Paper No 7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Inhomogeneous Compressible Numerical Analysis with Phase Change of the Mlixing
Phenomenon in TwoPhase Ejector using CO2
Masafumi Nakagawa, Ariel Marasigan* and Takanori Matsukawa
Department of Mechanical and Structural System Engineering, Toyohashi University of Technology,
11 Hibarigaoka, Tempakucho, Toyohashi City, Aichi, 4418580, Japan
*Email: ariel iinak: mech tut ac jp
Keywords: inhomogeneous, compressible, EulerianEulerian model, twophase ejector, phase change
The effectiveness of the twophase ejector is highly dependent on the mixing characteristics of highspeed motive flow and
suction flow. It should be further studied to make the use of ejector particularly in CO, refrigeration cycle more viable. In this
study, numerical simulations of the mixing and pressure recovery process inside a twophase CO, ejector were carried out at
different operating conditions and verified with experimental data. The effects of compressibility and mass transfer due to
phase change were included. The flow regime was consisted of dispersed droplets in continuous vapor. The numerical results
have shown good agreement with the experimental data. The pressure recovery profile and its magnitude were largely affected
by the motive nozzle inlet conditions. The suction inlet pressure yielded minimal effect on these parameters. The effect of
phase change was considerably small for the cases used in this research while the energy conversion efficiency of the motive
nozzle has increased for increasing inlet pressure and lower inlet temperature. It confirmed our conclusion in our previous
study that this efficiency, as well as the coefficient of performance of the system, increases as the inlet condition moves to the
left of a Ph diagram. The results of this study provide useful information for ejector optimization and better understanding of
the process taking place inside a highspeed twophase CO, ejector.
(Sriveerakul et al. 2007, Varga et al. 2009), air (Yaday and
Patwardhan 2008, Hemidi et al. 2009), and Freonbased
refrigerants (Rusly et al. 2009, Zhu et al. 2009, Bartosiewicz
et al. 2006). To our best knowledge, there has been very
limited published research on the CFD modeling of
twophase ejector using CO2. We have been researching on
CO? ejector through experimentation and numerical
modeling (Nakagawa et al. 2004, Nakagawa et al. 2009a,
Nakagawa et al. 2009b). One of our research objectives is to
model numerically the dynamics inside the ejector using
CO2 as the working fluid.
The geometrical features of an ejector, shown in Figure
1, for CO2 refrigeration application are similar to the ejector
using other working fluids. However, the inlet of the motive
nozzle for the case of a CO? ejector is at transcritical state,
as shown in Figure 2. The fluid then expands to twophase
state inside the motive nozzle. This process generates
highspeed dispersed droplets in continuous vapor flow. The
outlet flow of the motive nozzle then entrains and mixes
with the suction vapor flow from the evaporator to facilitate
pressure recovery from the mixing section to the diffuser. A
full CFD model of this whole process is desirable but it
involves complex processes such as the expansion from the
transcritical state which is an obstacle to attain that objective.
This study focuses on the mixing phenomenon inside the
ejector using CO2 including compressibility effects and
phase change. The numerical results were verified with the
experimental data. The effectiveness of atwophase ejector
is highly dependent on the mixing characteristics of
highspeed motive flow and suction flow. A numerical
model of this complex process is essential in designing an
optimized ejector.
Introduction
The simple design of ejectors and its potential in
utilizing available energy make it practical for wide range of
applications such as power plants and aircraft. In a standard
vapor refrigeration cycle, the available energy dissipated in
the throttling process of the working fluid can be utilized by
replacing the expansion valve with a twophase ejector. Its
application in this field extends for different working fluids
but due to the worsening problem of global warming, it is
desirable to focus the study of twophase ejector using
alternative refrigerants. The natural refrigerant CO?. which
has zero ozone depletion potential and very low global
warming potential, is seriously being considered today to
replace conventional refrigerants. However, significant
amount of energy is lost during expansion process when it is
used in conventional refrigeration system. The use of
twophase ejector can address this problem and it should be
studied further to make the use of alternative natural
refrigerant CO2 practical and efficient. Experimentation and
numerical modeling aid in achieving this goal. The latter is
desirable to give a better picture of the dynamics of the
ejector operation. However, any numerical modeling should
be verified with experimental data.
