Beruflich Dokumente
Kultur Dokumente
van Bommel
August 2016
Report number: P&E - 2713
Thermodynamic Model of a
Screw Compressor Master Thesis
Master of Science
in Mechanical Engineering – Sustainable Process and Energy Technology
3
4
Abstract
A Compressor Resorption Heat Pump (CRHP) is a potential contribution to energy reduction in
applications in which waste streams are upgraded, with limited energy addition, into high value
process streams for reuse in industry. Previous research concluded that a CRHP with a wet
screw compressor is a suitable option for many applications. An ammonia/water mixture was
found to be the most appropriate fit, in terms of thermodynamic behaviour, for such an
application.
A number of variations have been applied to the integrated model as examples of how to
evaluate options for improvements. Making use of the developed integrated model parameters
can be varied to show the influence on the compressor. The evaluations used a specific set of
boundary conditions from previous research, using the geometry specified by Zaytsev [1]. The
effects of three input parameters on the output and efficiency were evaluated: rotor length,
discharge port area and vapour quality. The main result of the evaluation is that per boundary
condition, the inputs from the geometry model have to be adjusted to achieve an optimal design
of the twin-screw compressor. Further research to find the optimal design can be done with the
help of the model that was developed for this thesis.
5
6
Content
Abstract 5
Nomenclature 9
Abbreviations 10
Glossary 10
1 Introduction 11
1.1 Heat Pumps 13
1.2 Compressor Process, Arrangement and Selection 14
1.3 Screw Compressor 16
1.4 Motivation of the Research 18
1.5 Research Question, Objective, Boundaries and Assumptions 18
1.6 Chapters 19
2 Models 20
2.1 Boundaries and Assumptions 20
2.2 Geometry Model Theory 21
2.3 Theory of the Thermodynamic Model 27
2.4 Theoretical Desired Performance Outputs 30
3 Thermodynamic Model Implementation 31
3.1 Historic Development 31
3.2 Input Implementation 32
3.2.1 Geometry Model 32
3.2.2 Thermo-Physical Properties 35
3.2.3 Input Values/Initial Values for the Implementation 36
3.3 Thermodynamic Model; Factors that Influence Ideal Behaviour 37
3.3.1 Mass Flows 37
3.3.2 Leakage Path Areas 39
3.3.3 Mass Flows of the Leakages 45
3.4 Desired Outputs 49
4 Validation 51
4.1 Inputs and Boundaries 51
4.2 Model Validation: Zaytsev 52
4.3 Adapting the model to the experimental data 54
5 Model Results and Discussion 57
5.1 Leakages 57
5.2 Boundary conditions ‘van de Bor’ 57
5.3 Geometry variation 58
6 Conclusions and Recommendations 65
6.1 Conclusions 65
6.2 Recommendations 66
Bibliography 67
Appendices 69
Appendix A: Envelope Method – Rotor Element Calculation 69
Appendix B: Conservation Equations of the Homogeneous Model 71
Appendix C: The Thermodynamic Model in Simulink 74
Appendix D: The Geometry Model in Matlab, van de Bor/Zaytsev [Matlab-Code] 79
Appendix E: Calculation Pressure Difference [Matlab-Code] 98
Appendix F: Calculation Efficiencies [Matlab-Code] 100
Appendix G: Calculation Shaft Rotation Angle 101
7
8
Nomenclature
Symbols Units
A Flow area m2
b Number of rotor lobes -
C Angle °
D Distance between rotor axes m
h Specific enthalpy J∙kg-1
H Enthalpy J
i Rotation speed ratio -
k Ratio + 1 -
L Length of the rotor mm
LL Length contact line, sealing line or height of the rotors mm
m Mass kg
𝑚 Mass flow rate kg∙s-1
n Speed of rotation rpm
p Pressure Pa/bar
Q Heat J
R, r Radius m/mm
t Time s
T Temperature K
u Flow velocity m∙s-1
v Specific volume m3∙kg-1
V Volume m3
W Work J
𝑊 Power J∙s-1
x Ammonia mole concentration mol∙mol-1
x, y, z Coordinates m
9
Subscript
0 Static coordinate system attached to rotor housing
1 Male rotor
2 Female rotor
c, comp Compressor (phase)
discharge Discharge phase
h Envelope radius
high Highest value
ideal Ideal process
in Flow in
is, s Isentropic process
low Lowest value, angle
out Flow out
real Real process
suction Suction phase
up Upper angle
vol Volumetric
w Wrap angle
Abbreviations
CFD Computational Fluid Dynamic
COP Coefficient of Performance
COP21 21st Conference of Parties
CRHP Compression Resorption Heat Pump
EDGAR Emission Database for Global Atmospheric Research
IEA International Energy Agency
IPCC Intergovernmental Panel on Climate Change
LULUCF Land Use, Land-Use Change and Forestry
NASA National Aeronautics and Space Administration
PBL Planning Board for Environment ‘Planbureau voor Leefomgeving’
UNFCCC United Nations Framework Convention on Climate Change
Glossary
Leading Cavity The leading cavity is the cavity volume with a time and comes
according to the time ‘in front’ of the main cavity.
Trailing Cavity The trailing cavity is the cavity that has a time shift and will
come according to the time ‘behind’ the main cavity.
10
1 Introduction
Sustainability is one of the words that is most commonly used when talking about global
warming. In 1987 the World Commission on Environment and Development [2] wrote a report
about the issue: ‘Our common future’. In this report the commission proposed a definition for
sustainability, which is now widely used and is cited below.
Last December the 21st climate conference was held in Paris, organised by the United Nations
Framework Convention on Climate Change (UNFCCC) [3]. During this yearly event, members
of the United Nations gather to discuss measures to fight global warming. During the 11th
edition in 1997, this resulted in the well-known Kyoto Protocol. During the latest edition in
Paris, the 21st Conference of Parties (COP21) [4], the 195 attending parties adopted the first-
ever universal, legally binding global climate deal and agreed on a global warming limit of 2
degrees Celsius.
Global warming is one of the largest threats to the world we live in. The highest increase in
global temperature has occurred during the last 35 years [5]. The data in Figure 1-1 show that
the earth’s temperature has increased by at least one degree since the average baseline of 1951-
1980. Recent data show that 2015 was the warmest year on record. The temperature rise limit
of 2 degrees agreed on COP21 will thus be a challenge, and drastic changes are needed to
accomplish the agreement. Research in global warming is done by many different
organizations. A few of these are: National Aeronautics and Space Administration (NASA),
Intergovernmental Panel on Climate Change (IPCC) and International Energy Agency (IEA).
Worldwide energy demand is high, and will only increase with growing population [6].
Something needs to be done to reach the agreed goal of the UNFCCC.
1,1
1
0,9
Temperature Anomaly (1951-1980 baseline)
0,8
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0
-0,1 1880 1895 1910 1925 1940 1955 1970 1985 2000 2015
-0,2
-0,3
-0,4
-0,5
-0,6
-0,7
-0,8
Year
Annual Mean
1 °C above 19th
Century
Figure 1-1 Temperature increase over the years from 1880 to 2015 [5].
11
In the Netherlands, the attention to global warming and its consequences is increasing and
becoming a major concern to both the government and the public as well [6, 7, 8]. The causes
and effects of climate change are intensively investigated by the IPCC [10], of which the Dutch
government is a member. The rising sea level is one of the most significant consequences of
global warming, and even more so for a country that is partially below sea level. The
Netherlands are already fighting the sea with their famous Delta works, but the question is to
what extent they can strengthen them to withstand the increasing rise of the sea level. Other
consequences include more extreme weather conditions and the pollution of sweet water rivers
with salt water, which complicates the utilization of river water for drinking, agricultural and
industrial cooling purposes.
The global warming effect is mostly linked to the emission of CO2. In Figure 1-2 the shares of
different greenhouse gases in the total emission of 2010 can be seen, both for the world and for
Europe. This figure clearly shows that the most significant greenhouse gas is CO2, which
comes from burning fossil fuel and from industrial processes. Reducing CO2 emission will thus
slow down global warming and might even decrease the global average temperature. The Dutch
organization Planning Board for Environment ‘Planbureau voor Leefomgeving’ (PBL) [9]
released a report in 2015 concerning the CO2 emission all around the world: ‘Trends in global
CO2 Emissions’ [11]. This report is based on the data from PBL and Emission Database for
Global Atmospheric Research (EDGAR) [12]. Burning of fossil fuels, generating electric
energy and generating heat emits most CO2. The increasing demand for energy by the growing
population will prove to be a challenge for the world in the struggle to reduce the energy usage
and thus decrease CO2 emission.
World Eu28
6% 2% CO2 fossil fuel and industrial
5% 3% processes
10%
64% N2O
76%
Fluorinated gases (F-gases): HFC, PFC
and SF6
Countries all over the world are striving to emit less CO2 by exploiting renewable energy
sources and reducing energy losses. The IEA concluded in 2011 [13] that 47% of energy
consumption in the world is thermal energy. The industry is the largest thermal energy
consumer worldwide, weighing in at 44% of total thermal energy consumption. In many
industrial processes a large proportion of this thermal energy is dissipated to the environment as
waste heat. In the Netherlands alone, the industry is estimated to squander over 250 PJ through
wasted thermal energy, of which an estimated 150 PJ/year can be reused in other processes
[13,14]. This thermal energy loss to the environment has a temperature range of 40-150 °C.
Therefore, the reuse of waste heat is considered an interesting possibility to reduce CO2
emissions.
One of the most well-known technologies for (waste) heat recovery is the so called heat pump,
which can extract thermal energy from a low caloric stream, to release it in a high caloric
12
stream. Therefore the heat pump is one of the most interesting devices to study when it comes
to waste heat reduction.
A heat pump can be used in many different situations; from heating up a building with
geothermal energy to the reuse of waste heat in an industrial process. For more details about
heat pumps, see Moran and Shapiro [16].
Q! out Q! out
Resorber (Condenser)
Resorber (Condenser)
Solu8on
Pump
W! c W! c
Valve Compressor
Valve Wet-Compressor
Q! in Q! in
(a) (b)
Figure 1-3 Schematic representation of two variations, for compression, of a heat pump cycle (a) compression
resorption heat pump with solution pump and (b) compression resorption heat pump with wet-compressor.
The refrigerant in a heat pump is used to facilitate the heat transfer. There are many different
refrigerants to work with. One of these options is the refrigerant-absorbent combination, which
consists of a mix of two elements: the refrigerant and an absorbent. Ammonia/water and water–
lithium bromide are two well-known refrigerant-absorption combinations. In the combination
of ammonia/water, water is the absorbent and ammonia the refrigerant. More information on
refrigerants can be found in Dinçer and Kanoglu [17].
13
Two types of compression resorption heat pump (CRHP) can be defined: based on the
Osnabrück cycle Figure 1-3(a) and wet compression Figure 1-3(b). In the Osnabrück cycle gas
and liquid phases are separated before entering the compressor and the liquid is pumped
parallel to the compressor; after the compression the compressed gas and liquid are mixed. The
type where liquid and gas from the desorber are not separated and are compressed as ‘wet’
phase and fed into the resorber is called wet compression.
According to van de Bor et al. [20], it is possible to use a wet compressor in a resorption cycle,
instead of the compressor/solution pump combination of the Osnabrück cycle, Figure 1-3. Van
de Bor et al. [20] have done numerical research of the performance of the heat pump, with
ammonia/water mixture as refrigerant, in 50 specific industrial cases. In van de Bor et al. [21]
three different heat pumps are compared for a certain situation: the recovery of heat from an
industrial cooling tower stream. The temperature of this waste stream is generally around 45-60
°C. According to the research, the compression resorption heat pump (CRHP) would be most
suited for the application of heat recovery from a cooling tower stream, Figure 1-4.
Resorber (Condenser)
Waste water W! c
Stream: 45-60 °C
Valve Wet-Compressor
Cold UAlity: 5 °C
Desorber (Evaporator)
Figure 1-4 CRHP heat recovery from waste stream cooling tower [21].
Due to its design, the heat pump can be used both for cooling systems and for heating systems.
In order to achieve the highest efficiency of the CRHP, the use of wet compression is most
advisable. In the research the water flow was split into two stream, one stream to be heated
above 110 °C and the other to be cooled down to 5 °C, as shown in Figure 1-4. Van de Bor et
al. [21] assume that the wet compressor has an isentropic efficiency of 0.7. However in order to
reach an efficiency of 0.7 or higher, further development of the wet compression is needed.
The two-phase fluid in the heat pump poses a problem at the compression side; the following
sections will therefore discuss the working of the compressor of the heat pump.
14
compressor is combined with a pump placed parallel to the compressor, to transfer liquid
stream of the ammonia/water mixture. According to the research of Mongey et al. [18] wet
compression was believed not to be feasible, due to the large liquid fraction that remains after
the desorber, which would enter the suction of the compressor. The other option is using a wet
compressor arrangement, Figure 1-4b. Van de Bor et al. [21] make use of the wet compressor,
which provides an advantage compared to dry compression with separate liquid circulation as
discussed above.
Information on wet compression is limited, and the aforementioned paper strongly recommends
further research into optimization of compressors for this purpose. Itard [22] did research on
the wet compression-resorption cycle in 1998, focusing mainly on the difference between dry
and wet compression. According to her research, wet compression has an advantage, for the
cases that were considered, that varies between 2.5% and 13% on the Coefficient of
Performance (COP) compared to dry compression. For the compression section during the
experiments a liquid ring compressor was used.
In 2003 Zaytsev [1] did more research on the type of wet compressor suitable for the CRHP. A
review has been done on the different compressor types available on the market and their
possibilities for wet compression. In the end he concluded that a twin-screw compressor would
be the best option for wet compression in CRHP. The choice for a twin-screw compressor was
based on the capability of this compressor to work in the two-phase regime of an
ammonia/water mixture, with a sufficiently high thermodynamic efficiency. According to data
the isentropic efficiency of a twin-screw compressor can reach up to 0.75, which would fit the
needed threshold as per van de Bor et al. [21]. The experiments of Zaytsev [1] showed an
isentropic efficiency of the compressor still relatively low compared to the required minimum
of 0.7. The twin-screw compressor can also operate over a wide range of pressure and
temperature, which makes it suitable for CRHP application.
The twin-screw compressor additionally has the capability to work under oil-free conditions. It
also has a medium risk to hydraulic locking compared to the other compressor type options.
The disadvantages of the twin-screw compressor are the built-in volume and a constraint that at
the moment there are, to current knowledge, no twin-screw compressors available on the
market that meet the specifications for use in the ammonia/water based CRHP.
Zamfirescu et al. [23] did in 2004 further research on Zaytsev’s results on the twin-screw
compressor for use as wet compressor. Zamfirescu et al. [23] developed a Computational Fluid
Dynamic (CFD) simulation with a non-homogeneous model and carried out experiments on the
twin-screw compressor. From the CFD model it was concluded that the available liquid-phase
fluid in the compressor is spread as a layer against the compressor housing because of the
centrifugal force. This liquid layer fills up the clearances between the screws and decreases the
leakage between the different cavities. This has the positive effect of increasing the isentropic
efficiency. Also the leakages at the bearings were researched. Several labyrinth seals were
studied and it was concluded that the use of an optimal labyrinth seal would improve the
thermodynamic performance of the compressor. Zamfirescu et al. [23] also did a prediction on
the heat and mass transfer during compression to improve the accuracy of the model. More
research has been done on the topic of liquid injection, which was proposed by Zaytsev [1] and
could increase the compressor efficiency even more. Zamfirescu et al. [23] also did research on
the effect of the rotational speed on the efficiency. According to their experiments the
mechanical efficiency would be optimal at a rotational speed between 2500 and 3000 rpm. The
mechanical losses will increase drastically at higher rotational speed.
15
Lets have a deeper look into the twin-screw compressor. With the research carried out to date
the type of screw, mechanical arrangement and process fluid selection provides quite a number
of unknown parameters. According to the information above more research is needed to
accomplish a working wet compressor suitable for the wanted specifications.
Figure 1-6 shows the three phases of the twin-screw compression process (within the model
these three phases will be divided into two, suction and compression/discharge). During the
suction and discharge phase the temperature and pressure are assumed to be constant in this
theoretical figure. The only change in these variables occurs during the compression phase. See
Wennemar [25] or Arbon [26] for more info on the working of the twin-screw compressor.
Figure 1-6 Idealized pressure-volume diagram for a screw compressor with well suited built-in volume ratio [25].
16
Literature research shows that most wet screw compressors are lubricated/injected with oil. Oil
is the most appropriate lubricant used. As a result there is a lot of literature about oil injected
screw compressors [27]. The disadvantage of using oil for lubrication of the system is that the
oil and gas need to be separated after the compression. Failing to do so results in oil staying
behind in the mixture when it is transported to the heat exchangers, the resorber and the
desorber. The oil forms an oil layer on the heat exchanger surface, resulting in a decreased
performance of the heat exchangers. Besides decreased heat exchanger performance, the
presence of oil in the mixture also requires extra equipment to separate the oil phase from the
gas phase. There are a few other reasons for using oil in the compressor next to lubrication.
The oil reduces the blowholes and clearances of the leakage paths between the rotors, as well
as, between the rotors and the housing. Application of oil as lubricant/sealant additionally
absorbs heat from the compression and keeps the temperature of the compressor low, as well as
reducing wear on the lobes of the rotors. Concluding, use of a suitable lubricant/sealant is an
important element in the design and use of a screw compressor.
The heat pump under study uses compression with an ammonia/water mixture. Such a mixture
excludes the use of oil as lubricant/sealant as it would lead to a three-phase liquid/vapour fluid,
which requires further treatment downstream to separate the oil from the liquid phase [23].
The aim is to use the liquid phase of the ammonia/water mixture as a lubricant [20, 27, 28].
Using the liquid phase as lubricant requires less equipment and prevents the oil problem
explained above. In order for the ammonia/water mixture to be effectively used as lubricant in
the compressor, two important premises need to be met. The liquid/vapour in the suction port
of the compressor needs to be ideally mixed in order for the liquid to be available on all the
surface area and in all the clearances of the compressor. Flow regimes like slug flow need to be
prevented by regulating the liquid intake at the suction port [23].
Screw compressors can operate either under dry or wet process conditions. Zaytsev [1] and van
de Bor et al. [21] have a slightly different approach to the desired working range of the
compressor. Zaytsev [1] chose for a process setup where the liquid part of the mixture is used
to lubricate the compressor. As a result, the inlet and the outlet contain large portions of liquid,
resulting in low efficiency. According to van de Bor et al. [21], the application of wet screw
compression is the best choice in the desired range, for an average temperature lift of 85 °C.
During the compression phase the temperature and pressure increase and the vapour/liquid
equilibrium changes. This may result in drying up of the last section of the compression and
discharge phase. Van de Bor et al. [21] researched the option to let the compressor run dry at
the end. According to the study this will result in a higher efficiency and an optimal
performance of the heat pump as the ideal vapour quality at the resorber inlet would be
approximately 100%.
There are a lot of different approaches to the problem of lubrication, but not an exact solution
for the problem. For this research report a homogeneous model is assumed, and leaves the
lubrication and the change in the liquid/vapour concentration open for further research. For this
research report the focus is on the physical simulation of the twin-screw compressor for
application in a CRHP.
To make a simulation model of the twin-screw compressor two different models need to be
combined. One model will describe the thermodynamics of the twin-screw compressor, while
the other model describes the geometry of the compressor.
The thermodynamic model will calculate the pressure and temperature increase during a whole
compression cycle. Conservation equations are used to calculate the compression phenomenon
17
in the compressor. Next to the thermodynamic model a geometry model will be used, which
describes the geometry profile of the two rotors, where the female rotor is described as a
function of the male rotor.
The geometry model will be used as a source for the inputs for the thermodynamic model, such
as the cavity volume. The combination of these two models allows the design of a suitable
geometry of the twin-screw compressor for every application. There are some papers about
these two types of models, especially on models that describe the geometry. Zaytsev made a
combination of a thermodynamic and geometry model [1], using the method of Sakun [30] for
the geometry. This method of modelling the geometry was also used and described by You [31]
in 1994. Zaytsev and Infante Ferreira [28] developed after the use of the method of Sakun a
different method for calculating the geometry of the rotors.
This new method was based on the meshing line, which is used as a starting value to model the
rest of the rotor profiles. Stosic et al. [27] worked on the modelling of a twin-screw compressor
as well, in 2005. The method Stosic et al. use is based on a rack line, which is used as a starting
position to generate the profiles. This rack line is generated by specific criteria. Modelling the
geometry of the rotors will be discussed in more detail in chapter 2.
To be able to define and design this wet compressor more knowledge about the parameters of
the compression process is needed. To evaluate these parameters a model of the twin-screw
compressor needs to be developed. This model will need to consist of two separate models, a
geometry model and a thermodynamic model, which will be based on Zaytsev’s [1] research.
Unfortunately the model Zaytsev developed in 2003 is out dated, based on C++ and not able to
run anymore.
