male rotor rotation of about 6.75° and

5
 allows for further pressure build up in the
chambers.
Case 3. Variable Pitch rotors and a discharge port opening
area same as that of Case 1 with Uniform Pitch
rotors. Since the wrap angle on both rotors is the
same, exposure to the discharge port will take place
at the same male rotor rotation angle.
Figure 11: Grid sections of variable geometry rotor, 3/5
‘Rotor generated demonstrator’ profile
The configuration of the compressor is as follows; Male
rotor with 4 lobes and a 60mm outer diameter, Gate rotor
with 5 lobes, L/D ratio 1.55, Rotor center distance 42mm
with Rack generated ‘N’ Profiles. The wrap angle on the
male rotor was 306°. The compressor was run at 10000 rpm
which gives a tip speed of 31.5m/s. An oil free case was
studied. The radial, interlobe and end clearances were
30µm. Due to oil free operation and low tip speed, leakages
are enormous and the clearances were reduced to 10 µm to
6
allow positive flow. The lead on the male rotor for the
uniform pitch rotors was 108.8 mm. For the variable lead
cases, the wrap angle on the rotors remains the same while
the lead was continuously changed from 125.0 mm at the
suction end to 92.6mm at the discharge end. For all three
cases, the rotor grids were generated with the same density.
The numerical grids for ports were generated externally.
Figure 12 shows the compressor domain decomposed into
three main regions; the Suction Port, the Discharge Port and
the Rotor domains.
Figure 12: Different parts of the compressor numerical
model
The flow at the suction and discharge of the compressor are
highly unsteady and therefore boundary conditions are
difficult to specify. To provide good boundary conditions,
constant pressure receivers were modelled at both ends.
Also the ports were reasonably extended to provide good
convergence of the flow and reduce the numerical
discrepancies arising from pressure pulsations and re-
circulations in the flow. Formulation of such constant
pressure receivers is described in detail in [6]. The pressure
in the suction receiver was 1.0bar absolute while the
Discharge receiver had pressure of 3.0bar absolute. The
convergence criteria for all equations were targeted at
1.0x10-3 and coefficient loops for every time step was set at
10. During solution, r.m.s residuals for all time steps were
between 1.0x10-3 and 5.0x10-3 for the momentum equation.
These were below 1.0x10-3 for continuity and energy
equations. The calculations were run sufficiently long to
allow a cyclic repetition of flow and pressure characteristics
at the boundaries. The fluid was air following an ideal gas
law with the molar mass of 28.96 kg kmol-1, Specific Heat
Capacity 1004.4 J kg-1 K-1, Dynamic Viscosity 1.831x10-5
kg m-1 s-1 and Thermal Conductivity 2.61x10-2 W m-1 K-1.
3.1 Results and Discussion favourable to the reduction of throttling losses in the
compressor.
3.1.1 Compression characteristics
Figure 13 shows the compression characteristic of the three
different cases for one full cycle. Case 3 with the variable
rotor pitch has a steeper rise in pressure than Case 1 and
Case 2 which have constant rotor pitch. At the same time
Case 1, with the same discharge area as that in Case 3,
builds up the least pressure before exposure to the discharge
pressure at about 246° male rotor rotation. Near this region,
Case 2, with a reduced discharge port area, continues to
compress before getting exposed to the discharge pressure.
This lag is about 6.75 ° of male rotor rotation.

Figure 15: Discharge Port Area gain with variable lead

rotor for the same delivery pressure

Figure 16 shows the mass flow rate at the suction and

Figure 13: Absolute Pressure variation with Male Rotation
discharge of the compressor at rotor ends. The average flow
was found to be similar in all three cases. The negative flow
at the beginning of the cycle is due to under-compression in
which case the discharge pressure is higher than the pressure
reached in the compressor chambers just before the opening
of the port. The highest negative flow was obtained for Case
1 with uniform rotor pitch, which developed the least
internal pressure (Figure 14).

Figure 14: Absolute Pressure variation near discharge

opening period of rotation
3.1.2 Discharge Port Area
Figure 15 shows an increase in the discharge port area with
the variable lead rotor. In Case 2, with uniform pitch and a
reduced discharge port opening area, the pressure rise was
about 2.3bar before exposure to a discharge pressure of
3.0bar and the opening area was 741.57 mm2. This pressure
rise was close to that of the variable pitch rotor of about
2.4bar, for which the opening area was 788.98 mm2. This
increment in the discharge area is of the order of 6% and is
Figure 16: Mass Flow Rate at Suction and Discharge (Two
flow cycles)
3.1.3 Sealing Line Length
The interlobe sealing line is the line of closest proximity
between the two rotors. The leakage of gas takes place
through this gap and is proportional to the length of the
sealing line and normal clearance [2], [4] and [14]. Contours
of pressure distribution on the rotors can be established from
numerical calculations in which case the line which divides
the high and low pressures represents the sealing line.

