You are on page 1of 11

Applied Thermal Engineering 52 (2013) 360e370

Contents lists available at SciVerse ScienceDirect

Applied Thermal Engineering


journal homepage: www.elsevier.com/locate/apthermeng

Study of ejector efciencies in refrigeration cycles


Fang Liu*, Eckhard A. Groll
School of Mechanical Engineering, Ray W. Herrick Laboratories, Purdue University, 140 S. Martin Jischke Drive, West Lafayette, IN 47907, USA

h i g h l i g h t s
< Ejector efciencies were determined using ejector model and measured data.
< Ejector efciencies vary with ejector geometry and operating condition.
< Empirical equations of ejector component efciencies were established.

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 29 September 2012
Accepted 6 December 2012
Available online 17 December 2012

This paper presents a method of determining internal ejector component efciencies using a two-phase
ow ejector model. The measured performance data external to the ejector in conjunction with the
ejector model were used to determine the isentropic efciencies of the motive and suction nozzles, and
the efciency of mixing section. It was found that ejector component efciencies vary with ejector
geometries and operating conditions, which are different from the constant efciencies assumed by
previous researchers. The efciencies of the controllable ejector motive nozzle, suction nozzle and
mixing section range from 0.50 to 0.93, from 0.37 to 0.90, and 0.50 to 1.00, respectively, in CO2 refrigeration cycles under different operating conditions. Empirical correlations were established and incorporated into a CO2 air conditioning system model to estimate ejector component efciencies at different
ejector geometries and operating conditions.
2012 Elsevier Ltd. All rights reserved.

Keywords:
Ejector efciencies
Controllable ejector model
Empirical correlations
Refrigeration cycle

1. Introduction
The expansion losses of an isenthalpic throttling process have
been identied as one of the largest irreversibilities of transcritical
refrigeration cycles, which contribute to the low efciency of such
cycles. An ejector expansion device was proposed to recover the
expansion losses and increase the cycle efciency. The ejector
expansion refrigeration cycle consists of an ejector, a separator,
a compressor, a gas cooler, an evaporator, and control valves as
shown in Fig. 1(a). Fig. 1(b) shows a schematic of an ejector, which
consists of motive nozzle, suction nozzle, mixing section and
diffuser. Fig. 1(c) shows the ejector working process in a CO2 pressure enthalpy diagram. The high pressure motive stream expands in
the motive nozzle and entrains the low pressure suction stream into
the mixing section. The motive nozzle and suction nozzle processes
are treated as the expansion processes of converging nozzles in
order to simplify the analysis. The two streams are then mixed in
the mixing section, and the mixed stream ows through the
diffuser increasing its pressure along the way.

* Corresponding author. Tel.: 1 765 496 2201; fax: 1 765 494 0787.
E-mail address: louisaiu@gmail.com (F. Liu).
1359-4311/$ e see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.applthermaleng.2012.12.001

The COP of the ejector cycle is very sensitive to the ejector


efciency [1]. However, there is very limited research on ejector
efciency. In most of the literature studies, values of 0.7e0.95 were
assumed for the individual ejector component efciencies as listed
in Table 1 [1e17]. Vereda et al. [2] used the xed values 0.85, 0.9 and
0.8 for ejector nozzle efciency, ejector mixing efciency, and
ejector diffuser efciency, respectively, in their simulation study of
an ejector-absorption (single-effect) refrigeration cycle. Li and Groll
[3] assumed the efciencies of ejector motive and suction nozzles
both of 0.9, and diffuser efciency of 0.8 in their model of an ejector
CO2 refrigeration cycle. Ksayer and Clodic [4] assumed the efciencies of motive and suction nozzles both of 0.85, and diffuser
efciency of 0.75 in a constant pressure-mixing-zone model for the
CO2 ejector. Deng et al. [6] assumed ejector motive nozzle efciency
of 0.7 and suction nozzle efciency of 0.8 in their theoretical
analysis of a transcritical CO2 ejector expansion refrigeration cycle.
Sarkar [7] assumed nozzle efciency of 0.8 and diffuser efciency of
0.75 in the mathematical modeling and simulation for ejector
expansion transcritical CO2 heat pump cycles. Elbel and Hrnjak [9]
assumed values of 0.8 for the individual ejector component efciencies in their study of ejector in CO2 systems. Sun [8] analyzed
the effect of ejector geometries on performance of a 5-kW steamjet refrigerator, and assumed nozzle efciency of 0.85 and

F. Liu, E.A. Groll / Applied Thermal Engineering 52 (2013) 360e370

Nomenclature
A
COP
D
h
L1
L2

_
m
P
Q_
RH
T
V
v
y

area (m2)
coefcient of performance ()
diameter (mm)
specic enthalpy (kJ/kg)
the distance measured to calculate the motive nozzle
throat area (mm)
the distance measured to calculate the distance
between motive nozzle outlet and mixing section
constant-area entry (mm)
mass ow rate (kg/s)
pressure (MPa)
cooling capacity (kW)
relative humidity (%)
temperature ( C)
velocity (m/s)
specic volume (m3/kg)
the distance between motive nozzle exit and mixing
section constant-area entry (mm)
density (kg m3)

