Experimental investigation and theoretical analysis of an ejector refrigeration

system
Daniel A. Pounds
a
, J.M. Dong
a, b
, P. Cheng
a
, H.B. Ma
a,
*
a
Department of Mechanical and Aerospace Engineering, University of Missouri e Columbia, Columbia, MO 65211, USA
b
Institute of Marine Engineering and Thermal Science, Marine Engineering College, Dalian Maritime University, Dalian 116026, China
a r t i c l e i n f o
Article history:
Received 22 October 2011
Received in revised form
31 October 2012
Accepted 3 November 2012
Available online 10 January 2013
Keywords:
Ejector refrigeration system
Phase change heat transfer
a b s t r a c t
The ejector refrigeration system can be powered from low-grade thermal energy such as solar generated
hot water or waste heat, especially when electricity supply is limited or does not exist. An experimental
investigation of an ejector refrigeration system was conducted to determine the effects of nozzle size,
axial nozzle location, high-temperature evaporator temperature, and refrigeration temperature. The
tested conditions include the effects of the high-temperature evaporator (HTE) temperatures ranging
from 120 to 135

C, low-temperature evaporator (LTE) temperatures ranging from 5 to 15

C, and
condenser temperatures of 7e30

C. It was found that an optimum nozzle location which can produce
a maximum coefcient of performance (COP) exists for a given set of operating conditions. At the same
time, a mathematical model has been developed to predict the system COP, which agrees well with
experimental data. The experimental results show that the ejector refrigeration system can achieve
a COP of 1.7, which is much higher than the results typically reported in the literature, but at the expense
of critical backpressure. Current investigation demonstrates that the ejector refrigeration system is a very
promising alternative to the status quo vapor compression systems.
2012 Elsevier Masson SAS. All rights reserved.
1. Introduction
Various forms of low-grade thermal energies such as waste heat
from industrial processes, vehicle exhaust, and solar generated hot
water could all be useful in powering a cycle known as the ejector
refrigeration system (ERS) [1]. In addition to energy savings, the
technology investigated herein could lead to a reduction in harmful
carbon dioxide emissions, that some believe to be linked to envi-
ronmental climate shifts, and that could be associated with power
plants that burn fossil fuels to generate electricity.
Even though the ERS is a valuable alternative to traditional vapor
compression refrigeration systems, the current ERS has a low ef-
ciency. As a result, the vapor compression refrigeration system
remains dominate in the market. In order to make a commercially
available ERS that is competitive in price and performance, further
research and enhancement need to be done in order to increase
efciency and reduce unit cost of production to a level suitable for
a consumer-based market. Additionally, the benets of the ERS
should be compared to that of absorption refrigeration, which has
developed into a commercially available product in recent years.
Fig. 1 shows a schematic diagram of a typical ejector refrigera-
tion system. The boiler, mechanical pump, and ejector in the ERS
are analogous to the mechanical compressor in a conventional
vapor compression refrigeration system. The primary uid owing
through the ejector is a saturated or superheated vapor formed in
the boiler from adding heat and evolving the liquid in the boiler
into a high-temperature, high pressure vapor. The primary uid is
accelerated through a nozzle in the ejector shown in Fig. 2 creating
suction on the evaporator and drawing the low-pressure secondary
uid into the ejector. As the secondary uid evaporates in the low-
pressure evaporator, it draws on heat from its surroundings and
creates useful refrigeration. The ejector exhausts its mixture of
uids into the condenser where it is liqueed at ambient temper-
ature. From there, the two liquids are separated and the primary
uid is pumped back into the boiler while the secondary uid is
drawn back into the evaporator. Often times the boundaries are set
on the operating conditions of the boiler, evaporator, and
condenser by the available heat input, the refrigeration needs, and
the local ambient conditions. Typically, a mechanical pump is
required to pump primary uid back into the high pressure boiler,
but recently signicant efforts to replace the mechanical pump
with a thermally-controlled pump have been made [2]. The cycles
efciency is measured by the coefcient of performance (COP)
given by:
* Corresponding author. Tel.: 1 573 884 5944.
E-mail address: mah@missouri.edu (H.B. Ma).
Contents lists available at SciVerse ScienceDirect
International Journal of Thermal Sciences
j ournal homepage: www. el sevi er. com/ l ocat e/ i j t s
1290-0729/$ e see front matter 2012 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ijthermalsci.2012.11.001
International Journal of Thermal Sciences 67 (2013) 200e209
COP
useful refrigeration
heat input to boiler power consumed by pump
: (1)
Eq. (1) is widely used as a design metric for designing such an
ERS. If the energy input to the mechanical pump is neglected, as it
usually is due to its relative magnitude compared to the energy
input to the high-temperature evaporator (HTE), Eq. (1) can be
expressed as:
COP
_ m
sf
$h
lv;sf
_ m
pf
$h
lv;pf
(2)
where h
fg
is the latent heat of vaporization, and _ m is the vapor
generation rates at the respective evaporators. If the secondary
uid is the same as the primary uid, Eq. (2) would nearly reduce to
being equivalent to the entrainment ratio, i.e., COPyuy_ m
sf
= _ m
pf
. If
the secondary and primary uids are different, the latent heats are
different, as are the properties of the chosen uids. If two immis-
cible uids are carefully selected, the COP could theoretically be
increased by a multiplier equal to the ratio of latent heats in Eq. (2),
and separation of the condensate would not be an issue. If the
secondary uids latent heat is much greater than the primary
uids latent heat, then the cycle efciency could be greatly
increased if the same entrainment ratios are achieved as in the
single uid ERC.
In order to predict the dynamics of the uid ow in the ejector,
a Quasi-1-Dimensional analysis was rst imposed by Keenan et al.
[3]. Keenan et al. [3] was able to nd very satisfactory agreement
between experimental data and analytical predictions and show
that the ideal gas assumption is correct. There are several ways to
improve the ERS theoretical models. It is believed that a large
portion of the irreversibility associated with the mixing process is
due to an energy loss from the shear force developed at the inter-
face of the two ow streams and consequently the shear mixing.
Chang et al. [4] proposed a novel petal nozzle, as opposed to the
conical nozzle used in the typical ERS. The nozzle was originally
developed to augment jet engine thrust. According to experimental
results, a normal, stream-wise vortex exists behind the exit of the
nozzle. Due to the KelvineHelmholtz instability, this vortex sheds
periodically from the trailing edge of the nozzle and directly
Nomenclature
A area [m
2
]
AR diffuser throat-to-nozzle throat area ratio
a sound speed [m/s]
C
p
heat capacity [kJ/kg K]
COP coefcient of performance
CR compression ratio
h enthalpy [kJ/kg]
h
lv
latent heat of vaporization [kJ/kg]
HTE high-temperature evaporator (boiler)
I current
LTE low-temperature evaporator (evaporator)
M mach number
_ m mass ow rate [kg/s]
MW molecular weight [kmol/kg]
NXP nozzle exit position (normalized)
P pressure [kPa]
R gas constant [J/kg K]
T temperature [K]
U velocity [m/s]
V RMS voltage [V]
Greek
convergence criteria
r density [kg/m
3
]
g specic heat ratio
h isentropic efciency
u entrainment ratio
Subscripts
1 NXP position
3 post-mixing, pre-shock wave
4 post-shock wave
