Applied Thermal Engineering 167 (2020) 114783

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Progress and challenges in utilization of ejectors for cryogenic cooling T

a,b,⁎ b
H.S. Cao , H.J.M. ter Brake
a
Key Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Energy and Power Engineering, Tsinghua University, 100084 Beijing,
China
b
Faculty of Science and Technology, University of Twente, 7500 AE Enschede, the Netherlands

HIGHLIGHTS

• Four configurations of ejector equipped JT cooling cycle are summarized.

• Analytical and CFD modelling of ejectors has been classified and compared.
• Both the macroscopic features and the local flow characteristics of ejectors should be adopted for validating ejector models.
• Key challenges for cryogenic ejector modelling are turbulent, two-phase and non-equilibrium flows.

ARTICLE INFO ABSTRACT

Keywords: In the field of cryogenic cooling, Joule–Thomson (JT) cooling is a mature technology but suffers from low
Ejector efficiency. Various efforts such as the optimization of compressors, cold stages, working fluids, and operating
Joule–Thomson conditions have been reported to develop more efficient JT cooling systems. As a further alternative, ejectors can
Cryogenic be used in JT cooling systems to create sub-atmospheric pressures and to recover expansion work, and thus
Turbulence models
achieve lower temperatures and higher system efficiency. This paper presents a comprehensive review of the use
Visualization techniques
of ejectors for cryogenic cooling. The research progress of cooling cycles with ejectors at the system-scale is
analyzed and compared, followed by the experimental investigation and numerical simulation of ejector per-
formance at the component-scale. The experimental studies mainly discuss the local flow characteristics of
ejectors captured through pressure measurements and flow visualization. The simulation studies introduce the
analytical modelling of ejectors although the focus is on the CFD modelling. Current challenges and suggestions
for future research on the use of ejectors in cryogenic cooling systems are summarized.

1. Introduction
Ejectors, also referred to as jet pumps or compressors, can be used to
support vacuum operations [1], or they can be operated as transfer and
mixing pumps [2–4]. Ejectors are fascinating devices, which have an
interesting history. Giffard invented a condensing ejector in 1858, and
his investigations were reviewed in detail by Kranakis [5]. In 1901, Sir
Charles Parsons invented an ejector which was used for removing air
from the condenser of a steam engine [6]. An ejector was later used by
Maurice Leblanc in a steam jet refrigeration system in 1912 [7]. Since
then, considerable efforts have been concentrated on the understanding
and design of ejectors. A historical development of ejectors was given
by Groll [8], Elbel [9] and Fischer [10]. In the articles of Groll and
Elbel, the references were limited to US sources, which was supple-
mented by Fischer in giving some important original sources.
Ejectors have been combined with various refrigeration systems for
improving their performance [11,12]. There are basically four utiliza-
tion categories of ejectors in refrigeration systems as shown in Fig. 1:
Firstly, an ejector can be a substitute for a compressor in a heat-driven
refrigeration system. In this case, the ejector is powered by a low grade
heat source such as solar energy and waste heat [13]. Secondly, an
ejector can be a substitute for a Joule–Thomson (JT) expansion valve in
a refrigeration system [14]. The JT expansion is an isenthalpic process,
which means that the kinetic energy produced during the expansion is
dissipated and eventually wasted [15]. The function of the ejector is to
partially recover this wasted energy. Thirdly, an ejector can be used in a
vapor compression refrigeration system, providing a second step of
compression, in order to decrease the input amount of energy required
by a mechanical compressor [16]. In this application, the primary flow
is liquid delivered by a pump, whereas the secondary flow is vapor fed
from a compressor. Fourthly, an ejector is used to drive a liquid re-
circulation loop, working as a pump. The fourth category is gaining a

⁎
Corresponding author at: Department of Energy and Power Engineering, Tsinghua University, 100084 Beijing, China.
E-mail addresses: HaishanCao@tsinghua.edu.cn, HaishanCao@gmail.com (H.S. Cao).

https://doi.org/10.1016/j.applthermaleng.2019.114783
Received 22 August 2019; Received in revised form 24 November 2019; Accepted 7 December 2019
Available online 11 December 2019
1359-4311/ © 2019 Elsevier Ltd. All rights reserved.
H.S. Cao and H.J.M. ter Brake Applied Thermal Engineering 167 (2020) 114783

Fig. 1. Schematic of four utilization categories

of ejectors in refrigeration systems. (a)
Utilization of low-grade energy as a heat-driven
compressor, adapted from [13]; (b) Utilization
of high-pressure gas to lower the evaporator
pressure and thus temperature, adapted from
[14]; (c) Utilization of high-pressure liquid as a
second stage compressor, adapted from [16]; (d)
Utilization of high-pressure gas as a liquid pump,
adapted from [17].

widespread interest in the refrigeration by using CO2, R134a and
R1234yf as the working fluids [17–19].
Previous reviews on ejectors mainly focused on the research efforts
and achievements on the use of ejectors in refrigeration systems and the
key contributions of several representative reviews are described in
Table 1. Ejectors, as expansion devices, can also be used in cryogenic
systems. To our knowledge, there is no review as yet on the progress
and challenges in the use of ejectors for cryogenic cooling, and this
review intends to fill this gap. This review extensively discusses the
ejector cryogenic cooling cycles at the system-scale, the ejector per-
formance at the component-scale, and the ejector fluid dynamics at the
local-scale.
2. Ejectors and their physics
An ejector is a work-producing device without moving parts, con-
sisting of three sections (see Fig. 2a): a suction section with a nozzle, a
mixing section, and a diffuser. Basic metrics of ejector performance
include the entrainment ratio ( µ ), defined as the ratio of the secondary
mass-flow rate (ms ) to that of the primary (mp ), and the pressure lift
ratio ( ), defined as the ratio of the back pressure ( pb ) (that is, the
ejector outlet pressure) to the secondary flow pressure ( ps ). The larger
the two parameters the better, however, there is a trade-off between the
two parameters. For a fixed-geometry ejector, the effect of the back
pressure on the entrainment ratio is illustrated in Fig. 2b, in which three
regions can be identified: double-choking flow, single-choking flow and
reversed flow. At low values of the back pressure, the primary flow and
the secondary flow are both choked and so the entrainment ratio is
constant. If the back pressure exceeds a critical value ( pb ), the sec-
ondary flow is no longer choked and the entrainment ratio decreases
sharply at increasing values of the back pressure. The entrainment ratio
reduces to 0 as the back pressure increases to the value of pb0 (see
Fig. 2b). Further increase in the back pressure will lead to reverse flow.
The critical value depends on the ejector geometry, nozzle type,
working fluid and operating conditions.
Two types of nozzles are often adopted in an ejector: a convergent