diffuser efciency of 0.85 in the optimizing calculations of ejector


geometry. During the modeling study of ejector cycle with R141b,
by xing motive nozzle efciency at 0.95, suction nozzle efciency
at 1, and diffuser efciency at 1, Ksayer [5] found that the mixing
efciency varies between 0.9 and 0.98 and depends on the diameter ratio of the nozzle throat and the constant area diameter. In the
literature, ejector efciencies were assumed as constant values by
other researchers as shown in Table 1. Measured individual ejector
component efciencies (the efciencies of motive nozzle, suction
nozzle, and mixing section) have not been reported in the literature. Few studies have been found in the literature on the empirical
correlations of ejector component efciencies. However, the ejector
geometries and operating conditions affect the ejector efciencies
signicantly [5,25].
Therefore, in this study, a method was introduced to determine
the efciencies of the ejector motive nozzle, suction nozzle, and
mixing section based on measured data using a validated ejector
model. This method is illustrated with a case study of a controllable ejector in transcritical CO2 air conditioning systems.
Furthermore, empirical equations were established to estimate
individual ejector component efciencies in refrigeration cycles
with various ejector geometries under different operating
conditions.
2. Method of determining the internal ejector efciencies

quality
efciency
_m
_ s =m
ejection ratio, 4 m

Subscripts
b
suction nozzle exit
c
critical, cooling
d
diffuser, discharge
f
uid
g
vapor
id
indoor
is
isentropic
m
motive nozzle
mix
mixing section
od
outdoor
s
suction nozzle
t
nozzle throat
tr
triple point
Superscripts
derivatives

2.2. Ejector model


In this study, a model developed to simulate a two-phase ow
ejector consists of four sub-models: motive nozzle ow, suction
nozzle ow, mixing section ow and diffuser ow models.
2.2.1. Model of motive nozzle ow
The motive nozzle of the ejector is modeled using the following
assumptions:
 The ow inside the motive nozzle is steady and onedimensional.
 The nozzle is a converging nozzle and its throat is at its exit.
 At the nozzle throat, the ow reaches the critical ow
condition.
 The isentropic efciency of the nozzle, hm, is given.
 The inlet ow velocity is neglected.
 The heat transfer between the uid and the nozzle wall is
neglected.
 The gravitational force effect on the ow is neglected.
Using the motive nozzle inlet pressure pm and temperature Tm,
and the isentropic efciency of the motive nozzle, the following
equations are solved for the motive nozzle outlet pressure pt and
velocity Vt.
The isentropic efciency of the motive nozzle is dened by:

2.1. Flow chart for determining ejector component efciencies


The method of determining ejector component efciencies is
based on the ejector model and measured parameters (pressures,
temperatures, and mass ow rates at the inlet to each of the motive
and suction nozzles, and ejector discharge pressure). The twophase ow ejector model was utilized to determine the efciencies of motive nozzle, suction nozzle, and mixing section using the
measured data. Fig. 2 shows an overall ow chart, which outlines
that the isentropic efciencies of motive nozzle and suction nozzle
are determined by matching the predicted and measured mass ow
rates through motive nozzle and suction nozzle, respectively, and
that the mixing section efciency is determined by matching the
predicted and measured diffuser discharge pressures.

361

hm

hm  ht
hm  ht;is

(1)

Using Eq. (1), the outlet enthalpy ht can be calculated for a given
nozzle efciency of hm. Then, the outlet velocity Vt can be determined by using the energy conservation equation between the inlet
and outlet of the motive nozzle as follows:

hm ht

Vt2
2

(2)

With an assumed value of the outlet pressure pt and the outlet


enthalpy ht determined from Eq. (1), the quality xt, can be determined since the two independent state points of pt and ht are

362

F. Liu, E.A. Groll / Applied Thermal Engineering 52 (2013) 360e370

Qgc
2

3
Control
Valve

Compressor

Wc

Gas Cooler

Ejector

Separator

6
Evaporator

Control
Valve

Qeevap
vap

(a) Schematic of ejector expansion refrigeration cycle

Motive
nozzle

Suction nozzle

Mixing section

2 Diffuser

Pt

Pm

Pd

Pmix

Pb
Ps
(b) Schematic of ejector expansion device
CarbonDioxide

130

P [b a r]

100

d
s
t

30
-300

mix

-200

-100

h[kJ/kg]
Note: The motive stream (from point m to t) expands from P m to Pt in the motive nozzle, while the suction stream
(from point s to b) expands from P s to Pb. Then, the two streams are mixed to become one stream in the mixing
section (from point t to mix and b to mix). This mixed stream flows throughout the diffuser (from point mix to d)
while increasing its pressure from P mix to Pd.

(c) Ejector working processes in a CO2 pressure-enthalpy diagram


Fig. 1. Ejector expansion refrigeration cycle.

F. Liu, E.A. Groll / Applied Thermal Engineering 52 (2013) 360e370


Table 1
Summary of literature e assumed ejector component efciencies in modeling
studies [1e17].
Authors

Fluid

hm

hs

Keenan et al. (1950)


Alexis and Rogdakis
(2003)
Sun (1996)

Air
Water

1.0
0.7

1.0

LiBreH2O/
H2OeNH3
Ammonia/
lithium
nitrate
R134a
R123
R141b
R142b
CO2
CO2
CO2
R141b
CO2
CO2
CO2
CO2
CO2

0.85

0.85

0.85

0.85

0.85e0.9
0.85
0.9
0.85
0.9
0.9
0.85
0.95
0.7
0.8
0.8
0.9
0.7

0.85e0.9
0.85

Vereda et al. (2012)