b backpressure (or exhaust pressure)
d diffuser throat
i isentropic conditions
m mixture
o stagnation conditions
pf, P primary uid
sf, S secondary uid
t nozzle throat
* critical conditions
Vapor generator
Expansion valve
Condenser
Evaporator
Ejector
Added Heat
Pump
Rejected Heat
Refrigeration
Fig. 1. Schematic diagram of a typical ejector refrigeration system. Fig. 2. Typical ejector geometry, pressure and velocity proles along ejector length.
D.A. Pounds et al. / International Journal of Thermal Sciences 67 (2013) 200e209 201
enhances the mixing process. Therefore, if the mixing process can
be done in a more energy conscience manner, more energy could be
recovered to the backpressure [4], and improve the compression
ratio. They concluded that the compression ratio and the entrain-
ment ratio of the ejector could indeed be enhanced if the petal
nozzle was used, but only when the area ratio, AR A
d
=A
t
, is high.
The use of the petal nozzle can increase the critical condenser
pressure for a larger AR regardless of operating conditions. There
also exists an optimum AR under which a maximum compression
ratio can be obtained from use of the petal nozzle.
Several attempts to understand the ow characteristics through
the ejector have been made. The well-known constant cooling
capacity was initially unexplained until the work of Munday and
Bagster [5]. Their researchfoundthat the restrictionof thesecondary
owwithinthe ejector is the primary reasonfor the constant cooling
capacity. Inaddition, MundayandBagster [5] were able to prove that
the liquid water found in a low-temperature evaporator was very
close to beinginequilibriumwithits vapor, therefore, conrmingthe
assumption that the pressure in the evaporator is taken as the uids
saturation pressure at the respective temperature.
Chunnanond and Aphornratana [6] were able to obtain results
similar toMundayandBagsters [5] breakback theory. Chunnanond
and Aphornratana [6] further explored Munday and Bagsters [5]
notion of an effective area in order to further understand the mix-
ing process and pressure prole across the ejector. When the satu-
rationtemperature andpressure inthe boiler were lowered, a smaller
primary uid (PF) mass exits the nozzle with a lower velocity. The PF
ow stream fans out with a lower momentum and, therefore,
a smaller expansion angle. This pushes the effective area further
downthe entrained duct, which therefore, increases the entrainment
ratio. Thus, the systemhas a higher cooling capacity and overall cycle
efciency. However, this also was found to result in a lower mixed
stream momentum, thereby causing the shocking position to move
upstream forcing the ejectors critical condenser pressure lower [6].
The ejectors geometry is the other primary inuence on ejector
performance. Keenan and Neumann [8] found that the axial posi-
tioning of the primary nozzle heavily inuenced the ERS cooling
capacity and COP. From their experimental investigation, it was
observed that moving the primary nozzle out of the mixing chamber
caused the system COP to increase at the expense of critical
condenser pressure decrease. Aphornratana and Eames [9] found
interesting results of the effects of the primary nozzle position on
system performance. For xed boiler and condenser conditions, the
evaporator temperature and pressure decreased as the primary
nozzle was moved towards the mixing chamber. There exists
a minimum temperature and further movement beyond the
minimum results in an increase in evaporator temperature. For
a xed nozzle position, the evaporator temperature varies propor-
tionally to the condenser pressure and inversely proportional to the
boiler pressure. The effects of boiler and condenser pressure tend to
reduce at positions further into the mixing chamber. Aphornratana
andEames [9] concludedthat theoptimumnozzlepositionis directly
dependent on the boiler and condenser pressures only. Increasing
the condenser pressure or decreasing the boiler pressure moved the
optimum position into the mixing chamber, and vice versa.
The previous discussions show how the ejectors efciency is
dependent on both the operating conditions as well as the ejectors
geometry. As the ejector is designed to operate at a specic set of
operating conditions, the conditions may change, which will drive
down the efciency of the ejector potentially causing it to cease
function totally. Sun [10] proposed a variable geometry ejector to
overcome this constraint, which gives an ejector the ability to
change geometry to perform optimally within a range of operating
conditions. Their theoretical results indicated that a much better
efciency could be gained from such a device but did not
thoroughly describe the device or the mechanism for the geometry
change. The variable geometry ejectors performance characteris-
tics changed slightly in that when the condenser pressure
increased, the efciency did not drop abruptly as in the xed area
ejector. Instead, there was more of a gradual decrease in perfor-
mance when operating conditions changed from the design point.
The aim of this investigation was to better understand the uid
ow and refrigeration mechanisms occurring in the ejector refrig-
eration system through both theoretical analysis and experimental
investigation. A general mathematical model was developed by
implementing different aspects of several current models and
theories. Based on previous investigations, as presented above,
a new ejector system was developed. Using the experimental
system developed herein, the prototype of the ejector refrigeration
systemwas investigated. The investigation offers a newapproach to
the design of a more efcient ejector refrigeration system.
2. Theoretical modeling
The model is based on the Quasi-1-Dimensional Gas Dynamics
model, which can be found in a number of literature studies [2e12],
but differs from most models found elsewhere in that it is general
enough to design a multi-uid ejector rather than only a typical
single uid ejector. However, when the single uid assumption is
applied, the model takes on a similar form as to what has been
derived in the past. The model presented herein is based on the
following assumptions.
The primary and secondary streams expand isentropically
through the nozzles. Also, the mixture stream compresses
isentropically in the diffuser.
The primary and secondary uid streams are saturated vapor
and their inlet velocities are negligible.
Velocity of the compressed mixture leaving the diffuser is
negligible.
Constant isentropic expansion exponent and ideal gas behavior
is assumed.
The mixing of the primary and secondary vapor takes place in
the mixing chamber and is completed before the presence of
the shock wave.
The walls of the ejector are considered an adiabatic boundary.
Friction losses are dened in terms of isentropic efciencies in
the nozzle, diffuser, and mixing chamber.
The ejector ow is one-dimensional with variable cross-
sectional area (Quasi-1-D), and operating at steady-state
conditions.
At position 1, or the NXP, the static pressure of the two ow
streams is assumed to be uniform.
For the control volume shown in Fig. 3, the mass ow rates into
the given control volume is equal to the sum of the mass ow rates
out of the same control volume, i.e.,
_ m r
1
A
1
U
1
r
2
A
2
U
2
(3)
where _ m is the mass ow rate of any ow stream in or out of the
control volume, r is the uids local density, A is the cross-sectional
area of the control volumes two boundaries, U is the local velocity
at the control volume boundary, and subscripts 1 and 2 denote the
entrance and exit of the control volume, respectively. For the same
control volume, the well-known conservation of momentum
equation is given by
P
1
A
1
_ m
1
U
1
P
2
A
2
_ m
2
U
2
(4)
where P is the local static pressure. Additionally, the conservation of
energy is given by
D.A. Pounds et al. / International Journal of Thermal Sciences 67 (2013) 200e209 202
h
1