Table 1
Overview of representative reviews on ejector refrigeration.
Reference Key contributions

[112] Summaries the one- and two-dimensional design methods of ejectors for compressible fluids and the application of ejectors in refrigeration before 1995.
[113] Describes fundamental characteristics of shock-train and pseudo-shock phenomena in ejectors.
[114] Introduces several ejector refrigeration cycles at the system level; little information on physics of ejectors at the component level.
[115] Summaries the studies of ejectors and their applications in refrigeration from 1995 to 2005; CFD analysis of ejectors has been carried on since then.
[116,9] Focuses on the transcritical R744 two-phase ejector refrigeration cycle at the system level.
[3] Describes solar-driven ejector refrigeration systems at the system level and working fluids can be used in these cycles.
[117] Presents the development of single ejector refrigeration and hybrid ejector refrigeration with a mechanical compressor at the system level.
[118] Discusses the use of two-phase ejectors in refrigeration through thermodynamics modeling at the system level; little mention of the characteristics of a two-phase
ejector at component level.
[119] Provides a review on the developments in ejectors and their applications in refrigeration from 2003 to 2013, involving ejector design fundamentals, refrigerants
selection and system optimization.
[120] Describes various ejector refrigeration cycles at the system level, and emphasizes that sophisticated ejector models are desired to guide the ejector design.
[121] Presents a review linking ejector flow phenomena with component- and system-level performance; suggestions for future research focus on the local flow phenomena
including shock phenomena, two-phase flow, and visualization capturing the two phenomena.
[72] Discusses the developments in the use of ejectors for expansion work recovery in vapor-compression systems.
[122,123] Introduces the relations among ejector fluid dynamics at the local-scale, ejector behavior at the component-scale and the overall performance at the system-scale;
collects the past ejector research activities, describes the present achievements and proposes future research directions.

2
H.S. Cao and H.J.M. ter Brake
Fig. 2. Schematic (a) and operational modes (b) of a fixed-geometry ejector. Adapted from [73].
Fig. 3. Shock structures of jets in case of over-expansion (a) and under-ex-
pansion (b). Over-expansion only occurs in convergent-divergent nozzle (in
black colour), whereas under-expansion occurs in both convergent nozzle (in
purple colour) and convergent-divergent nozzle (in black colour). (For inter-
pretation of the references to colour in this figure legend, the reader is referred
to the web version of this article.)
nozzle and a convergent-divergent nozzle [20]. In a nozzle, a high-
pressure primary flow expands and its enthalpy converts into kinetic
energy. The secondary flow gets entrained into the ejector due to the
momentum exchange through a turbulent shear layer between the
primary and secondary flows. The primary flow typically forms a su-
personic jet in the mixing section, and the jet structure is determined by
the nozzle outlet pressure and the back pressure of the nozzle. The
nozzle outlet pressure can be greater than or equal to the back pressure
of the nozzle in a convergent nozzle, whereas the nozzle outlet pressure
can be greater than, equal to, or less than the back pressure of the
nozzle in a convergent-divergent nozzle. Therefore, over-expansion
only occurs in convergent-divergent nozzles, whereas under-expansion
occurs in both convergent nozzles and convergent-divergent nozzles
(see Fig. 3). When the flow is over-expanded, an oblique and a normal
shock waves form at the outlet plane of the nozzle and the gases un-
dergo compression as they pass through the shock waves. Meanwhile,
the normal shock creates a Mach disk in the exhaust flow. The reflected
3
Applied Thermal Engineering 167 (2020) 114783
oblique shock wave hits the free jet boundary where the outer edge of
the primary flow meets the secondary flow. At the reflected shock
wave-free jet boundary intersection, Prandtl–Meyer expansion [20]
occurs, followed by Prandtl–Meyer compression. If the compression
waves are strong enough, they will merge into an oblique shock wave
and form another Mach disk. This process repeats itself to create the
diamond pattern as shown in Fig. 3. Since the jet interacts with the
secondary flow, the wave patterns disappear eventually. When the flow
is under-expanded, the sequence of compression and expansion is
identical to that described for an over-expanded flow except that it
begins with the creation of a Prandtl–Meyer expansion fan rather than
oblique shock waves. As the mixed flows pass through the diffuser, and
convert kinetic energy into enthalpy, the pressure further increases and
the velocity decreases.
Four different ejector configurations are shown in Fig. 4: three cir-
cular and one planar. Fig. 4a represents a typical ejector lay-out, in
which a nozzle is attached to a main ejector body and the nozzle po-
sition is adjustable. Johnsn and Daggett [21] adopted the ejector design
in a closed-cycle cryocooler to provide cooling at temperatures below
4.2 K. Different from configuration a, the primary flow can be tuned in
configuration b by using a control-pin. Giffard invented this type of
ejector design (see Fig. 4b), that can be inserted into the ejector over a
greater or lesser extent to control the mass-flow rate through the ejector
[22]. This type of ejector has been used in a variety of air-conditioning
and refrigeration systems [9]. Little et al. [23,24] introduced a micro-
miniature ejector fabricated in a glass wafer (see Fig. 4c). The ejector
was a planar design, having a nozzle with a throat 0.28 mm wide and
0.25 mm deep. The planar configuration allows it to be fabricated by
micromaching technology which offers advantages such as high fabri-
cation accuracy and the possibility of batch processing. The dis-
advantage of this planar configuration was the additional frictional
losses due to the close proximity of the walls, which lowered the ejector
efficiency. Little et al. intended to integrate the micro-miniature ejector
with a miniature cryocooler in order to realize a cold-end temperature
below 77 K. Fig. 4d describes a two-stage ejector, in which the mo-
mentum of the mixed stream exhausted from the first stage is employed
to drive the second entrained stream into the second stage ejector,
consequently leading to a better performance over a single-stage ejector
[25]. Grazzini et al. investigated the use of a two-stage ejector in a
steam ejector refrigeration cycle [26].
3. Utilization of ejectors in JT cryocoolers
A JT cryocooler is based on the Linde–Hampson cycle (see Fig. 5a)
[27]. The cycle represented in the pressure versus enthalpy diagram
shown in Fig. 5b describes the cooling cycle of a 100 K JT cryocooler
operating with nitrogen gas between 0.78 MPa and 10 MPa. The high-
pressure nitrogen gas flows through a counter flow heat exchanger
(CFHX) (1 2), and undergoes JT expansion when it passes through a
H.S. Cao and H.J.M. ter Brake Applied Thermal Engineering 167 (2020) 114783