Domanski (2005)
Yapici and Ersoy (2005)
Yu and Li (2006)
Yu et al. (2007)
Elbel and Hrnjak (2004)
Li and Groll (2005)
Ksayer and Clodic (2006)
Ksayer (2007)
Deng et al. (2007)
Sarkar (2008)
Elbel and Hrnjak (2008)
Sun and Ma (2011)
Eskandari Manjili and
Yavari (2012)

hmix

0.85

0.85
0.95
0.9
0.9
0.85
1.0
0.7
0.8
0.8
0.9
0.7

0.9e0.98

0.95

 The heat transfer between the uid and the nozzle wall is
neglected.
 The gravitational force effect on the ow is neglected.

d
0.8

0.9

0.8

0.7
0.85
0.85
0.85
0.9
0.8
0.75
1
0.8
0.75
0.8
0.8
0.8

Once the mass ow rate through the motive nozzle is deter_ s through the suction nozzle can be
mined, the mass ow rate m
determined as:

_ s 4m
_m
m

Vc

hs

hs  hb
hs  hb;is

(3)

(4)

where the ow density rt at the nozzle throat can be calculated as:

rt

xt

rg;t

1
1  xt

(5)

rf ;t

In the motive nozzle model, it is implicitly assumed that the exit


stream is in the two-phase region. It is also assumed that the ow
becomes critical at the nozzle exit. For each simulation, these two
assumptions are checked for their validity. This is done through an
iteration procedure depicted in Fig. 3. It can be seen from Fig. 3 that
the model searches for the throat pressure pt at pressures below the
critical pressure pc and at pressures above the critical pressure to
match the outlet velocity Vt with the speed of sound Vc in twophase ow for the given input parameters.
2.2.2. Model of suction nozzle ow
In real ejectors, the suction nozzle process typically occurs in
a suction chamber with a complicated geometry. To simplify the
analysis, the expansion process of the suction nozzle is modeled in
the same way as the expansion process of a converging nozzle using
the following assumptions:
 The ow is steady and one-dimensional.
 The inlet ow velocity is neglected.

(7)

By initially assuming the outlet pressure pb, the outlet enthalpy


hb can be calculated for a given isentropic efciency hs from Eq. (7).

The speed of sound Vc is then compared with the outlet velocity


Vt, and the outlet pressure pt is updated until the speed of sound Vc
and the outlet velocity Vt are converged.
With the outlet velocity determined from Eq. (1), the mass ow
rate through the motive nozzle can be determined for the given
throat area of At by using the following equation:

_ m rt At Vt
m

(6)

where 4 is the ejection ratio.


Using the suction nozzle inlet pressure ps and enthalpy hs, the
suction nozzle isentropic efciency hs and outlet area Ab, the outlet
pressure pb and velocity Vb can be determined through the
following procedure.
The isentropic efciency of the suction nozzle is dened as:

known. Once the quality xt is determined, the speed of sound Vc, can
be calculated using the following Eq. (3) as [18]:

!1


2
v2mix hg  hf
 0




vg  vf hmix  vmix  v0mix hg  hf

363

Fig. 2. Flow chart to determine ejector component efciencies.

364

F. Liu, E.A. Groll / Applied Thermal Engineering 52 (2013) 360e370

Fig. 3. Flow chart of motive nozzle simulation.

Then, the outlet ow velocity Vb can be determined by using the


energy conservation equation between the inlet and outlet of the
suction nozzle as follows:

hs hb

Vb2
2

(8)

With the density rb determined from the assumed outlet pressure pb and the enthalpy hb determined from Eq. (7), the outlet
velocity Vb can be determined from the mass conservation
equation:

_ s rb Ab Vb
m

(9)

Finally, the two different values of the outlet velocity Vb from


Eqs. (8) and (9) are compared to each other, and the outlet pressure
pb is updated until the two values are converged. It should be noted
that under typical operating conditions, critical ow will not occur
in the suction nozzle because the pressure differences between the
inlet and outlet of the suction nozzle are small.
2.2.3. Model of mixing section ow
The mixing section of the ejector begins at the outlets of the
motive and suction nozzles and ends at the inlet to the diffuser as
shown in Fig. 1(b). To simplify the modeling of the mixing section,
the following assumptions are made:
 At the inlet plane 1, the motive stream has a velocity Vt,
a pressure pt, and occupies the area At.

 At the inlet plane 1, the suction stream has a velocity Vb,


a pressure pb, and occupies the area Ab.
 At the outlet plane 2, the ow becomes uniform and has
a velocity Vmix and a pressure pmix.
 The heat transfer between the uid and the mixing section wall
is neglected.
 The gravitational force effect is neglected.
Using these assumptions, the model is solved for the mixing
stream velocity Vmix and pressure pmix, and enthalpy hmix after the
motive stream velocity Vt and pressure pt, and the suction stream
velocity Vb and pressure pb are calculated.
The mass conservation equation between the inlet and outlet
planes is as follows:

rt At Vt rb Ab Vb rmix Amix Vmix

(10)

The mixing section efciency hmix is assumed to account for the


friction losses in the mixing chamber [19]. With the assumed
mixing section efciency, the momentum conservation equation
between the inlet plane and the outlet plane is reduced as:

pt At hmix rt At Vt2 pb Amix  At hmix rb Amix  At Vb2


2
pmix Amix rmix Amix Vmix

(11)

The energy conservation equation between the inlet plane and


the outlet plane is:

F. Liu, E.A. Groll / Applied Thermal Engineering 52 (2013) 360e370

_ m ht
m

Vt2
2

!
_ s hb
m

Vb2
2

!
_ m m
_ s hmix
m

2
Vmix
2

(12)