U
2
1
2
h
2

U
2
2
2
(5)
where h is the local enthalpy, and the ideal gas law is given by
P rRT (6)
When the primary uid ows through the nozzle, the contact
time is very short. Heat transfer between the uid and nozzle can
be neglected, and if the frictional loss is not considered, the isen-
tropic relationships have been established. Using the conservation
equations, and the isentropic relations previously established, the
local pressure, temperature, and density can be related to their
stagnation conditions by a series of isentropic ow functions. The
performance of an ejector can be dened in terms of the entrain-
ment ratio, or the ratio of the entrained vapor mass ow rate to the
mass ow rate of the primary uid stream, i.e.,
u
_ m
s
_ m
p

P
1;s
P
1;p
A
1;s
A
1;p
_
T
o;p
T
o;s
_
1=2
_
R
p
R
s
_
1=2
f
_
g
s
; M
1;s
_
f
_
g
p
; M
1;p
_ (7)
where f is the function dened as
f g; M M
_
g
_
1
g 1
2
M
2
__
1=2
: (8)
Considering the isentropic efciency of [14]
h
p

h
o;p
h
1;p
h
o;p
h
1i;p
(9)
and h C
p
T it can be found as:
T
1i;p
T
o;p
1
1
h
p
_
1
T
1;p
T
o;p
_
(10)
Recall that with ideal gas law and the denition of the Mach
number, the mass ow rate through the primary nozzle can be
expressed as
_ m
p

P

T
p AM
_
g
p
R
p
_
1=2
(11)
Now, utilizing mass conservation through the nozzle, a relation-
ship for the NXP area to the nozzle throat area can be expressed as:
A
1;p
A
t