Fig. 4. Four types of ejector configuration. (a-2) copied from [21], (b-2) copied from [22], and (c-2) copied from [23,24].

flow restriction resulting in a lower temperature (2 3). In the steady
state operation, liquid forms after the expansion and stores in the
evaporator. The liquid absorbs heat from the device to be cooled and
evaporates (3 4), the low-pressure cold gas then flows back through
the CFHX (4 5) thereby precooling the incoming high-pressure gas.
The cold-end temperature of the JT cryocooler is determined by the
boiling temperature of the liquid inside the evaporator, that is 100 K @
0.78 MPa. The cooling effect of gas expansion only occurs if the initial
temperature of the gas, prior to expansion, is below the inversion
temperature of the gas. The inversion temperature is the temperature
where the JT coefficient ( µJT ), that is, the change in temperature re-
sulting from an isenthalpic pressure drop ( T / p)h is zero. The max-
imum inversion temperature is about 10 times the normal boiling point
of the fluid [28].
For an adiabatic JT expansion process, the process is isenthalpic and
the energy of the high-pressure gas is not utilized as work or dis-
charged, but is dissipated as heat. This cancels out a large amount of the
expansion cooling effect of the gas, and only the JT cooling effect

Fig. 5. Schematic of the Linde–Hampson cooling cycle (a) and corresponding pressure versus enthalpy diagram (b).