Since the density rmix in Eqs. (10) and (11) is a function of the
pressure pmix and the enthalpy hmix, the only unknowns in Eqs.
(10)e(12) are the pressure pmix, velocity Vmix and enthalpy hmix,
which can be determined by solving the three equations simultaneously. The pressure pmix and enthalpy hmix are then used to
determine the quality xmix. The speed of sound of the two-phase
mixing stream is determined using Eq. (3) in order to check
whether the ow in the two-phase mixing section becomes critical
or not.
2.2.4. Model of diffuser ow
While the mixed stream ows through the diffuser, the static
pressure of the ow rises as the kinetic energy is converted to static
pressure. By assuming that the mixed stream at the outlet of the
mixing section is in homogeneous equilibrium, the pressure
recovery coefcient, Ct, is dened as:

Ct

pd  pmix
1
r V2
2 mix mix

(13)

A correlation proposed by Owen et al. [20] is used to calculate


the pressure recovery coefcient:

#


 " 2
xmix
Amix 2
1  xmix 2
Ct 0:85rmix 1 

rg;mix
rf ;mix
Ad

Fig. 4. Experimental controllable ejector expansion device.

a thread mechanism (L1 was measured to calculate the motive


nozzle throat area); and the motive nozzle exit position relative to
the mixing section inlet and thus, the suction nozzle throat area is
adjustable through a thread mechanism (L2 was measured to
calculate the distance between motive nozzle outlet and mixing
section constant-area entry). The developed investigated twophase ow ejector, which has a converging motive nozzle only,
was designed for a CO2 air conditioning systems with 10 kW cooling
capacity. For other refrigerants, such as R410A, and other cooling
capacities, the ejector internal geometries may be different.
This controllable ejector expansion device was tested in the CO2
air conditioning system with various ejector geometries and under
different operating conditions.
3.2. Simulation procedure

(14)

By neglecting the heat transfer from the ejector to the environment, the diffuser outlet enthalpy hd can be determined from
the energy conservation equation across the whole ejector as
follows:

_ m hm m
_ s hs m
_mm
_ s hd
m

365

(15)

Then, the diffuser outlet quality xd can be determined from


thermodynamic property relations since the outlet pressure pd and
enthalpy hd are already specied.
By combining the models of motive nozzle ow, suction nozzle
ow, mixing section ow, and diffuser ow, a simulation model of
a two-phase ow ejector was developed. The specied parameters
for the ejector model are motive nozzle throat area and efciency,
suction nozzle throat area and efciency, cross-sectional area of the
mixing section, mixing section efciency, and the diffuser outlet
area. This two-phase ejector model can be used to simulate an
ejector in trans-critical refrigeration cycles (e.g. CO2) and subcritical refrigeration cycles (e.g. R410A).
3. Application
3.1. A controllable ejector expansion device in CO2 air conditioning
system
A controllable ejector expansion device was designed, constructed, installed into an existing transcritical CO2 air conditioning
system with a cooling capacity of 10.3 kW [21]. Fig. 4 depicts this
experimental stainless-steel controllable ejector expansion device.
The motive nozzle throat diameter is 2.7 mm, the mixing section
constant-area diameter is 4.0 mm, the constant-area length of the
mixing section is 6.5 times of the constant-area diameter, the
diffuser diameter ratio is 3, and the diffuser angle is 7. A needle was
designed to change the throat area of the motive nozzle. The ejector
expansion device has two control options: the throat area of the
motive nozzle is adjustable by positioning a needle in the nozzle via

The measured parameters are the CO2 pressures and temperatures, and the mass ow rates at the inlet to the motive nozzle and
the suction nozzle as well as the CO2 pressure at the ejector outlet.
The two-phase ow ejector model was used to determine the
motive nozzle, suction nozzle and mixing section efciencies based
on the measured data. CO2 thermodynamic and transport properties were calculated using the source code and dynamic link library
developed by Span and Wagner [22].
Figs. 5e7 show the details of how the motive and suction nozzle
isentropic efciencies and the mixing section efciency are calculated using the two-phase ow ejector model. Using the measured
pressure, temperature and mass ow rate at the inlet to motive
nozzle, the isentropic efciency of the motive nozzle was determined by matching the measured motive nozzle mass ow rate to
the motive nozzle mass ow rate predicted using the two-phase
ow ejector model as shown in Fig. 5(a) and (b) for critical ow
and non-critical ow, respectively. Using the measured pressure,
temperature and mass ow rate at the inlet to suction nozzle, the
isentropic efciency of the suction nozzle was determined by
matching the measured suction nozzle mass ow rate to the
suction nozzle mass ow rate predicted using the two-phase ow
ejector model as shown in Fig. 6. With the determined motive and
suction nozzle efciencies as well as measured data, the mixing
section efciency was determined by matching the measured
ejector outlet pressure to the predicted ejector outlet pressure
using the two-phase ow ejector model as shown in Fig. 7.
3.3. Uncertainties of ejector efciencies
Table 2 presents the isentropic efciencies of the motive and
suction nozzles as well as the mixing section efciency that were
determined from the experimental data for one test run. The
absolute and relative uncertainties of each of the three determined
efciencies are listed in Table 2. In addition, the individual uncertainty contributions of the measured data to each of the three
determined efciencies are listed. The uncertainties of the efciencies were determined using a standard error analysis in EES

366

F. Liu, E.A. Groll / Applied Thermal Engineering 52 (2013) 360e370

Fig. 6. Flow chart to calculate suction nozzle efciency (non-critical ow).

distance L2 as shown in Fig. 4, respectively. Through adjusting the


distance L2, the distance between motive nozzle outlet and mixing
section constant-area entry (y) was adjusted. However, for the
mixing section efciency, the two largest uncertainty contributors
that are equally signicant are the motive nozzle inlet pressure pm
and the suction nozzle inlet pressure ps.