_
2
g
p
1
_
g
p
1=2g
p
1
M
1;p
_

_
1
1
h
p

1
h
p
_
1
g
p
1
2
M
2
1;p
_
_

_
g
p
1=2g
p
1
(12)
It can be shown that Eq. (11) can be rearranged to show the
exiting Mach number of the primary uid as a function of the
pressure ratio across the nozzle as:
M
1;p

_
2
g
p
1
_
_

_
h
p
_
1
_
P
1
P
o;p
_
g
p
1=g
p
_
1 h
p
_
1
_
P
1
P
o;p
_
g
p
1=g
p
_
_

_
(13)
where the subscript p has been dropped fromthe pressure termat
position 1 (NXP) due to the assumption of uniform static pressure
at this location. Applying the energy conservation equation
between the LTE and the nozzle exit plane, it can be shown that the
secondary ow stream Mach number is expressed as
M
1;s

_
2
g
s
1
___
P
o;s
P
1
_
g
s
1=g
s
1
_

(14)
In order to determine the Mach number of the mixture,
considering conservation of momentum on a control volume
enclosing the mixing chamber gives
h
m
_
_ m
p
U
1;p
_ m
s
U
1;s

_ m
m
U
3
(15)
where the subscript 3 denotes the position at the exit of the
mixing chamber, just before the shock wave and h
m
is the mixing
efciency. Considering the entrainment ratio and continuity
equation, Eq. (15) can be rewritten as
U
3
h
m
_
U
1;p
uU
1;s
1 u
_
(16)
Substituting the denition of the critical sound speed, Eq. (16)
can be expressed as
Fig. 3. Mixing chamber control volume for quasi-1-D analysis.
M
*
3
h
m
_

_
M
*
1;p
uM
*
1;s

g
s
g
p
R
s
R
p
T
o;s
T
o;p
_
g
p
1
g
s
1
_

1 u
p

g
p
1
g
p
R
p
g
m
1

g
p
R
p
g
m
1
_
g
p
1
_ u
g
s
R
s
g
m
1
g
s
1
_
T
o;s
T
o;p
_

_
_

_
(17)
D.A. Pounds et al. / International Journal of Thermal Sciences 67 (2013) 200e209 203
where the characteristic Mach number is related to the actual Mach
number by [13]
M
*

M
2
g 1
M
2
g 1 2

(18)
Using Eqs. (17) and (18), the Mach number after the mixing
process can be obtained. After the mixing process, the shock wave
takes place. The conditions after the shock wave can be determined
from typical gas dynamics relations, i.e.,
M
2
4

M
2
3

2
g
m
1
_
2g
m
g
m
1
_
M
2
3
1
(19)
P
4
P
3

1 g
m
M
2
3
1 g
m
M
2
4
(20)
where subscript 3 denotes pre-shock position and 4 denotes
post-shock position.
Similarly, as in the primary nozzle, by using the denition of
isentropic efciency, the pressure lift ratio across the diffuser can be
derived from the momentum equation as follows:
P
b
P
4

_
h
d
g
m
1
2
M
2
4
1
_
g
m
=g
m
1
(21)
where P
b
is the exhaust pressure of the ejector and h
d
is the isen-
tropic efciency of the diffuser. Then considering Eq. (7), the ratio of
the areas of the ow stream exits at the NXP in terms of the
entrainment ratio can be readily found as
A
1;s
A
1;p

_
T
o;s
T
o;p
_
1=2
_
R
s
R
p
_
1=2
f
1
_
g
p
; M
1;p
_
f
1
_
g
s
; M
1;s
_ u (22)
Mass conservation requires that
_ m
p
_ m
s
_ m
m
(23)
where subscript m denotes the uid mixture post-mixing. Often
when designing the constant pressure mixing ejector, it is useful to
express the ratio of mixing chamber throat-to primary nozzle
throat area in terms of the entrainment ratio. Similarly, as in the
case with Eq. (22), using Eqs. (7) and (23), the ratio of mixing
chamber throat-to-nozzle throat area can be derived as
A
d
A
t
1 u
P
o;p
P
3

T
o;p
_

R
m
T
o;3
_

g
p
h
n
R
p
_
2
g
p
1
_
g
p
1=g
p
1

_
f
1
g
m
; M
3

(24)
The stagnation temperature of the mixed stream, in Eq. (24), can
be determined by an alternate form of the energy equation, and
expressed as
T
o;3

