Purdue University
Purdue e-Pubs
06(40$6,10$.1/24(55140*,0((4,0*10)(4(0&( &+11.1)(&+$0,&$.0*,0((4,0*
CFD Analysis of Oil Flooded Twin Screw
Compressors
Sham Ramchandra Rane
Centre for Compressor Technology, City University London, United Kingdom5+$/4$0(&,6:$&7-
Ahmed Kovacevic
Centre for Compressor Technology, City University London, United Kingdom$-18$&(8,&&,6:$&7-
Nikola Stosic
Centre for Compressor Technology, City University London, United Kingdom05615,&&,6:$&7-
1..196+,5$0'$'',6,10$.914-5$6
+<25'1&5.,%274'7(('7,&(&
;,5'1&7/(06+$5%((0/$'($8$,.$%.(6+417*+74'7((7%5$5(48,&(1)6+(74'7("0,8(45,6:,%4$4,(5.($5(&106$&6(27%5274'7(('7)14
$'',6,10$.,0)14/$6,10
1/2.(6(241&((',0*5/$:%($&37,4(',024,06$0'10',4(&6.:)41/6+($:#(44,&-$%14$614,(5$6 +<25(0*,0((4,0*274'7(('7
(44,&-8(06514'(4.,6+6/.
$0( +$/$/&+$0'4$18$&(8,&+/('$0' 615,&,-1.$0$.:5,51),..11'('!9,0 &4(91/24(55145
International Compressor Engineering Conference. $2(4
+<25'1&5.,%274'7(('7,&(&
1023, Page 1
23
rd
International Compressor Engineering Conference at Purdue, July 11-14, 2016
CFD Analysis of Oil Flooded Twin Screw Compressors
Sham RANE*, Ahmed KOVACEVIC, Nikola STOSIC
City University London, Centre for Compressor Technology,
London, EC1V 0HB, UK
sham.rane@city.ac.uk
* Corresponding Author
ABSTRACT
Modelling of screw compressors using Computational Fluid Dynamics (CFD) offers better insight into the working
chamber of twin screw machines when compared with chamber models. As shown by authors in earlier publications,
CFD models predict performance of dry gas and refrigeration compressors fairly accurately. However numerical flow
models used for modelling of oil flooded twin screw compressors are still at the development stage. This is mainly
due to the lack of understanding of the flow complexity and the techniques used for solving coupled equations that
represent interactions between the gas and the oil in such machines.
This paper presents the modelling approach used for calculation of the performance of an oil flooded screw
compressor. It requires a structured numerical mesh which can represent all moving parts of the compressor in a single
numerical domain. Such mesh is generated by SCORG
TM
using novel boundary distribution technique called casing-
to-rotor conformal boundary mapping. A test oil injected twin screw compressor with rotor configuration 4/5 and 127
mm main rotor diameter was measured in the compressor rig of the Centre for Compressor Technology at City
University London. Measurements of the chamber pressure history and integral parameters of the compressor such as
mass flow rate of gas and oil, indicated power and temperatures are used for the comparison with CFD results. The
analysis showed a close match in the prediction of the mass flow rates of gas. The indicated power obtained by CFD
predictions matched well with the measured shaft power. The model provided an exceptional visualization of the
interaction of gas and oil inside the compression chamber. The mixing of the phases, distribution of oil, heat transfer
between gas and oil and also effects on sealing due to high oil concentration in leakage gaps were well captured.
1. INTRODUCTION
Oil is injected into the working chamber of twin screw compressors mainly for three reasons. The main reason is to
control the gas temperature during the compression. For high pressure ratios between the suction and the discharge
the temperature rise in the absence of internal cooling may be significant and can result in thermal deformation of the
rotors and the casing, eventually leading to seizure. Since early times, the adopted practice for twin screw compressors
and many other positive displacement compressors is to inject oil in the compression chamber for cooling. Secondly,
the injected oil reduces clearance gaps in the working chamber, thereby improving volumetric efficiency. Thirdly, the
injected oil lubricates rotors and bearings. The rule of thumb in screw compressor industry is that oil contribution by
mass to cooling, sealing and lubrication is in the proportion of 100:10:1 respectively. Although oil injection has several
benefits and is essential for functioning of the compressor, the presence of oil has negative effects. Being of high
viscosity, the shear of oil in clearances contributes to the additional mechanical losses. The bulk of the oil with density
several orders of magnitude higher than the compressed gas has a high momentum and consumes some of the input
power during injection and transport through the chambers. Deipenwisch and Kauder (1999) used a mathematical
model for predicting oil induced losses in twin screw compressors shown in Figure 1 as a representative of contribution
of the losses associated with oil injection in a typical screw compressor at a male rotor tip speed of 35 m/s. As such,
oil injection needs to be controlled so that optimum quantity of oil is used and maximum effectiveness is achieved.