Fig. 5. Flow chart to calculate motive nozzle efciency.

[23]. It can be seen from Table 2 that all three efciencies were
determined with an uncertainty of 6% and that the suction nozzle
isentropic efciency has the largest uncertainty. It turns out that for
the ejector motive and suction nozzles, the largest uncertainties
comes from the motive nozzle inlet temperature Tm and the

Fig. 7. Flow chart to calculate mixing efciency.

F. Liu, E.A. Groll / Applied Thermal Engineering 52 (2013) 360e370


Table 2
Uncertainty analysis of ejector components.
Measured data Value Absolute uncertainty hm

hs

hmix

Uncertainty contributions
54.03
L1a (mm)
38.3
L2b (mm)
12.855
Pm (MPa)
3.748
Ps (MPa)
50.88
Tm ( C)

21.63
Ts ( C)
_ m (kg/s)
0.18
m
_ s (kg/s)
0.07
m
Pd (MPa)
4.499
Calculated results
Absolute uncertainty
Relative uncertainty

0.2
0.2
0.019
0.019
0.5
0.5
0.0009
0.00035
0.019

0.01481% 0.00%
0.01472%
0.00%
97.64%
0.00%
17.72%
0.00%
47.16%
0.00%
0.8649% 47.15%
69.33%
0.00%
3.279%
0.00%
0.05455% 0.00%
12.94%
0.00%
0.1062%
0.00%
1.444%
1.607%
0.00%
0.00%
0.6791%
0.986
0.972
0.882
0.01056
0.05734
0.02696
1.071%
5.9%
3.058%

The distance was measured to calculate the motive nozzle throat area.
The distance was measured to calculate the distance between motive nozzle
outlet and mixing section constant-area entry.
b

3.4. Results and discussion


3.4.1. Determined ejector efciencies
Fig. 8(a) and (b) shows the determined motive nozzle efciency
as a function of the pressure ratio Dt/Dmix and Pm/Ps, respectively;
Fig. 8(c) and (d) shows the determined suction nozzle efciency as
a function of the pressure ratio Pm/Ps and ejection ratio, respectively. Fig. 8(e) and (f) shows the mixing section efciency as
a function of the pressure ratio Dt/Dmix and ejection ratio, respectively. In each gure, a total of 24 different test runs that are
combinations of two different values of y (y 1.5 Dmix and 6 Dmix),
four different values of Dt/Dmix (Dt/Dmix 0.45, 0.625, 0.65, 0.675),
and three different outdoor air temperatures of Tod (Tod 27.8  C,
35  C and 37.8  C) are presented. Indoor air temperature, Tid, was set
at 26.7  C and relative humidity, RHid, was set at 50%.
Fig. 8(a) and (b) shows that among the cases with the same
value of y, the motive nozzle efciency decreases and the pressure
ratio increases as the Dt/Dmix ratio decreases (i.e., the ejector throat
diameter decreases). For instance, at y of 6 Dmix, the motive nozzle
efciency decreases from approximately 90e50% when the Dt/Dmix
ratio decreases from 0.675 to 0.45. Fig. 8(a) also shows that at the
xed ejector geometries with low Dt/Dmix ratio of 0.45, the motive
nozzle efciency was not affected by outdoor air temperature; but
at large Dt/Dmix ratio, the motive nozzle efciency was slightly
affected by outdoor air temperature. The isentropic efciencies of
motive nozzle can range from 0.50 to 0.93 with various ejector
geometries under different operating conditions. Motive nozzle
efciency reaches 0.5 at y of 1.5 Dmix or 6 Dmix when Dt/Dmix ratio is
0.45; around 0.75 at y of 1.5 Dmix when Dt/Dmix ratio is 0.625 or 0.65;
and around 0.9 at y of 1.5 Dmix when Dt/Dmix ratio is 0.675, or at y of
6 Dmix when Dt/Dmix ratio is 0.625, 0.65 or 0.675. Large Dt/Dmix ratio,
y of 6 Dmix, and low outdoor air temperature result in high motive
nozzle efciencies.
Fig. 8(c) shows that overall suction nozzle efciencies at high
pressure ratio are relatively high. Small Dt/Dmix ratio and y of 1.5
Dmix sometimes lead to relatively high suction nozzle efciency.
Fig. 8(c) also shows that the motive nozzle exit position not only
affects the suction nozzle efciencies at certain Dt/Dmix ratio but
also affects the pressure ratio of the motive nozzle inlet to the
suction nozzle inlet. Furthermore, it is shown that the suction
nozzle efciency somewhat depends on the outdoor air temperature. Fig. 8(d) shows that ejection ratio increases as suction nozzle
efciency increases. The isentropic efciencies of suction nozzle
can range from 0.37 to 0.90 with various ejector geometries under
different operating conditions.
Fig. 8(e) shows that at the given Dt/Dmix ratio, the mixing section
efciency are higher at y of 1.5 Dmix than at y of 6 Dmix. At the xed