g
p
R
p
g
p
1
T
o;p
u
g
s
R
s
g
s
1
T
o;s
1 u
g
m
R
m
g
m
1
(25)
The equations presented above may be used for a multi-uid
system, but if a single uid system is desired, the equations
reduce to the commonly used set of equations presented in virtu-
ally most ejector model publications, such as the model presented
by Eames et al. [7]. It is assumed that the initial pressure and
temperature of the two uids are known, as well as the owrates of
both uids. Due to the lack of a closed form solution for the system,
an iterative process, as shown in Fig. 4, can be used to solve the
system of equations. The following procedure, outlined by Eames
et al. [7], can be used in order to determine the exhaust pressure of
the ejector.
1. First, guess a value of the pressure at the NXP, P
nxp
. Calculate
the Mach numbers of the primary and secondary uids at the
NXP, M
p,1
and M
s,1
, using Eqs. (13) and (14).
2. Calculate the Mach number of the uid mixture, M
3
, using Eqs.
(17) and (18).
3. Calculate the Mach number of the uid mixture after the shock
wave, M
4
, using Eq. (19).
4. Calculate the pressure lift ratio, P
4
/P
3
, across the shock wave
using Eq. (20).
5. Calculate a pressure lift ratio across the diffuser using Eq. (21).
6. Using the pressure ratios and the known pressures, the exhaust
pressure, P
b
, can now be determined.
7. Go back to step 1 and repeat process using a new value of P
nxp
until P
b
is equal to the design parameter.
8. Calculate the ejector cross-sectional areas utilizing Eqs. (11),
(12), (22), (24), and (24).
The isentropic efciencies for the nozzle, the mixing chamber,
and the diffuser should be chosen based on experimental data. If
the surfaces inside the ejector are very smooth, then experimen-
tation has shown the efciencies to be approximately 0.95, 0.85,
and 0.85 for the nozzle, diffuser, and mixing chamber, respectively.
Fig. 4. Equations solution procedure.
D.A. Pounds et al. / International Journal of Thermal Sciences 67 (2013) 200e209 204
It should be noted, however, that this design method is only
accurate when the ejector is operated at the precise designed
conditions and operating at critical backpressure.
3. System design and experimental setup
Table 1 contains experimental data reported by Eames et al. [7],
and the far right columns show the error analysis for the current
theoretical model under investigation. To make a comparison,
using the operating conditions specied and the COP that was
measured, the area ratio of diffuser throat-to-nozzle throat area
was calculated and compared to the actual experimental ejector
used. The model shows good agreement with their experimental
data.
Based on the theoretical model developed above, the ejector
refrigeration system and the experimental setup as shown in
Figs. 5e7 were developed. The system shown consists of a boiler
simulating the temperature evaporator (HTE), a low-temperature
boiler creating low-temperature evaporation, a condenser, a pres-
sure-tap subsystem, a pumping subsystem, a reservoir, and the
ejector. Both boilers were equipped with a 5 kWelectric immersion
heater and were controlled using a variable voltage transformer.
The condenser used was a shell and tube style water-cooled
condenser. Inside each boiler, there was a type T thermocouple,
a pressure transducer with a range of 0e350 kPa absolute (HTE)
and 0e35 kPa absolute (LTE), 8% of the range, and a liquid level
switch to control the pumping subsystem.
Using the model developed herein, the ejector with the
dimensions as shown in Table 2 was designed. The design was
similar to the design discussed by Chunnanond and Aphornratana
[6] with modications. A movable nozzle was also designed so that
the axial position of the nozzle could be easily moved between
testing periods with very little effort and no need for disassembly.
Using this design, the theoretical model could be validated and the
experimental results could then be compared to results presented
in the current investigation.
4. Results and discussion
Tests on the experimental prototype, with steam as the single
working uid, were conducted to determine the effects of the vapor
temperature in the HTE, the evaporation temperature in the LTE,
and condenser temperature on system COP. To perform each test,
the systemwas allowed to run until it had reached equilibriumwith
respect to all temperature measurements to ensure heat inputs
were steady. Data from the temperature and pressure sensors and
the power levels were then measured and averaged over a 30-min
time period. The power inputs were measured in real time by
Table 1
Theoretical model validation with experimental data from Eames et al. [7].
Temperature [

C] Pressure [kPa] Theoretical

area ratio
Error
(%) LTE HTE Condenser Condenser COP
5 120 26.5 3.4 0.4044 83.0 7.8
125 27.8 3.7 0.3442 88.7 1.4
130 30.8 4.4 0.2756 86.1 4.3
135 33.4 5.1 0.2513 85.1 5.4
140 34.4 5.4 0.1773 93.2 3.6
7.5 120 27.3 3.6 0.5004 79.7 11.4
125 29.5 4.1 0.4189 80.9 10.1
130 31.5 4.6 0.3553 83.2 7.6
135 33.4 5.1 0.2965 86.6 3.8
140 35.3 5.7 0.2334 88.8 1.3
10 120 28.3 3.8 0.5862 76.8 14.7
125 30.0 4.2 0.5374 80.6 10.4
130 31.9 4.7 0.4734 83.1 7.7
135 34.0 5.3 0.3832 84.5 6.1
140 36.3 6.0 0.3093 85.2 5.3
The experimental ejector had an actual area ratio of 90.
Fig. 5. Photographs of the prototype test rig.
D.A. Pounds et al. / International Journal of Thermal Sciences 67 (2013) 200e209 205
sampling the RMS voltage drop, V, across each heater, as well as the
current, I, through each. With these measurements, the systemCOP
can be calculated by
COP
V$I
LTE
V$I
HTE
(26)
although this calculation by Eq. (26) includes unwanted heat gains/
losses to the environment. Because of this, the actual system COP
should be higher than the one based on electrical measurements
dened by Eq. (26). But the system was very well insulated;
therefore, the COP measurement is close to the actual system COP
with a measurement error of less than 4%.
Following the procedure previously described, tests on the
prototype were conducted in order to determine the effects of the
vapor temperature in HTE, the evaporation temperature in the LTE,
and the condensing temperature on the system COP. Careful
temperature measurements were taken along the insulated
surfaces and free convection correlations [15] were used to esti-
mate the system heat losses and gains. The estimation shows that
the heat losses for both the HTE and LTE are less than 3% of the total
heat input. The pressure losses in the ducts to the primary and
secondary nozzles were estimated as well. Using the pressure
sensors in the HTE vessel and another in the chamber just behind
the primary nozzle, the pressure losses for the primary streamwere
estimated to be less than 5%. The pressure losses in the suction line
to the LTE were measured in a similar fashion resulting in a pres-
sure loss measurement of 3e5%.
Unless otherwise noted, the data presented hereafter is
a representation of using the nozzle with a throat diameter of
1.2 mm and 1.6 mm, a diffuser with a throat diameter of 19 mm,
and a constant area mixing section with a length of 90 mm. The
ERS performance was measured over a range of HTE saturation
temperatures of 120e135