One of the classical 1dimensional models of ejectors
was formulated by Keenan and Neumann (1942) using air as
the working fluid. Since then, a number of researches based
on thermodynamic and dynamic numerical modeling of
ejectors from 1dimensional to 3dimensional approaches
were published (He et al. 2009) In the advancement of
computational machines and codes, the use of computational
fluid dynamics (CFD) technique is deemed as more
appropriate to capture the actual flow inside the ejector.
Most of the researches based on CFD have used steam
1 Gas cooler
L '
5 o
Figure 3: Ejector assembly used in the experiment
Table 1: Locations of the thermocouple along the along the
longitudinal axis of the ejector with x=0 at mixing outlet
No. x (mm) No. 15mm
1, 2 suction nozzle 11 3 (mix)
3 21.35 (conv) 12 6.7 (mix)
4 18.75 (conv) 13 10.3 (mix)
5 15.55 (conv) 14 14.6 (mix out)
6 9.9 (div) 15 19.1 (dif)
7 8.05 (div) 16 25.6 (dif)
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
EXPERIMENTAL METHOD
In this study, an ejector with rectangular crosssection
WaS employed. This research was carried out for academic
7 purposes and this ejector design is easier to fabricate as
compared to circular type. The exploded view of the
ejector assembly is shown in Figure 3. The bottom part
WaS a 20mm thick bronze plate where holes were drilled as
Suction and motive nozzles inlets and an ejector outlet. 19
Ktype thermocouples were placed on the top wall along
the longitudinal axis of the ejector. These were fixed using
epoxy inside the 00).8mm holes drilled on the top Bakelite.
During experimentation, these thermocouples measured the
static temperature of the fluid trapped inside the 00).8mm
holes. Theoretically, the temperature readings, specifically
from mixing section to diffuser, can be used to obtain the
pressure profile in these sections since the flow is in
twophase state. In order to verify this, pressure gauges
were fixed instead of thermocouples and the system was
run at similar operating conditions. The results showed that
the saturation pressure based on thermocouple readings
were comparable to the pressure transducer measurements.
The horizontal positions of the thermocouples used in this
study are shown in Table 1 with origin set at the inlet of
the mixing section.
Paper No
MOOE
FLOW
DIOUENR
Figure 1: Illustration of a twophase ejector
 Ejector Cycle
 cntorwntnl Ccol*
CO,
Figure 2: Ph diagram of an ejector refrigeration
cycle using CO2
p. steel
SBakelite
SBake le
Bronze
x J U
z:
Epoxy
Symbol
th
g
h
IAD
P
T
TPC
U
Subscript
1
2
3
4
Quantity
nozzle mass flow rate
entrainment ratio
enthalpy
inlet section from motive
nozzle
inlet section from suction
nozzle
pressure
slip ratio
temperature
Thermal Phase Change
axial velocity
quality
density
efficiency
motive nozzle inlet
motive nozzle outlet
suction nozzle inlet
suction nozzle outlet
mixing inlet
diffuser inlet
diffuser outlet
gas cooler outlet
converging nozzle
diffuser
diverging nozzle
motive nozzle
maximum
evaporator
ejector
profile
recovery
suction nozzle
SI Unit
kg/s
kJ/kg
EJE top plate
EJE center plate
EJE bottom plate 4
SUCTION IN
MOTIVE IN
SUCTION IN
kg/m3
S10
2 8
!4
Enthalpy (kJ/kg)
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
pressure and temperature.