90% of the oil is injected through the oil ports that are timed accurately with the compression cycle such that for the
given injection pressure and compressor internal pressure the desired quantity of oil mass moves into the compression
process.
1023, Page 2
23
rd
International Compressor Engineering Conference at Purdue, July 11-14, 2016
Figure 2 shows an example of the oil injection ports. It
is possible to have oil injection systems where the
compressor discharge pressure is utilized to inject the
oil. In order to improve reliability or to account for high
variations in loading many systems use an independent
oil pumping system which often provides better control
of oil injection but comes at the cost of additional power
loss. It is essential to have a guideline for a compression
system to vary the oil injection pressure for varying rotor
speed or discharge pressure and a computational model
can effectively provide this type of data. Additionally
the oil temperature at the point of injection is important
in achieving effective heat transfer.
Figure 1: Oil power loss contributions in leakages
(Reproduced from Deipenwisch and Kauder, 1999)
If the oil temperature is very close to the gas temperature at injection point then there will be low convection; if the
oil temperature is very low compared to gas temperature then due to a short residence time in the compression
chamber, the heat transfer will continue in the discharge port thus increasing the load on the oil cooling heat exchanger
downstream, without any benefit inside the compression chamber. It is required to develop a computational model
that can provide a complete description of the flow process inside the oil flooded compression chamber for its design
and optimisation.
Figure 2: Oil injection ports and injection angle on a compression volume curve
CFD can provide a useful means of modelling the oil injection and study the physical phenomenon in detail in order
to understand the complete behaviour of oil flooded twin screw compressors. Kovačević (2002, 2005) successfully
used an algebraic grid generation method with boundary adaptation and transfinite interpolation which has been
implemented in the program SCORG. Kovačević et al. (2007) have reported CFD simulations of twin screw machines
to predict flow, heat transfer, fluid-structure interaction, etc. Kovačević (2002) also reported a test case study of an oil
injected compressor using source terms in transport equations with a segregated pressure based solver. Vande Voorde
et al. (2005) used an algorithm for generating block structured mesh from the solution of the Laplace equation for
twin screw compressors and pumps using differential methods. Reports on analysis of dry air screw compressors and
twin screw expanders with real gas models are available in literature using these techniques (Papes et al., 2013).
Recently, Arjeneh et al. (2014) have presented the analysis of flow through the suction port of a screw compressor
with water injection. It was reported that it was difficult to stabilize the solver in a full 3D analysis with both deforming
rotor domains and multiphase models. Although several attempts have been made in the recent past to extend the CFD
technology to oil injected compressors, it has proven to be difficult to achieve the desired grid structure and the
modelling conditions that can provide stability to the numerical solvers. Rane (2015) in his thesis proposed a new
analytical approach for grid generation that can independently refine the interlobe region of the screw rotors. It was
demonstrated that such grid refinement improves the prediction of mass flow rates. The same algorithm has been
extended to produce a rotor grid that eliminates the interface between two rotors and thereby providing a desirable
grid for oil injected or multiphase modelling. An attempt has been made in the presented work to utilize the latest
developments in CFD technology for modelling oil injected twin screw compressors.