C. The LTE conditions were also varied
from 5 to 15

C. For each set of conditions, the critical back
pressure was found. Fig. 8 shows the effects of back pressure, area
ratio, HTE and LTE operating conditions on the system COP. This
performance map shown in Fig. 8 could be used to design an
ejector for a particular application in that it shows how the
system can be operated optimally with any changes in environ-
mental conditions, i.e., the condensing temperature. Typically the
condensing temperature is restricted to a minimum temperature
to that of the surroundings. If the surrounding temperature
changes, then the system must adjust in order to maintain
optimal operation. If the condensing temperature is reduced,
then the system can be adjusted to operate optimally in one of
two ways.
1. If the cooling capacity is to remain constant, the HTE temper-
ature should be reduced in order to establish a newcritical back
pressure that correlates to the lowered condensing tempera-
ture since the system is operating under choked conditions.
This means that the COP will increase due to a reduced heat
input required to the HTE, and the LTE temperature can
decrease for the same cooling capacity.
2. If the LTE temperature is to remain constant, the HTE temper-
ature should be reduced even further than in the previous
operation method (#1) which will produce the same effect as
Fig. 6. Schematic of test rig.
Fig. 7. Schematic of ejector prototype design.
Table 2
Detailed dimensions of the ejector. ERS II parts list (all dimensions listed in [mm]).
Primary nozzle 1 2
Throat diameter 1.2 1.6
Exit diameter 8 8
Diffuser
(b) Throat diameter 19 19
(c) Exit diameter 88.9 42.49
(f) Divergent angle (total) 10

10

Secondary nozzle
(a) Inlet diameter 56.01
Constant area section
(b) Diameter 19 19
(e) Length 90 50
D.A. Pounds et al. / International Journal of Thermal Sciences 67 (2013) 200e209 206
mentioned above, except that the further reduction in HTE
temperature will allow for the LTE temperature to rise back to
the original conditions.
It should be noted that if the effects of nozzle position are
considered, and the ejector could be designed so that the nozzle is
self-adjusting with changing conditions, then the systemoperation
could maintain consistency and only a change in nozzle position
would be required. Effects of nozzle positioning will be presented
hereafter.
The cooling capacity of the experimental ejector is presented in
Fig. 9. The cooling capacity at higher LTE temperatures is nearly the
same for the two different area ratios tested. However, at lower HTE
temperatures, the cooling capacity changes much more signi-
cantly. If the ejector is made to operate at its critical backpressure,
then the cooling capacity seems to be only dependent on the LTE
operating condition for a given area ratio.
Table 3 shows a comparison between the model prediction and
experimental results. The measured COP was used with the oper-
ating conditions of the vessels to calculate the area ratio of diffuser
throat area to nozzle throat area. Then the error between the
calculated area ratio and the actual area ratio of the experimental
ejector was calculated. The error was 17% on average, and never
over 30%.
The experimental results in Fig. 10 showthat the primary nozzle
position, which is the distance from the entrance of the mixing
chamber to the tip of the nozzle normalized by the length of the
mixing chamber, signicantly affects the systems cooling capacity
and there exists an optimum position. The optimal NXP is depen-
dent only on the high and low-temperature evaporator conditions.
However, dependence on HTE conditions is much greater than the
dependence on LTE conditions, which appeared to be slight, but
a more thorough investigation should be conducted. It should be
noted that for a given HTE operating condition, the input power
was not dependent on the NXP but only the HTE operating condi-
tion itself. Therefore, as far as the COP is concerned, only the cooling
capacity is affected by the NXP. It was found that changing the HTE
operating condition will shift the location of the optimum NXP, but
it is unclear at this time what that shift will look like. In column 6 of
Table 4, it can be seen also that the critical back pressure increase
with each change in NXP becomes less as the nozzle is positioned
further into the mixing chamber, and even eventually begins to
taper off.
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3 4.8
C
O
P
Critical Backpressure [kPa]
Performance Map
Area Ratio: 141
Area Ratio: 251
Fig. 8. Performance map of experimental data from prototype tests.
0.25
0.45
0.65
0.85
1.05
1.25
1.45
1.65
1.85
0.8 1.3 1.8 2.3 2.8 3.3 3.8 4.3 4.8
C
o
o
l
i
n
g

C
a
p
a
c
i
t
y

[
k
W
]
Critical Backpressure [kPa]
Cooling Capacity
Area Ratio: 141
Area Ratio: 251
10C
LTE: 15C LTE: 15C
Fig. 9. Cooling capacity at critical backpressure.
Table 3
Comparison of experimental data with theoretical prediction.
Temperature [