367

ejector geometries, the mixing section efciency varies slightly


with outdoor air temperature. Fig. 8(f) shows that the ejection ratio
increases as the Dt/Dmix ratio decreases beyond a certain value. The
efciencies of mixing section can range from 0.50 to 1.00 with
various ejector geometries under different operating conditions.
Overall the variation ranges of the determined ejector component efciencies of CO2 refrigeration cycles in this study are larger
than the assumed ejector component efciencies in the literature
(Table 3). Ejector component efciencies can be lower than 0.7,
minimum assumed value found in the literature, at certain ejector
geometries and operating conditions.
3.4.2. Empirical correlations of ejector efciencies
Based on the determined efciencies of ejector components at
various ejector geometric parameters and various operating
conditions, three empirical correlations of ejector efciencies were
developed. Typically, an optimal design is used when fabricating an
ejector. In this study, the ejector geometry parameters were
determined based on the authors simulation study and the literature review. Usually, the ow reaches fully developed ow when
the mixing section length is 7e8 times the constant-area diameter.
To make the motive stream and suction stream mix completely in
the mixing section, 6.5 times the constant-area diameter was
chosen as the length of the mixing section. Owen et al. [20] suggested that the optimum diffuser angle for diffusing the two-phase
air/water mixture ow is 7. Therefore, this diffuser angle was
chosen for this study. As a result of xing these geometric parameters, the effects of motive nozzle convergence angle, diffuser
angle, and mixing length on ejector component efciencies are not
considered in this study. Usually, the mixing pressure is not easy to
be measured. Thus this parameter was not included in the empirical equations for the ejector component efciencies.
Eqs. (16)e(18) show the correlations for the motive nozzle
isentropic efciency hm, the suction nozzle isentropic efciency hs
and the mixing section efciency hmix, respectively, as the functions
of ejector geometry, pressure ratio, and ejection ratio. Ejection ratio
varies with motive nozzle position and outdoor air temperature.
Both the suction nozzle isentropic efciency and mixing section
efciency vary with ejection ratio, which is affected by y. Eqs. (17)
and (18) reect the effects of y on ejector component efciencies
through the ejection ratio.

 
Pm
Ps
 2

 3

Pm
Pm
Dt
1:161
0:106
212:320
Ps
Ps
Dmix

2
3

Dt
Dt
 355:359
196:035
Dmix
Dmix

hm 36:137  4:160

(16)

 2
 
Pm
Pm
 314:471
Ps
Ps
 3
 4
 5
Pm
Pm
Pm
79:521
12:222
0:814
Ps
Ps
Ps

hs  3173:171 934:102

694222:14  295614542 795045343


 11432727044 668915545  649905:1Z
2647000Z 2  6885025Z 3 9627161Z 4  5490126Z 5

 0:02 
Pm
Z 4
Ps
(17)

F. Liu, E.A. Groll / Applied Thermal Engineering 52 (2013) 360e370

Motive nozzle efficiency

test 1
test 4
test 7
test 10
test 13
test 16
test 19
test 22

0.9

0.8
0.7

test 2
test 5
test 8
test 11
test 14
test 17
test 20
test 23

test 3
test 6
test 9
test 12
test 15
test 18
test 21
test 24

Motive nozzle efficiency

368

0.6

0.5

0.8
0.7

0.5

0.6

0.7

0.8

0.9

0.4

Suction nozzle efficiency

Suction nozzle efficiency

0.8

0.4

test 2
test 5
test 8
test 11
test 14
test 17
test 20
test 23

test 3
test 6
test 9
test 12
test 15
test 18
test 21
test 24

0.2

0.8

0.6

test 1
test 4
test 7
test 10
test 13
test 16
test 19
test 22

0.4

0.2
2

Pressure ratio Pm/Ps

0.2

0.4

test 2
test 5
test 8
test 11
test 14
test 17
test 20
test 23

test 3
test 6
test 9
test 12
test 15
test 18
test 21
test 24

0.6
0.4

0.6

0.8
test 1
test 4
test 7
test 10
test 13
test 16
test 19
test 22

0.6
0.4

0.8

(e)
y
6Dmix
6Dmix
6Dmix
6Dmix
6Dmix
6Dmix
6Dmix
6Dmix
6Dmix
6Dmix
6Dmix
6Dmix

0.2

0.4

0.6

test 2
test 5
test 8
test 11
test 14
test 17
test 20
test 23

test 3
test 6
test 9
test 12
test 15
test 18
test 21
test 24

0.8

Ejection ratio mm/ms

Diameter ratio Dt/Dmix

Run No.
1
2
3
4
5
6
7
8
9
10
11
12

0.2

0.2

0.4

0.8

1.2

Mixing section efficiency

Mixing section efficiency

0.8

test 3
test 6
test 9
test 12
test 15
test 18
test 21
test 24

(d)
test 1
test 4
test 7
test 10
test 13
test 16
test 19
test 22

0.6

test 2
test 5
test 8
test 11
test 14
test 17
test 20
test 23

Ejection ratio mm/ms

(c)
1.2

(b)

test 1
test 4
test 7
test 10
test 13
test 16
test 19
test 22

Pressure ratio Pm/Ps

(a)

0.6

test 3
test 6
test 9
test 12
test 15
test 18
test 21
test 24

0.5

Diameter ratio Dt/Dmix


1

test 2
test 5
test 8
test 11
test 14
test 17
test 20
test 23

0.6

0.4
0.4

test 1
test 4
test 7
test 10
test 13
test 16
test 19
test 22

0.9

(f)
Dt/Dmix
0.675
0.675
0.675
0.65
0.65
0.65
0.625
0.625
0.625
0.45
0.45
0.45

27.8
35.0
37.8
27.8
35.0
37.8
27.8
35.0
37.8
27.8
35.0
37.8

Run No.
13
14
15
16
17
18
19
20
21
22
23
24

y
1.5Dmix
1.5Dmix
1.5Dmix
1.5Dmix
1.5Dmix
1.5Dmix
1.5Dmix
1.5Dmix
1.5Dmix
1.5Dmix
1.5Dmix
1.5Dmix