This thesis will describe the new model and will be based on Zaytsev’s simulation of the
geometry and thermodynamics.
Objective: To develop a dynamic model that describes the thermodynamic process and includes
the input from a twin-screw compressor geometry model. The CRHP has the aim to transform
large industry waste streams like cooling water discharge into valuable re-usable energy
carriers.
18
The model and variables include the following assumptions and constraints.
• The model will be based and build further on the work done by Zaytsev [22].
• The thermodynamic model will be based on a homogeneous model.
• The model will take as boundary conditions the heat pump process as described by van
de Bor et al. [21].
• The process is based on an ammonia/water mixture of given composition.
• The applied compressor is a wet twin-screw compressor where the geometry is
described in ref. [22, 29, 30].
• The compressor should have an isentropic efficiency of at least 0.7.
• The temperature to be achieved in the discharge of the compressor should be at least
115 °C, van de Bor et al. [21].
1.6 Chapters
The thesis report describes the steps carried out to develop a working model that integrates
screw compressor geometry with thermodynamic behaviour of the selected ammonia/water
composition and physical properties.
Chapter 2 explains the models that are used to simulate the geometry and thermodynamics of
the twin-screw compressor. This chapter starts with the boundaries and assumptions followed
by explaining the model. The geometry model will be explained in detail and will be followed
by an explanation of the theory behind the thermodynamic model.
Chapter 3 explains the development of the thermodynamic model in Simulink. It includes the
input from the geometry model, design inputs and physical properties. It describes the
implementation of the theories in the model structure developed in Simulink. The
implementation also includes empirical correction factors for reality with internal leakage paths
as the most influential ones.
Chapter 5 describes the model with a set of boundaries as specified by van de Bor et al. [21]
and with the geometry from Zaytsev [1]. The rotor length, discharge port area and number of
lobes were varied. Results show improvement of isentropic efficiency and to achieve the
wanted discharge pressure and temperature.
Chapter 6 includes the conclusions of the development of the thermodynamic integrated model.
Recommendations to optimise the compressor model are given.
19
2 Models
For the dynamic modelling of the twin-screw compressor a thermodynamic model has been
developed during this thesis in which existing work and application of compressor geometry
have been included. In this chapter both models will be described and explained.
In this chapter the boundaries and assumptions used for the development of the models for the
wet twin-screw compressor are clarified first. In the next sections the thermodynamic theory
and the theory of the geometry will be explained in detail. In the last section the required
outputs are described.
The physical properties of the ammonia/water fluid will be invoked from Refprop [32] via
Fluidprop [33]. A combination of Matlab [34] and Simulink, a tool within Matlab, will be used
to model the wet twin-screw compressor.
Resorber (Condenser)
Valve Compressor
Boundary
Desorber (Evaporator)
As can be seen in Figure 2-1, the compressor inlet comes from the desorber. The composition
of the fluid entering the compressor is a mixture of ammonia and water leaving the desorber.
The stream is assumed to be a two-phase fluid consisting mainly of vapour with a dispersed
liquid phase. Wet compression will be applied to attain the higher temperatures that are
required for the resorber. From a thermodynamic point of view, the most efficient application
of the wet compressor is when the heat of vaporization is fully used and the discharge fluid is
fully vaporised and at its condensation point.
The boundary conditions of the compressor are determined by the required conditions for the
resorber/desorber process. The resorber inlet receives a waste water stream of 45-60 °C, which
needs to be increased to 110 °C, van de Bor et al.[21]. The desorber cools the waste water
stream down from 45-60 °C to 5 °C, as illustrated in Figure 1-4.
20
It is assumed that for industrial applications a temperature of 110 °C from the resorber can be
utilised, as well as a waste water temperature of 5 °C from the desorber. To reach the required
waste water conditions, the heat exchanger requires a ∆T, driving force, of approximately 5 K
to occur effectively. The compressor inlet temperature and outlet temperature become 40-55 °C
and above 115 °C respectively. In Table 2-1 these properties are listed and are dependent on the
heat pump cycle and the specific application. The values in the table are based on van de Bor et
al. [21].
The compression will ideally be an isentropic process where the entropy remains constant.
The compression is assumed to be an adiabatic process, where no heat transfer through the
boundaries of the compressor to the environment will occur.
Table 2-1 Assumed conditions of inlet and outlet of the twin-screw compressor boundaries.
The properties in the table above are used as starting values for the simulation model of the
compressor. The concentrations are set at these values during this research. The properties in
the study need to be considered as variables. The concentrations are based on the optimal of the
research of van de Bor et al. [20].
The screw compressor process can be divided into two phases: suction (inlet) and compression,
within the compression phase the discharge takes place (the outlet). The geometry of the lobes
and screws influences the performance of the compressor significantly. The geometry of the
lobes can be adapted to fit the requirements of the process (pressure and temperature) and gas
characteristics (composition, molecular weight, density, etc.). In literature and patents different
kinds of geometries for the lobes of the screw compressor have been proposed, each with their
own purpose. For most cycles the compressor efficiency is important for the cycle, and needs to
be as high as possible. To achieve these requirements the geometry needs to be adjusted until
the efficiency meets the optimum conditions. The desired outputs will be further explained in
section 2.4.
The change in volume of the cavity can be assumed to function as a batch type process. This
batch process can be characterised and modelled. Together with its geometry each twin-screw
compressor can be characterised by its cavity volume profile.
21
A mathematical model has been developed to simulate the geometry of the screw compressor;
the different methods, which have been proposed in literature, for this model are listed in Table
2-2. Although these methods make use of different starting positions to generate the geometry
of the rotors, all are essentially based on the envelope theory.
The envelope theory adapted to the geometry generation was used by Deng and Shu 1988 [35]
and Rinder 1979 [36]. Stosic et al. [27], Sakun [30], Zaytsev [1] and You [31] make use of the
same method but have a different approach towards describing the rotor profiles.
The envelope theory is used to generate the rotor profiles of the male and female rotor. In the
envelope theory the male rotor makes a curve that is used to generate the female rotor curve,
shown in Figure 2-2, where the rotor profiles are illustrated as circles. Sakun [30] and Zaytsev
[1] applied the envelope theory in such a way that the female profile is derived from the male
profile. The male rotor only spins around its centre and remains static. The female rotor spins
around its centre as well as turning around the male rotor, resulting in two separate movements.
The curve of the male rotor conjugates the curve of the female rotor. The female rotor profile
can be derived from the conjugated curve. The way the two rotors turn around each other
produces a curve for each of the rotors. You [31] inverted the envelope theory, using the female
rotor to calculate the male rotor. For a detailed explanation of the envelope method see You
[31].
R1h
R2h
The difference between Stosic et al. [27] and Sakun [30], You [31] , and Zaytsev [22] is that
the profiles of both the male and the female rotor are calculated from an original profile of
infinite radius called the rack The rack is generated on a static coordinate system. The rack is
what distinguishes the method of Stosic et al. [27] from the others, even though it is based on
the envelope theory as well.
22
Zaytsev and Infante Ferreira [28] developed another methods of calculating the rotor profiles,
based on the work of Sakun [30], but with a different starting view. Zaytsev and Infante
Ferreira [28] further developed a method based on the meshing line, the curve where the two
profiles of the rotors meet. This meshing line is used to calculate both the male and the female
rotor. In the method of Stosic et al. [27] the meshing line is an output variable. The meshing
line is a crucial parameter for the performance of a compressor. Zaytsev and Infante Ferreira
[28] developed their own model based on the meshing line as an input for the profile
generation. Zaytsev and Infante Ferreira [28] choice of using the meshing line as an input
parameter results in the meshing line becoming the main variable to describe the geometry of
the compressor rotor as well as the profile of the lobes.
The model used in this thesis is based on the original model developed by Sakun [30] and
Zaytsev [1].
The model describes the relation between the two rotor profiles through angle relations.
Coordinate systems were defined to describe the angle-based relation between the rotors. For
each rotor there is one static coordinate system and one rotating coordinate system. Subscript 1
represents the male rotor, 2 the female rotor and 0 the static coordinate system. The coordinate
systems are shown in Figure 2-3.
The coordinate systems can be described mathematically. The relation of change of the angles
is defined as the relation between the male and the female rotor profiles. The equations used in
the geometry model to calculate the male and female profiles are shown below. For more
details of these geometry equations see Zaytsev [1]. The rotor turning angle φ is used to
calculate the x and y coordinates of the male and female rotors. In these equations the r1h is the
male rotor radius from the origin to the beginning of the male lobe and the r2h is the female
rotor radius from the origin of the female rotor to the lobe tip of the female rotor minus the tip
radius of the female lobe, as can be seen in Figure 2-5.
𝜑! 𝜔! 𝑏! 1 𝑟!"
= = = 𝑖!" = = (2.2.1)
𝜑! 𝜔! 𝑏! 𝑖!" 𝑟!"
2. The relation between the rotation and axial motion, the z-axis can be calculated with this
formula to make the profile 3D:
23
𝜏 (2.2.2)
𝑧=𝐿∙
𝜏!
3. The male rotor profile in parametrical equations. ψ is the profile parameter:
𝑥! = 𝑥! 𝜓 (2.2.3)
𝑦! = 𝑦! 𝜓 (2.2.4)
4. Male to female:
5. Female to male:
Symbol Units
Envelope radius male rotor r1h mm
Radius male rotor lobe r mm
Radius male rotor is R1=r1h+r
Radius tip female lobe r0 mm
𝒃𝟐
Envelope radius female rotor r2h, calculated by 𝒓𝟐𝒉 = 𝒓𝟏𝒉 ∙
𝒃𝟏
Radius female rotor is R2=r2h+r0
Number of lobes male rotor b1 -
Number of lobes female rotor b2 -
Wrap angle of male rotor τw °
Length of the rotors L mm
Clearance Clearance mm
The equations above describe the relation between the geometry of the rotors. Extra
calculations are necessary to generate a representation of the rotor profiles. An example of such
geometry is given in Figure 2-4.
24
50
40
30
20
Figure 2-4 Example geometry of the rotor profiles: left is the male rotor with 5 lobes, right is the female rotor with 6
lobes.
Figure 2-4 shows an example of the profiles of the male and female rotor. These profiles are
calculated by means of different segments that form the rotor profiles together. First the male
rotor areas are calculated, then the female rotor areas are calculated with the equations to go
from one coordinate system to the other, as shown above. Six areas and segments are
generated in total for each rotor lobe, which form one lobe of the male rotor together, as shown
in Figure 2-5. The segments of the areas for the male rotor are: D1C1, C1A1, A1I1, I1L1, L1F1 and
F1D1, for the female rotor the same script is used but with subscript 2. All lobes of the male
50
rotor are identical, after generating one; the total profile of the male rotor can be generated. The
same goes for the female rotor, as shown in Figure 2-4. For more details on the areas and
segments see Sakun [30] and for the English version You [31].
40
D1
30
20
r2h
R2
L2 F2
20 F1
30 r0 r0 D2
L1
r0 I
2 r0
40 A1 A2
10 I1
C1
50 r
D2
0 O1 O2
D1
60
r1h R1
-10
-10 0 10 20 30 40 50 60
Figure 2-5 The angles and segments of the male and female rotor.
25
The areas and segments are all calculated separately, the area D1C1 is explained in detail to
clarify the calculation of the areas. The D1C1 segments (from the O1 to D1 over the curve
segment D1C1 and from C1 back to the origin) is shown in Figure 2-6 and the equation for the
area is given below, 2.2.12.
Area D1C1:
𝑟 (2.2.12)
𝐷! 𝐶! = − ∙ 𝑟!" ∙ 𝑠𝑖𝑛 𝜃!" − 𝑠𝑖𝑛 𝜃!"# + 𝑟 ∙ 𝜃!" − 𝜃!"#
2
The equations of the area consist out of r, r1h, 𝜃!"# and 𝜃!" . The angles are named up and low,
describing the direction the angle is going (from the lower angle to the upper angle), the angles
are calculated by using the cosine rule. 𝜃!"# becomes zero, for this segment, because this is the
starting line and does not make any angle with a next point. 𝜃!" as can be seen in Figure 2-6
makes an angle between D1, C1 and the origin of the male rotor O1. These angles are used
together with the radii to calculate the area of the segments.
50
This method is used for calculation of all the areas, all with their own lower and upper angles.
In the end all the six segment curves are generated with the area and the upper and lower angle.
The segment curves are added together and from the total curve that can be seen in Figure 2-5.
40 The equations that are used to calculate the areas of the other segments can be found in
D1
Appendix A.
30
r2h
L2 F2 R2
20 F1
r0 r0 D2
L1
r0 I
2 r0
A1 A2
10 I1
C1
r1h
r
Θup
0 O1 2
R1
D1 O
Figure 2-6 Profile segment D1C1 of the male rotor.
-10
-10 Parameters
0 that define
10 the geometry
20 of the twin-screw
30 compressor
40 are calculated
50 using the
60
geometry model. The main parameters are the cavity volume, suction port area and discharge
port area. These three parameters are used as input variables for the thermodynamic model.
Cavity volume:
The cavity volume is the main parameter when defining the compression part in the compressor
geometry model. The total cavity volume determines the change in volume. Cavity volume is
defined as the change of volume over the shaft rotation angle and is determined by the
geometry of the compressor, it is the main input for the thermodynamic model.
26
The volume calculation is separated into two parts, the suction calculation and the compression
calculation (the discharge is part of the compression). In Figure 2-7 the red line represents the
cavity volume. The increasing part of the graph represents the suction phase, the decreasing
part the compression phase. The two phases together give the cavity volume change for the
total compressor working cycle of one cavity. The unit of the cavity volume is mm3.
Suction area:
The suction area allows the mass flow to enter the compressor, and is calculated in mm2. The
suction port is represented by the first blue curve. The up slope of the curve represents the
period during which the suction port is opening. Once the suction port is fully opened, the area
remains constant for a certain amount of ‘time’, described by the shaft rotation angle,
represented in Figure 2-7 by the flat part of the curve. Eventually the suction port starts closing,
which is represented in the figure by the down slope of the curve. The moment the suction port
is closed the compression phase starts, which is represented by the down slope of the cavity
volume, represented by the red line.
Discharge area:
Like the suction port, the discharge port opens and closes again, represented by the second blue
curve in Figure 2-7. Less port area is needed to empty the cavity volume, as the fluid in the
compressor has been compressed and the volume has decreased. The discharge area is
calculated in mm2 as well.
#10 -4
1000 2
Cavity Volume in m3
Port Area in mm 2
0 0
0 100 200 300 400 500 600 700
Shaft Rotation Angle in Degrees
Figure 2-7 Schematic representation of the cavity volume, shown in red. The suction area and discharge area are shown
in blue.
27
the complete CRHP process. The thermodynamic model of the compressor is described as a
process with two phases, each with its own conditions and parameters: suction (inlet) and
compression, discharge (outlet) phases. The compression takes place between the inlet and
outlet, where reduction of the cavity volume results in increase of both pressure and
temperature. Besides pressure, temperature and cavity volume, the other significant variables
for the compressor are specific volume, enthalpy and mass flow. Some of these variables are
inputs and others are calculated in the thermodynamic model. To develop a dynamic
thermodynamic model, a homogeneous model that is reproducible with current modelling tools
needs to be developed first. This is done with the use of the homogeneous model developed by
Zaytsev [1]. For the simulation, the modelling tool Simulink is used extended with Matlab [34]
and Refprop [32] via Fluidprop [33] to calculate the required physical properties.
The thermodynamic model is influenced by the geometry of the wet twin-screw compressor.
The geometry of the compressor needs to be integrated into the model, in order to produce a
dynamic model that simulates the compressor as realistically as possible. The model used in
this thesis is based on the geometry model developed by Zaytsev [1], which has been explained
in section 2.2.
Inlet Outlet
Upstream Downstream
Compressor
Figure 2-8 Schematic compressor.
The compressor inlet is expected to be in the two-phase flow regime. This is a given property
from the chosen heat pump cycle, as described in section 2.1. Within the two-phase regime the
pressure and temperature are dependent. In this regime the physical state cannot be modelled
using a fixed temperature and pressure only. For the outlet of the compressor saturated vapour
is assumed for achieving the highest possible efficiency of the heat pump cycle. Deviating from
an exact full vaporisation reduces the efficiency and this will in reality require practical process
control settings to reach optimal operation.
In the homogeneous pT-model the liquid and vapour are at equilibrium at any given moment in
time. This allows for writing the conservation equations for the working mixture inside the
cavity volume. The cavity volume is defined in the compressor geometry model, section 2.2.
The equations of Zaytsev [1] were written based on the shaft rotation angle 𝜑, the derivation of
the conservation equations can be found in appendix B. In order to apply the modelling tool
combination Matlab/Simulink, the conservation equations need to be converted to a time basis
instead of the shaft rotation angle 𝜑 basis used by Zaytsev [1].
Rewriting the conservation equations is possible because the time for each degree of rotation
can be determined. From a known rotational speed, the time difference per degree rotation
angle can be calculated. With the calculated time per rotation, the same conservation equation
can be used replacing the change in shaft rotation angle ∆𝜑 by the change in time ∆𝑡. The
changes to the time dependent input values for the conservation equations will be explained in
chapter 3.
28
The four conservation equations for the homogeneous thermodynamic model of Zaytsev [1],
!"
are written as a function of pressure, concentration, temperature and mass flow over time: !" ,
!!! !" !"
!"
, !" and !" . For more details of the derivation of the conservation equations see Zaytsev [1]
and Appendix B. These four functions describe the thermodynamic behaviour of the mixture in
the cavity volume.
!!
If it is assumed that no liquid is separately injected, the concentration change over time !"! can
be assumed to be zero. The only moment that this function will have a different value than zero
will be when external fluid is added during compression. In the twin-screw compressor model
external injection will not be applied and hence the concentration change will stay zero during
!!
the whole compression cycle. The function !"! will therefore drop out of the conservation
equations as will the conservation equation for change of concentration itself.
Therefore three conservation equations remain. The remaining three conservation equations can
be combined into two. The combination can be accomplished as the pressure and temperature
!" ! 𝑑𝑚 𝑑𝑚 !"
are dependent of the change in mass over time !" = ! 𝑛𝑘=1 𝑑𝑡 − 𝑙𝑘=1
𝑑𝑡
. The
!"
can 𝑜𝑢𝑡 𝑖𝑛
𝑘 𝑘
!" !"
therefore be added to the !" and !" relations so that the equations result in two conservation
equations that are applied in the model. The rewriting leads to the following two simplified
conservation equations as applied in the homogeneous model.
𝑑𝑝 1 𝑣
!
𝑑𝑚!"#
!
𝑑𝑚!" 1 𝑑𝑉 𝜕𝑣 𝑑𝑇
𝑑𝑡
=
𝜕𝑣 𝑚 𝑑𝑡
−
𝑑𝑡
+
𝑚 𝑑𝑡
−
𝜕𝑇
(2.3.1)
! ! !,! 𝑑𝑡
!!! !!!
𝜕𝑝 !,!
!"
Equation 2.3.1 shows the mass conservation rewritten to pressure change over time !"
in units
!"
of !
. Equation 2.3.2 shows the energy conservation rewritten to temperature change over the
!" ! !" !" !!
time !"
in units of ! . The values for the physical parameters !" !,!
, !" !,!
and !" !,!
are
calculated in the homogeneous model with a combination of two fluid property programmes:
Refprop [32] via Fluidprop [33].
!" !"
The change in cavity volume over time !" is an input together with the mass flows !"
of the
inlet, outlet and the leakages within the compressor.
29
2.4 Theoretical Desired Performance Outputs
To identify the theoretical performance of the modelled compressor, it is necessary to
determine which outputs are required to describe that performance. The main outputs are
isentropic efficiency, compressor shaft power and volumetric efficiency. These outputs are
required to compare the different geometrical designs of the compressors and to select the most
efficient one.
𝑊!",!"#$% ℎ!! − ℎ!
𝜂!" = = (2.4.1)
𝑊!"#$ ℎ! − ℎ!
The ideal line of the compression in the equation above goes from point 1 to 2s, respectively
the inlet and outlet of the compressor, and follows the isentropic line that represents constant
entropy. The real work of the compression, point 1 to 2, is increased compared with the ideal
work; this is according to the prediction that more energy is needed if the losses of friction and
flow are added. The work is representing the compressor shaft power (ideal or real) in the
isentropic efficiency calculations.
Volumetric efficiency:
Next to the isentropic efficiency, the volumetric efficiency must be calculated. The volumetric
efficiency gives the ratio of the ‘real’ or total volume capacity of the compressor and the ‘ideal’
actual volume of the fluid in the compressor, shown in equation 2.4.2.
𝑉!"#$
𝜂!"# = (2.4.2)
𝑉!"#$%
The three desired outputs as explained above are used in the model to evaluate the efficiency of
the modelled screw compressor. The implementation of the two efficiency equations into the
thermodynamic model is explained in section 3.4.