4
H.S. Cao and H.J.M. ter Brake
remains below the inversion temperature. Ejectors can use the energy of
the high-pressure gas, which is completely wasted as heat in the normal
JT expansion, to entrain and compress a low-pressure gas to a higher
intermediate pressure. This results in two beneficial effects: more
cooling by extracting work from the high-pressure gas, as well as use of
this work to reduce the input power. Ejectors can be used to vacuumize
an evaporator, reducing the pressure in the evaporator and thereby
lowering the boiling temperature of the evaporating liquid. Thus, this
avoids the use of a room-temperature vacuum pump at sub-atmospheric
pressure and eliminates air leakage. On the other hand, ejectors can
also be used to lift the pressure at the compressor inlet and reduce the
required pressure ratio of the compressor, resulting in less compression
work. Thus, ejectors can be used either to lower the pressure of the
evaporator of a JT cryocooler, or reduce the required compression input
power. In the latter case also a higher compression output pressure can
be established at unchanged input power. Lower input power and
higher output pressure both result in a higher cooler efficiency.
The utilization of an ejector in a JT cooler was first proposed by
Rietdijk [29] to create a sub-atmospheric pressure in a liquid helium
evaporator. In his patent, Rietdijk presented a number of cooling cycle
configurations with one or more ejectors, for producing cold and/or
liquefying gases [30]. Rietdijk later updated his patent with two further
ejector designs: one ejector having the inside walls of its nozzle hea-
table for volatizing contamination from the working fluid deposited on
the walls [31]; and the other ejector with a nozzle made of heat in-
sulating material for preventing the conduction of heat through the
walls of the nozzle [31]. Nicholds [32] presented a JT type cooling
apparatus, in which an ejector was adopted at the warm end of a JT
cooler to reduce the pressure in the evaporator of the cooler. Daney
et al. [33] added an ejector to a JT cooler with nitrogen gas as the
working fluid. An assemble type ejector was used to characterize the
performance of ejectors with different dimensions. Both the primary
nozzle and the mixing tube-diffuser including a secondary nozzle were
exchangeable. There were three primary nozzles with diameters of
0.58, 0.57 and 0.46 mm and three secondary nozzles with diameters of
6.35, 7.54 and 8.73 mm. Several combinations were tested and the best
ejector gave a suction pressure of 0.27 bar and a corresponding eva-
porator temperature of 67.7 K. Wu et al. [24] used an ejector in a he-
lium JT cycle to create sub-atmospheric pressure for achieving a tem-
perature below 4 K. The ejector under test had a nozzle throat diameter
of 2.24 mm and a diameter ratio (the ratio of the diameter of the dif-
fuser throat to that of the nozzle throat) of 3. Wu et al. emphasized that
a sub-atmospheric pressure created through an ejector can save power
and permit the use of smaller heat exchangers in the cooling system.
Von Rohr and Trepp [34] incorporated an ejector into a commercial
helium cooler. The measurement results indicated that Laval nozzles
with cylindrical end parts had higher entrainment ratios than the ones
without cylindrical end parts at a given pressure lift ratio, and that the
temperature of the primary flow affected the entrainment ratio.
Table 2
Summary of the utilization of an ejector in a JT cooling cycle at cryogenic temperatures. The cooling cycle configurations a, b, c and d correspond to the four types
shown in Fig. 6. Dt refers to nozzle throat diameter.
Reference Cycle configuration Dt Working fluid
(–) (mm) (–)
[29] a N/A Helium
[32] b N/A N/A
[33] a 0.58/0.57/0.46 Nitrogen
[24] a 2.24 Helium
[34] a 7.2 Helium
[21] a 0.15 Helium
[35,36] c N/A N2/N2-CH4 mixture
[37] c N/A R13-R24 mixture
[38] c N/A Air, CO, N2, Propane, R13, R508B
[39] d N/A N2/CH4
[40] a N/A N2
5
Applied Thermal Engineering 167 (2020) 114783
Johnson and Daggett [21] presented a helium ejector design that can be
used in a JT cooler to provide refrigeration at temperatures below 4.2 K.
The ejector consisted of a non-adjustable primary converging–diverging
nozzle assembly which was attached to a main ejector body. The pri-
mary nozzle with a throat diameter of 0.15 mm was fabricated by
electrode-discharge machining (EDM). The ejector was characterized
both at room and cryogenic temperatures and the relations between
input pressures and entrainment ratios were obtained. The room-tem-
perature relations were used to anticipate the ejector performance at
cryogenic temperatures. But the actual difference in the ejector physics
happening inside ejectors between room and cryogenic temperatures
was not given. Yu et al. [35] compared the coefficient of performance
(COP) of JT coolers without and with an ejector by theoretical calcu-
lation. It was found that the COP of the cooler with an ejector operating
with pure N2 increased by 25–56% and with a N2-CH4 mixture increase
by 18–47%. Yu et al. [36] further analyzed the JT cooler with an ejector
by using the exergy method and reconfirmed that applying an ejector
was an effective approach to improve the performance of the JT cooler.
Moreover, Yu et al. [37] theoretically investigated the performance of a
JT cooler with an ejector using the non-azeotropic refrigerant mixture
R14/R23 and found a COP increase by 24–42% compared to a basic JT
cooler. Rashidi et al. [38] analyzed the performance improvement of a
JT cooler due to the introduction of an ejector into the cycle using a
combined first and second law approach. The COP and exergy effi-
ciency of six different working fluids (air, carbon monoxide, nitrogen,
propane, R13 and R508B) were compared. The analysis results showed
that both the COP and exergy efficiency increased as the pressure lift
ratio of the ejector increased. Cao et al. [39] compared the COP of a
basic JT cycle with an ejector equipped JT cooling cycle. The analysis
results confirmed the COP improvement for both nitrogen and methane.
Energy conservation analysis showed that the required entrainment
ratio for a given pressure lift ratio increased as the ejector position was
shifted from the cold end to the warm end of the cooler. Since the
working fluid may condense and thus disturb the performance of the