WATER IN WATER OUT
Paper No
The main ejector assembly was composed of three
stainless plates and the schematics of the center plate are
shown in Figure 4. The motive nozzle had a throat width of
0.4mm, divergence angle of 0.970 and divergence length of
11.5mm. This nozzle design was based on our previous
experimental data and numerical analysis using isentropic
homogenous equilibrium model (Nakagawa and Morimune
2003). The ejector mixing area ratios which is defined as
the ratio of mixing area to the motive nozzle outlet area
was 4.22. The mixing length was 15mm while the diffuser
section had divergence angle and length of 5.20 and
30.5mm, respectively. The basis of this design was also
from the result of our previous study using 1dimensional
3velocity numerical model of the mixing process for R12
ejector system (Nakagawa et al. 1994). The whole ejector
assembly was tested for leakage at a pressure of at least
10MPa before installing it in the main system.
O O O O
R3 5
15 O
re In millimeter
O
x
( O
dimensions a
Figure 4: Schematics of the center ejector metal plate
The experimental setup used in this study, as shown in
Figure 5, was a modified conventional refrigeration system
with heat exchanger. An internal heat exchanger with an
effectiveness of about 35% was employed between the
inlet of the compressor and outlet of the gas cooler.
Pressure transducers and thermocouples were placed
between system components to monitor and maintain the
desired operating conditions and to obtain the required
parameters to evaluate the fluid properties. The suction
flow rate sh,~ was obtained based on the enthalpy
difference of the air passing through the evaporator. The
required parameters for the air flow were measured using
thermocouples, differential pressure transmitter and
hygrometer. The motive flow rate shz was calculated
using heat balance in the gas cooler where circulating
water was used as heat sink. The evaporator fan and
compressor were controlled using inverters. The accuracies
of these measuring devices are summarized in Table 2. The
experiments were carried out at different operating
AIR IN I~T
S: pressure transducer T* : thermocouple
Figure 5: Schematic diagram of the experimental setup
Table 2. Accuracy of measuring devices used in the
experiment
Measuring Device Accuracy
Pressure transducer +().25% full scale
Thermocouple +().5oC actual reading
Differential pressure transmitter 1.()% full scale
Hygronieter 2% actual reading
NUMERICAL MODEL
Domain Setup
The numerical calculation was carried out using the
commercial software ANSYS CFX11. The details of the
numerical models that will be presented can be found from
the software's documentation (ANSYS User's Guide v.11).
The inhomogeneous steadystate EulerianEulerian
approach was used and the flow regime was set as dispersed
liquid droplets in continuous vapor. The Particle model was
used for modeling interphase transfer. The droplet was
assumed to have a constant diameter of 10pLm which was
estimated from the data of Padorski et al. (1980). The
kepsilon and Schiller Naumann models were employed to
model the turbulence effects and interphase momentum
transfer, respectively. The Total Energy heat transfer model
was employed to include compressibility effects. The CO,
fluid properties were obtained using Redlich Kwong
equation state provided in CFX. The simulations were
carried out for both cases of with and without mass transfer
'vhich was modeled using Thermal Phase Change Model
and the Two Resistance Model was used to calculate the
heat transfer coefficient.
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Paper No
INS1
IND m
IN S2
Figure 6: Computational mesh
The computational mesh, as shown in Figure 6, was
generated using the ANSYS ICEM CFD code. The control
volume of the CFD model was primarily consisted of
mixing section and diffuser to focus on the dynamics of the
mixing phenomenon of the motive and suction flows. This
domain is represented by the region enclosed by broken
lines in Figure 4. The mesh was constructed such that the
orthogonality, expansion factor and aspect ratio were within
the acceptable range. Several geometrical meshes with
increasing number of elements were first used in this study
until grid independence was obtained. Symmetric
assumption, which is not obvious because of the geometric
design, was not employed in this study to better capture the
actual physics of the flow.