C] Backpressure
[kPa]
COP Theoretical
area ratio
Error
(%) HTE LTE
135 15 2.8 1.18 195 16
10 2.4 0.69 212 8
5 1.7 0.36 294 27
130 15 2.6 1.36 190 18
10 2.3 0.77 194 16
5 1.6 0.40 267 16
120 15 2.2 1.72 173 25
10 1.7 1.06 207 10
The experimental ejector had an actual area ratio of 251.
1.00
1.10
1.20
1.30
1.40
1.50
1.60
1.70
1.80
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
C
o
o
l
i
n
g

C
a
p
a
c
i
t
y

[
k
W
]
NXP Position
Cooling Capacity
HTE 130C, LTE15C, @ CriticalBackpressure
M
i
x
i
n
g

C
h
a
m
b
e
r

E
n
t
r
a
n
c
e
Fig. 10. Variations in measured cooling capacity with different nozzle exit positions.
Table 4
Cooling capacities and back pressures for various NXP positions.
NXP
position
COP Cooling
capacity
[W]
Temperature [

C] Critical
backpressure
[kPa]
HTE LTE
0.21 1.186 1480 130.0 15.0 2.24
0.38 1.332 1670 130.0 15.0 2.46
0.56 1.27 1570 130.0 15.0 2.67
0.73 1.139 1400 130.0 15.0 2.37
0.97 0.953 1200 130.0 15.0 1.87
D.A. Pounds et al. / International Journal of Thermal Sciences 67 (2013) 200e209 207
Fig. 11 illustrates the effects of the LTE and HTE operating
temperature and pressure on the system COP. Increasing the LTE
temperature will increase the units cooling capacity, the COP, and
raise the critical backpressure. Typically, the cooling temperature
should be chosen to meet the requirements of the refrigerated
space or ow stream. Note that the higher allowable temperature
will improve the system in all critical aspects. It was suggested by
Chunnanond and Aphornratana [6] that the higher saturation
pressure put more of a compression effect on the expansion wave
from the primary nozzle and caused the effective mixing area to
move further downstream. This means, the larger duct created
between the primary ow stream and ejector wall could have
a higher capacity for an increased amount of entrained refrigerant
vapor. In addition, if indeed, the secondary uid is choking at the
local sound speed, then this could explain the increased critical
backpressure. With the increase in saturation temperature, the
local sound speed is increased, and hence an increase in absolute
velocity. Higher absolute velocity means more momentum for the
mixed ow stream and an increase in critical backpressure.
Fig. 12 illustrates the effect of the area ratio of the diffuser
throat-to-nozzle throat on the system COP. Two area ratios have
been tested and have shown intriguing results. The largest
impact observed with different area ratios is the COP and crit-
ical back pressure. With a larger area ratio, a higher COP can be
achieved. With a smaller area ratio, a higher critical back
pressure can be achieved, as shown in Fig. 12. However, one
interesting thing to note is the variation in cooling capacity
with area ratio. Referring back to Fig. 9, the lower area ratio had
a cooling capacity only slightly higher than the larger area ratio
at higher LTE temperatures. This suggests that a more thorough
investigation of the area ratios effect on cooling capacity is
necessary. The cooling capacity seems to be weakly inuenced
by the area ratio at higher LTE temperatures and slightly more
at lower LTE temperatures, but this may not be the major
contributing factor.
Fig. 13 shows the effect of the diffuser exit diameter on the
system COP. The test was conducted by setting the HTE tempera-
ture to 130

C and the LTE temperature to 15

C and varying the
cooling capacity with incremental increases in backpressure from
below to above critical conditions in the condenser. One of the
differences in geometry fromwhat Chunnanond, and Aphornratana
[6] reported was the exit diameter of the diffuser. It can be
concluded the COP was not inuenced by the diffusers enlarged
exit diameter, as the COP is the same for both diffuser sizes. There is
a slight decrease in critical backpressure, which can be attributed to
the function of the diffuser itself.
0.000
0.200
0.400
0.600
0.800
1.000
1.200
1.400
1.600
1.800
C
O
P
Back Pressure [kPa]
HTE 135C, NXP = 0.5
LTE15C
LTE10C
LTE5C
0.000
0.200
0.400
0.600
0.800
1.000
1.200
1.400
1.600
1.800
1.000 1.500 2.000 2.500 3.000 3.500 4.000
1.000 1.500 2.000 2.500 3.000 3.500 4.000
C
O
P
Back Pressure [kPa]
LTE 15C, NXP = 0.5
HTE 135C
HTE 130C
HTE 120C
(b) (a)
Fig. 11. (a) Effects of LTE and condenser conditions on the system COP (b) Effects of HTE and condenser conditions on the system COP.
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
00.0
1.25 1.75 2.25 2.75 3.25 3.75
C
O
P
Back Pressure [kPa]
Effect of Diffuser Throat to Nozzle Throat Area Ratio
HTE 130C, LTE 10C, NXP 0.5
AR = 141
AR = 251
Fig. 12. Effects of area ratio on system performance.
0.000
0.200
0.400
0.600
0.800
1.000
1.200
1.400
1.600
1.500 1.700 1.900 2.100 2.300 2.500 2.700 2.900 3.100 3.300 3.500
C
O
P
Back Pressure [kPa]
HTE 130C, LTE 15C, NXP = 0.33
Diffuser Exit Diameter 42 mm
Diffuser Exit Diameter 88.9 mm
Fig. 13. System performance variations with diffuser size.
D.A. Pounds et al. / International Journal of Thermal Sciences 67 (2013) 200e209 208
5. Conclusions
A mathematical model, similar to the model proposed by
Chunnanond and Aphornratana [1, 6], Eames et al. [7], Keenan and
Neumann, [8] and Aphornratana and Eames [9], was developed to
predict the performance of the ejector refrigeration system devel-
oped herein. The main difference is that the current model can be
directly used to predict the performance of a two-uid system. In
order to verify the model developed herein, the experimental data
reported by other investigators is compared with the prediction
and found that the average error is less than 7%. Using the same
model, the system COP was calculated and compared with the
experimental data investigated herein and found that the average
error is less than 17%. Both comparisons show that the model is
acceptable for a useful tool to design an ERS. However, the results
predicted by the modeling are typically higher than the experi-
mental data, and it can be concluded that an improved model is
needed for accurate prediction.
Based on theoretical analysis, the prototype was developed
including the experimental system. The prototype was tested over
a range of HTE temperatures of 120e135