ejector at low temperatures, the optimum ejector position remained an
open question. Lee et al. [40] designed an ejector-equipped JT cooler
operating at 68 K using nitrogen as the working fluid. The analysis
showed that the corresponding evaporation pressure of 28 kPa could be
realized through the ejector. Compared to the JT cycle with an external
room-temperature vacuum pump, the cooling cycle with an ejector was
considered to be more attractive for sub-atmospheric operation.
The above-mentioned studies are summarized in Table 2. Most of
these studies focused on conceptual designs or experimental in-
vestigations of the use of ejectors in JT cooling cycles. Yu et al. [35–37],
Rashid et al. [38], Cao et al. [39] and Lee et al. [40] analyzed the
improved performance of JT cooling cycles with an additional ejector.
In these analyses, entrainment ratios and the pressure lift ratios of
ejectors were assumed or derived based on mass, momentum and en-
ergy conservation, taking into account the assumed efficiencies of the
Cold-end temperature Pressure
(K) (MPa)
4.2/3.6 0.1/0.05
N/A N/A
67.7 0.27
3.9 N/A
3.55 0.5
3.0 N/A
80–130 N/A
208–238 N/A
75–135 N/A
63.3–90.0(N2),90.7–145(CH4) 0.013–0.36(N2),0.012–0.823(CH4)
68 0.28
H.S. Cao and H.J.M. ter Brake
Fig. 6. Schematics and corresponding pressure versus enthalpy diagrams of four configurations of ejector equipped JT cooling cycle. ‘p’, ‘s’ and ‘m’ refer to primary,
secondary and mixed flows, respectively.
nozzle, mixing section and diffuser. The approach how to realize an
ejector with the assumed performance, however, needs further in-
vestigation.
Four typical configurations of ejector equipped JT cooling cycles are
schematically depicted in Fig. 6. In configuration a ejectors have the same pressure lift ratio of 1.88 (ratio of 1.47 MPa to
[29,41,24,34,33,21,40], the high-pressure gas first passes the CFHX
(1 2), then expands through the ejector (2 3) prior to the JT valve to
entrain the vaporized refrigerant from the evaporator (6 3). After
passing the ejector, the refrigerant enters a phase separator. The satu-
rated liquid in the phase separator further flows through the throttling
valve to the evaporator (4 5), while the saturated vapor in the phase
separator returns to the CFHX (7 8). In configuration b [32], one
stream of the high-pressure gas passes through the CFHX (1,2 3) and
then expands through the valve (3 4) and the other stream flows to the
ejector as the primary flow (1,2 7) to entrain the vaporized refrigerant
from the CFHX (6 7). The outlet gas of the ejector returns to the
compressor. In configuration c [35,36,38], the high-pressure gas flows
through CFHX I (1 2), after which it is split into two streams. One
stream passes CFHX II (2 3) and expands through the JT valve (3 4)
and the other stream flows to the ejector as the primary flow (2 6) to
entrain the vaporized refrigerant from the evaporator (5 6). The outlet
gas of the ejector returns to CFHX II. In configuration d [39], after
passing CFHX I (1 2), one stream further flows through CFHX II (2 3)
and expands through the JT valve (3 4), while the other stream flows
to the ejector as the primary flow (2 7). The vaporized refrigerant
from the evaporator first passes CFHX II (5 6), and then is entrained to
the ejector (6 7). The outlet gas of the ejector returns to CFHX I (7 8).
In principle, the four typical configurations are applicable for all
cryogenic fluids such as helium, hydrogen, neon, nitrogen, etc. To
realize Joule–Thomson cooling, it is required that the outlet tempera-
tures of the compressors are below the corresponding inversion tem-
peratures of these cryogenic fluids. The performances of the basic JT
cooling cycle shown in Fig. 5 and the four configurations shown in
Fig. 6 are analyzed and compared based on the following assumptions:
(i) The working fluid is nitrogen gas; (ii) The outlet temperatures of the
6
Applied Thermal Engineering 167 (2020) 114783
compressors are fixed at an ambient temperature of 300 K; (iii) The
outlet pressures of the compressors are fixed at 10 MPa; (iv) The CHFXs
are ideal heat exchangers; (v) The working fluid flowing out the eva-
porator is saturated nitrogen gas at 0.78 MPa, that is, 100 K; (vi) The
0.78 MPa) and inlet and outlet velocities; (vii) the ejectors are adia-
batic.
For configuration a, the pressure at point 4 is 1.47 MPa, and the
enthalpy difference between points 5 and 6 (h56 ) is determined as de-
scribed in Fig. 6a. Based on the law of conservation of energy, the
cooling power of the configuration a can be expressed as ms·h56 or mp·
h18 , and thus the entrainment ratio of the ejector ( µ = ms / mp ) is fixed at
h18/ h56 , which in this case equals 0.12. It is assumed that the ejectors of
configurations b, c and d have the same entrainment ratio for the
purpose of comparison. For configuration b, the states of the inlets of
the primary and secondary flows (points 1,2 and 6) are fixed under the
assumption that the CHFX is an ideal heat exchanger. For configuration
c, the state of the secondary flow inlet (point 5) is saturated vapor at
0.78 MPa, whereas the state of point 2 can be adjusted. For config-
uration d, the states of the inlets of the primary and secondary flows
(points 2 and 6) have the same temperature based on the assumption
that the CHFX II is an ideal heat exchanger. Due to this constraint, the
temperatures of points 2 and 6 are fixed at 108 K based on the law of
conservation of energy. Table 3 gives the thermodynamic states of
points 1–8 of the four cooling cycle configurations in details.
The COP of configuration a shown in Fig. 6 is defined as µ h56/ g81,
where g81 is the Gibbs free energy difference between points 8 and 1;
whereas the COPs of configurations b, c and d are defined as
µ h 45/(g81 (µ + 1)), where h45 is the enthalpy difference between
points 4 and 5. Thermodynamic analysis results show that the config-
urations a, c and d have the same COP of 0.0955, which is higher than
0.0784, the COP of the basic JT cooling cycle; whereas the COP of the
configuration b is 0.0197, which is less that of the basic JT cooling
cycle, because the energy flowing out the ejector (that is point 7 or 8
shown in Fig. 5b) is not recovered. The analysis is based on the
H.S. Cao and H.J.M. ter Brake Applied Thermal Engineering 167 (2020) 114783