Dt/Dmix
0.675
0.675
0.675
0.65
0.65
0.65
0.625
0.625
0.625
0.45
0.45
0.45

27.8
35.0
37.8
27.8
35.0
37.8
27.8
35.0
37.8
27.8
35.0
37.8

Fig. 8. Determined ejector component efciencies in CO2 refrigeration cycles.

hmix  6869:077 19308:18Z 0  18089:31Z 02 5649:417Z 03




Z0

Dt
Dmix

0:1

1 40:35
(18)

These correlations should be used within the following


boundaries:

8:0 MPa < Pm < 14:0 MPa; 2:5 MPa < Ps < 5:0 MPa;
40 + C < Tm < 60 + C; 15 + C < Ts < 26+ C;
0:1g=s < mg < 0:25 g=s; 0:05 g=s < Ts < 0:07 g=s; 1:8 mm
< Dt < 2:7 mm; Dmix 4 mm:
It is found that each correlation predicts determined ejector
component efciencies within a correlation error of 5%. The
established empirical equations of the ejector component

F. Liu, E.A. Groll / Applied Thermal Engineering 52 (2013) 360e370


Table 3
Comparison of the determined CO2 ejector component efciencies in this study with
those assumed in the literature.

369

Table 4
Statistical data of deviations between the model predictions and the measured data
for the ejector expansion transcritical CO2 system.

Ejector efciency

This study

The literature

Item

hm
hs
hmix

0.50e0.93
0.37e0.90
0.50e1.00

0.7e0.9
0.7e0.9
0.95

Mean
deviation (%)

Standard
deviation (%)

Maximum
deviation (%)

COP (e)
Q_ c;ref (kW)

2.49
3.32

3.14
3.97

7.57
11.23

efciencies can be used in the simulation study to determine the


performance of CO2 air conditioning systems under certain
boundary conditions.
3.5. Validation of results
This two-phase ow ejector model (using empirical correlations
to determine the ejector component efciencies) was incorporated
with the existing validated compressor model, gas cooler model and
evaporator model [24,25] into an ejector expansion air conditioning
system simulation model. The ejector expansion CO2 air conditioning system was tested to validate this simulation model of the
ejector expansion transcritical CO2 air conditioning system at
various ejector geometries (Dt 2.7, 2.6, 2.5 and 1.8 mm; y 1.5 Dmix
and 6 Dmix) and different operating conditions (Tod 27.8, 35.0 and
37.8  C; Tid 26.7  C; RHid 50%). Comparisons of the measured and

Predicted COP

1.6

1.2

0.8

data

0.4

0
0

0.4

0.8

1.2

1.6

Measured COP

(a) Comparison between predicted and measured COP

Predicted cooling capacity

20

16

12

8
data

predicted COP and cooling capacity are shown in Fig. 9(a) and (b),
respectively. It can be seen from Fig. 9(a) and (b) that the simulation
model predicts the COP within 8% of the measured COP and the
cooling capacity within 12% of the measured cooling capacity. A
statistical analysis for the deviations between the model predictions
and the measured data is shown in Table 4.
4. Summary
This paper introduces a method of determining ejector
component efciencies in refrigeration cycles. It is based on an
ejector model and measured performance data. A two-phase ow
ejector model, consisting of sub-models for motive nozzle ow,
suction nozzle ow, mixing section ow and diffuser ow, was
developed. The measured data in conjunction with the ejector
simulation model were used to determine the isentropic efciencies of the motive and suction nozzles, and the efciency of mixing
section.
The application of this method is illustrated with a case study of
a controllable ejector in transcritical CO2 air conditioning systems
at outdoor air temperatures of 27.8  C, 35  C and 37.8  C. Study
results show that ejector geometries and operating conditions
affect ejector component efciencies signicantly: 1) the motive
nozzle efciency is very sensitive to ejector throat diameter and it
ranges from 0.50 to 0.93; 2) the suction nozzle efciency is affected
by motive nozzle throat diameter, motive nozzle exit position and
outdoor air temperature, and it ranges from 0.37 to 0.90; and 3) the
efciency of mixing section is affected by the motive nozzle exit
position and outdoor air temperature, and it ranges from 0.50 to
1.00. Small motive nozzle throat diameter leads to low motive
nozzle efciency and high suction nozzle efciency. The motive
nozzle efciency and suction nozzle efciency cannot reach high
values at the same time. The motive nozzle placed at a distance of
1.5 times Dmix from the mixing section inlet results in a little bit
higher mixing section efciency than at a distance of 6 times Dmix.
The ejector component efciencies can be much lower than the
assumed values in the literature at certain ejector geometries and
operating conditions.
Empirical correlations were established to estimate ejector
component efciencies at different ejector geometries and operating conditions. With empirical correlations instead of assumed
constant values of ejector efciencies as in the literature, the twophase ejector model was incorporated into a CO2 air conditioning
model to predict the performances of ejector expansion CO2 air
conditioning systems. The system simulation model predicts the
COP within 8% of the measured COP and the cooling capacity
within 12% of the measured cooling capacity.

Acknowledgements
0
0

12

16

20

Measured cooling capacity

(b) Comparison between predicted and measured cooling capacity


Fig. 9. Comparison between predicted and measured cooling COP and cooling capacity
of ejector expansion transcritical CO2 cycles.