30
3 Thermodynamic Model Implementation
The thermodynamic model was developed to include and integrate inputs derived from
geometry and physical property models. To develop the model, a number of approaches and
trial attempts have been made. Explanation of the historic development will support the current
modelling approach. This chapter includes the implementation of the equations and the inputs
of the thermodynamic model. In the model implementation diagram, Figure 3-1 the flow of
information over the several models and inputs is shown. The inputs for the thermodynamic
model are divided into three sections, inputs from the geometry model, physical properties and
the input values/initial values. These inputs are explained in section 3.2 ‘Input
Implementation’. In section 3.3 the thermodynamic model is described and includes equations
and correlations to describe real life compressor energy loss parameters such as friction and
leakages. In section 3.4 the implementation of the theoretical desired outputs, the isentropic en
volumetric efficiencies, are explained.
Geometry
Volumetric
Model
(Matlab) Efficiency
Pressure
Physical Thermodynamic Isentropic
Proper6es Model Efficiency
(Fluidprop/Refprop) (Simulink)
Temperature
Important
ShaC Power
Input Requirement
Values
Matlab/Simulink [34] is the tool selected that should be able to include the integration between
the geometry model and thermodynamic model. The integration also required calculated
property derivatives through an integrated physical property model. In the present
implementation Refprop [32] via Fluidprop [33] is accessed from Matlab to calculate the
required properties and its derivatives.
31
The geometry model developed by Zaytsev has been modified to fit the requirements of the
present model. It contains the conservation equations and the full geometric modelling of the
compressor in the 2 phases of inlet, compression/outlet. The rotation angle based geometry
model that was originally developed, has been transferred into a time-based model that
provides input to the time dependent thermodynamic model.
The integrated model was built in Matlab/Simulink. The combination of Refprop and Fluidprop
into Matlab/Simulink gave the possibility to integrate the generation of physical properties for
input, compressor internal process and outputs. The thermodynamic model is now able to
describe the compressor behaviour in time (dynamic) and generate output parameters and
values for compressor design and optimisation and can be included in a full CRHP model.
Inputs
Ini)al Values
(Pressure and Temperature)
Physical Proper)es
(Input: Pressure and Temperature)
If t < ttotal cycle
Conserva)on Equa)ons
(Thermodynamic Model)
Pressure
&
Temperature
If t = ttotal cycle
Outputs
32
Matlab and needs to be combined with the thermodynamic model in Simulink. In this thesis a
specific case of geometry is considered. The input variables that are used for the geometry
model are shown in Table 3-1 below, the values are variable and can be changed if needed.
Table 3-1 Input values used in the geometry model, these are the inputs of the geometry model as used by Zaytsev [1].
As already explained in chapter 2, the geometry model is based on the shaft rotation φ. The
thermodynamic model requires time-based inputs. To achieve the inputs based on time the total
time of one compression cycle is calculated with equation 3.2.1.
1
𝑡𝑖𝑚𝑒 = 𝑛 ∙ 𝜑
∙ 360 (3.2.1)
60
The time will be in seconds. The compression cycle in this case consists of a shaft rotation
angle of 777 degrees; the total time of one compression cycle becomes therefore 0.03695s. The
shaft rotation angle is specific for this case and will vary for different geometries, dependent on
the wrap angle, the number of lobes and the geometry of the rotors. Calculation of the shaft
rotation angle is shown in detail in Appendix G.
!"
To calculate the volume change !" in time from the cavity volume curve, the change over time
dt needs to be calculated. To model the geometry as time based, each compression cycle has
been defined as an array of individual one degree angle rotations from 0-777 degrees. The
rotation angle can be transformed to time with equation 3.2.1, adding one shaft rotation degree
to this equation the time per degree becomes 4.76 ∙ 10!! s.
0.008
1.2
0.006
1 0.004
Volume Change [m /s]
3
0.002
Volume [m 3 ]
0.8
0
0.6 -0.002
-0.004
0.4
-0.006
0.2
-0.008
0 -0.01
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time [s] Time [s]
(a) (b)
Figure 3-3 Volumes of the compressor over time (a) the cavity volume and (b) the change in volume.
33
The most important property used in the thermodynamic model is the cavity volume and it is
displayed in Figure 3-3(a). To decrease the run time of the thermodynamic model, the volume
!"
change !" will be calculated in the geometry model and applied as input for the thermodynamic
!"
model instead of the cavity volume curve. The curve of the volume change over time, !"
, is
shown in Figure 3-3(b).
!"
The volume change !" in Figure 3-3(b) shows the rate of increase or decrease of the volume.
The volume change changes from positive: the increase of the cavity volume, to negative: the
decrease of the cavity volume. The switch from positive to negative is on the maximum value
of the cavity volume, at time 0.02186 s.
The other two output variables of the geometry for the homogeneous model are the suction port
area and discharge port area. The two areas can be represented as in Figure 3-4. In Figure
3-4(a) the suction area is shown. In Figure 3-4(b) the opening and closing area of the discharge
port are shown.
As expected the discharge port area is smaller than the suction port area. Less area is needed for
the discharge due to the compressed fluid leading to increased density and therefore requiring
less volume per mass flow and as such less area. The discharge port area in Figure 3-4(b)
shows that the port closes two times during a single cycle. This phenomenon is redundant, the
discharge port area should, like the suction port area, open and close just once. The so called
close-trapped volume [1] causes an extra curve at the end of the discharge. The close-trapped
volume is a volume that is formed between the two rotors after the discharge port is closed. The
close-trapped volume is included in the calculated discharge port area. The influence on the
thermodynamic model is however negligible and has been filtered out in the thermodynamic
model in Simulink.
700 700
600 600
500 500
Area [mm ]
Area [mm 2 ]
400 400
300 300
200 200
100 100
0 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time [s] Time [s]
(a) (b)
Figure 3-4 Port areas in mm2 (a) is the suction port area and (b) is the discharge port area.
The port areas of the suction and discharge are used as inputs to calculate the mass flows into
and out of the compressor. The suction port area will be used to calculate the mass flow of the
inlet of the compressor and the discharge port will be used to calculate the mass flow of the
outlet of the compressor. This will be explained in detail in section 3.3.1.
34
3.2.2 Thermo-Physical Properties
Next to the geometry model, input of thermo-physical properties is required. Eight thermo-
physical properties have been defined for the model. The eight thermo-physical properties that
are used are shown in Table 3-2. These are the physical properties that are needed for the
conservation equation calculations (explained in section 2.3) and mass flow calculations (which
will be explained in section 3.3.1). The thermo-physical properties are invoked through
equation of state calculation tools. To invoke the thermo-physical properties two inputs are
needed for a given composition, the pressure and the temperature. The composition is pre-set
for the used ammonia/water mixture. The thermo-physical properties of the ammonia/water
mixture are invoked from Refprop via Fluidprop by a Matlab/Simulink file. The pressure and
temperature inputs are calculated by the conservation equations. The change in temperature
divided by the change in time is integrated over time and results in the temperature evolution
during the compression cycle. For the pressure the same calculation method is applied. The
Simulink blocks are shown in Figure 3-5 and the needed Matlab code in Figure 3-6.
Figure 3-5 shows a block diagram where the green blocks are the inputs, pressure and
temperature, the orange block calculates the thermo-physical properties through Matlab and the
!"
red block the output !" in this case.
!
𝒅𝒗
Figure 3-5 Example of a Simulink block diagram to invoke physical properties for .
𝒅𝒑 𝑻
The orange block in the Simulink block diagram above calls the Matlab file to invoke the
thermo-physical properties. For this specific example the Matlab code is given in Figure 3-6.
!"
The code is to invoke the physical properties and to calculate !" , which is one of the inputs
!
!"
for the conservation equations. To achieve the value for !! !
a small variation is needed in the
pressure (dp), in this case assumed 𝑑𝑝 = 𝑝 ∙ 10!! Pa. In this code example the specific volume
is invoked twice. Once with the pressure and once with the pressure added with a small
difference dp, both the invoked specific volume values are at constant temperature.
35
%% Diff Specific volume/pressure constant T [(m3/kg)/Pa]%%
% With constant Temperature and concentration %
dp =1;
v = invoke(FP,'SpecVolume','PT',p, T);
v_dp = invoke(FP,'SpecVolume','PT',p+dp, T);
vp = (v_dp-v)/(dp);
output(1)=vp;
end
Figure 3-6 Example of the Matlab code to invoke the physical properties from Refprop via Fluidprop and calculate
𝒅𝒗
.
𝒅𝒑 𝑻
More inputs are invoked from Refprop via Fluidprop and are listed in Table 3-2. In this table
!"
the first three physical properties in the table need to be calculated as explained above, ( !" ,
!,!
!" !!
!" !,!
,
!" !,!
). The Table also lists the required values for calculating these thermo-physical
properties. The last four properties are invoked similarly, but without additional calculations.
The only inputs needed are the pressure and temperature.
The input values are determined by the process as selected for the heat pump and are related to
the optimal values for the conditions imposed by the compression resorption heat pump cycle.
36
The two most important input values are the input values that are required for integration of the
!" !"
output values of two of the conservations equations !" and !" . The integration block in
Simulink carries out the integration.
• Flow friction
• Leakage through several paths.
• Heat transfer within the screw compressor housing.
The above inefficiencies influence the thermodynamic model. Other inefficiencies will be more
related to mechanical losses and will not be dealt with in this thesis.
The ODE solver that is used in the thermodynamic model is the ‘Ode23tb’. The ‘Ode23tb’ is a
stiff one-step method and is based on the Runge-Kutta method with two stages. The first stage
is a trapezoidal rule step while the second stage uses a backward differentiation formula of
order 2.
Section 3.3 is divided into three sub-sections; 3.3.1 implementation of the mass flows, 3.3.2
leakage path areas and 3.3.3 mass flows of the leakages. The heat transfer within the screw
compressor housing is expected to have limited effect and will not be taken into account in the
model and will not be explained in this chapter.
𝑑𝑚
= 𝜁𝜌𝐴𝑢 (3.3.1)
𝑑𝑡
For equation 3.3.1 an area A is needed as input. For calculating the mass flow in and the mass
flow out of the working cavity, the areas of the suction port and discharge port are needed. The
ζ is an empirical flow coefficient, ρ is the density and u is the flow velocity that will be
calculated with equation 3.3.2.
!!!"!
𝑢= 2∙ 𝑣𝑑𝑝 (3.3.2)
!!"#
37
For the calculation of the flow velocity u, integration is done between the high pressure and the
low pressure of the fluid respectively entering and leaving the specific flow areas. The specific
volume v is calculated with Fluidprop whilst it is assumed that the process is ‘isentropic’.
The inputs for the continuity equation 3.3.1 and the flow velocity equation 3.3.2 are generated
by the geometry model (the area A) and by the thermodynamic model (the density ρ and the
specific volume v). The compressor is defined from inlet port to outlet port. Two parameters
(boundaries of the inlet pressure suction and outlet pressure discharge) are added to the
thermodynamic model as input for the input mass flow and output mass flow calculations, see
Figure 3-7. The first parameter is the pressure difference over the suction port of the
compressor. The pressure difference for the inlet is defined as the difference between the heat
pump system outlet Phigh and the compressor inlet Plow, Phigh – Plow. The second parameter is the
pressure difference over the discharge port of the compressor. The pressure difference between
the compressor outlet Phigh and the heat pump system inlet Plow, Phigh-Plow.
Compressor Boundary
Suc1on Discharge
Figure 3-7 Pressure differences of the mass flow of the suction and discharge of the compressor boundary. The pressure
difference is needed for the calculation of the mass flows in and out of the compressor.
One of the inputs of the continuity equations 3.3.1 is an empirical flow coefficient. The
empirical flow coefficient is a correction factor for the non-isentropic effects (irreversible
effect), for example the friction in the flow area. For the empirical flow coefficient different
values are given. The reason is that the flow coefficient is dependent on each particular
application such as the working fluid and the geometry. For the application in this case, an
empirical coefficient is applied that is based on a constant mass flow in and out through the
cross sectional area. It has to be acknowledged that this value is a variable that can and needs to
be adapted for different applications with change of fluid, geometry, and actual clearance
resulting from compressor manufacturing, etc.
Table 3-3 Empirical flow coefficient values.
The flow coefficient values applied in this thesis originate from Zaytsev’s literature review [1],
Table 3-3. The empirical flow coefficient determined by Prins and Infante Ferreira [37] have
been experimentally obtained.
In section 3.2.1 the cross sectional port areas of the suction and the discharge have been
introduced which will become the input for calculating the mass flows in and out of the
compressor. The port area of the suction and the discharge are generated with the geometry
model based on the input values listed in Table 3-1.
38
In figures 3-8(a-b) the mass flow is graphically shown as time dependent of the rotor turning a
full cycle from opening of the inlet port to closing of the outlet port. The calculated mass flow
for the suction within the thermodynamic model is shown in Figure 3-8(a). The mass flow
increases and decreases together with the opening and closing of the suction port. In Figure
3-8(b) the mass flow leaving the working cavity volume via the discharge port is shown. The
mass flow follows the curve of the port area. The curves in Figure 3-8 are exemplary for the
mass flows calculated by the model.
0.18
1.2
0.16
1
0.14
0.8
0.1
0.6
0.08
0.06 0.4
0.04
0.2
0.02
0 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
Time [s] Time [s]
(a) (b)
Figure 3-8 (a) Mass flow into the compressor through the suction port area, calculated with the input suction port area
from the geometry model, Figure 3-4(a). (b) Mass flow out of the compressor through the discharge port area,
calculated with the input discharge port area from the geometry model, Figure 3-4(b).
The flows that enter and leave the cavity volume are divided into three flows: the suction flow,
discharge flow and the internal leakage flows between the different cavity volumes. The mass
flows from the suction and discharge ports are explained above. The internal leakage path areas
will be explained in sections 3.3.2. These internal leakages are based on a single cavity volume.
The calculation approach of the internal leakages will be explained in the section 3.3.3 and will
also be based on the isentropic converging nozzle equations. The empirical flow coefficients
used for the internal leakages calculations are listed in Table 3-3, Zaytsev.
One cavity volume has been modelled and will be applied throughout the compressor model
and is defined as the main cavity volume. For the single defined main cavity volume leakage
from the leading cavity volume (advanced in time) and to the trailing cavity volume (delayed in
time) will be modelled. All three cavity volumes are illustrated in Figure 3-10. The leakages
mass flows will however depend on the increased pressure over the sequenced cavities and
39
hence change in density. These internal flow phenomena, together with the leakage mass flow
calculation, will be explained in the next section 3.3.3.
Leakage is one of the more significant mechanisms that influence the efficiency of the
compressor according to Zaytsev [1]. Zaytsev modelled five leakage paths. That is one path
less when comparing to Fleming and Tang [38] where six paths were identified and defined. In
Figure 3-9 the reduction of efficiency of the six leakage paths of Fleming and Tang [38] is
shown. As can be seen path 5 has the lowest efficiency reduction and has been excluded by
Zaytsev [1] for implementation of the leakage paths in the thermodynamic model. Path 5 of
Fleming and Tang [38] represented the leakage of the suction end clearance of the rotors.
Figure 3-9 Leakage paths efficiency, based on 3000 rpm and an evaporating temperature of 253.15 K [38]
The five paths that are defined are renumbered 1-5. The five paths that are modelled in the
geometric model are:
The five paths summarised above are illustrated in Figure 3-10, Figure 3-11, Figure 3-15 and
Figure 3-16 and will be further explained in detail.
The five leakage path areas are calculated in the geometry model and will be used as inputs for
the thermodynamic model to calculate the inflow and outflow of the working cavity volumes.
For calculation of the leakage flows, the same calculation as for the mass flows, section 3.3.1,
is used and the implementation of the leakage flows will be explained in detail in section 3.3.3.
In the geometry model the flow areas of the leakage paths 1, 2 and 5 are calculated with
equation 3.3.3. The clearance is multiplied by the length of the contact line for path 1, the
sealing line for path 2 or the height of the lobes (r+r0) for path 5, shown in Figure 3-16. The
contact line, sealing line and the lobe height are represented by LL in equation 3.3.3 and will
result in the cross sectional areas for paths 1, 2 and 5. The variable clearance that is used in this
application is 0.01 mm, Table 3-1.
𝐴 = 𝑐𝑙𝑒𝑎𝑟𝑎𝑛𝑐𝑒 ∙ 𝐿𝐿 (3.3.3)
40
The blowhole areas are used for path 3 and 4. For path 3 the cusp blowhole is used and for path
4 the start blowhole is used. For the calculation of the blowhole areas the definitions of Singh
and Bowman [39] are used.
In Figure 3-10 the first three paths are illustrated. In Figure 3-11 path 1 is shown in more detail
and in Figure 3-15 the two blowholes for path 3 and path 4 are shown. The discharge end
clearance of the rotors for path 5 is illustrated in Figure 3-16 and shows two different kinds of
paths that are used for path 5.
Suc$on
Leakage Path 1
Leakage Path 2
Leakage Path 3
Compression
Discharge
Figure 3-10 Illustrated leakage paths 1, 2 and 3. View from rotors above.
Leakage path 1
Leakage path 1 shown in Figure 3-10 is the leakage path across the contact line between the
male and the female rotor. The area of the leakage path is modelled and calculated in the
geometry model and used as an input for the thermodynamic model. The isentropic converging
nozzle equation and the leakage path 1 area are used to calculate the amount of fluid flow
across the contact line between the two rotors. In Figure 3-11 a side-view of the female rotor is
shown. In this figure the contact line is shown with leakage path 1 divided into five smaller
leakage flows numbered from 1 to 5. In the geometry file these five cross sectional areas of the
leakage flow are added together to form one cross sectional area. As can be seen, path 1 creates
a leakage from the compression side to the suction side.
2 3 5
1 4
Figure 3-11 Path 1: in this figure the female rotor is shown from the side. The line in the middle is the contact line and
the arrows crossing the line illustrate the leakage direction of path 1 through the cross sectional area which is created
by the contact line and the clearance. The mass flow of leakage path 1 flows from the compression side to the suction
side [1].
This leakage path is modelled in the geometry model in two different paths. One is the leakage
through the contact line during the suction phase, Figure 3-12(a), and the second is the leakage
through the cross sectional area during the compression phase of the compressor, Figure
41
3-12(b). The suction phase and compression phase are related to the cavity volume, the red line
in Figure 2-7, where the increasing slope represents the suction phase and the decreasing slope
represents the compression phase.
Figure 3-12(a-b) show each three curves, the total curve and two curves representing different
segments of the contact line. The total leakage area of path 1 is the summation of the two
segment curves together. The first segment is the part of the segment line from segment 1 to 5.
The second segment is the leakage path area from segment 5 to 1. In the implementation of the
leakage path areas, path 1 is implemented as the two separate segment curves. This is caused
by the different pressure difference for each segment during the mass flow calculations of the
leakage flow and will be described in detail in the next section.
Leakage Area Path 1 - Contact Line - Suction Phase Leakage Area Path 1 - Contact Line - Compression Phase
14 14
Total Path Area Total Path Area
Segment 1-5 Segment 1-5
Segment 5-1 Segment 5-1
12 12
10 10
8 8
6 6
4 4
2 2
0 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time [s] Time [s]
(a) (b)
Figure 3-12 The leakage area of path 1, this is the leakage area created by the contact line between the two rotors. (a)
Path 1 during the suction phase. (b) Path 1 during the compression phase.
Leakage path 2
Leakage path 2 is determined by the cross sectional area of the sealing line between the rotor
tips and the housing, Figure 3-10. This path is calculated in the geometry model similar to path
1, see equation 3.3.3. Figure 3-13 shows the leakage area from the leading cavity to the main
cavity with the solid line and the leakage area from the main cavity to trailing cavity with the
dotted line. These two paths are the summation of the leakage areas of the female rotor and
male rotor.
50
40
Leakage area [mm2 ]
30
20
10
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time [s]
Figure 3-13 Leakage area of path 2 is the leakage through the sealing line between the rotors and the housing.
42
Leak path 3
Leakage path 3 is the leakage flow area in and out of the cusp blowholes, Figure 3-10 and
Figure 3-15. The cusp blowholes are the spaces where the contact line does not reach the
housing, the space between the contact line and the sealing line. These holes are called
blowholes and have a triangular shape.
Leakage Area Path 3 - Cusp Blowhole Compression Side Leakage Area Path 4 - Start Blowhole Suction Side
1.5 60
From Leading Cavity From Leading Cavity
To Trailing Cavity To Trailing Cavity
50
1 40
30
0.5 20
10
0 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time [s] Time [s]
(a) (b)
Figure 3-14 Both path 3 and 4 are the leakage areas through a blowhole. In figures (a) for path 3 and (b) for path 4 are
the leakage areas shown. Path 3 represents the leakage through the cusp blowholes on the compression side and path 4
represents the leakage through the start blowholes on the suction side.