Boundary Conditions
The inlet section IND of the fluid flowing from the
motive nozzle was located 2mm before the mixing inlet or at
the location of the last thermocouple along the motive
nozzle's longitudinal axis. The inlet condition for this
section was set to mixed subsonic and supersonic flow. In
this case, the temperature and pressure must be set which
were obtained from the experimental data. Inlet velocity of
the fluid was also set at this section. The mean velocity of
the each phase was calculated based on the following
equation:
U, (1)
(s2 1)x +1
Where, the sub scripts 1 and 2s correspond to the actual inlet
and isentropic outlet of the motive nozzle, respectively. The
parameters h, s, iln, and x denote enthalpy, slip ratio, motive
nozzle isentropic efficiency and quality. A constant slip ratio
equal to 1 was assumed at this inlet section. Several values
of motive nozzle efficiency were used in the calculation to
obtain the pressure profile and compare it with experimental
data. The vapor void fraction was calculated using the
following relation:
a = (2)
1+1 xX JP s ;7
In order to better represent the fluid velocity at IND, a
nonuniform velocity was employed using the 1/7" power
law which
expressions:
was calculated
based on the following
U (y. z) = Umax
, u = ( i+
Where t and b are the thickness and base, respectively, of
section ID and n was set to 7.
Meanwhile, the location of the inlet section INS which
corresponds to the inlet of the fluid flowing to the domain
from the suction nozzle was decided based on some initial
calculations. At first, the whole actual suction nozzle was
modeled and the results were compared with that of using
the grid presented in Figure 6. The results showed negligible
difference so that the latter was used for other calculations.
The inlet condition for this section was set to mass flow rate
using the experimental dataliz~and the flow was assumed to
be pure vapor. The inlet temperature at this section was also
set based on experimental data.
The outlet boundary condition was defined using the
pressure obtained from experiment. The wall was set to
adiabatic and the wall influence on the continuous and
dispersed phases was set to no slip and free slip, respectively.
The convergence was judged by lowering the residuals until
le5, monitoring pressure and temperature at different
points until it stabilizes, and until the mass imbalance
reaches the value of le5 kg/s. However, recirculation zone
near the outlet of the diffuser section was obtained at high
inlet pressure P,,, which made convergence based on
residuals difficult to achieve. In such case, the stability of
the values of monitored points and imbalance can be used as
convergence criteria (ANSYS USER's Guide).
RESULTS AND DISCUSSION
The CFX results were validated by comparing the
magnitude of suction inlet pressure and the pressure
recovery profile inside the ejector with the experimental
data. In a trancritical CO, ejector refrigeration cycle, the
three operating parameters; motive nozzle inlet pressure or
the gas cooler pressure P,,,, gas cooler outlet temperature T,
and evaporator temperature Te are important in accessing the
performance of the system. The numerical simulation for
each set of operating conditions was carried out for the cases
Of without mass transfer (noTPC) and with mass transfer
L
IIIIIIIIIII
0 10 20 30 40 50
x [mm]
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Paper No
using Thermal Phase Change Model (TPC). In addition to
that, each case was performed using different motive nozzle
inlet velocities which were calculated based on the assumed
values of motive isentropic efficiency im. The following
graphs show the comparison of experimental suction
pressure and experimental pressure recovery profile with the
calculated results. These values for numerical results were
connected from suction inlet to mixing section inlet for
better visualization. The other lines indicate the centerline
velocities of gas and droplet phases as a function of the axial
distance and only the case of with mass transfer model was
included in this research.
u 7 U.1.1 TPC Papmm 11 5%.nTPC
u =1,70%, TPC .Ppme ilm=75% nTPc
  ~~ u=70% TPC m%nTc
O3
I Pn=9MPa
~~To=45oC Te=5"C


300
250
200 C
150 &i
100
so
4.4
a 4.2
4.0
30 40 50
0 10 20
x [mm]
(a)
Varying Motive N~ozzle Inlet Pressure P,,,
_ u~~~% TPC
u .=80% TPC
 U~ 1=80%. TPC
* Papenm 1=5%, noTPc
 P r~ '1m=85%. TPC
 P~, ror =80%, noTPC
The results for varying inlet pressure Pn from 9MPa to
11MPa are shown in Figures 7a to 7d. In this case, the gas
cooler outlet temperature To and evaporator temperature Te
were kept constant at 450C and 50C, respectively. The
numerical results show good agreement with respect to the
experimental pressure data. The difference between the
results of with and without mass transfer was almost
negligible for both pressure and velocity profiles. It suggests
the minimal effect of mass transfer in the mixing process
inside a twophase CO2 ejector.