The authors would like to express their appreciation to the Air


Conditioning and Refrigeration Technology Institute for providing
nancial support for this study. In addition, the authors would like
to thank Dr. Daqing Li, Dr. Stefan Bertsch and Mr. Joe Poland as well
as the shop staff of the School of Mechanical Engineering at Purdue
University for their contributions to this project.

370

F. Liu, E.A. Groll / Applied Thermal Engineering 52 (2013) 360e370

References
[1] P.A. Domanski, Minimizing throttling losses in the refrigeration cycle, in:
Proceedings of 19th International Congress of Refrigeration, vol. 1Vb, 1995,
pp. 766e773.
[2] C. Vereda, R. Ventas, A. Lecuona, M. Venegas, Study of an ejector-absorption
refrigeration cycle with an adaptable ejector nozzle for different working
conditions, Appl. Energy 97 (2012) 305e312.
[3] D. Li, E.A. Groll, Transcritical CO2 refrigeration cycle with ejector-expansion
device, Int. J. Refrigeration 28 (2005) 766e773.
[4] E.B. Ksayer, D. Clodic, Enhancement of CO2 refrigeration cycle using an ejector:
1D analysis, in: Proceeding of International Refrigeration and Air Conditioning
Conference at Purdue, Purdue University, USA, 2006. Paper No. R058.
[5] E.B. Ksayer, Study and design of systems with improved energy efciency
operating with CO2 as refrigerant. Ph.D. thesis, Ecole de Mines, Paris, France;
2007.
[6] J. Deng, P. Jiang, T. Lu, W. Lu, Particular characteristics of transcritical CO2
refrigeration cycle with an ejector, Appl. Therm. Eng. 27 (2007) 381e388.
[7] J. Sarkar, Optimization of ejector-expansion transcritical CO2 heat pump cycle,
Energy 33 (2008) 1399e1406.
[8] D.W. Sun, Variable geometry ejectors and their applications in ejector
refrigeration systems, Energy 21 (1996) 919e929.
[9] S. Elbel, P. Hrnjak, Experimental validation of a prototype ejector designed to
reduce throttling losses encountered in transcritical R744 system operation,
Int. J. Refrigeration 31 (2008) 411e422.
[10] J.H. Keenan, E.P. Neumann, F. Lustwerk, An investigation of ejector design by
analysis and experiment, J. Appl. Mech. 17 (1950) 299e809.
[11] G.K. Alexis, E.D. Rogdakis, A verication study of steam-ejector refrigeration
model, Appl. Therm. Eng. 23 (2003) 29e36.
[12] R. Yapici, H.K. Ersoy, Performance characteristics of the ejector refrigeration
system based on the constant area ejector ow model, Energy Convers.
Manage. 46 (2005) 3117e3135.

[13] J. Yu, Y. Li, A theoretical study of a novel regenerative ejector refrigeration


cycle, Int. J. Refrigeration 30 (2007) 464e470.
[14] J. Yu, Y. Ren, H. Chen, Y. Li, Apply mechanical subcooling to ejector refrigeration for improving the coefcient of performance, 48 (2007) 1193e1199.
[15] S.W. Elbel, P.S. Hrnjak, Effects of internal heat exchanger on performance of
transcritical CO2 systems with ejector, in: Proceedings of 10th International
Refrigeration and Air Conditioning Conference at Purdue, Purdue University,
USA, 2004. Paper No. R166.
[16] F. Sun, Y. Ma, Thermodynamic analysis of transcritical CO2 refrigeration cycle
with an ejector, Appl. Therm. Eng. 31 (2011) 1184e1189.
[17] F. Eskandari Manjili, M.A. Yavari, Performance of a new stage multiintercooling transcritical CO2 ejector refrigeration cycle, Appl. Therm. Eng.
40 (2012) 202e209.
[18] A. Attou, J.M. Seynhaeve, Steady-state critical two-phase ashing ow with
possible multiple choking phenomenon Part 1: physical modelling and
numerical procedure, J. Loss Prev. Process. Indust. 12 (1999) 335e345.
[19] B.J. Huang, J.M. Chang, C.P. Wang, V.A. Petrenko, A 1-D analysis of ejector
performance, Int. J. Refrigeration 22 (1999) 354e364.
[20] I. Owen, A. Abdul-Ghani, A.M. Amini, Diffusing a homogenized two-phase
ow, Int. J. Multiphase Flow 18 (1992) 531e540.
[21] F. Liu, Y. Li, E.A. Groll, Performance enhancement of CO2 air conditioner with
a controllable ejector, Int. J. Refrigeration 35 (2012) 1604e1616.
[22] R. Span, W. Wagner, A new equation of state for carbon dioxide covering the
uid region from the triple-point temperature to 1100 K at pressure up to
800 MPa, J. Phys. Chem. Ref. Data 25 (1996) 1509e1596.
[23] S.A. Klein, Engineering Equation Solver, F-Chart Software, 2004.
[24] M.T. Ortiz, D. Li, E.A. Groll, Evaluation of the Performance Potential of CO2 as
a Refrigerant in Air-to-Air Air Conditioners and Heat Pumps: System Modeling
and Analysis. Final Report ARTI-21CR/611-10030, Ray W. Herrick Labs
Research Report HL 2003-20, Purdue University, West Lafayette, IN, 2003.
[25] F. Liu, E.A. Groll, Recovery of Throttling Losses by a Two-phase Ejector in
a Vapor Compression Cycle. Final Report ARTI-10110-10201, Ray W. Herrick
Labs, Purdue University, West Lafayette, IN, 2008.

You might also like