7

Figure 17: Sealing Line on Uniform Pitch Rotor
Figure 18: Sealing Line on Variable Pitch Rotor
Figure 17 shows the sealing line obtained on the uniform
pitch rotors (Case 1 and Case 2) and Figure 18 shows the
sealing line obtained on the variable pitch rotor (Case 3).
The projection of the sealing line on the rotor normal plane
shows the difference more clearly in Figure 19. The sealing
line on the uniform pitch rotor is of the same length for each
interlobe space along the rotor. However, on the variable
pitch rotors the sealing line is longer at the suction end. It
decreases towards the discharge end of the rotor.
Figure 19: Comparison of Interlobe Sealing Line Length
Table 1 presents the variation in the sealing line lengths
between the uniform and variable pitch cases at one of the
rotor positions indicating the magnitude of change along the
rotors. At the suction end the sealing line on variable pitch
8
rotor is 3mm longer but at the discharge end it is 3.5mm
shorter. This helps to reduce leakage as the largest pressure
difference across the sealing line is at the discharge.
Additionally, the total length of the sealing line is reduced
by 2mm in the variable pitch rotors.
Table 1: Comparison of Interlobe Sealing Line Length
4. CONCLUSION
The grid generation for twin screw compressors with
variable pitch rotors was formulated and implemented
successfully in this paper. The framework defined in this
research is suitable for rotors with variable profiles.
Examples of grids with variable lead have been presented
and CFD analysis has shown the flow characteristics in the
machines.
The analysis showed that by varying the rotor lead
continuously from the suction to the discharge, it is possible
to improve compression characteristics with a steeper
internal pressure build up. The analysis also shows that
varying the rotor lead allows a larger size of discharge port
area, thereby reducing throttling losses, and provides
increase in volumetric efficiency by reducing the sealing
line length in the high pressure zone. The latest
enhancements in the grid generation open new opportunities
for further investigation of the flow behavior and
performance predictions for variable lead and variable
profile screw machines by the use of CFD.
NOMENCLATURE
L – Rotor Length
D – Male Rotor Outer Diameter
Φw – Male Rotor Wrap Angle
α – Male rotor rotation angle
Δα – Increment in Male rotor angle
Z – Axial distance along the rotors
ΔZ – Increment in Axial distance
ps – Starting Pitch
pe – Ending Pitch
z1 – Number of lobes on the Male rotor
z2 – Number of lobes on the Female rotor
r.p.m – Male rotor speed
t – Time
Vi – Built in Volume Index
 REFERENCES Grid Generation, Int. J. Numer. Meth. Fluids, 15, 297-
311, 1992
[1] P.R. Eiseman, J. Hauser, J.F. Thompson and N.P.
Weatherill. Numerical Grid Generation in [14] N. Stosic, I.K. Smith and A. Kovacevic. Screw
Computational Field Simulation and Related Fields, Compressors: Mathematical Modeling and
Proceedings of the 4th International Conference, Performance Calculation, Springer Verlag, Berlin,
Pineridge Press, Swansea, Wales, UK, 1994 ISBN: 3-540-24275-9, 2005
[2] J.S. Fleming and Y. Tang. The Analysis of Leakage in [15] J.F. Thompson, B. Soni, N.P. Weatherill. Handbook of
a Twin Screw Compressor and its Application to Grid generation, CRC Press, 1999
Performance Improvement, Proceedings of IMechE,
Part E, Journal of Process Mechanical Engineering,
209, 125, 1994
[3] J.W. Gardner. US Patent No 3,424,373 – Variable
Lead Compressor. Patented 1969
[4] K. Hanjalic and N. Stosic. Development and
Optimization of Screw machines with a simulation
Model – Part II: Thermodynamic Performance
Simulation and Design Optimization. ASME
Transactions. Journal of Fluids Engineering. 119,
664, 1997
[5] A. Kovacevic, N. Stosic, I.K. Smith. Grid Aspects of
Screw Compressor Flow Calculations, Proceedings of
the ASME Advanced Energy Systems Division, 40, 83,
2000
[6] A. Kovacevic. Three-Dimensional Numerical Analysis
for Flow Prediction in Positive Displacement Screw
Machines, Ph.D. Thesis, School of Engineering and
Mathematical Sciences, City University London, UK,
2002
[7] A. Kovacevic. Boundary Adaptation in Grid
Generation for CFD Analysis of Screw Compressors,
Int. J. Numer. Methods Eng., 64(3): 401-426, 2005
[8] A. Kovacevic, N. Stosic and I.K. Smith. Screw
compressors - Three dimensional computational fluid
dynamics and solid fluid interaction, ISBN 3-540-
36302-5, Springer-Verlag Berlin Heidelberg New
York, 2007
[9] V.D. Liseikin. Grid Generation Methods, ISBN 3-540-
65686-3, Springer-Verlag, 1999.
[10] A.J. Samareh, R.E. Smith. A Practical Approach to
Algebraic Grid Adaptation, Computers & Mathematics
with Applications, 24(5/6): 69-81, 1992
[11] T.I.P. Shih, R.T. Bailey, H.L. Ngoyen and R.J Roelke.
Algebraic Grid Generation For Complex Geometries,
Int. J. Numer. Meth. Fluids, 13, 1-31, 1991
[12] B.K. Soni, Grid Generation for Internal Flow
Configurations, Computers & Mathematics with
Applications, 24(5/6): 191-201, 1992
[13] E. Steinthorsson, T.I.P. Shih and R.J. Roelke.
Enhancing Control of Grid Distribution In Algebraic

View publication stats