The motive nozzle efficiency rlm was varied from 70%
to 90% for the range of Pn used in this section. This was
done since the motive nozzle outlet velocities of gas and
droplet phases are not well established. The basis of the
assumed values of rlm was the experimental suction pressure
data and the experimental pressure recovery profile. It can
be seen from the graphs that the total pressure recovery Pre,
which is the difference between the suction inlet and ejector
outlet, changes with respect to the value of im. The total
pressure recovery increased as rlm increased because of
higher energy input at higher values of im. The value of rm
which would best fit the experimental data curve has
increased for increasing value of Pn. It implies the higher
efficiency of motive nozzle in converting pressure energy to
kinetic energy at higher inlet pressure.
In the values of Pn used for this section, the numerical
simulation for 10MPa and 10.5MPa yielded excellent results
in comparison with the experimental data. At lower
P,,=9MPa, the numerical results followed the experiment
data except the region near the mixing outlet and diffuser
inlet. The pressure profile from numerical calculation was
higher than the experimental data which means that the
kinetic energy at this region was converted to pressure
energy more efficiently than the actual process. The trend
was different for higher P,, as shown in Figure 7d. In this
case, the numerical simulation yielded better results except
the inside region of the mixing section. It means the
conversion from kinetic to pressure energy was insufficient
using the numerical simulation.
The results for centerline velocity profile for the cases
of with and without mass transfer were almost the same so
that only the numerical results with mass transfer model are
shown in the graphs. In all of the values of P,, used in this
section, the centerline velocities of liquid and gas phases
had negligible difference. Small velocity difference was
obtained near the inlet of the mixing section but this has
disappeared for higher Pan
250
200
c
150 3
100
50
 P~~ 1m=80%, TPc
Ic IXn= 10MPa
u, T = 450C Te= 50C
0 4.2
an
3.8 E
.. u ~l=9 %, TPC
 U n =90%. TPC
 Up a =80%. TPC
 u om=80%. TPC
e penment
fror~, r~m=904
 P or, 11 =904
 P~ orl, a=809
250
200
C
150
100
So
4q.2
04.0
1
30 40 50
0 10 20
x [mm]
(c)
 U. 11=90%. TPC
 U =90%. TPC
u.p=80%,TPC
 4 n,,=80%, TPC
* P
 P~~ ,,=904
P ~~=0
250
200
150 .
100
so
,2
,4.0
30 40 500
0 10 20
x [mm]
Figure 7: Comparison of numerical results and experimental
data and the calculated horizontal centerline gas velocity U,
 U. 11m=90%, TPC experiment
 u~, =90%. TPC  P,.,ir m=90%, noTPc 
 U up =80%, TPC Ppror m,=90%, TPc
 u~,m=80% TPC ',i ~8)%. noTPC 
. ~1,=0. C
'z r I
j~ jP p~= 10MPa
W. To=420C T= 20C
 
0 10 20 30 40 5
x [mm]
(a)
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Paper No
and liquid velocity UI for varying motive nozzle inlet
pressure Pn at (a) 9MPa, (b) 10MPa, (c) 10.5MPa and (d)
11MPa
 ul,, =90%. TPC
 U Em=90%,. TPC
 u.1=80%. TPC
 U~ Im=80%,TPC
 P,
Pn
xpenment
ror '1m=90%. noTPC
ror 1im=90%. TPC
ror 1m=80%, noTPC
ror Em=80%. TPC
P= 10MPa
= 420C T="
250
200
150
100
so
a
24.4
4.2
4.0
Varying Evaporator Temperature Te
The results for varying evaporator temperature Te from
20C to 80C at P,=10MPa and T,=420C are shown in Figures
7a to 7d. The numerical results again show good agreement
with respect to the experimental pressure data. The
negligible effect of including the mass transfer in the
calculation was also obtained. This trend was obtained for
both pressure and velocity profiles.