In Figure 3-14(a) the cross sectional area of the leading cavity to the main cavity (solid line)
and the cross sectional area from the main cavity to the trailing cavity (dotted line) are
displayed. Like path 2 the pressure difference needed for this leakage flow calculation is the
pressure difference between the leading, main and trailing cavities.
Figure 3-15 The two different blowholes. In this figure the female rotor is shown together with the contact line. In the
compression side is the cusp blow-hole shown and in the suction side is the compression start blow-hole shown [1].
Leakage path 4
Leakage path 4 is created by leakage cross sectional areas for the inflow and outflow of the
compression start blowhole and is illustrated in Figure 3-15. In this figure the side-view of the
female rotor is shown together with the contact line. The compression start blowhole is like the
cusp blowhole (path 3) the difference is that the compression start blowhole is located at the
suction side of the rotors. The leakage area of path 4 is shown in Figure 3-14(b).
43
Leakage Paths 5
Discharge cavity
to suc5on cavity
Leading cavity to
trailing cavity
Leakage path 5
Leakage path 5 is the leakage through the discharge end clearance and is shown in Figure 3-16.
The leakage is defined between the end plate, the rotor end face and the discharge end. In this
area there are two different kinds of paths where the fluid can go. The first path is the area
between the ‘discharge’ cavity (at the compression side of the rotors) and the ‘suction’ cavity
(at the suction side of the rotors). The second path is the cross sectional area between the
leading cavity, main cavity and the trailing cavity. Both the two different cross sectional areas
are generated by the geometry model and are shown in Figure 3-17(a-b).
Leakage Area Path 5 - Discharge End Clearence - Discharge/Suction Leakage Area Path 5 - Discharge End Clearence - Leading/Trailing
1.6 3.5
From Discharge From Leading Cavity
To Suction To Trailing Cavity
1.4
3
1.2
2.5
Leakage area [mm2 ]
Leakage area [mm2 ]
1
2
0.8
1.5
0.6
1
0.4
0.5
0.2
0 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time [s] Time [s]
(a) (b)
Figure 3-17 Cross sectional areas of leakage path 5. This is the leakage through the discharge end clearance of the
rotors. (a) Is the leakage path area between the compression side and suction side. (b) Is the leakage path area between
the leading cavity, main cavity and trailing cavity.
Figure 3-17(a) corresponds with the grey line in Figure 3-16, the solid line is the area from the
discharge to the discharge end clearance, the dotted line is from the discharge end clearance to
the suction. Figure 3-17(b) corresponds with the orange leakage flows in Figure 3-16. The solid
line is the cross sectional area from the leading cavity to the discharge end clearance and the
dotted line is the cross sectional area from the discharge end clearance to the trailing cavity.
This together with delays between the area curves explains why there are four different cross
sectional areas needed.
44
The areas of the leakage flows are used for the leakage flow calculations. The leakage flows are
calculated in a similar way as the mass flows of the suction and discharge of the compressor in
section 3.3.1.
The thermodynamic model is running over time. The pressure difference curve for calculating
the mass flow of the leakage paths is based on the pressure difference between the advance
pressure curve, the main pressure curve and the delay pressure curve. The three pressure curves
need to be generated by different shifts in time. The main pressure curve is generated by the
conservation equations in the thermodynamic model. The main pressure curve is used to
generate the advanced pressure curve and the delay pressure curve. The shift needs to be done
before the model starts running. The first pressure curve (main) is generated by the
homogeneous model and is used as starting value. This pressure curve will be delayed and
shifted in advance for the pressure difference calculations that are needed during the run of the
thermodynamic model with the leakage paths. The pressure difference calculations are
performed in a separated matlab file, see Appendix E.
In this Matlab file the main pressure curve is used to calculate the advanced pressure and the
delayed pressure, these three curves are used to calculate the pressure difference. The generated
main pressure curve will be used for the calculation of the pressure difference by shifting the
main pressure curve back and forward in time. Because the calculations have to be done back
in time all three pressure curves will have a delay, see Figure 3-18. By adding the difference
between the advanced curve and the main curve and the same difference between the main
curve and the delayed curve results in the same delay and advance to the main curve.
Advanced Curve
Delayed Curve
Pressure [Pa]
Main Curve
dφ dφ
Delay
Delay + dφ
Delay + 2 · dφ
Time [s]
Figure 3-18 Illustration of the pressure curve shift and the difference between the curves.
The delay and the difference 𝑑𝜑 between the three curves are added and the pressure difference
can then be calculated. The advanced curve minus the main curve and the main curve minus the
delay curve generates two pressure difference curves. The last step before these curves can be
45
used is: The delay that is added to all the pressure curves is subtracted from the beginning of
the pressure difference curve. The two pressure difference curves need to become the same
length as the main pressure curve. The end of these curves is subtracted till the curves have the
same length.
To model the leakages in the thermodynamic model three different time shifts of the pressure
curve need to be calculated. Each of the three are used for different leak paths:
The three pressure differences will be next explained in detail. The data in the plots are an
example of how the pressure curves and pressure difference curve will look like. The inputs of
the geometry file used are based on the experimental conditions reported by Zaytsev [1], Table
3-1. The initial values used in the thermodynamic model are a suction pressure of 3.7 bar,
suction temperature of 62.8°C and a discharge pressure of 9.08 bar. These initial values will
likewise be used in chapter 4 during the validation, Table 4-1.
The pressure difference between the leading, main and trailing cavity is calculated by 360°
divided by the number of lobes of the male rotor, b1. In this case (based on Zaytsev [1]) the
male rotor has 5 lobes and results in a shift of 72°, each degree has a time step of 4.76 ∙ 10!! s
with a rotational speed of 3500 rpm. In Figure 3-19(a) the pressure shift is shown. Figure
3-19(b) shows the pressure difference between the advanced – main pressure curve and the
main – delay pressure curve. This graph is used in the thermodynamic model for the calculation
of the mass flow of the leakage paths 2, 3, 4 and part of path 5. This part of path 5 is defined as
the leakage between the leading/main and main/trailing cavity at the discharge end clearance,
Figure 3-16. These four paths are divided into two different areas, explained in section 3.3.2.
One area of the two represents leakage flow into the main cavity; in this case the advance/main
pressure difference is implemented. The second area represents the leakage flow out of the
main cavity and this mass flow is connected with the pressure difference between the main -
delay pressure difference curve.
°
Pressure difference
#105 Pressure Curves Advanced/Main/Delayed of 360/b1 °
11 #105 Advanced/Main/Delayed of 360/b1
6
Original Pressure Curve
Advanced-Main
10 Advanced
Main-Delayed
Main
5
Delayed
9
4
8
Pressure [Pa]
Pressure [Pa]
7 3
6
2
5
1
4
3 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
Time [s] Time [s]
(a) (b)
Figure 3-19 (a) Graph of the pressure shift of 360/b1 in this case 72°, 0.003427 s. In this graph the original pressure
curve is given. (b) The pressure difference between the leading - main and trailing – main cavity.
The next pressure difference that is needed is the pressure difference between the compression
side and the suction side, Figure 3-15. This pressure difference is over the contact line. To
calculate this pressure difference, a pressure curve shift is made by 360°, as can be seen in
Figure 3-20(a). In Figure 3-20(b) the resulting pressure difference is shown. The pressure
46
difference in this case is calculated the same way as done above. This pressure difference is
applied in the mass flow calculations of the leakages as part of path 1 and part of path 5. The
areas of Path 1 are explained in section 3.3.2 and for this pressure difference the area of
segment 1 to 5 of the contact line is used. The leakage area of path 5 from the discharge to the
suction is used together with this pressure difference with pressure shift of 360° to calculate the
mass flow leakage of discharge side to suction side. The pressure differences are added in the
same way as explained for the pressure shift of 72°.
Pressure [Pa]
4
7
3
6
2
5
4 1
3 0
0 0.05 0.1 0.15 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
Time [s] Time [s]
(a) (b)
Figure 3-20 (a) Graph of the pressure shift of 360°. (b) Pressure difference between the compression side and the suction
side.
There is one leakage area that needs a different pressure shift than the two pressure differences
mentioned above. This leakage path area is the area of path 1, the contact line is divided into
two, segment 1 to 5 and segment 5 to 1. This last segment 5 to 1 is at the contact line between
the two cavity volumes. This requires adding 72° to the 360° pressure shift, Figure 3-21(a).
Segment 5 to 1 is divided into compression and suction side and needs the two pressure
differences as shown in Figure 3-21(b).
Pressure Curves
° ° Pressure difference
#105 Advanced/Main/Delayed of 360 +360/b1 ° °
11 #105 Advanced/Main/Delayed of 360 +360/b1
Original Pressure Curve 7
Advanced Advanced-Main
10 Main-Delayed
Main
6
Delayed
9
5
8
Pressure [Pa]
Pressure [Pa]
4
7
3
6
2
5
4 1
3 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
Time [s] Time [s]
(a) (b)
Figure 3-21(a) Graph of the pressure shift of 360 degrees + 360/b1 in this case 72 degrees. (b) Pressure difference
between the shifted pressure curves.
With the generated pressure differences, the leakage mass flows can be calculated.
Implementation of the mass flows of the leakages to the compressor model is done in Simulink
equal to the mass flows for the suction and discharge as based on the leak path calculation
approach as explained above. To have a total overview of the thermodynamic model in
Simulink, see Appendix C.
47
The leakage mass flows in and out of the main cavity volume influence the mass flow from
suction to discharge. Like path 1, the leakage flow over the contact line flows out of the main
cavity volume is added to the mass flow out of the conservation equation (like the discharge
mass flow) in the thermodynamic model. Path 2, 3, 4 and 5 have two different flows, into the
main cavity volume and out of the main cavity volume. Out of the main cavity volume is added
to the mass flow out and mass calculation is done as per path 1. The leakage flows of these
paths into the main cavity volume are added to the normal mass flow into the cavity volume
(like the suction mass flow), and are connected to the conservation equations.
The quantities of the 5 mass flows of the leakage paths are shown in Figure 3-22. In this graph
first the total leakage flow is shown together with the inflow (positive) and outflow (negative)
of the main cavity volume. After the total leakage flow the individual paths are shown in the
graph divided into total, in and out of the main cavity.
0,02
0,00
-0,02 Total Mass Flow
-0,04 Inflow main Cavity
-0,06
OuElow main Cavity
-0,08
5
2
4
hs
h
h
h
t
Pa
t
t
t
Pa
Pa
Pa
Pa
Pa
e
ag
ak
Le
Figure 3-22 The graph shows the mass flow in kg/s. The total mass flow and the division: into the main cavity and out of
the main cavity.
As can be seen in Figure 3-22 is the leakage mass flows of path 2 through the sealing line form
the largest amount in total into the cavity volume. Path 1 has the largest leakage flow out of the
main cavity. The totalized leakage flow where all the paths are summed reaches almost the
0.023 kg/s. In Figure 3-23 the total leakage paths are added together with the suction mass flow
and discharge mass flow. The suction mass flow reaches over 0.085 kg/s and the leakage mass
flows less than 0.023 kg/s. The calculation is done with a clearance of 0.1 mm. Changing the
clearance will influence the amount of the leakage mass flow. Increasing the clearance will
result in an increase of the leakage mass flow.
Mass Flows
0,10
0,08
0,06
Mass flow [kg/s]
0,04
Total Mass Flow
0,02
0,00 Inflow main Cavity
-0,02 OuHlow main Cavity
-0,04
-0,06
-0,08
Suc-on Discharge Leakage Paths
Figure 3-23 The suction mass flow, the discharge mass flow and the leakage paths mass flows are illustrated in this
graph.
48
The leakages calculation within the thermodynamic model will be done through iteration. First
the thermodynamic model will be run without leakage paths. Second, the pressure of the
thermodynamic model without leakages is used as input for the thermodynamic model with
leakage, the leakages are then added as explained before. The result of the model with the
leakages is used again into the model with the leakages. This iteration process will continue
until the results have converged.
Matlab calculates these efficiencies after running the thermodynamic model in Simulink that
provides the needed inputs for the theoretical desired outputs. The Matlab script for calculation
of the efficiencies can be seen in Appendix F.
Isentropic Efficiency
The equation below is used for the calculation of the isentropic efficiency as proposed by
Zaytsev [1]. Equation 3.4.2 is substituted into equation 3.4.1.
Equation 3.4.1 shows the formula for isentropic efficiency [1]. In this equation the suction and
discharge enthalpies are respectively the same as stage 1 and 2 in equation 2.4.1. 𝑊!"#$ is the
shaft work that is needed to turn the rotors of the compressor for compression. The equation
needed to calculate the shaft work is given in 3.4.2, where ω is the shaft rotational speed in
rotations per second and b1 the number of male lobes.
The isentropic efficiency consists of the ideal power and the real power. The ideal power is the
mass flow of the compressor multiplied with the difference between the isentropic enthalpy and
the suction enthalpy. The isentropic enthalpy is extracted from Refprop with the starting values
at the suction side used to calculate the ideal discharge enthalpy with the same entropy, caused
by the ideal and therefore isentropic compression.
The suction enthalpy is extracted from the thermodynamic model. The suction enthalpy is the
enthalpy in the suction. The enthalpy is in J/kg and the mass flow in kg/s, the calculated ideal
power is in J/s.
The number of male lobes, b1, the rotation speed per second, 𝜔, and the change in pressure by
changing cavity volume calculates the real power. The number of male lobes, rotation speed
and cavity volume change are taken from the geometry model. The cavity volume is in m3 and
the pressure is in Pa, the calculated real power is in J/s. The isentropic efficiency becomes
dimensionless.
49
The higher the isentropic efficiency the less energy is lost. According to van de Bor et al. [21]
the wet screw compressor is expected to perform with an isentropic efficiency of 0.7. This is
within van de Bor’s needed conditions. During the research of Zaytsev [1] the predicted
isentropic efficiency by a model of the screw compressor was around 0.67. The actual
efficiency from experiments was less than 0.1. This difference between the two isentropic
efficiencies was caused by various of reasons; one of the reasons was that in the first model
(isentropic efficiency of 0.67) there was no labyrinth seal added to the model. Other reasons
were the position of injection (in the suction), leakages in the compressor and damage of the
lobes of the compressor, Zamfirescu [23]. The isentropic efficiency of less than 0.1 is way too
low to be a functional option for a heat pump. The leakages in the screw compressor will
influence the isentropic efficiency. For instance the leakages transport energy from the main
cavity to other cavity volumes; this increases the total energy loss within the compressor.
Volumetric Efficiency
The second efficiency that is calculated as an output is the volumetric efficiency. The
volumetric efficiency will change if there is a volume change within the cavity volume. In this
case the efficiency is mainly influenced by the leakage flows that are implemented into the
model. Without leakage paths and friction the volumetric efficiency of the homogeneous
thermodynamic model will have an ideal volume flow and the efficiency will be identical to
one. If the leakage flows are added, making use of the empirical coefficients, the volumetric
efficiency will decrease and result in a lower efficiency compared to the model without
leakages. The volumetric efficiency decreases even further if the injection is implemented in
the model, this is the case for the results by Zaytsev [1]. The volumetric efficiency is calculated
with equation 3.4.3.
𝑑𝑚
𝑣!"#$%&' 𝑑𝑡 𝑑𝑡 (3.4.3)
𝜂!"# =
𝑉!"#$%
Implementing the volumetric efficiency into Matlab, requires the following variables. The
!"
vsuction is the specific volume at the suction side in m3/kg. The !"
𝑑𝑡 is calculated by taking
the maximum mass leaving the main cavity in kg. The Videal is the maximum value of the cavity
volume in m3. This makes the volumetric efficiency dimensionless.
50
4 Validation
Validating the twin-screw compressor model is essential for knowing if the developed
integrated model is working correctly. The validation of the developed model will be done with
the model by Zaytsev [1] and with the measured data from his experimental set-up. Zaytsev’s
model case A will be used. The inputs of geometry and boundaries need to be set equal to the
inputs of the model/experiments. Both models are based on a homogeneous phase. First the
developed model will be validated with the model of Zaytsev [1] and after that, the output data
will be validated with the experimental data. Validation with the model from Zaytsev case A
and the experimental data is assumed to be adequate.
Zaytsev used the empirical flow coefficient to calculate the mass flows shown in Table 3-3: the
empirical flow coefficient for the mass flow at the suction (intake) port and discharge port and
the leakage mass flows. Zaytsev experimented with different clearances in a range of 0.06 mm
and 0.14 mm. The used clearance is 0.1 mm see Table 3-1.
In Figure 4-1 two different plots are shown calculated with the integrated model. In both plots
the pressure curve, with leakages and without leakages is plotted. Figure 4-1(a) shows the
pressure curve plotted versus time and in Figure 4-1(b) the pV-diagram of the two different
pressure curves are shown. The pV-diagram gives a view of the pressure increase compared to
the volume change in the compressor. The leakage flows between the cavities have an impact
on the pressure curve and on the efficiency of the compressor. The leakages flow into the
cavity volume during the suction phase from the leading cavity volume in compression phase.
The pressure increase starts earlier in the model with the leakages compared to the model
without the leakages. The cavity volume is full before the suction port closes and this causes
the compression to start earlier. The leakages leaking into the cavity volume during the suction
period have an effect on the amount of fluid that can be added to the cavity volume during the
suction and therefore affects the isentropic efficiency.
51
Pressure Curve: pV-Diagram:
#10 5 Data With Leakage and Without Leakage #10 5 Data With Leakage and Without Leakage
11 11
With Leakage With Leakage
Without Leakage Without Leakage
10 10
9 9
8 8
Pressure [Pa]
Pressure [Pa]
7 7
6 6
5 5
4 4
3 3
2 2
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0 0.2 0.4 0.6 0.8 1 1.2 1.4
Time [s] Cavity Volume [m 3 ] #10 -4
(a) (b)
Figure 4-1 (a) Shows the pressure curve without and with the leakages plotted to the time. (b) pV-diagram of the
pressure curve without and with leakages.
The calculated isentropic efficiency, volumetric efficiency and the needed shaft power from the
integrated model are listed in Table 4-2. As expected the isentropic efficiency and the
volumetric efficiency are higher for the model without the leakages and the needed shaft power
is lower. This difference is only caused by the leakages. The Isentropic efficiency decreases by
21%, the volumetric efficiency by 5% and the shaft power increases by 27%. In both the
models friction losses caused by mass flow are taken into account by means of the empirical
flow coefficient. The mechanical losses are not taken into account in the model.
Table 4-2 The calculated outputs of the integrated model with and without leakages.
The computational model that includes leakages used by Zaytsev [1] resulted in an isentropic
efficiency of around 0.67. Looking at the isentropic efficiency this model and the model used
by Zaytsev are in the same range.
From the boundary conditions (Table 4-1) with the geometry used by Zaytsev [1] a
compression ratio of 2.45 can be calculated. This value will be the pressure ratio that can be
reached with this geometry. Increasing the pressure ratio would require a different geometry or
reducing the discharge port area.
52
pV-Diagram:
#10 5 "Case A" calculated and measured
10
Calculated "Case A"
Measured "Case A"
Pressure [Pa]
7
3
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Cavity Volume [m 3 ] #10 -4
Zaytsev produced a pressure and volume curve as his main outcome. Figure 4-3 shows the
comparison between the pressure curve generated by Zaytsev (calculated and measured) and
the pressure curves generated with the current integrated model (with and without the
leakages). Figure 4-3(a) shows the pressure curve plotted against the time and in Figure 4-3(b)
the pV-diagram is plotted.
9 9
8 8
Pressure [Pa]
Pressure [Pa]
7 7
6 6
5 5
4 4
3 3
2 2
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0 0.2 0.4 0.6 0.8 1 1.2 1.4
Time [s] Cavity Volume [m 3 ] #10
-4
(a) (b)
Figure 4-3 (a) Shows the pressure curve without leakages, with the leakages, calculated and measured pressure by
Zaytsev [1] plotted to the time. (b) pV-diagram of the pressure curve without leakages, with leakages, calculated and
measured pressure by Zaytsev [1].
The maximum pressure of the integrated model increases further during a longer period of time
compared to the calculated results by Zaytsev [1]. This can be caused by insufficient mass flow
leaving the compressor. Adjusting the discharge port area would be an option to solve this
problem. It can be observed from Figure 4-3 that the cavity volume of the integrated model is
smaller compared to the cavity volume used by Zaytsev [1], while the inputs of the models are
equal. More detail on adjusting the model to the experimental data is given in section 4.3.
The integrated models and the calculated model case A by Zaytsev [1] are homogeneous. The
difference between the integrated models and the models of case A, calculated and measured,
53
are a labyrinth seal, mechanical losses and the injection of the liquid into the compressor (has a
large impact on the volumetric efficiency), see section 4.3.
The integrated model is validated to the calculated model of Zaytsev [1]. As can be seen in
Figure 4-2 the calculated model and measured results differ, this difference will be evaluated in
the next section, together with the potential causes. The effect of a too small discharge port and
a smaller cavity volume will be evaluated with the integrated model of this research.