1023, Page 3
23
rd
International Compressor Engineering Conference at Purdue, July 11-14, 2016
2. APPROACHES FOR MODELLING OIL INJECTION IN SCREW COMPRESSORS
2.1 Thermodynamic chamber model approach
Stošić et al. (1992, 2005) have presented an oil injected twin screw compressor model by considering the working
chamber, suction and discharge process as an open thermodynamic system; mass flow varying with time and gas
defined via equation of state. In such type of lumped parameter models, oil droplets are assumed to have a mean Sauter
diameter and the heat exchange between the spherical droplets and the gas via convection can be balanced with the
droplet temperature rise. The differential form of this energy balance is given as Equation 1.
Using,
, heat transfer coefficient for Stokes flow.
And integration of the equation gives the oil droplet temperature at each time step.
is the oil droplet temperature at the previous time step.
, with
, being the non-dimensional time constant of the droplet.
For a given mean Sauter diameter
, the non-dimensional time constant is
(1)
If tends to zero, the oil and gas temperatures will be equal, while for finite values of , the gas and oil temperatures
will differ. The above described approach is based on the assumption that the oil-droplet time constant is smaller
than the droplet travelling time through the gas before it hits the rotor or casing wall, or reaches the compressor
discharge port.
Figure 3: Oil injection modelling using chamber models
This means that the heat exchange is completed within the droplet residence time through the gas during the
compression process. This prerequisite can be fulfilled by appropriate atomization of the injected oil which produces
sufficiently small droplet sizes producing sufficiently small droplet time constant, as well as by choosing adequate
nozzle angle and to some extent the initial oil spray velocity. Figure 3 shows a representative result of the gas and oil
temperature variation calculated using a multi-chamber thermodynamic model.
2.2 CFD approach using source term formulations
Computational fluid dynamic solvers calculate the solution of the conservation of mass, momentum, energy and other
transported quantities by numerically integrating the governing transport equations over the entire computational
domain. Equation 2 is the general transport equation in which the last term is a generic source term and can be
formulated in order to account for additional physicals effects that are not directly modelled in the solver.
1023, Page 4
23
rd
International Compressor Engineering Conference at Purdue, July 11-14, 2016
(2)
where, is the volume of CV, S is the surface enclosing this CV, is the unit vector orthogonal to S and directed
outwards, is the fluid velocity inside CV in a fixed coordinate system, is fluid density and is time. Here, the
transient term represents rate of change of property in CV, the convection term accounts for the net advective
transport of across surface S and diffusive transport is described by the gradient. is the diffusivity for the quantity
while
represents source or sink of in the CV. In the approach of modelling oil injection using source term
formulation, it is possible to model the thermal and sealing effects of the injected oil while still calculating the main
flow field as a single gas phase. Even in conditions where the mass of injected oil is 10 times the mass of gas, the
volume occupied by oil to that of gas is in the order of 1:100. Hence a passive scalar transport equation can be solved
to model the distribution and mixing of oil in the gas. The numerical solver will solve for continuity of the gas, its x,
y and z momentum components and total energy equation while the physical effects of oil injection will come from
applied at the oil concentration regions calculated by passive scalar transport or at selected regions of the domain.
Thermal effect: The source term formulation for energy equation is such that if the gas temperature exceeds the
injection oil temperature, an energy sink will become active. The assumption in this approach is that the heat transfer
between gas and oil is instantaneous and the mass of oil is sufficient enough to completely get into thermal equilibrium
with the gas. This energy source can be limited to the rotor domains and defined as in Equation 3.
,
(3)
is a source coefficient and is required in order to linearize the source in discretized equations.
is set at the oil
injection temperature. The source term is such that if the gas temperature exceeds oil injection temperature, heat is
removed from the computational cell. In this case a lower temperature limit source is required to stabilize the solution
process. Figure 4a presents the temperature field in the rotor domain produced by such an energy source formulation.
As seen the gas temperature is almost clipped to the oil injection temperature everywhere in the domain after the
compression process begins. A careful calibration of such formulation is required to get basic results such as pressure
rise and indicated power from the analysis. The accurate prediction of gas temperature rise and oil losses is challenging
and dependant on the calibration of the source term.
Figure 4: CFD approach for oil injection with source formulations
Sealing effect: A momentum source term in x, y, and z direction can be added to the momentum transport equations.
The formulation is presented in Equation 4 and it can be applied in the leakage gaps to reduce the velocity of the gas.
,
,
,
(4)