The numerical simulation for this section was only
carried out for values of rlm equal to 80% and 90%. This was
enough to obtamn results comparable to the experimental
data. It shows that the suction pressure has minimal effect
on the motive nozzle performance which is logical since the
inlet conditions of the motive nozzle were not varied.
However, it is interesting to point out that at these different
values of suction pressure, the entrainment ratio had
negligible difference as obtained from the experiment.
c eh
ST.
30 40 50
I_
0 10 20
x [mm]
Figure 8. Comparison of numerical results and experimental
data and the calculated horizontal centerline gas velocity U,
and liquid velocity UI for varying evaporator temperature Te
at (a) 20C, (b) 50C and (c) 80C
The difference between the experimental and numerical
results of pressure profile can be seen near the central region
of the mixing section for all values of Te. It should be noted
that although the inlet pressure is at 10MPa, the temperature
To is lower as compared to the case of varying pressure Pn.
The profiles of both experimental and numerical had almost
maintained a constant shape even though the suction
preSsure has changed. The only difference is the magnitude
of the minimum and maximum pressure for each graph. It is
in contrast for the case of varying Pn where the shape of
pressure recovery had changed for a different value of Pn.
Also, the difference between the centerline velocities for
each phase was negligible.
Varying Gas Cooler Outlet Temperature To
The actual inlet temperature of the ejector was lower
than the gas cooler outlet temperature Te since internal heat
exchanger was employed between these two points. The
average temperature drop at the high pressure side of the
heat exchanger was about 7 degrees. The effect of
temperature To can be obtained by analyzing Figure 7b and
8b. In these two graphs, the pressure Pn and temperature Te
were kept at 10MPa and 50C, respectively while the
temperature To was changed from 420C to 450C.
The value of rlm which gave better numerical results in
comparison with the experimental total pressure recovery
was higher than for the case of lower Tc. However, the
numerical results for higher To yielded better approximation
of the pressure curve. At lower T,, the numerical results of
the pressure recovery profile inside the mixing section had
sginificant deviation with the experimental data. The slip
velocity near the inlet of mixing section was negligible for
lower Tc. The results in this section showed that not only Pn
had significant effect on the mixing and pressure recovery
process. The temperature To also had considerable effect on
this process.
The numerical results have shown good agreement
with experimental data in all the conditions used in this
research with some deviation in some cases. The numerical
modeling can be improved by establishing a better
4.2
4.0
~23.8
S3.6
3.4
3.2
4.6
250
200
C
150~
50
0
0
250
200
150 C
100
so
 U. 1m=9 %. TPC
U~.m=90%. TPC
 U. m=80%, TPC
 U =180%. TPC
experiment
P~.relm=90%, noTPc
pre PEm,1=90%. TPC
 Pr~, 11=80%, noTPC
 Pr~, Em=80%, TPc
P ,= 10MPa
To=420C Te= 50C
`L.
30 40 500
0 10 20
x [mm]
(b)
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
Nakagawa, M., Harada, A. and Berana, M. Analysis of
Expansion Waves Appearing in the Outlets of TwoPhase
Flow Nozzles. ASHRAE HVAC&R Research, Vol.15,
pp.10651079 (2009a)
Nakagawa, M., Marasigan, A.R. and Matsukawa, T.
Experimental analysis on the effect of internal heat
exchanger in transcritical CO2 refrigeration cycle with
twophase ejector. International Joumnal of Refrigeration. In
Press (2009b)
Nakagawa, M. and Morimune, Y. Subsequent Report on
Nozzle Efficiency of TwoPhase Ejector Used in Carbon
Dioxide Refrigerator, Thermal Sci. and Eng'g, Vol. 11,
pp.1011 (2003)
Padovski I.S., Trelin Y.S. and Zaitsey V.P., Experimental
Investigation of the Sound Speed of Dispersion in Moist
Carbon Dioxide Vapor, Translated from
InzhenemnoFizicheskii Zhumnal, Vol. 39, pp. 10771083
(1980)
Rusly, E., Lu, A., Charters, W. and Ooi, A. CFD analysis of
ejector in a combined ejector cooling system. International
Journal of Refrigeration. Vol.28, pp.1092101 (2005)
Sriveerakul T., Aphomnratana, S. and Chunnanond, K.