Table 3
The thermodynamic states of points 1–8 of the four cooling cycle configurations shown in Fig. 6. The thermodynamic states of points marked with asterisk∗ are
adjustable.
Point Configuration a Configuration b Configuration c Configuration d

T p T p T p T p
(K) (MPa) (K) (MPa) (K) (MPa) (K) (MPa)

1 300 10 300 10 300 10 300 10

2 157 10 300 10 200∗ 10 108 10
3 110 1.47 158 10 102 10 102 10
4 110 1.47 100 0.78 100 0.78 100 0.78
5 100 0.78 100 0.78 100 0.78 100 0.78
6 100 0.78 300 0.78 156∗ 1.47 108 0.78
7 110 1.47 286 1.47 177∗ 1.47 110 1.47
8 300 1.47 286 1.47 300 1.47 300 1.47

assumption that the ejectors have the same entrainment ratio and Table 4
pressure lift ratio. From the pressure versus enthalpy diagrams of four Summary of the representative working fluids for ejector applications. The
configurations, we can see that the states of the inlets and outlets of the temperatures without asterisk∗ refer to the normal boiling points of the cor-
ejectors are different from each other, which affects the ejector beha- responding working fluids, and the temperature marked with asterisk∗ refers to
vior at the component-scale, and thus also the performance of the four the triple point temperature of CO2.
configurations at the system-scale. Reference Working fluid Temperature (K)

[34] Helium 4.2 K

4. Experimental investigation of ejectors [33] Nitrogen 77.2 K
[17] R32 221.2 K
The system-scale performance of the ejector equipped JT cooling [124] R125 224.8 K
[125] R22 232.1 K
systems shown in Fig. 6 is determined by the component-scale behavior
[126] Ammonia 239.6 K
of the ejector, which is characterized by the complicated local-scale [127] R12 221.2 K
fluid dynamics [42]. The ejectors applied in the JT cooling systems can [128] R1234yf 243.8 K
be two-phase ejectors or single-phase ejectors, depending on the ther- [52] R134a 246.8 K
modynamic states of the inlets and outlet of the ejectors. This section [79] R152a 248.8 K
[129] R600a 261.1 K
focuses on the experimental investigation of the component-scale
[79] R142b 263.7 K
ejector behavior and the involved local-scale fluid dynamics. [46] R123 300.6 K
An extensive number of experimental studies has been performed to [50] R141b 304.8 K
explore the effects of the operating conditions, geometrical factors and [91] Water 372.8 K
[130] CO2 216.6 K∗
working fluids on the component-scale behavior of the ejector char-
acterized by: the entrainment ratio and the pressure lift ratio. The op-
erating conditions include temperatures and pressures of the primary
(PIV) [54]. A typical set-up for measuring total and wall static pressures
flow, the secondary flow and the outlet flow. The following ejector
with a measurement range of 0–20 bar is given in Fig. 7.
geometrical factors have been investigated: the primary nozzle throat
In shadowgraph photography [55,56] and Schlieren photography
diameter [43,44], the area ratio of the mixing section to the primary
[57–66], the flow motion is revealed by changes in the refractive index
nozzle throat [45,46], the primary nozzle exit position (NXP) [47–49],
of the fluid. A Schlieren image shows the gradient of the refractive
the mixing section length [50,51], and the mixing section converging
index, whereas a shadowgraph shows the change in this gradient. Dif-
angle [50]. At given operating conditions, the mass flow rate through a
fering from shadowgraph and Schlieren photography, Mie scattering
nozzle operating at choked condition is determined by the nozzle throat
[67] and Rayleigh scattering [68,69] can visualize and distinguish two
diameter. The optimum area ratio for the maximum entrainment ratio
interacting flows by tracing particles that are added to the flow. De-
depends mainly on the primary flow pressure. Yan et al. [52] found that
sevaux et al. [68,69] investigated Rayleigh scattering and laser induced
the optimum area ratio increased almost linearly with an increase of the
fluorescence for visualizing the mixing zone in an ejector qualitatively.
primary flow pressure. The optimum mixing section length should be
Tracers used in Rayleigh scattering were water droplets present in the
selected to be equal to or slightly greater than the pseudo-shock length,
humid air flow and fluorescent tracer particles were selected for laser
depending on the similarity of the viscosities in the primary and sec-
induced fluorescence. PIV [70] not only can visualize and distinguish
ondary flows (which can be represented by the Tanimoto coefficient).
two interacting flows but also measure their velocities. Bouhanguel
Once the mixing section is longer than the optimum length, the effect of
et al. [70] explored velocity profiles of a supersonic air ejector using
NXP is not prominent [48]. The optimum mixing section converging
PIV. The quantitative PIV technique is attractive for the validation of
angle is very sensitive to the primary flow pressure, and a slight var-
CFD simulations in supersonic ejectors. Fig. 8 shows the flow features in
iation in the angle has a great influence on the ejector performance near
a supersonic ejector by using Schlieren photography and laser Mie
its optimum working point [50]. Various working fluids with normal
scattering techniques. The Schlieren image clearly shows the shock
boiling points from 4.2 K to 372.8 K that have been investigated for a
cells, mixing layer, and the non-mixed region of the ejector. After
varity of applications are listed in Table 4.
passing through the non-mixed region, the flow enters a region of in-
The two component-scale parameters cannot adequately reflect the
creased mixing with large scale turbulent structures. Absence of shock
essence of an ejector; besides, the local-scale flow characteristics are
structures in this region indicates the presence of subsonic flow. The
essential to understand the ejector performance in multi-scale view
laser scattering image shows that the seeded primary flow is made
[53]. Local flow characteristics have been examined experimentally
visible while the secondary flow remains dark. The shock cells, mixing
through local pressure measurements and various flow visualization
layer, and the non-mixed region of the ejector are also clearly seen in
techniques, such as shadowgraph photography, Schlieren photography,
this image.
Mie scattering, Rayleigh scattering and Particle Image Velocimetry