C, LTE temperatures of 5e
15

C and condenser temperatures of 7e30

C. The experimental
results show that the ERS can achieve a COP of 1.7 e much higher
than the results typically reported in the literature but at the
expense of critical backpressure. It was found that the primary
nozzle positioning signicantly affects the system COP and an
optimum NXP exists only for a given set of HTE and LTE conditions.
Studying the effects of the diffuser throat-to-nozzle throat area ratio
revealed the tradeoff betweencritical backpressure andsystemCOP.
The diffuser exit diameter appears to increase the critical back-
pressure only bya small amount, proportional tothe area of the exit.
In addition, the nozzle size effect on the system COP was investi-
gated as well, and the experimental data show that using a smaller
nozzle can result in a higher COP; however, the critical condensing
pressure is signicantly reduced. Whenthe area ratio decreased, the
cooling capacity is nearlyconstant whichsuggests that the inlet area
of the secondary ow might be a factor for this constant cooling
capacity, but further investigation is needed to verify. A systemCOP
map of the experimental data over a range of evaporator operating
conditions was developed, which could be used to design a new
ejector refrigeration system for a given application.
Acknowledgment
The work presented in this article was supported by Leonard
Wood Institute under Grant No. 0024733.
References
[1] K. Chunnanond, S. Aphornratana, Ejectors: application in refrigeration tech-
nology, Renewable and Sustainable Energy Reviews 8 (2004) 129e155.
[2] P. Srisastra, S. Aphornratana, A circulating system for a steam jet refrigeration
system, Applied Thermal Engineering 25 (14e15) (2005) 2247e2257.
[3] J.H. Keenan, E.P. Neumann, F. Lustwerk, An investigation of ejector design
by analysis and experiment, Journal of Applied Mechanics 72 (1950)
299e309.
[4] Y. Chang, Y. Chen, Enhancement of the steam-jet refrigerator using a novel
application of the petal nozzle, Experimental Thermal and Fluid Science 22
(2000) 203e211.
[5] J.T. Munday, D.F. Bagster, A new ejector theory applied to steam jet refriger-
ation, Industrial and Engineering Chemistry Process Design and Development
16 (4) (1977) 442e449.
[6] K. Chunnanond, S. Aphornratana, An experimental investigation of a steam
ejector refrigerator: the analysis of the pressure prole along the ejector,
Applied Thermal Engineering 24 (2004) 311e322.
[7] I.W. Eames, S. Aphornratana, H. Haider, A theoretical and experimental study
of a small-scale steam jet refrigerator, International Journal of Refrigeration 18
(6) (1995) 378e386.
[8] J.H. Keenan, E.P. Neumann, A simple air ejector, Journal of Applied Mechanics
(June 1942) 75e81.
[9] S. Aphornratana, I. Eames, A small capacity steam-ejector refrigerator:
experimental investigation of a system using ejector with movable primary
nozzle, International Journal of Refrigeration 20 (5) (1997) 352e358.
[10] D. Sun, Variable geometry ejectors and their application in ejector refrigera-
tion systems, Energy 21 (10) (1996) 919e929.
[11] A. Arbel, A. Shklyar, D. Hershgal, M. Barak, M. Sokolov, Ejector irreversibility
characteristics, Journal of Fluids Engineering 125 (2003) 121e129.
[12] H. El-Dessouky, H. Ettouney, I. Alatiqi, G. Al-Nuwaibit, Evaluation of steam jet
ejectors, Chemical Engineering and Processing 41 (2002) 551e561.
[13] J.D. Anderson, Modern Compressible Flow: with Historical Perspective, third
ed,, McGraw-Hill, Boston, 2003.
[14] R. Yapici, H. Ersoy, Performance characteristics of the ejector refrigeration
system based on the constant area ejector ow model, Energy Conversion and
Management 46 (18e19) (2005) 3117e3135.
[15] F.P. Incropera, D.P. Dewitt, T.L. Bergman, A.S. Lavine, Introduction to Heat
Transfer, fth ed., Wiley, Hobokenm, NJ, 2007.
D.A. Pounds et al. / International Journal of Thermal Sciences 67 (2013) 200e209 209