Performance prediction of steam ejector using
computational fluid dynamics: Part 1. Validation of the CFD
results. International Joumnal of Thermal Sciences, Vol. 46
(8), pp.812822 (2007)
Varga, S., Oliveira, A.C. and Diaconu, B. Numerical
assessment of steam ejector efficiencies using CFD.
International Joumnal of Refrigeration, Vol. 32(6),
pp.12031211 (2009)
Yaday, R.L. and Patwardhan, A.W. Design aspects of
ejectors: Effects of suction chamber geometry. Chemical
Engineering Science, Vol. 63(15), pp.38863897 (2008)
Zhu, Y., Cai, W., Wen, C. and Li, Y. Numerical investigation
of geometry parameters for design of high performance
ejectors. Applied Thermal Engineering, Vol. 29 (56), pp.
898905
(2009)
Paper No
correlation of the droplet size and by obtaining more
accurate motive nozzle outlet velocity. These potentials for
improvement are the subject of our ongoing research. The
numerical results also confirmed our conclusion in our
previous study that the motive nozzle and ejector becomes
more efficient as the motive inlet moves to the left of the
Ph diagram and also that the suction pressure has minimal
effect on ejector performance as compared to P,,, and T,.
The results of this numerical study aid in better
understanding the highspeed dynamics inside the CO2
twophase ejector are shown in
CONCLUSIONS
A numerical simulation using commercial CFD code
was used in this research to model the mixing and pressure
recovery process inside a twophase CO2 two phase ejector.
The pressure profiles obtained from CFD model have shown
good agreement with experimental data. The pressure
recovery profile and the magnitude of pressure recovery are
largely affected by the motive nozzle inlet condition. The
suction pressure yielded minimal effect on these parameters.
The effect of phase change was considerably small for the
cases used in this research. The centerline velocity of gas
and liquid phases had negligible difference particularly
during the start of pressure recovery. The energy conversion
efficiency of the motive nozzle has increased for increasing
inlet pressure and lower inlet temperature. The numerical
results confirmed our conclusion in our previous study that
this efficiency, as well as the experimental coefficient of
performance of the system, increases as the inlet condition
moves to the left of a Ph diagram. The results of this study
provide useful information on how to realize an optimum
CO2 ejector refrigeration system.
REFERENCES
ANSYS USER's Guide CFX release 11.0
Bartosiewicz, Y., Aidoun, Z. and Mercadier, Y. Numerical
Assessment of Ejector Operation for Refrigerant
Applications Based on CFD," Applied Thermal Engineering,
26, pp. 604612 (2006)
He, S., Li, Y. and Wang, R.Z. Progress of mathematical
modeling on ejectors. Renewable and Sustainable Energy
Reviews, Vol.13 (8), pp.17601780 (2009)
Hemidi, A., Henry, F., Leclaire, S., Seynhaeve, J. and
Bartosiewicz, Y. CFD analysis of a supersonic air ejector.
Part I: Experimental validation of singlephase and
twophase operation. Applied Thermal Engineering, Vol. 29
(89), pp.15231531 (2009)
Keenan, J.H. and Neumann, E.P. A simple air ejector. ASME
Journal of Applied Mechanics, Vol. 64, pp.7581 (1942)
Nakagawa, M., Berana, M. and Kishine, A. Supersonic
twophase flow of CO2 through convergingdiverging
nozzles for the ejector refrigeration cycle. International
Journal of Refrigeration, Vol. 32 (6), pp.1951202 (2009)