8
Pressure [Pa]
2
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Cavity Volume [m 3 ] #10
-4
Figure 4-4 Shows the pressure results of the integrated model with leakages, the calculated results (Zaytsev [1]) and the
experimental measured results (Zaytsev [1]).
The integrated model shows a higher maximum pressure in the pV curve compared to the
measured case A. To adapt the integrated model to the measured case A, as suggested above, it
is required to decrease the maximum pressure and allow the flow to increase for the same
amount of mass out of the compressor. The total mass balance remains unchanged. This can be
achieved by for instance increasing the discharge port area.
Adapting the empirical flow coefficient compensates the flow behaviour in the compressor due
to leakages and irreversibilities. Adapting these inputs results in modified inputs of the
integrated model compared to the inputs used by Zaytsev [1]. In Figure 4-5 the original
discharge port area (a) and the adapted discharge port area (b) are shown. The adapted
maximum discharge port area has been increased by 41%, which will result in the same mass
through the discharge port at a lower pressure difference.
54
Discharge Port Area Increased Discharge Port Area
700 700
600 600
500 500
400 400
Area [mm2 ]
Area [mm2 ]
300 300
200 200
100 100
0 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time [s] Time [s]
(a) (b)
Figure 4-5 (a) Original discharge port area used in section 4.2. (b) The adapted and increased discharge port area.
What also has been observed in Figure 4-4 is that the cavity volume in case A from Zaytsev [1]
is larger than the cavity volume of the integrated model in this research. This is caused by the
geometry model that is used. To solve this difference the cavity volume of the integrated model
needs to be increased. Changing the inputs of the geometry model can result in an increase of
the cavity volume; in this case an increase of the rotor length. The compressor length will be
increased from 172.5 mm to 175.45 mm to increase the maximum cavity volume from
1.3129 ∙ 10!! m3 to 1.33524 ∙ 10!! m3 which is comparable with the maximum cavity volume
in case A used by Zaytsev [1].
Adapting the model with the two changes discussed above will result in the pressure curve,
shown in Figure 4-6. The model is calculated first without leakages to adjust the curve to the
measured results before adding the leakages. In this way calculation time is saved. The curve
without leakages has been adapted with an increased discharge port area and has reached
similar values for cavity volume and pressure compared to the calculated case A by Zaytsev
[1].
pV-Diagram:
#10
5 Adapting the integrated model to the calculated results "Case A"
11
Without Leakage
Without leakage adapted discharge/compressor length
Calculated "Case A"
10 Measured "Case A"
8
Pressure [Pa]
3
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Cavity Volume [m 3 ] #10
-4
Figure 4-6 Adapted integrated model of this research closer to the calculated pressure curve used by Zaytsev [1] (The
integrated model is without the leakages and the calculated “Case A” by Zaytsev is with leakages).
55
Zamfirescu et al. [23] carried out research on improving the prediction of the calculated results
to the measured result used by Zaytsev [1] see Figure 4-2. Comparing these two results it needs
to be kept in mind that the calculated results are from a homogeneous model and the measured
is heterogeneous (Zaytsev [1]). Zamfirescu et al. [23] stated that not all leakage paths were
considered and an underestimation of clearance areas was made. Also it is mentioned that the
predicted maximum pressure at the discharge end phase is higher than the maximum pressure
of the experimental measurements and that a faster increase of the pressure at the beginning of
the compression phase occurs by the calculated model of Zaytsev [1]. Varying the empirical
flow coefficient of the leakages, Zamfirescu et al. [23] adjusted the calculated model to the
measured model. The empirical coefficient for the leakages has been adapted from 0.7 to 1.2.
The result after adjusting the empirical flow coefficient is shown in Figure 4-7.
pV-Diagram:
#10 5 Adapting the integrated model to the measured results "Case A"
11
With leakage adapted discharge/compressor length and Empirical Coefficient of 1.2
Calculated "Case A"
10 Measured "Case A"
7
Pressure [Pa]
1
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Cavity Volume [m 3 ] #10 -4
Figure 4-7 The adapted integrated model to the measured results of the experiments used by Zaytsev [1].
As can be seen in Figure 4-7 the pressure curve of this research is close to the measured results
of the experiments used by Zaytsev [1]. The pressure peak at the discharge phase is still too
high and the pressure increases slightly faster than the measured results. This can be the
explanation for the isentropic efficiency difference. The isentropic efficiencies are listed in
Table 4-3. Decreasing the pressure curve peak to get the curve even closer to the measured
values can be done by making use of: increasing the discharge port area and increasing the
empirical flow coefficient of the discharge mass flow calculation. The empirical flow
coefficient needs to be established with experiments, this would be a subject of further
research.
Table 4-3 The isentropic efficiency and shaft power of the adapted model.
56
5 Model Results and Discussion
The validation of the model has been carried out in chapter 4 and the integrated model can now
be used for further research. In this chapter a number of variations will be evaluated illustrative
for the possibilities of the integrated model, a step-up for further development. For the
evaluation of variations, the integrated model of section 4.3 will be used however without
leakages. The integrated model from chapter 4.3 with the geometry of Zaytsev [1] has been
adapted with the boundaries used by van de Bor et al. [21]. This has been done to be able to
evaluate the integrated model consistent with the requirements of this research as explained in
section 1.5.
After adapting the model to the boundary conditions by van de Bor et al. [21], the rotor length,
the discharge port area and the vapour quality will be varied to evaluate what the influences are
of these parameters on the compressor and the isentropic efficiency. As the variations are
illustrative for the use of the model, it should be realised that these subjects require further
investigation.
5.1 Leakages
Evaluation will be done without the leakages. Leakages are amongst others determined by the
clearances and the empirical flow coefficients. The influence of the leakages is that the
isentropic efficiency decreases drastically with the larger size of the clearances and the high
leakage empirical flow coefficients of Zaytsev [1] and Zamfirescu et al. [23]. The clearances
and the leakage empirical flow coefficients are dependent on the design and the fabrication of
the screw compressor and selection of fluid. The empirical flow coefficients need to be
obtained through experiments. To allow a clear evaluation of the integrated thermodynamic
model the leakages have been taken out. Additional advantage is that less calculation time is
needed for the evaluation without leakages.
The pV-diagram and temperature curve with the boundary conditions by van de Bor et al. [21]
are shown in Figure 5-1.
57
pV-Diagram: Temperature Curve:
#10 4 Boundary Conditions van de Bor et al. [21] Boundary Conditions van de Bor et al. [21]
6 345
5.5
340
4.5 335
Temperature [K]
Pressure [Pa]
4
330
3.5
3 325
2.5
320
2
1.5 315
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
Cavity Volume [m 3 ] #10
-4
Time [s]
(a) (b)
Figure 5-1(a) pV-diagram and (b) Temperature curve, of the compressor with the geometry used by Zaytsev[1] and the
boundary conditions of van de Bor et al. [21], without leakages.
As can be observed in Figure 5-1, with the defined input values, the discharge pressure has
been adjusted manually to 0.55 bar. The pressure has been established through iteration from
manual adaption of the discharge pressure in the integrated model. The pressure is the highest
possible discharge pressure for the selected boundary conditions where back flow from the
discharge system into the compressor is close to zero. This means that the compressor reaches
this pressure by compression. This pressure is in the same range when using the compression
ratio as has been used with the geometry by Zaytsev see section 4.1. In Figure 5-1(b) the
temperature curve is shown: the starting temperature is 318.15 K (45 °C) and the temperature
reaches 340 K (66.9 °C) during compression. The 340 degrees is much lower than the required
388.15 K (115 °C, section 1.5). The calculated isentropic efficiency and shaft power are listed
in Table 5-1. The shaft power is very low compared to the values in chapter 4. This means less
power is needed during the compression. In this case the pressure difference is just 0.35 bar,
which is a small pressure difference according to the pressure difference during the validation
in chapter 4 of around 5.3 bar. This explains the lower needed shaft power in this case with the
boundaries of van de Bor et al. [21].
Table 5-1 Efficiency and shaft power of the compressor using the geometry used by Zaytsev [1] and the boundaries of
van de Bor et al. [21].
To achieve the required results of the compressor model for the case of van de Bor et al. [21],
some research and adaption of the input values have to be done. For this case more pressure
difference is needed to achieve the required higher temperature of 388.15 K (115 °C). This can
be achieved by for instance adjusting the geometry input.
58
Rotor Length
The rotor length has been adapted during the validation in chapter 4. By varying the rotor
length in the model while keeping the other inputs constant, the effect of the rotor length on the
efficiency can be described. Increasing the rotor length from 130 mm to 230 mm in steps of 10
mm will have influence on the cavity volume of the compressor since the rotor length is a part
of the cavity volume calculation. In Figure 5-2 the different cavity volumes related to the
different rotor lengths are plotted versus time. In this figure it can be observed that by
increasing the rotor length it results in an increase of the cavity volume as expected.
Rotor Length:
#10 -4 Cavity Volume
1.8
130 mm
140 mm
1.6 150 mm
160 mm
170 mm
180 mm
1.4 190 mm
200 mm
210 mm
1.2 220 mm
230 mm
Cavity Volume [m 3 ]
0.8
0.6
0.4
0.2
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
Time [s]
The pV-diagram with the different rotor lengths is shown in Figure 5-3. In this pV-diagram it
can be seen that the longest rotor length, 230 mm, has the highest pressure increase and the
shortest rotor length, 130 mm, the lowest pressure increase. This pressure increase is related to
the cavity volume. The cavity volume is the representation of the volume in the cavity of the
conservation equations in section 2.3. An increase of the cavity volume results in a steeper
volume curve and an increase of the mass that can be compressed by the compressor. The
steeper volume curve results in more pressure increase during the compression.
Rotor Length:
#10
4 pV-Diagram
6.5
130 mm
140 mm
6 150 mm
160 mm
170 mm
5.5 180 mm
190 mm
200 mm
5 210 mm
220 mm
230 mm
4.5
Pressure [Pa]
3.5
2.5
1.5
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Cavity Volume [m 3 ] #10 -4
59
In Figure 5-4 the isentropic efficiency is plotted versus the rotor length. As can be seen,
increase of the rotor length results in an increase in efficiency. The isentropic efficiency
increase between the rotor length of 130 mm and 230 mm is 64 %.
Rotor Length:
Rotor Length - Isentropic Efficiency
0.65
0.6
0.55
Isentropic Efficiency
0.5
0.45
0.4
0.35
130 140 150 160 170 180 190 200 210 220 230
Rotor Length [mm]
Figure 5-4 Isentropic efficiency plotted against the rotor length in mm.
The relation between the isentropic efficiency and the maximum pressure and maximum
temperature are plotted in Figure 5-5. As can be seen in the two plots there is an ‘asymptotic’
increase between the pressure/temperature and the isentropic efficiency.
342
6
5.9 341.5
5.8
341
5.7
340.5
5.6
5.5 340
0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.35 0.4 0.45 0.5 0.55 0.6 0.65
Isentropic Efficiency Isentropic Efficiency
(a) (b)
Figure 5-5(a) Pressure and the isentropic efficiency, (b) Temperature and the isentropic efficiency, with the different
rotor length.
These results show that increasing the rotor length will increase the isentropic efficiency; all the
inputs and boundary conditions are kept constant. The length of the rotors also influences some
of the leakage paths, which will be enlarged by the increasing rotor length. For example
increase of the rotor length will increase the contact line and the sealing line leakage path areas
and will have a negative effect on the isentropic efficiency of the compressor, as an increased
leakage area will result in an increased leakage mass flow.
60
part is taken, for example 1/3, of the maximum cavity volume that is used in this geometry. The
maximum cavity volume in this ‘case’ is 1.335 ∙ 10!! m! . In Figure 5-6 all the different
discharge port areas are presented that are used during the evaluation. The discharge port areas
vary from 1/3 to 1/7 of the cavity volume in steps of 1/0.5.
Discharge Port Areas
900
1/3 * V max
1/3.5 * V max
800
1/4 * V max
1/4.5 * V max
1/5 * V max
700
1/5.5 * V max
1/6 * V max
500
400
300
200
100
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time [s]
As can be seen in Figure 5-6 with the increase of the discharge port area, the opening of the
discharge port shifts earlier in time. The closing of the discharge port area is for all the
discharge port areas constant. This will have effect on the pressure increase in the compressor.
The pressure curves of the different discharge port areas are shown in a pV-diagram, Figure
5-7. As can be seen in this figure, the smaller the discharge port area the higher the maximum
pressure in the compressor will become. The discharge port areas of 1/3 to 1/4 have a steeper
pressure curve; the increase of the pressure takes place at the discharge port opening. The
discharge pressure in this case is set on 0.55 bar. The early opening of the discharge port results
in advanced pressure increase and is not done by the compression but caused by back flow. It
can be concluded that these discharge port areas are too large and open too early for this
compression case. The other smaller discharge port areas 1/4.5 to 1/7 are small enough and
open later in time such that the compressor is capable of compressing the fluid to above 0.55
bar before opening of the discharge port. It is a possibility to make the discharge pressure equal
to the maximum pressure in the compressor. This could lead to an optimum between the
discharge port area and discharge pressure.
Discharge Port Area:
#10 4 pV-Diagram
7
1/3 * V max
1/3.5 * V max
1/4 * V max
6 1/4.5 * V max
1/5 * V max
1/5.5 * V max
1/6 * V max
5 1/6.5 * V max
1/7 * V max
Pressure [Pa]
1
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Cavity Volume [m 3 ] #10
-4
61
The relation between the different discharge port areas and the isentropic efficiency is plotted
in Figure 5-8. In this figure it can be observed that there is an optimal peak in the isentropic
efficiency, between the discharge port of 1/4 and 1/4.5 of the cavity volume. This is the
moment that the port opens when the pressure has reached the discharge pressure of 0.55 bar,
Figure 5-7.
Discharge Port Area:
Discharge Port Area - Isentropic Efficiency
0.565
1/3 * V max
1/3.5 * V max
0.55
0.545
0.54
0.535
0.53
1.5 2 2.5 3 3.5 4 4.5
Discharge Port Area [mm 2 ] #10 4
Figure 5-8 The discharge port area plotted to the isentropic efficiency.
From this evaluation it can be concluded that per case a specific discharge port area is needed.
The discharge port should open at the moment that the compressor reaches the discharge
pressure. Adapting each discharge port to the required case will increase the isentropic
efficiency of the compressor. In this case using the correct discharge port area the isentropic
efficiency can increase by 3.7 %.
This analysis is with a constant discharge pressure of 0.55 bar with different discharge port
areas. Changes in the discharge pressure in relation with the discharge port area can be
considered.
Vapour Quality
The vapour quality is a parameter that is expected to have influence on the wet twin-screw
compressor in view of the 2-phase fluid used. The vapour quality, here defined as the fraction
of vapour of the ammonia/water mixture, is going to be varied and will be discussed in this
chapter.
The discharge pressure was set at 0.55 bar, in this analysis the discharge pressure will be
adapted to the pressure curve with the varying vapour quality, this will be explained further on.
With the used geometry and suction pressure of 0.2 bar the discharge pressure will be in line
with the compression ratio as explained in chapter 4.1 and will vary slightly.
The discharge pressure has to be established manually by adapting the discharge pressure that
is a constant in the integrated model. The model needs to be run several times to verify the
maximum pressure at opening of the discharge. With an increase of the maximum pressure,
flow from the discharge system flows back in the discharge port, this indicates that the
discharge pressure is too high for this case. The discharge pressure is set at the moment that
there is no back flow into the compressor; this is the ‘optimal’ discharge pressure.
As can be seen Figure 5-9, the vapour quality has been varied from 0.5 to 1. Varying the
vapour quality is done by adapting the suction temperature to the suction pressure of 0.2 bar.
62
For each vapour quality a different suction temperature is used and is listed in Table 5-2. In this
table also the suction pressure and the respective optimal discharge pressure to the vapour
quality is added.
Table 5-2 The suction temperature, suction pressure and related discharge pressure.
The pressure curves and temperature curves of the varied vapour quality are shown in Figure
5-9. The pressure curve is plotted together with the cavity volume curve in the pV-diagram (b)
and the temperature is plotted versus time (b).
Vapour Quality: Vapour Quality:
#10 4 pV-Diagram Temperature versus Time
6 410
0.5 0.5
0.625 0.625
5.5 0.75 400 0.75
0.875 0.875
1 1
390
5
380
4.5
370
Temperature [K]
Pressure [Pa]
360
3.5
350
3
340
2.5
330
2
320
1.5 310
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
Cavity Volume [m 3 ] #10 -4 Time [s]
(a) (b)
Figure 5-9 (a) pV-diagram of the varied vapour quality. (b) Temperature versus Time of the varied vapour quality.
From Figure 5-9(b) it can be observed that the temperature of the discharge increases less for
higher vapour quality. The model shows that the optimal discharge pressure reduces with
increased vapour quality, Figure 5-9(a). The results for a vapour quality of 1 deviates from the
results at lower vapour qualities. The phenomena cannot be fully explained and will require
further research. The various interpretations of these phenomena will be reflected below.
With increase of the vapour quality, the suction temperature is in equilibrium with the suction
pressure that has been set constant at 0.2 bar. This results in a higher temperature in the suction
with increasing vapour quality. This temperature adjustment also results in a smaller
temperature difference between suction and discharge. With a vapour quality of 1 a much
higher optimal discharge temperature is observed. This can be explained by the absence of the
liquid. However the optimal discharge pressure result seems lower than expected. The
homogeneous model is developed based on a 2-phase system. The homogeneous model is
limited by the fact that the pressure and temperature are the same for both the two phases (gas
and liquid). A heterogeneous model calculates the temperature of liquid and gas phase
individually with the same pressure. The difference allows a different temperature for each
63
phase throughout the compressor and can lead to a different result. The homogeneous model
could be the limitation and therefore using a heterogeneous model could solve the effect of the
unexpected results.
Conclusion
The evaluation on the rotor length, discharge port area and the vapour quality shows there is
room for optimisation using this integrated model. The rotor length and the discharge port area
each have a different influence on the compressor efficiency. The increase of rotor length and
the optimal discharge port area related to the discharge pressure has significant effect on the
isentropic efficiency. The results of these two varied parameters conclude that each parameter
needs to be adjusted to the required boundary conditions to achieve/design the optimal screw
compressor for the required application. For the third parameter, the vapour quality, the
unexpected phenomena need to be understood before coming to a conclusion. These three
analyses are examples and all three need further research to evaluate the relations between the
parameters.
64
6 Conclusions and Recommendations
This research describes the development of a thermodynamic model with integrated geometry
model of a wet twin-screw compressor. The objective of this research as described in 1.5 is: ‘to
develop a dynamic model that describes the thermodynamic process and includes the inputs
from a twin-screw compressor geometry model’. In this chapter the conclusions and
recommendations will be discussed.
6.1 Conclusions
A thermodynamic model for a wet twin-screw compressor has been developed in Simulink and
is based on a homogeneous model by Zaytsev [1]. The model has been transformed from a
rotation-based model to a time-based model, integrated with the geometry model developed by
Zaytsev [1], and includes the physical properties continuously calculated by Refprop via
Fluidprop. Chapter 4 discusses how validation of the developed integrated model was carried
out. The integrated model is compared with the homogeneous model by Zaytsev [1] and the
measured data from the experiment done by Zaytsev [1]. It can be concluded that this
integrated model is working properly for further use in optimising the compressor.
The thermodynamic model has been extended with leakage paths based on the work by Zaytsev
[1] and the work of Zamfirescu et al. [23]. Zamfirescu et al. [23] did research on the calculated
and measured data used by Zaytsev [1], which resulted in the use of a higher empirical flow
coefficient. During the validation of the developed integrated model it has been concluded that
the leakages are dependent on the clearances and the empirical flow coefficients of the
geometry model. Data and information for both these values are too scarce to provide adequate
values to describe the leakages in the actual screw compressor.
The thermodynamic model conditions have been set on the boundary conditions described by
van de Bor et al. [21] as described by the research objectives. The geometry of the screw
compressor is based on the inputs used by Zaytsev [1]. With this combination of inputs and
boundaries a discharge pressure of 0.55 bar, discharge temperature of 66.9 °C and an isentropic
efficiency of 0.56 have been calculated, which represents the integrated model without
leakages. The isentropic efficiency decreases even further when the leakages are taken into
account. As a result, it can be concluded that the required isentropic efficiency of 0.7 and the
discharge temperature of 115 °C cannot be reached for this set of inputs.