7
H.S. Cao and H.J.M. ter Brake
Fig. 7. Locations of total and wall static pressure measurements. Adapted from [66].
Compared to two component-scale parameters, visualization cap-
tures the local flow phenomena like pseudo-shocks and phase change.
Ejectors with the same component-scale behavior may have different
local flow phenomena. Therefore, the agreement between modelling
and experimental results in component-scale parameters does not au-
tomatically imply adequate understanding of the local flow phenomena
within the ejector. In this sense, it is essential to measure the local flow
characteristics for validating models of ejectors.
5. Numerical simulation of ejectors
Compared to the experimental investigation, the numerical simu-
lation takes less time and cost, and offers more characteristics of the
simulated flow. Numerical simulation can be used to guide ejector de-
sign, optimize ejector performance and interpret experimental results.
Rapid advances in computer hardware and numerical methodology
have had a profound effect on ejector modeling.
There are typically two ways to simulate the flow in ejectors: ana-
lytical modelling and dynamic modelling [71]. The review given by He
et al. [71] outlined the evolution process of the mathematical models of
ejectors, mainly on analytical modelling. Analytical models are also
Fig. 8. Schlieren (a) and laser scattering (b) images of the air flow through a supersonic air ejector operated with a primary pressure of 7.69 bars, a secondary
pressure of 0.86 bars and an exit pressure of 1.29 bars. Copied from [110].
8
Applied Thermal Engineering 167 (2020) 114783
called thermodynamic models by He et al. [71], zero-dimensional
models by Elbel et al. [72] and one-dimensional models by Huang et al.
[73]. In analytical modelling, the ejector is divided into a motive
nozzle, a mixing section, and a diffuser [74,75,12]. The flow para-
meters including temperatures, velocities and pressures of the separate
sections are obtained through mass, momentum, and energy con-
servation equations for each section. Detailed velocity, pressure, and
temperature distributions along ejectors are not considered. The effects
of frictional and mixing losses are taken into account by using corre-
sponding coefficients introduced in the governing equations. However,
these coefficients need to be determined experimentally or by using
dynamic models. Detailed information along the ejectors can be ob-
tained from dynamic models that may be two or three-dimensional.
Besides, the position of the shock waves can also be tracked. After
discussing the analytical modelling in Section 5.1, this review focuses
more extensively on dynamic modelling in Section 5.2.
5.1. Analytical modelling
The general operation process of ejectors can be explained through
analytical modelling. Many assumptions are made, such as steady-state
H.S. Cao and H.J.M. ter Brake
operation, adiabatic flow, zero velocity at inlets of primary and sec-
ondary flows, uniform temperature and velocity profile at each cross-
section [76]. Analytical models are expressed in explicit algebraic
equations, which can be divided into single- and two- phase flow
models, respectively discussed in 5.1.1 and 5.1.2.
5.1.1. Single-phase flow models
In analytical models, the flow in ejectors can be divided into three
sections: the suction chamber with nozzle, the mixing section and the
diffuser as shown in Fig. 2a. For the nozzle and the diffuser, it was
assumed that the fluid expands or compresses isentropically and the
friction losses are accounted for by using isentropic efficiency coeffi-
cients. The mixing process can be described by using momentum, en-
ergy and continuity equations and ideal gas relations with a coefficient
accounting for the frictional loss [73]. However, the difficulty lies in
deriving a reasonable analytical solution for those equations during
mixing. To obtain a closed form solution, extra hypotheses need to be
stated including: the starting point of mixing (nozzle exit, or down-
stream of the nozzle exit), and the mixing conditions. Keenan et al. [57]
introduced two types of mixing conditions: constant pressure mixing
and constant area mixing. In addition to the hypothesis of the constant
pressure mixing or constant area mixing, Keenan et al. assumed that the
primary and secondary flows started to mix at the primary nozzle exit.
In the model proposed by Huang et al. [73], the mixing was assumed to
have both constant-area and pressure. Differing from Keenan et al.’s
assumptions, Huang et al. assumed the mixing to start somewhere
downstream of the nozzle exit. In both models of Keenan et al. and
Huang et al., a normal shock wave was assumed in the mixing section.
The flow through a shock is adiabatic but non-isentropic, which means
the process is not a reversible process. The process can be described by
using continuity, momentum, energy equations and two thermo-
dynamic relations which are the enthalpy-temperature relations and the
equation of state. More details can be found in the reference [77]. A
normal shock wave has no heat and momentum transfer, which means
it has the same energy equation as Fanno flows (constant area duct
flows with friction and without heat transfer) and the same momentum
equation as Rayleigh flows (constant area duct flows with heat transfer
and negligible friction). Therefore, the intersection of the Fanno and
Rayleigh lines shown in enthalpy–entropy diagrams (which is a Mollier
diagram) indicates what happens in the shock [78].
Most analytical models are based on the ideal gas assumption which
does not reflect some phenomena occurring in ejectors as, for instance,
the JT effect, which is due to intermolecular forces between real gas
particles. Rogdakis and Alexis [78] developed a model, in which con-
tinuity, momentum, and energy equations were applied to the nozzle,
the mixing and the diffuser sections of the ejector. Instead of using ideal
gas properties, the thermodynamic and transportation properties of real
gases were adopted in the model. However, among others two as-
sumptions of this model need to be verified. One is that the ratio of the
total mass-flow rate to the cross section area was assumed to be con-
stant; and the other is that the occurrence of a shock wave was not
considered. Yu et al. [79] developed a similar model for real gases, in
which the velocity of the secondary flow in the mixing section was
assumed to be negligible.
5.1.2. Two-phase flow models
As shown in Fig. 6, the thermodynamic states of the flow in the
ejector may cross the equilibrium saturation line. Therefore, two-phase
flow should be considered in these cases. Unfortunately, due to the
complexity of two-phae flow, there are no well-established two-phase
flow ejector models for analyzing or designing an ejector. Liu and Groll
[80] presented a two-phase flow ejector model for the transcritical CO2
cycle and the model predicted the pressure, quality and mass flow rate
of the outlet flow for given inlet conditions of the primary and sec-
ondary flows. The expression for the speed of sound of the two-phase
flow proposed by Attou and Seynhaeve [81] was adopted in the study.
9
Applied Thermal Engineering 167 (2020) 114783
The two-phase flow was assumed to be in homogenous equilibrium
(instantaneous heat transfer between the phases) and the quality of the
two-phase flow was introduced to evaluate the thermodynamic prop-
erties. In the homogeneous equilibrium model (HEM), vapor and liquid
are at saturation conditions, having the same pressure, temperature,
and velocity. It is, however, problematic that the expansion process
occurs too rapidly (1–10 µ s [82]) for the two-phase flow to reach hy-
drodynamic and thermodynamic equilibrium. To account for the me-
tastability, Feburie et al. [83] proposed the delayed equilibrium model
(DEM), which described the metastable flow as composed of three
phases: saturated liquid, saturated vapor and metastable liquid or
vapor, depending on condensation or evaporation. The three phases
had the same pressure and velocity, whereas the temperature of the
metastable liquid or vapor was different from that of the saturated li-
quid and vapor. Lorenzo et al. found DEM was more reliable than HEM
in predicting the critical pressure and critical mass flux of two-phase
flow [84]. HEM was based on the belief that the momentum, energy,
and mass transfer between phases occurred rapidly and the equilibrium
reached instantaneously. Considering the shortcomings of HEM, the
homogeneous relaxation model (HRM) was proposed, in which the
mechanical equilibrium is assumed, however, the non-equilibrium
phase change is taken into account. Differing from DEM, HRM has the
same equilibrium state as HEM. Angielczyk et al. [85] compared HEM
and HRM for supersonic CO2 two-phase flow through an ejector motive
nozzle and found that HRM fidelity was slightly better in terms of speed
of sound predictions and thus also in predicted the mass flow rates. The
speed of sound varies due to phase change, which affects shock for-
mation. Lear et al. [86] explored the choking of the secondary flow as a
result of the flow entering the two-phase regime.
The general operation process of ejectors can be understood by
using analytical models, which can also provide preliminary perfor-
mance estimations. However, analytical models cannot correctly de-