Additional to the development of the integrated model an evaluation has been done with the
boundary conditions by van de Bor et al. [21] and the inputs from the geometry used by
Zaytsev [1]. The evaluation has been done by varying the rotor length, discharge port area and
vapour quality. The evaluation should be seen as an example for the possibilities of varying the
parameters and use of the integrated model. This evaluation clearly shows that these three
parameters can be optimized to improve (the efficiency of) the screw compressor. Increasing
the rotor length can have a positive effect on the increase of the pressure, temperature and the
isentropic efficiency. The discharge port area needs to be chosen such to obtain the most
optimal area and opening at the correct time to increase the isentropic efficiency and the
increase of the pressure and temperature during the compression. The result for the varied
vapour quality will require additional thoughts to explain within constraints of the integrated
model. The results of the analysis have been modelled without inclusion of the leakages.
Including the leakages will likely result in a different outcome. The next step in optimising the
screw compressor should be to add the correct leakages, the correct clearances, and the
empirical flow coefficient.
65
The integrated model shows that variations can be evaluated both in geometry as well as in
thermodynamic behaviour. As a result, it can be concluded that for every new situation of
boundary conditions a compressor with different specifications and geometry is needed.
6.2 Recommendations
For every CRHP situation an optimal screw compressor geometry can be defined with the help
of the model developed for this research. This will require restructuring of the parameters and
their influence on the compressor thermodynamics, efficiency and geometry. The efficiency as
stated by van de Bor cannot be achieved within the set boundary conditions. Further research
needs to be done and adaptions need to be made to the screw compressor in order to match the
requirements of van de Bor for use of the screw compressor in the CRHP.
In this thesis the leakages have been modelled consistent with previous work (Zaytsev [1]).
Further work research is needed on the clearances within the compressor, as well as on the
empirical flow coefficients. These two inputs have a large influence on the compressor-specific
model and the values therefore need to be adjusted to the compressor that is going to be used.
Investigation into the tightest clearance that can be achieved as well as determining the actual
empirical flow coefficient from experiments is needed. Having the actual clearances and the
empirical flow coefficients corresponding to the actual screw compressor the leakages can be
adapted into the model and the optimisation becomes more realistic.
In this research a thermodynamically homogeneous model has been developed. To develop the
model closer to the actual compressor a heterogeneous model should be considered, to achieve
more realistic results. This model will need to include the changing equilibrium conditions
between the liquid and the gas phase throughout suction, compression and discharge, and will
result in a different temperature at constant pressure for each phase.
The labyrinth seal, heat exchange between the compressor and the fluid, and the mechanical
friction have not been modelled. Adding these three parameters will result in a more realistic
model and more realistic experiments as well.
With the integrated model, working in current modelling software, the geometry and
thermodynamic aspects can be further evaluated. Variations in rotor length and discharge port
area have been evaluated as illustrations to show how the integrated model can be used.
Adapting the rotor geometry to the boundary conditions is a real challenge and needs more
insight into the dependency between the parameters. Important for this area of improvement is
that the discharge pressure needs to be adjusted together with the change in geometry, and is
expected to result in higher pressures than modelled in section 5.3.
66
Bibliography
[1] D. Zaytsev, ‘Development of wet compressor for application in compression-resorption
heat pumps’, TU Delft, PhD Thesis, 2003.
[2] World Commission on Environment and Development, ‘Report of the World
Commission on Environment and Development: Our Common Future (The Brundtland
Report)’, Med. Confl. Surviv., vol. 4, no. 1, p. 300, 1987.
[3] UNFCCC, ‘Climate Change Newsroom from the UNFCCC’. [Online]. Available:
http://newsroom.unfccc.int. [Accessed: 01-Apr-2016].
[4] COP21, ‘United nations conference on climate change’. [Online]. Available:
http://www.cop21.gouv.fr. [Accessed: 01-Apr-2016].
[5] NASA, ‘Data: GISS Surface Temperature Analysis’, http://data.giss.nasa.gov/gistemp/.
[Online]. Available: http://data.giss.nasa.gov. [Accessed: 01-Apr-2016].
[6] IEA, ‘Energy and climate change’, World Energy Outlook Spec. Rep., pp. 1–200, 2015.
[7] Overheid, ‘Gevolgen klimaatverandering’. [Online]. Available:
https://www.rijksoverheid.nl. [Accessed: 01-Apr-2016].
[8] KNMI, ‘Effecten klimaatverandering steeds duidelijker’. [Online]. Available:
https://www.knmi.nl. [Accessed: 01-Apr-2016].
[9] PBL, ‘Planbureau voor de Leefomgeving’. [Online]. Available: http://www.pbl.nl.
[Accessed: 01-Apr-2016].
[10] IPCC, ‘Intergovernmental Panel on Climate Change’. [Online]. Available: http://ipcc.ch.
[Accessed: 01-Apr-2016].
[11] J. G. J. Olivier, G. Janssens-Maenhout, M. Muntean, and J. A. H. W. Peters, ‘Trends in
global CO2 emissions 2015 report’, 2015.
[12] EDGAR, ‘Emission Database for Global Atmospheric Research’. [Online]. Available:
http://edgar.jrc.ec.europa.eu. [Accessed: 01-Apr-2016].
[13] IEA, ‘Co-generation and renewables, Solutions for a low-carbon energy future.’, 2011.
[14] W. G. Haije, J. B. J. Veldhuis, K. Brandwagt, and J. W. Dijkstra, ‘Solid Sorption Based
Heat Transformers for Application in Industrial Processes : a Techno-Economic
Evaluation’, Proc. Int. Sorption Heat Pump Conf., pp. 469–472, 2002.
[15] S. Spoelstra, W. G. Haije, and J. W. Dijkstra, ‘Techno-economic feasibility of high-
temperature high-lift chemical heat pumps for upgrading industrial waste heat’, Appl.
Therm. Eng., vol. 22, pp. 1619–1630, 2002.
[16] M. J. Moran and H. N. Shapiro, Fundamentals of Engineering Thermodynamics. Wiley,
2006.
[17] I. Dincer and M. Kanoglu, Refrigeration Systems and Applications. Wiley, 2010.
[18] B. Mongey, N. J. Hewitt, J. T. Mcmullan, P. C. Henderson, and G. A. Molyneaux,
‘Performance trends and heat transfer considerations in an ammonia-water resorption
cycle’, Int. J. Energy Res., no. July 2000, pp. 41–51, 2001.
[19] L. C. M. Itard, ‘Wet compression versus dry compression in heat pumps working with
pure refrigerants or non-azeotropic mixtures’, Int. J. Refrig., vol. 18, no. 7, pp. 495–504,
1995.
[20] D. M. van de Bor, C. A. Infante Ferreira, and A. A. Kiss, ‘Optimal performance of
compression–resorption heat pump systems’, Appl. Therm. Eng., vol. 65, no. 1–2, pp.
219–225, 2014.
[21] D. M. van de Bor, C. A. Infante Ferreira, and A. A. Kiss, ‘Low grade waste heat
recovery using heat pumps and power cycles’, Energy, vol. 89, pp. 864–873, 2015.
[22] L. C. M. Itard, ‘Wet compression-resorption heat pump cycles: thermodynamic analysis
and design’, TU Delft, PhD Thesis, 1998.
[23] C. Zamfirescu, N. Nannan, M. Marin, and C. A. Infante Ferreira, ‘Oil free two phase
ammonia ( water ) compressor’, TU Delft, Final Report, 2004.
67
[24] A. Lysholm and E. Al., ‘Rotary Compressor’, 2,111,568, 1938.
[25] J. Wennemar, ‘Dry Screw Compressor Performance and Application Range’, MAN
Turbo AG, Germany, 1985.
[26] I. M. Arbon, The design and application of rotary twin-shaft compressors in the oil and
gas process industry. London : Mechanical Engineering Publications, 1994.
[27] N. Stosic, I. Smith, and A. Kovacevic, Screw Compressors: Mathematical Modelling
and Performance Calculation. Springer, 2005.
[28] D. Zaytsev and C. A. Infante Ferreira, ‘Profile generation method for twin screw
compressor rotors based on the meshing line’, Int. J. Refrig., vol. 28, no. 5, pp. 744–755,
2005.
[29] C. A. Infante Ferreira, C. Zamfirescu, and D. Zaytsev, ‘Twin screw oil-free wet
compressor for compression-absorption cycle’, Int. J. Refrig., vol. 29, pp. 556–565,
2006.
[30] I. A. Sakun, ‘Screw Compressors’, Leningrad, Mashinostroenie.1970 (In Russian). .
[31] C. X. You, ‘A theoretical study of rotor forces and torques in helical twin screw
compressors’, University of Strathclyde, B.Sc, M.Sc, 1994.
[32] ‘Refprop 9.0’. US Department of Commerce, NIST, E.W. Lemmon et al., 2010.
[33] ‘FluidProp 2.4’. TU Delft, T. van der Stelt, P. Colonna, 2010.
[34] ‘MATLAB 2015a’. The MathWorks Inc., M.A. Natick, 2015.
[35] D. G. Deng and P. C. Shu, ‘Rotary Compressors’, Mechanical Industry Press, Beijing
(In Chinees), 1988. .
[36] L. Rinder, Schraubenverdichter (in German). Springer, 1979.
[37] J. Prins and C. A. Infante Ferreira, ‘Feasibility and design of leakage experiments on a
running twin screw compressor’, ‘From Thermo-economics to Sustanability’ Ed. by
Hirs, G.G. Part 2 ECOS 2000 Proceedings, Univ. Twente, Netherl, pp. 869–880, 2000.
[38] J. S. Fleming and Y. Tang, ‘The analysis of leakage in a twin screw compressor and its
application to performance improvement’, IMechE 1995, vol. 209, pp. 125–136, 1995.
[39] P. J. Singh and J. L. Bowman, ‘Calculation of blow-hole area for screw compressors’,
Int. Compress. Eng. Conf., vol. 786, pp. 938–948, 1988.
68
Appendices
Appendix A: Envelope Method – Rotor Element Calculation
Figure A-1 The angles and segments of the male and female rotor [30]
𝑟𝑜𝑐 = (𝑟1ℎ 2 + 𝑘𝑟 ! )
𝑘 = 1 + 𝑖21
Area D1C1
𝜃!"# = arccos 𝑅! ! − 𝑟 ! − 𝑟!" ! /(2 ∙ 𝑟 ∙ 𝑟!" )
𝜃!" = arccos 𝑟𝑜𝑐 ! − 𝑟 ! − 𝑟!" ! /(2 ∙ 𝑟 ∙ 𝑟!" )
𝑟
𝐷! 𝐶! = − ∙ (𝑟!" ∙ 𝑠𝑖𝑛 𝜃!" − 𝑠𝑖𝑛 𝜃!"# + 𝑟 ∙ 𝜃!" − 𝜃!"# )
2
Area C1A1
𝜃!"# = 𝑖!" ∙ arccos 𝐷! + 𝑟!" ! − 𝑟𝑜𝑐 ! /(2 ∙ 𝐷 ∙ 𝑟!" )
𝜃!" = 𝑖!" ∙ arccos 𝐷! + 𝑟!" ! − 𝑟!" ! /(2 ∙ 𝐷 ∙ 𝑟!" )
1 1+𝑘
𝐶! 𝐴! = 𝜃!" − 𝜃!"# ∙ 𝐷! + 𝑟!" ! ∙ 𝑘 − ∙ 𝐷 ∙ 𝑟!"
2 𝑖!"
Area A1I1
69
𝑟!
𝜃!"# = arccos
(2 ∙ 𝑟!" )
𝜃!" =0
𝑟!
𝐴! 𝐼! = ∙ (𝑟!" ∙ (sin( 𝜃!" ) − sin( 𝜃!"# )) − 𝑟! ∙ (𝜃!" − 𝜃!"# )
2
Area I1L1
𝜃!"# = −𝜃! − 𝜃!
𝜃!" = −𝜃! − 2 ∙ 𝛼! + 𝜃!
1
𝐼! 𝐿! = (𝑟!" − 𝑟! )! ∙ (𝜃!" − 𝜃!"# )
2
Area L1F1
𝜃!"# = 0
𝑟!
𝜃!" = 𝑎𝑟𝑐𝑜𝑠
(2 ∙ 𝑟!" )
𝑟!
𝐿! 𝐹! = ∙ (𝑟! ∙ 𝜃!" − 𝜃!"# − 𝑟!" ∙ (sin 𝜃!" − sin 𝜃!"# )
2
Area F1D1
𝜃!"# = 𝑖!" ∙ arccos 𝐷! + 𝑟!" ! − 𝑟!" ! /(2 ∙ 𝐷 ∙ 𝑟!" )
𝜃!" = 𝑖!" ∙ arccos 𝐷! + 𝑟!" ! − 𝑅! ! /(2 ∙ 𝐷 ∙ 𝑟!" )
1 𝜃!" − 𝜃!"# ∙ −𝐷! − 𝑟!" ! ∙ 𝑘 + (1 + 𝑘)
𝐹! 𝐷! =
2 𝑖!" ∙ 𝐷 ∙ 𝑟!" ∙ (sin 𝜃!" ∙ 𝑖!" − sin 𝜃!"# ∙ 𝑖!" )
70
Appendix B: Conservation Equations of the Homogeneous Model
The conservation equations of the homogeneous model are based on Zaytsev [1]. In this
homogeneous model the liquid/vapour phases are in equilibrium during the compression cycle.
In this appendix a detailed derivation of the conservation equations that have been applied in
the time dependent model is given. The following three equations B.1, B.2 and B.3 have been
!" !! !"
rewritten to time dependent conservation equations !" , !"! and !" .
In the equations above v is the specific volume and is a function of the pressure p, temperature
T and the overall concentration x0 . The equation B.4 is differentiated by the time and shown in
B.5.
𝑑𝑚 1 𝑑𝑉 𝑉 𝜕𝑣 𝑑𝑝 𝜕𝑣 𝑑𝑇 𝜕𝑣 𝑑𝑥!
= − + + (B.5)
𝑑𝑡 𝑣 𝑑𝑡 𝑣 ! 𝜕𝑝 !,! 𝑑𝑡 𝜕𝑇 !,! 𝑑𝑡 𝜕𝑥! !,!
𝑑𝑡
Equation B.1 is differentiated by the time. It becomes the mass flow equation of the inflow and
outflow of the compression:
! !
𝑑𝑚 𝑑𝑚!" 𝑑𝑚!"#
= − (B.6)
𝑑𝑡 𝑑𝑡 𝑑𝑡
!!! !!!
The mass conservation equations B.6 and B.5 can be combined after rearranging the terms to
!"
!"
and becomes:
71
! !
𝑑𝑝 1 𝑣 𝑑𝑚!"# 𝑑𝑚!" 1 𝑑𝑉 𝜕𝑣 𝑑𝑇 𝜕𝑣 𝑑𝑥! (B.7)
= − + − −
𝑑𝑡 𝜕𝑣 𝑚 𝑑𝑡 ! 𝑑𝑡 ! 𝑚 𝑑𝑡 𝜕𝑇 !,! 𝑑𝑡 𝜕𝑥! !,!
𝑑𝑡
!!! !!!
𝜕𝑝 !,!
The differentiated conservation equation of the ammonia mass B.2 written as time dependent:
! !
𝑑𝑥! 𝑑𝑚 𝑑𝑚!" 𝑑𝑚!"#
𝑚 + 𝑥! = 𝑥!!",! − 𝑥! (B.8)
𝑑𝑡 𝑑𝑡 𝑑𝑡 ! 𝑑𝑡 !
!!! !!!
The equation for the concentration change over time is obtained after adding and rearranging
!"
equations B.6 and B.8. These two equations result in the equation !"! used for calculating the
change in concentration of the ammonia/water mixture, B.9.
! !
𝑑𝑥! 1 𝑑𝑚!" 𝑑𝑚!"
= 𝑥!!",! − 𝑥! (B.9)
𝑑𝑡 𝑚 𝑑𝑡 ! 𝑑𝑡 !
!!! !!!
The conservation equations have been rewritten to make the ammonia/water concentration
!"
dependent over time. This means that !"! will change over time when the concentration of the
inflow of the compressor differs from the outflow of the compressor. That situation will for
example occur if liquid of the fluid is injected downstream of the inlet stream in the
compressor.
The energy conservation equation B.3 is differentiated to the time and for the enthalpy change
!" !! !"
!"
= 𝑚 !" + ℎ !" is added, shown in B.10.
! !
𝑑ℎ 𝑑𝑚 𝛿𝑄 𝑑𝑚!" 𝑑𝑚!"# 𝑑𝑝 (B.10)
𝑚 +ℎ = + ℎ!",! −ℎ +𝑉
𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡 ! 𝑑𝑡 ! 𝑑𝑡
!!! !!!
The change of specific enthalpy divided by the change in time, B.10, can be considered as a
function of pressure, temperature and concentration:
!! !! !" !! !" !! !!!
!"
= !" !"
+ !" !"
+ !! !"
. Substituting this formula in equation B.10 leads
!,! !,! ! !,!
B.11 as the result.
𝜕ℎ 𝑑𝑝 𝜕ℎ 𝑑𝑇 𝜕ℎ 𝑑𝑥! 𝑑𝑚
𝑚 + + +ℎ (B.11)
𝜕𝑝 !,!
𝑑𝑡 𝜕𝑇 !,! 𝑑𝑡 𝜕𝑥! !,!
𝑑𝑡 𝑑𝑡
! !
𝛿𝑄 𝑑𝑚!" 𝑑𝑚!"# 𝑑𝑝
= + ℎ!",! −ℎ +𝑉
𝑑𝑡 𝑑𝑡 ! 𝑑𝑡 ! 𝑑𝑡
!!! !!!
!"
Equation B.7 is added to equation B.11 to include !"
. The next partial derivative for the
!! !"
enthalpy by pressure is used for the thermodynamic relation: !" !,!
=𝑣−𝑇 !" !,!
. The
conservation equation of energy B.3 results after substituting and rearranging the equation for
!"
the change of temperature divided by the change of time !" , shown in equation B.12.
72
𝜕𝑣 𝑣 ! 𝑑𝑚!"# ! 𝑑𝑚!" 1 𝑑𝑉 𝜕𝑣 𝑑𝑥!
𝑇 !!! − !!! + −
𝑑𝑇 𝜕𝑇 !,! 𝑚 𝑑𝑡 ! 𝑑𝑡 ! 𝑚 𝑑𝑡 𝜕𝑥! !,!
𝑑𝑡
=
𝑑𝑡 𝜕𝑣 𝜕ℎ 𝜕𝑣 !
+𝑇
𝜕𝑝 𝜕𝑇
!,! 𝜕𝑇 !,!
!,! (B.12)
𝛿𝑄 𝑑𝑚!" 𝜕ℎ 𝑑𝑥!
+ !!!! ℎ!",! − ℎ −𝑚
𝑑𝑡 𝑑𝑡 ! 𝜕𝑥! !,!
𝑑𝑡
− !
𝜕ℎ 𝑚𝑇 𝜕𝑣
𝑚 +
𝜕𝑇 !,! 𝜕𝑣 𝜕𝑇 !,!
𝜕𝑝 !,!
The equations B.7, B.9 and B.12 are the adapted conservation equations that give the
information for pressure, concentration and temperature within the compressor in the
homogeneous model.
73
Appendix C: The Thermodynamic Model in Simulink
The thermodynamic model has been developed in Simulink. In this appendix the block-
calculations in Simulink will be explained.
The block calculations show the integration of physical properties in the thermodynamic
model. The inputs are defined by the conservation equations. Each conservation equation is
explained in the figures below:
Figure C-1 The conservation equation of the temperature change: dT. Gets its inputs from the physical
property calculations, mass flows and other calculated values.
74
Figure C-2 The conservation equation of the pressure change: dP. Gets its inputs from the physical
property calculations, mass flows and other calculated values.
Figure C-3 The integration blocks for calculating the pressure and temperature during the compression
cycle. The inputs come from the conservation equations Figure C-1 and Figure C-2.
75
Figure C-4 Input of the conservation equations, this is the volume change over time calculated in the
geometry model.
Figure C-5 The physical property calculation Refprop [32] via Fluidprop [33] is explained in detail in
section 3.2.2.
Figure C-6 The calculated mass is needed as input for the conservation equations. The inputs for the mass
calculations are the mass flow in and out of the cavity volume. The mass flows are integrated to the mass by
the integrator block. The 0.00001 is added because the mass cannot become zero, this would result in an
error at the conservation equation (divided by zero). The mass flows come from the suction/ discharge mass
flows and the leakage mass flows.
76
Figure C-7 The calculation of the suction mass flow, similar to the discharge and leakage mass flows
calculation (the inputs will differ for each different mass flow). The suction port area is imported from the
geometry model; the constant is the empirical flow coefficient.
Figure C-8 The calculated pressure difference for the mass flow calculation at the suction and discharge.
The constant blocks are the pressure values of the pressure outside the compressor boundaries before
entering or leaving the compressor.
77
Figure C-9 The implemented calculated pressure difference that is calculated in the pressure delay Matlab
file . These are the pressure differences used in the calculation for the leakage mass flows, explained in
detail in section 3.3.3.
78
Appendix D: The Geometry Model in Matlab, van de Bor/Zaytsev [Matlab-Code]
format compact
clear all
clc
tic
global fdischarge
global fsuction
global V_dt
%%% geo_in.txt
r1h=24; % radius male mm
r=16; % radius mm
r0=1.5; % radius mm
m1=4; % number of lobes male rotor
m2=6; % number of lobes female rotor
thau1zd = 300; % wrap angle of male rotor in degrees
L = 200; % length of rotor mm
epsilonv = 4; % Needed during the Discharge port area calculation
clearance = 0.01; % the clearance
geometry.r2h = r2h;
geometry.A = A;
geometry.i12 = i12;
geometry.i21 = i21;
geometry.k = k;
geometry.M = M;
geometry.R1 = R1;
79
geometry.R2 = R2;
geometry.theta2 = theta2;
geometry.theta3 = theta3;
geometry.theta1 = theta1;
geometry.theta4 = theta4;
geometry.theta5 = theta5;
geometry.theta6 = theta6;
geometry.alpha0 = alpha0;
geometry.beta = beta;
geometry.n = n;
80
psifl2up = 0;
f2l2 = F2L2_(psifl2low, psifl2up, geometry, phi1);
xb1 = R1 * cos(2*pi/m1);
yb1 = R1 * sin(2*pi/m1); %%% coordinates of point d1_s in the male rotor dynamic coord.
system
xa1 = xa2;
ya1 = ya2;
[xa1,ya1] = FemaleDynamicToMaleDynamic(xa1, ya1, geometry, phi1);
xa1vect(i) = xa1;
ya1vect(i) = ya1;
xa2vect(i) = xa2;
ya2vect(i) = ya2;
moment_slice_m(i) = MomentOfSlice(xa1, ya1, xb1, yb1, geometry);
end
if ((phi1 >= theta2) && (phi1 < (theta2 + theta5))) %%% stage2
f(i) = 0.5 * (z2l2 + l2i2_s + i2a2_s + a2d2_s + d2f2_s + f2l2_s + l2sh + hd1s + d1c1 +
c1a1 + a1i1 + i1l1 +l1z1);
xa2 = -r2h * cos(theta4 + theta6) - r0 * cos(theta4 + theta6 - psiz2);
ya2 = r2h * sin(theta4 + theta6) + r0 * sin(theta4 + theta6 - psiz2); %%% coordinates
of "contact" point z2 (segment f2l2) in the female rotor dynamic coord. system
moment_slice_f(i) = MomentOfSlice(xa2, ya2, xb2, yb2, geometry);
xa1 = xa2;
ya1 = ya2;
[xa1,ya1] = FemaleDynamicToMaleDynamic(xa1, ya1, geometry, phi1);
moment_slice_m(i) = MomentOfSlice(xa1, ya1, xb1, yb1, geometry);
xa1vect(i) = xa1;
ya1vect(i) = ya1;
xa2vect(i) = xa2;
ya2vect(i) = ya2;
end
if ((phi1 >= (theta2 + theta5)) && (phi1 < (theta2 + 2*alpha0 - theta5))) %%% stage3
81
f(i) = 0.5 * (z2i2_s + i2a2_s + a2d2_s + d2f2_s + f2l2_s + l2sh + hd1s + d1c1 + c1a1 +
a1i1 + i1z1);
f(i) = 0.5 * (z2a2_s + a2d2_s + d2f2_s + f2l2_s + l2sh + hd1s + d1c1 + c1a1 + a1z1);
xa2 = -r2h * cos(theta3 + theta6 - 2*pi/m2) - r0 * cos(theta3 + theta6 - psiz2 -
2*pi/m2);
ya2 = -r2h * sin(theta3 + theta6 - 2*pi/m2) - r0 * sin(theta3 + theta6 - psiz2 -
2*pi/m2); %%% coordinates of "contact" point z2 (segment i2a2_s) in the female rotor dynamic
coord. system
moment_slice_f(i) = MomentOfSlice(xa2, ya2, xb2, yb2, geometry);
xa1 = xa2;
ya1 = ya2;
[xa1,ya1] = FemaleDynamicToMaleDynamic(xa1, ya1, geometry, phi1);
moment_slice_m(i) = MomentOfSlice(xa1, ya1, xb1, yb1, geometry);
xa1vect(i) = xa1;
ya1vect(i) = ya1;
xa2vect(i) = xa2;
ya2vect(i) = ya2;
end
if ((phi1 >= (theta2 + 2*alpha0)) && (phi1 < (2*pi/m1))) %%% stages5,6
tz1(i) = fzero(@(x) G5(x,psiad2low, geometry, phi1),[tca1low,tca1up]);
c1z1 = C1A1_(tca1low, tz1(i),geometry);
xa2 = -r2h * cos(theta3 + theta6 - 2*pi/m2) - r0 * cos(theta3 + theta6 - psiia2up -
2*pi/m2);
ya2 = -r2h * sin(theta3 + theta6 - 2*pi/m2) - r0 * sin(theta3 + theta6 - psiia2up -
2*pi/m2); %%% coordinates of contact point a2 (segment i2a2_s) in the female rotor dynamic
coord. system
if (phi1 < (2*pi/m1 - beta)) %%% stage5
f(i) = 0.5 * (a2d2_s + d2f2_s + f2l2_s + l2sh + hd1s + d1c1 + c1z1);
moment_slice_f(i) = MomentOfSlice(xa2, ya2, xb2, yb2, geometry);
xa1 = xa2;
ya1 = ya2;
[xa1,ya1] = FemaleDynamicToMaleDynamic(xa1, ya1, geometry, phi1);
moment_slice_m(i) = MomentOfSlice(xa1, ya1, xb1, yb1, geometry);
else %stage 6
tz2 = fzero(@(x) G6(x,geometry,phi1),[tdf2low,tdf2up]);
d2z2_s = D2F2_S(tdf2low, tz2, geometry, phi1);
f(i) = 0.5 * (a2d2_s + d2z2_s + d1c1 + c1z1);
xb2 = xb1;
yb2 = yb1; %%% coordinates of contact point d1_s in the male rotor dynamic coord.
system
[xb2, yb2]= MaleDynamicToFemaleDynamic(xb2, yb2, geometry, phi1);
moment_slice_f(i) = MomentOfSlice(xa2, ya2, xb2, yb2, geometry);
xa1 = xa2;
ya1 = ya2;
[xa1,ya1] = FemaleDynamicToMaleDynamic(xa1, ya1, geometry, phi1);
moment_slice_m(i) = MomentOfSlice(xa1, ya1, xb1, yb1, geometry);
end
xa1vect(i) = xa1;
ya1vect(i) = ya1;
xa2vect(i) = xa2;
ya2vect(i) = ya2;
end
end
xa1vect = xa1vect';
ya1vect = ya1vect';
xa2vect = xa2vect';
82
ya2vect = ya2vect';
tlsl2low = theta4 + theta6;
tlsl2up = -theta3 - 2 * i21 * alpha0 + theta6;
ls2l2 = L2I2_(tlsl2low, tlsl2up, geometry, phi1);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% suction
alpha = i12 * (pi - acos((R1^2 - r2h^2 - A^2) / (2 * A * r2h))) - theta1;
for i=1:1:floor((2*pi/m1 + alpha + beta) * 180/pi+1) %changed starting angle from 0 to 1 and
added ending angle to +1
phi1 = (i-1) * pi /180 - beta; %%% turning angle of the male rotor in radians, when i=0
formation of closed volume is begining (phi1=-beta)
phi1d(i) = (i-1) - betad;
%%%phi1 runs from -0.62 rad to 1.94 rad
%%% female rotor segments
l2i2 = L2I2_(tli2low, tli2up, geometry, phi1);
i2a2 = I2A2_(psiia2low, psiia2up, geometry, phi1);
a2d2 = A2D2_(psiad2low, psiad2up, geometry, phi1);
d2f2 = D2F2_(tdf2low, tdf2up, geometry, phi1);
f2l2 = F2L2_(psifl2low, psifl2up, geometry, phi1);
if ((phi1 >= -beta) && (phi1 < 0)) %%% closed volume stage
z1d1 = F1D1_(tz1(i), tfd1up, geometry);
tz2 = fzero(@(x) GCV(x, geometry, phi1),[tdf2low,tdf2up]);
z2f2 = D2F2_(tz2, tdf2up, geometry, phi1);
fs(i) = 0.5 * (z2f2 + z1d1);
xa1 = R1;
ya1 = 0; %%% coordinates of point D1 in the male rotor dynamic coordinate system
xa2 = xa1;
ya2 = ya1;
[xa2,ya2] = MaleDynamicToFemaleDynamic(xa2, ya2, geometry, phi1);
moment_slice_f_s(i) = MomentOfSlice(xa2, ya2, xb2, yb2, geometry);
xb1 = xb2;
yb1 = yb2;
[xb1,yb1]=FemaleDynamicToMaleDynamic(xb1, yb1, geometry, phi1);
moment_slice_m_s(i) = MomentOfSlice(xa1, ya1, xb1, yb1, geometry);
xb1vect(i) = xb1;
yb1vect(i) = yb1;
xb2vect(i) = xb2;
yb2vect(i) = yb2;
xf(i) = xb1vect(i)-A;
xfm(i) = xb1vect(i);
yf(i) = yb1vect(i);
end
if ((phi1 >= 0) && (phi1 < alpha)) %%% stage1
psiz_2(i) = fzero(@(x) G1S(x, geometry, phi1),[psiad2low,psiad2up]);
tn1(i) = asin((-A * sin(phi1) + r2h * sin(k * phi1) + r * sin(psiz_2(i) - k * phi1)) /
R1) + phi1;
83
d1n1 = -D1D1S_(phi1, tn1(i), geometry);
fs(i) = 0.5 * (z2d2 + d2f2 + z1d1 + d1n1);
xa2 = -r2h + r*cos(psiz_2(i));
ya2 = -r*sin(psiz_2(i)); %%% coordinates of point z2 (segment a2d2) in the female rotor
dynamic coordinate system
moment_slice_f_s(i) = MomentOfSlice(xa2, ya2, xb2, yb2, geometry);
xb1 = xb2;
yb1 = yb2;
[xb1, yb1] = FemaleDynamicToMaleDynamic(xb1, yb1, geometry, phi1);
moment_slice_m_s(i) = MomentOfSlice(xa1, ya1, xb1, yb1, geometry);
xb1vect(i) = xb1;
yb1vect(i) = yb1;
xb2vect(i) = xb2;
yb2vect(i) = yb2;
xf(i) = xb2vect(i);
xfm(i) = xb2vect(i)+A;
yf(i) = yb2vect(i);
end
if ((phi1 >= alpha) && (phi1 < theta2)) %%% stage2
z1d1 = F1D1_(tz1(i), tfd1up, geometry);
fs(i) = 0.5 * (i2a2 + a2d2 + d2f2 + z1d1 + d1h1 + h1i2);
xa2 = -r2h * cos(theta3 + theta6) - r0 * cos(theta3 + theta6 - psiia2low);
ya2 = -r2h * sin(theta3 + theta6) - r0 * sin(theta3 + theta6 - psiia2low); %%%
coordinates of point i2 (segment i2a2) in the female rotor dynamic coordinate system
moment_slice_f_s(i) = MomentOfSlice(xa2, ya2, xb2, yb2, geometry);
xb1 = xb2;
yb1 = yb2;
[xb1, yb1] = FemaleDynamicToMaleDynamic(xb1, yb1, geometry, phi1);
moment_slice_m_s(i) = MomentOfSlice(xa1, ya1, xb1, yb1, geometry);
xb1vect(i) = xb1;
yb1vect(i) = yb1;
xb2vect(i) = xb2;
yb2vect(i) = yb2;
xf(i) = xb2vect(i);
xfm(i) = xb2vect(i)+A;
yf(i) = yb2vect(i);
end
if ((phi1 >= theta2) && (phi1 < theta2 + theta5)) %%% stage3
psiz1 = asin(r1h / r0 * sin(theta2 + theta5 - phi1)) - (theta2 + theta5 - phi1);
z1f1 = L1F1_(psiz1, psilf1up, geometry);
psiz2 = (theta4 + theta6 - phi1 * i21) + asin(r2h / r0 * sin(theta4 + theta6 - phi1 *
i21));
f2z2 = F2L2_(psifl2low, psiz2, geometry, phi1);
fs(i) = 0.5 * (i2a2 + a2d2 + d2f2 + f2z2 + z1f1 + f1d1 + d1h1 + h1i2);
xb2 = -r2h * cos(theta4 + theta6) - r0 * cos(theta4 + theta6 - psiz2);
yb2 = r2h * sin(theta4 + theta6) + r0 * sin(theta4 + theta6 - psiz2); %%% coordinates
of "contact" point z2 (segment f2l2) in the female rotor dynamic coord. system
moment_slice_f_s(i) = MomentOfSlice(xa2, ya2, xb2, yb2, geometry);
xb1 = xb2;
yb1 = yb2;
[xb1, yb1] = FemaleDynamicToMaleDynamic(xb1, yb1, geometry, phi1);
moment_slice_m_s(i) = MomentOfSlice(xa1, ya1, xb1, yb1, geometry);
xb1vect(i) = xb1;
yb1vect(i) = yb1;
xb2vect(i) = xb2;
yb2vect(i) = yb2;
xf(i) = xb2vect(i);
xfm(i) = xb2vect(i)+A;
yf(i) = yb2vect(i);
end
if ((phi1 >= (theta2 + theta5)) && (phi1 < (theta2 + 2*alpha0 - theta5))) %%% stage4
z1l1 = I1L1_(tz1(i), til1up, geometry);
tz2 = -theta3 - theta4 - 2 * i21 * alpha0 + i21 * phi1;
l2z2_s = L2I2_S(tli2low, tz2, geometry, phi1);
fs(i) = 0.5 * (i2a2 + a2d2 + d2f2 + f2l2 + l2z2_s + z1l1 + l1f1 + f1d1 + d1h1 + h1i2);
xb2 = -R2 * cos(tz2 + 2*pi/m2);
yb2 = R2 * sin(tz2 + 2*pi/m2); %%% coordinates of contact point z2 (segment l2i2_s) in
the female rotor dynamic coord. system
moment_slice_f_s(i) = MomentOfSlice(xa2, ya2, xb2, yb2, geometry);
xb1 = xb2;
yb1 = yb2;
[xb1, yb1] = FemaleDynamicToMaleDynamic(xb1, yb1, geometry, phi1);
moment_slice_m_s(i) = MomentOfSlice(xa1, ya1, xb1, yb1, geometry);
xb1vect(i) = xb1;
yb1vect(i) = yb1;
xb2vect(i) = xb2;
yb2vect(i) = yb2;
xf(i) = xb2vect(i);
xfm(i) = xb2vect(i)+A;
84
yf(i) = yb2vect(i);
end
if ((phi1 >= (theta2 + 2*alpha0 - theta5)) && (phi1 < (theta2 + 2*alpha0))) %%% stage5
psiz1 = asin(r1h / r0 * sin(theta1 + theta5 - 2*pi/m1 + phi1)) - (theta1 + theta5 -
2*pi/m1 + phi1);
z1i1 = A1I1_(psiz1, psiai1up, geometry);
psiz2 = (theta3 + theta6 + phi1 * i21 - 2*pi/m2) + asin(r2h / r0 * sin(theta3 + theta6
+ phi1 * i21 - 2*pi/m2));
i2z2_s = I2A2_S(psiia2low, psiz2, geometry, phi1);
fs(i) = 0.5 * (i2a2 + a2d2 + d2f2 + f2l2 + l2i2_s + i2z2_s + z1i1 + i1l1 + l1f1 + f1d1
+ d1h1 + h1i2);
xb2 = -r2h * cos(theta3 + theta6 - 2*pi/m2) - r0 * cos(theta3 + theta6 - psiz2 -
2*pi/m2);
yb2 = -r2h * sin(theta3 + theta6 - 2*pi/m2) - r0 * sin(theta3 + theta6 - psiz2 -
2*pi/m2); %%% coordinates of "contact" point z2 (segment i2a2_s) in the female rotor dynamic
coord. system
moment_slice_f_s(i) = MomentOfSlice(xa2, ya2, xb2, yb2, geometry);
xb1 = xb2;
yb1 = yb2;
[xb1, yb1] = FemaleDynamicToMaleDynamic(xb1, yb1, geometry, phi1);
moment_slice_m_s(i) = MomentOfSlice(xa1, ya1, xb1, yb1, geometry);
xb1vect(i) = xb1;
yb1vect(i) = yb1;
xb2vect(i) = xb2;
yb2vect(i) = yb2;
xf(i) = xb2vect(i);
xfm(i) = xb2vect(i)+A;
yf(i) = yb2vect(i);
end
if ((phi1 >= (theta2 + 2*alpha0)) && (phi1 < (2*pi/m1))) %%% stage6
z1a1 = C1A1_(tz1(i), tca1up, geometry);
tl2 = phi1 * i21 - theta4 -theta6;
l2_i2 = -L2SH_(tl2, th1i2up, geometry);
h1l2 = -L2SH_(th1i2low, tl2, geometry);
fs(i) = 0.5 * (i2a2 + a2d2 + d2f2 + f2l2 + l2i2_s + i2a2_s + z1a1 + a1i1 + i1l1 + l1f1
+ f1d1 + d1h1 + h1l2 +l2_i2);
xb2 = -r2h * cos(theta3 + theta6 - 2*pi/m2) - r0 * cos(theta3 + theta6 - psiia2up -
2*pi/m2);
yb2 = -r2h * sin(theta3 + theta6 - 2*pi/m2) - r0 * sin(theta3 + theta6 - psiia2up -
2*pi/m2); %%% coordinates of contact point a2 (segment i2a2_s) in the female rotor dynamic
coord. system
moment_slice_f_s(i) = MomentOfSlice(xa2, ya2, xb2, yb2, geometry);
xb1 = xb2;
yb1 = yb2;
[xb1, yb1] = FemaleDynamicToMaleDynamic(xb1, yb1, geometry, phi1);
moment_slice_m_s(i) = MomentOfSlice(xa1, ya1, xb1, yb1, geometry);
xb1vect(i) = xb1;
yb1vect(i) = yb1;
xb2vect(i) = xb2;
yb2vect(i) = yb2;
xf(i) = xb2vect(i);
xfm(i) = xb2vect(i)+A;
yf(i) = yb2vect(i);
end
if ((phi1 >= (2*pi/m1)) && (phi1 < (2*pi/m1 + alpha))) %%% stage7
tn1(i) = tn1(i - 360/m1);
n1d1 = -D1D1S_(tn1(i), (phi1 - 2*pi/m1), geometry);
psiz_2(i) = psiz_2(i - 360/m1);
a2z2_s = A2D2_S(psiad2low, psiz_2(i), geometry, phi1);
fs(i) = 0.5 * (i2a2 + a2d2 + d2f2 + f2l2 + l2i2_s + i2a2_s + a2z2_s + n1d1 +d1c1 + c1a1
+ a1i1 + i1l1 + l1f1 + f1d1 + d1h1 + h1i2);
xb2 = -r2h * cos(2*pi/m2) + r*cos(psiz_2(i) + 2*pi/m2);
yb2 = r2h * sin(2*pi/m2) - r*sin(psiz_2(i) + 2*pi/m2); %%% coordinates of point z2
(segment a2d2_s) in the female rotor dynamic coordinate system
xb_end(i) = xb1;
yb_end(i) = yb1;
xb2vect(i) = xb2;
yb2vect(i) = yb2;
xf_end(i) = xb2vect(i);
xfm_end(i) = xb2vect(i)+A;
yf_end(i) = yb2vect(i);
end
85
end
%%% dV/dphi1 and cavity geometrical moment for complete working cycle
for i = 1:1:floor((2*pi + thau1z + 2*pi/m1 + beta) * 180/pi+1) %changed starting angle from 0
to 1 and added ending angle to +1
phi1 = (i-1) * pi /180 - beta;
86
if ((phi1 < (2*pi - beta)) && (phi1 < (thau1z + 2*pi/m1 + alpha))) %%% suction
dV_dphi1(i) = dV_dphi1suck(i);
moment_m(i) = moment_m_s(i);
moment_f(i) = moment_f_s(i);
end
if ((phi1 >= (thau1z + 2*pi/m1 + alpha)) && (phi1 <= (2*pi - beta))) %%% suction is
finished, but compression is not begun
dV_dphi1(i) = 0;
moment_m(i) = 0;
moment_f(i) = 0;
end
if ((phi1 >= (2*pi - beta)) && (phi1 <= (thau1z + 2*pi/m1 + alpha))) %%% simultaneous
suction and compression
dV_dphi1(i) = dV_dphi1suck(i) + dV_dphi1comp(i - 360);
moment_m(i) = moment_m_s(i) + moment_m_c(i - 360);
moment_f(i) = moment_f_s(i) + moment_f_c(i - 360);
end
if ((phi1 > (2*pi - beta)) && (phi1 > (thau1z + 2*pi/m1 + alpha))) %%% compression
dV_dphi1(i) = dV_dphi1comp(i - 360);
moment_m(i) = moment_m_c(i - 360);
moment_f(i) = moment_f_c(i - 360);
end
end
87
alphacd = thau1zd - phi1cd;
alphac = alphacd * pi / 180.0; %%% transfer into radians
flag = 1;
elseif ((G1W(t, geometry, phi1, alphac) * G1W(t1, geometry, phi1, alphac))< 0)
&& abs(t - t1) > 0.00001
tu = t;
elseif ((G1W(t, geometry, phi1, alphac) * G1W(t1, geometry, phi1, alphac))> 0)
&& abs(tu - t) < 0.00001
flag = 1;
else
t1 = t;
end
end
if (phi1 == -alphac)
tz2 = 0;
else
tz2 = t; % result of solution
end
88
z2f2_s = D2F2_S(tz2, tdf2up,geometry, phi1);
a2z2_w = A2D2_W(psiad2low, psiz2w,geometry, alphac);
end
if (phi1 < (-alphac + 2 * alpha0 - 2 * theta5)) %%% stage1
ti1sl1wlow = -phi1 + theta2 + 2 * alpha0 - theta5;
ti1sl1wup = alphac + theta2 + theta5;
i1sl1w = (r1h - r0)^2 * (ti1sl1wup - ti1sl1wlow);
b = 0.5 * (d1c1 + c1a1 + a1i1 + i1sl1w + l1f1_w + f1d1_w + d1wd1s + i2a2_w + a2z2_w
+ z2f2_s + f2l2_s + l2si2w);
end
if ((phi1 >= (-alphac + 2 * alpha0 - 2 * theta5)) && (phi1 < (-alphac + 2 * alpha0)))
%%% stage2
psiz1 = fzero(@(x) G2W(x,geometry,phi1, alphac),[psiai1up,psiai1low]);
psiz1w=psiz1;
a1z1 = A1I1_(psiai1low, psiz1, geometry);
z1f1_w = L1F1_(psiz1w, psilf1up, geometry);
b = 0.5 * (d1c1 + c1a1 + a1z1 + z1f1_w + f1d1_w + d1wd1s + i2a2_w + a2z2_w + z2f2_s
+ f2l2_s + l2si2w);
end
if ((phi1 >= (-alphac + 2 * alpha0)) && (phi1 < (-alphac + 2 * alpha0 + 2 * (theta1 -
acos(xc1 / Rc1))))) %%% stage3
tz_1 = fzero(@(x) G3W(x, geometry, phi1, alphac),[psiai1up,psiai1low]) ;
tz1w = tz_1;
c1z1 = C1A1_(tca1low, tz_1, geometry);
z1d1_w = F1D1_(tz1w, tfd1up, geometry);
b = 0.5 * (d1c1 + c1z1 + z1d1_w + d1wd1s + i2a2_w + a2z2_w + z2f2_s + f2l2_s +
l2si2w);
end
if (phi1 >= (-alphac + 2 * alpha0 + 2 * (theta1 - acos(xc1 / Rc1)))) %%% stages 4, 5, 6
tz1w = fzero(@(x) G4W(x, geometry, phi1, alphac),[tfd1low,tfd1up]);
psiz1 = acos(A^2 / (2 * r1h * r) + r2h^2 / (2 * r1h * r) - A * r2h / (r1h * r) *
cos(i21 * tz1w) - r1h / (2 * r) - r / (2 * r1h));
z1d1_w = F1D1_(tz1w, tfd1up, geometry);
d1z1 = D1C1_(psidc1low, psiz1, geometry);
if (phi1 < (-alphac + theta1 + theta2)) %%% stage 4
b = 0.5 * (d1z1 + z1d1_w + d1wd1s + i2a2_w + a2z2_w + z2f2_s + f2l2_s +
l2si2w);
end
if ((phi1 >= -alphac + theta1 + theta2) && (phi1 < (-alphac + theta1 + theta2 + 2 *
theta5))) %%% stage 5
psiz2 = fzero(@(x) G5W(x,geometry, phi1, alphac),[psifl2up,pi/2]);
psiz2w=psiz2;
i2z2_w = I2A2_W(psiia2low, psiz2w, geometry, phi1);
z2l2_s = F2L2_S(psiz2, psifl2up, geometry, phi1);
b = 0.5 * (d1z1 + z1d1_w + d1wd1s + i2z2_w + z2l2_s + l2si2w);
end
if (phi1 >= (-alphac + theta1 + theta2 + 2 * theta5)) %%% stage 6.1
b = 0.5 * (d1z1 + z1d1_w + d1wd1s);
end
end
end
if (phi1 >= (-alphac + 2*pi/m1)) %%% after point D1s intersects point d1w
b = 0;
end
if (tha2sup >= tha2slow) %%% until point A2s intersects point H
c=0;
end
if ((tha2sup < tha2slow) && (phi1 <= (2*pi/m1))) %%% stages 6.2 - 9
%%% numerical solution for psiz2
i;
t1 = psiad2up;
tu = psiad2low;
flag = 0;
while flag == 0
t= 0.5*(t1+tu);
if G6W(t,geometry,phi1)*G6W(t1,geometry,phi1)<0 && abs(t-t1)<0.00001
flag =1;
elseif G6W(t,geometry,phi1)*G6W(t1,geometry,phi1)<0 && abs(t-t1)>0.00001
tu =t;
elseif G6W(t,geometry,phi1)*G6W(t1,geometry,phi1)>0 && abs(tu-t)<0.00001
flag =1;
else
t1=t;
end
end
psiz2 = t; % result of solution
89
tz2w = acos(xz2 / R1) - phi1;
z2wh = D1D1S_(tz2w, beta, geometry);
a2sz2 = A2D2_S(psiad2low, psiz2,geometry, phi1);
if (phi1 <= (2*pi/m1 - acos(xc1 / Rc1) - p0o1c1s)) %%% stage 6.2
c = 0.5 * (z2wh + ha2s + a2sz2);
end
if ((phi1 > (2*pi/m1 - acos(xc1 / Rc1) - p0o1c1s)) && (phi1 <= (2*pi/m1))) %%% stages
7-9
if (phi1 <= (2*pi/m1 - beta))
%%% numerical solution for psin1 on stages 7-8
psin1 = fminbnd(@(x) G7NW(x,geometry, phi1),psidc1low,psidc1up);
tnw = asin((-A * sin(phi1) + r1h * sin(2*pi/m1) - r * sin(psin1 - 2*pi/m1)) /
r2h) + phi1;
hnw = HA2S_(tha2slow, tnw, geometry);
n1c1s = D1C1_(psin1, psidc1up, geometry);
end
if (phi1 <= (2*pi/m1 - theta1)) %%% stage 7
%%% numerical solution for tz_1
tz_1 = fminbnd(@(x) G7ZW(x,geometry, phi1),tca1up,tca1low);
tz1w = asin(-A / r2h * (sin(phi1) - sin(tz_1 - theta1 + 2*pi/m1)) - sin(k *
tz_1 - theta1 + 2*pi/m1)) + phi1;
c1sz1 = C1A1_(tca1low, tz_1,geometry);
z1wa2s = HA2S_(tz1w, tha2sup, geometry);
c = 0.5 * (z2wh + hnw + n1c1s + c1sz1 + z1wa2s + a2sz2);
else
%%% numerical solution for tz_1
tz_1 = fzero(@(x) G5(x,psiad2low, geometry, phi1),[tca1up,tca1low]);
c1z1 = C1A1_(tca1low, tz_1,geometry);
if (phi1 <= (2*pi/m1 - beta)) %%% stage 8
c = 0.5 * (z2wh + hnw + n1c1s + c1sz1 + a2sz2);
else %%% stage 9
z2wd1s = D1D1S_(tz2w, (2*pi/m1 - phi1),geometry);
c = 0.5 * (z2wd1s + d1c1 + c1z1 + a2sz2);
end
end
end
end
fc(i) = b + c;
fc(alphacd + 360/m1) = 0;
end
for i = 1:1:floor(thau1zd + 360/m1 - phi1cd)%%% discharge port area
if ((i-1 + phi1cd - thau1zd + betad) < 0)
fd(i) = f01 + f02 - fc(i);
else
fd(i) = f(i + phi1cd - thau1zd + betad) - fc(i); %%% equation (2) chapter "Discharge
flow..."
if (fd(i) < 0)
fd(i) =0;
end
end
end
%%% female rotor cavity area calculation according to Sakun equations (352), (359.2) for
selfconfidence
phif = i21 * acos((A^2 + R1^2 - r2h^2)/(2 * A * R1));
f02Sakun = r2h^2/2 * (0.5*r^2/r2h^2*(pi - theta3) + theta3 - sin(theta3) + A^2/r2h^2*(phif -
sin(phif)) + R1^2*M/r2h^2*phif - A*R1/r2h^2*(1 + 2*i21)*sin(i12*phif) + A*R1/r2h^2*sin(M*phif)
- A/r2h*sin(theta4) + theta4) + 0.5*(theta3 +theta4)*(R2^2 - r2h^2);
integral_moment_m = 0;
integral_moment_f = 0;
integral_moment_cavity = 0;
for i = 1:1:floor((2*pi + thau1z + 2*pi/m1 + beta) * 180/pi+1)
integral_moment_m = integral_moment_m + moment_m(i);
integral_moment_f = integral_moment_f + moment_f(i);
integral_moment_cavity = integral_moment_cavity + (moment_m(i) + i21 * moment_f(i));
end
90
integral_moment_m = integral_moment_m / (i + 1);
integral_moment_f = integral_moment_f / (i + 1);
integral_moment_cavity = integral_moment_cavity / (i + 1);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% code from sealline.cpp %%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% suction
for i = 1:1:floor((thau1z + 2*pi/m1 + alpha + beta) * 180/pi+1)
phi1 = (i-1) * pi /180 - beta; %%% turning angle of the male rotor in radians, when i=0
formation of closed volume is begining (phi1=-beta)
91
thaumax = -theta2 - 2*alpha0 + theta5 + phi1;
if (thaumax < 0)
thaumax = 0;
end
if (thaumax > thau1z)
thaumax = thau1z;
end
thaumin = phi1;
if (thaumin < 0)
thaumin = 0;
end
if (thaumin > thau1z)
thaumin = thau1z;
end
phi1 = (i-1) * pi /180 - beta; %%% turning angle of the male rotor in radians, when i=0
compression begins (phi1=-beta)
%%% path1
%%% segment 5-1
92
%%% male rotor
thaumin = beta + phi1;
if (thaumin < 0)
thaumin = 0;
end
if (thaumin > thau1z)
thaumin = thau1z;
end
thaumax = thau1z;
path2min_c(i) = (thaumax - thaumin) * sqrt(R1^2 + (L / thau1z)^2);
93
end
end
else
path1s(i) = 0;
sgmns1_5ofpath1s(i) = 0;
sgmn51ofpath1s(i) = 0;
end
if (i <= 360)
path1c(i) = 0;
csgmns1_5ofpath1c(i) = 0;
csgmn51ofpath1c(i) = 0;
path2inc(i) = 0;
path2outc(i) = 0;
path3inc(i) = 0;
path3outc(i) = 0;
else
path1c(i) = path1_c(i - 360);
csgmns1_5ofpath1c(i) = segments1_5(i - 360);
csgmn51ofpath1c(i) = s51_c(i - 360);
path2inc(i) = path2in_c(i - 360);
path2outc(i) = path2out_c(i - 360);
path3inc(i) = path3in_c(i - 360);
path3outc(i) = path3out_c(i - 360);
end
%%% leak in
if (phi1 >= (2*pi - beta - 2*pi/m1))
path4inc(i) = COMP_START_BLOW_HOLE(geometry, phi1, ang_hop);
else
path4inc(i) = 0;
end
%%% path 6: between the end plate and the rotor end face at the discharge end
94
%%% leak out to the trailing cavity
if ((phi1 >= (thau1z + 2*pi/m1)) && (phi1 < (2*pi + 2*pi/m1 + thau1z)))
path6outt(i) = clearance * 2 * (r + r0);
else
path6outt(i) = 0;
end
%%% leak out to suction
if ((phi1 >= (2*pi + thau1z)) && (phi1 < (2*pi + thau1z + 2*pi/m1)))
path6outs(i) = clearance * (r + r0);
else
path6outs(i) = 0;
end
end
save(filename)
i = 1;
Vmax = 0;
iVmax = 0;
V = V * 1e-9;
while (i < length(V))
if (V(i) > Vmax)
Vmax = V(i);
iVmax = i; %%% when i = iVmax, compression begins
end
i=i+1;
end
95
y = A * sin(i21*phi1) - temp_x * sin(k*phi1) + temp_y * cos(k*phi1);
blow_hole = 0;
dthau = 0.001;
olddistance = 0;
thau1 = phi1 - ang_hop;
%%%_________________________________
%%% tangent plane will be found at point (xf, yf, zf) (psi, thau2)
xf = xm;
yf = ym;
zf = L * thau2 / thau1z;
if (zf < L)
aa = -r * L / thau1z * cos(psi - i21*(phi1 - thau2));
bb = r * L / thau1z * sin(psi - i21*(phi1 - thau2));
cc = -r2h * r * i21 * sin(psi);
dd = -(aa * xf + bb * yf + cc * zf); %%% tangent plane coefficients
distance = (aa*xm + bb*ym + cc*zm + dd) / sqrt(aa*aa + bb*bb + cc*cc);
else
distance = L - zm;
end
blow_hole = blow_hole + (distance + olddistance)*0.5 * dthau * R1;
olddistance = distance;
thau1 = thau1 + dthau;
end
%% Properties Constants %%
x0 = 0.3; %Overall ammonia concentration
dx = 0.001; %Small differences in concentration
96
%% Set fluidprop fluid NH3-H2O x0=0.3
global FP
FP = actxserver ('FluidProp.FluidProp');
Msg='FluidProp: COM object created';
disp(Msg);
invoke(FP,'SetUnits','SI','','',''); %Convert to SI-units
Msg = 'RefProp, Ammonia - Water';
disp(Msg);
Model = 'RefProp';
nCmp = 2;
Cnc=[x0,0;(1-x0),0];
Cmp1 = 'ammonia,water';
ErrorMsg = invoke(FP,'SetFluid_M',Model,nCmp,Cmp1,Cnc);
function [f_discharge]=f_discharge(G)
global fdischarge
G = floor((G/(0.043722222222222/787))+1);
if G<787
f_discharge=fdischarge(G);
else
f_discharge=fdischarge(787);
end
end
function [f_suction]=f_suction(G)
global fsuction
G = floor((G/(0.043722222222222/787))+1);
if G<787
f_suction = fsuction(G);
else
f_suction=fsuction(787);
end
end
function [V_test]=V_test(G)
global V_dt
G = floor((G/(0.043722222222222/787))+1);
if G<787
V_test=V_dt(G);
else
V_test=V_dt(787);
end
end
97
Appendix E: Calculation Pressure Difference [Matlab-Code]
%% Pressure difference calculation
load('Pressure.mat')
load('Time.mat')
Time_H=t_h_short';
P_h_x = P_h';
global P_h_x
global dPT
global dPL
global dPT2
global dPL2
global dPT3
global dPL3
v1 = 3.7*10^5;
v2 = 9.08*10^5;
n1 = 360/m1;
n2 = 360;
n3 = 360 - n1;
n4 = 360 + 2 * n1;
n5 = 360 + n1;
k1_1=repmat(v1,1,n1);
k1_2=repmat(v2,1,n1);
k2_1=repmat(v1,1,n2);
k2_2=repmat(v2,1,n2);
k3_1=repmat(v1,1,n3);
k3_2=repmat(v2,1,n3);
k4_1=repmat(v1,1,n4);
k4_2=repmat(v2,1,n4);
k5_2=repmat(v2,1,n5);
P_delay_1 = [k1_1,k1_1,k1_1,P_h_x];
P_main_1 = [k1_1,k1_1, P_h_x, k1_2];
P_advance_1 = [k1_1,P_h_x, k1_2, k1_2];
P_delay_2 = [k2_1,k2_1,k2_1,P_h_x];
P_main_2 = [k2_1,k2_1, P_h_x, k2_2];
P_advance_2 = [k2_1,P_h_x, k2_2, k2_2];
P_delay_3 = [k3_1,k4_1,k2_1,P_h_x];
P_main_3 = [k2_1,k2_1, P_h_x, k5_2];
P_advance_3 = [k3_1,P_h_x, k4_2, k2_2];
dP_Trailing_z2 = P_main_2-P_delay_2;
dP_Leading_z2 = P_advance_2-P_main_2;
dP_Trailing_z3 = P_main_3-P_delay_3;
dP_Leading_z3 = P_advance_3-P_main_3;
dP_Trailing_z1(dP_Trailing_z1<100)=0;
dP_Leading_z1(dP_Leading_z1<100)=0;
dP_Trailing_z2(dP_Trailing_z2<100)=0;
dP_Leading_z2(dP_Leading_z2<100)=0;
98
dP_Trailing_z3(dP_Trailing_z3<100)=0;
dP_Leading_z3(dP_Leading_z3<100)=0;
dPT = dP_Trailing_z1(2*n1+1:end-n1-1);
dPL = dP_Leading_z1(2*n1+1:end-n1-1);
dPT2 = dP_Trailing_z2(2*n2+1:end-n2-1);
dPL2 = dP_Leading_z2(2*n2+1:end-n2-1);
dPT3 = dP_Trailing_z3(2*n2+n1+1:end-n2-1);
dPL3 = dP_Leading_z3(2*n2+n1+1:end-n2-1);
99
Appendix F: Calculation Efficiencies [Matlab-Code]
clear all
clc
%% Efficiency Calculations %%
Cavity_V = load('Cavity_volume.mat');
dV_dt = load('dV_dt.mat');
% plot(Time_G.time,Cavity_V.V);
b1 = 5;
n = 3500;
h = load('Enthalpy.mat');
P = load('Pressure.mat');
MD = load('Discharge.mat');
Mass_out = load('Mass_in_discharge.mat');
Mass_in = load('Mass_in_suc.mat');
Mass_out_max = max(Mass_out.Mass_out_SD);
Mass_in_max= max(Mass_in.Mass_in_SD);
mdischarge = sum(MD.MFD_h)
h_suc = h.H_h(1);
h_id = 3.6606* 10^5; %% Caculate from fluidprop! (concentratie in mol/mol)
x=size(P.P_h);
x = x(1);
Work(1)=0;
for i = 2:x-1
p0(i) = P.P_h(i);
p1(i) = P.P_h(i+1);
V(i) = dV_dt.V_dt(i);
dP(i) = p1(i)-p0(i);
V_dP(i) = V(i)*dP(i);
Work(i) = Work(i-1) + V_dP(i);
end
Work_t = sum(Work)
Eff_is = W_ideal/(Work_t/mdischarge)
%% Volumetric Efficiency %%
S_V = load('Specific_volume.mat');
vs = S_V.Vsv_h(1);
Vmax = max(Cavity_V.V);
Eff_vol = ((Mass_out_max)*vs)/Vmax
100
Appendix G: Calculation Shaft Rotation Angle
The shaft rotation angle is based on the wrap angle of the male rotor, number of male lobes and
the rotor geometry. The inputs as used by Zaytsev [1] and described in chapter 1-4 are used in
this appendix as inputs, Table G-0-1.
Table G-0-1 Inputs of the shaft rotation angle calculation used by Zaytsev [1].
The shaft rotation angle is necessary in the calculations for the transfer from the shaft angle to
the time. The maximum shaft rotation angle is calculated by equation G.1. The input values
!"#
will be in radians and are multiplied by ! to calculate to angle degrees. The one added in the
end of the equation results from the total shaft rotation that starts at 0 whilst Matlab only can
start at 1, one extra step needs to be added.
50
2𝜋 180
𝑆ℎ𝑎𝑓𝑡 𝑅𝑜𝑡𝑎𝑡𝑖𝑜𝑛 𝐴𝑛𝑔𝑙𝑒 = 2𝜋 + 𝜏! + +𝛽 ∙ + 1 (G.1)
40 𝑚! 𝜋
In the equation of the shaft rotation angle, angle β is used. Angle β is calculated with the cosine
rule,10 is given in equation G.2 and illustrated in Figure G-1. Angle β represents the angle from
the first touch between the male rotor and the female rotor and the ‘A-line’. The A-line is line
0 used as start point of the rotation angle of the rotors (𝐴 = 𝑟!" + 𝑟!" in m).
-1
0 50 𝐴! + 𝑅! ! − 𝑟!" !
-1
0 𝛽 = 𝑎𝑐𝑜𝑠 (G.2)
2 ∙ 𝐴 ∙ 𝑅!
0
40
10
30
20
r2h
R2
20
30
R1
R2h
40
10
50 r
β
0
O1 O2
A
60
r1h R1
-10
-10 0 10 20 30 40 50 60
Figure G-1 Illustration of angle β. β is used in the shaft rotation angle calculation.
101