scribe the flow physics locally along ejectors. This is attributed to the
nature of analytical models that ignore local behavior and only consider
a set of parameters at a few cross-sections along the ejector [87].
Analytical models have limited functionality, and cannot be used for
ejector design independently, because all the loss coefficients in-
troduced in the model are specific to working fluid and geometry, and
need to be determined experimentally. Analytical models have been
revised to make them more accurate, such as real gas models and two-
phase flow models, which consider the real properties of the fluid in
ejectors. However, these revised models are still not reliable and ac-
curate enough for actual design due to the ambiguous loss coefficients.
More details on the analytical ejector models can be found in a previous
review given by Little and Garimella [76].
5.2. CFD modelling
Dynamic modelling provides insights into the local flow physics
along ejectors, which is more reliable and accurate for ejector design
[88–90]. Dynamic models are expressed in partial differential equa-
tions, which are solved by using numerical methods, such as finite
difference method, finite element method, and finite volume method.
Dynamic models have attracted more and more attention because their
modelling results are closer to the actual physics of ejectors and because
computing power over the years has dramatically increased. The typical
Mach number distribution in an ejector simulated by Computational
Fluid Dynamics (CFD) modelling is shown in Fig. 9.
5.2.1. Single-phase flow models
The flow in an ejector is compressible with choked flow, turbulent
mixing and shock waves, in which significant changes in velocity,
pressure and temperature are involved. The compressible flow is gov-
erned by the fluid flow conservation equations (continuity, momentum,
and energy equations). Crucial inputs to CFD modelling are the pressure
and temperature-dependent properties of the working fluid. It is
H.S. Cao and H.J.M. ter Brake
Fig. 9. Simulated Mach number evolution based on Realizable k- model along the axis length of a typical ejector. Copied from [111].
reasonable to use the properties of ideal gas to predict the ejector
performance when the relevant temperatures are much higher than the
critical temperature [91]. However, once ejectors are used in JT cryo-
genic cooling systems, the deviations from ideal gas behavior are no
longer negligible. Therefore, in this case gas properties of real gases are
required [92,93]. The specific heat can be derived by using the equation
of state, whereas other properties such as thermal conductivity and
viscosity need additional equations, which for instance can be found in
the references included in Refprop [28]. Another crucial point of CFD
modelling is the choice of turbulence models. Turbulent flows are
characterized by unsteady, irregular (aperiodic) motion in which
transported quantities such as momentum, energy, and species con-
centration fluctuate in time and space. Turbulence modelling has been
aimed at developing tractable mathematical models that can accurately
predict properties of turbulent flows. However, there is not yet a single,
practical turbulence model that can reliably predict all turbulent flows
with sufficient accuracy. For practical estimations of the ejector per-
formance, the Averaged Navier–Stokes based models including Eddy
viscosity based models and Reynolds stress models (RSM) are generally
adopted (see Table 5).
By using CFD modelling, Lamberts et al. [94] analyzed the location
of the transfers between the primary and the secondary streams, and
the irreversibilities within supersonic ejectors, and their connection
with the entrainment. The analysis results showed that the mixing be-
tween both streams was not yet completed when they entered the dif-
fuser if the ejector was operated in the on-design region (see Fig. 2). The
exergy of the secondary stream was the highest under the condition that
exergy losses by heat transfer and viscous dissipation was minimized
and the exergy transfer was maximized. Sierra-Pallares et al. [95] in-
vestigated the mechanisms of entropy generation with the aid of CFD
based analysis. Using this method, the locations of irreversibility sites
were identified, the irreversibility origin were extracted, and geometry
improvements were suggested. Banasiak et al. [96] presented a CFD-
based analysis of the flow irreversibility in R744 ejectors and assessed
the contribution of the local irreversibilities to the overall entropy
generation. Based on this analysis, it was suggested that future ejector
optimization procedures should concentrate on the entire ejector geo-
metry rather than on a single component.
In general, CFD modelling offers more complete flow field in-
formation for ejector simulations and produces a better prediction
performance than analytical modelling. The choice of an appropriate
turbulence model for the modelling of an ejector is crucial. In the
previous studies summarized in Table 5, Averaged Navier–Stokes based
turbulence models have been adopted to predict the ejector perfor-
mance including Spalart–Allmaras, standard, realizable and RNG k-
models, standard and SST k- models, Transition SST model, and RSM.
The thermal and hydrodynamic fluid properties were evaluated by
using the ideal gas model and real gas models including Refprop
10
Applied Thermal Engineering 167 (2020) 114783
database equations and other equations of state. Compared to the ideal
gas model, real gas models provide better results. The global perfor-
mance of ejectors in terms of the entrainment ratio and the critical back
pressure can be reasonably well estimated by using Averaged Na-
vier–Stokes models, and there is no consistent conclusion on the pre-
ferred turbulence models (see the turbulence models marked with as-
terisk∗ in Table 5). Hemidi et al. [53] found that the preferred
turbulence models were operating conditions dependent: standard k-
model was preferred for on-design conditions, whereas SST k- model
was preferred for off-design conditions. In general, the differences in
global performance predictions between turbulence models are small.
However, SST k- shows better performance in terms of the predictions
of local flow phenomena such as non-mixing length [97], shock wave
structure [98], and static pressure distribution [99].
5.2.2. Two-phase flow models
Next to the challenges facing the single-phase flow, CFD models of
the two-phase flow ejectors are still faced with difficulties such as two-
phase flow modelling, mass-transfer mechanism across the interface
and the prediction of speed of sound. Although advances in computa-
tional fluid mechanics have provided the basis for insight into the dy-
namics of multi-phase flows, the modelling of two-phase flow in ejec-
tors is clearly still under development. In terms of two-phase flow
models, the mixture model and the Eulerian model have been studied
for ejector modelling (see Table 6). The mixture model uses a single-
fluid approach to solve the continuity equation, the momentum equa-
tion, and the energy equation for the mixture, and allows the phases to
move at different velocities, using the concept of slip velocities. The
Eulerian model solves one set of momentum, continuity, and energy
equations for each phase separately. In terms of mass transfer me-
chanism across the interface of equilibrium and non-equilibrium two-
phase flows, besides the above-mentioned HEM and HRM, evaporation/
condensation models based on the dynamics of bubble growth and the
kinetic theory of phase change [100], evaporation/condensation
models based on Rayleigh-Plesset equation [101], Lee evaporation
model [102], Shan condensation model [103] and Sideman condensa-
tion model [104] have been investigated. Karathanassis et al. [105]
compared different mass-transfer models across the interface based on
HEM, HRM, the kinetic theory of gases (Hertz-Knudsen equation), and
bubble-dynamics considerations using the Zwart-Gerber-Belamri (ZGB)
model for the prediction of flashing flows in nozzles. The numerical
results showed that the mass-transfer model based on the kinetic theory
of gases had the best prediction accuracy, whereas the validity of the
HRM and ZGB models was conditional. Moreover, it was found that the
numerical results were highly sensitive to the assumed distribution of
the activated nucleation sites, and its accurate determination was a
prerequisite condition for the predictions of the phase-change process.
The speed of sound in the two-phase flow is essential to model the
 H.S. Cao and H.J.M. ter Brake

Table 5
Single-phase flow CFD models of ejectors. The ejector configuration presented in the Table is type a shown in Fig. 4. The subscripts A, B and C of pressure (p) and temperature (T) refer to the primary inlet flow, secondary
inlet flow and the outlet flow, respectively. The preferred turbulence models are marked with asterisk∗. Dt refers to nozzle throat diameter.
Reference Dt Working fluid Operating pressures Operating temperatures Gas property model Turbulence model Secondary mass Primary mass Entrainment ratio
(pA,pB,pC) (TA,TB,TC) flow flow
(mm) (–) (MPa) (MPa) (–) (–) (g/s) (g/s) (–)

[131] 7.9 Air 0.13–0.60; 0.10; 0.1–0.18 300; 300; 300 Density (Ideal gas) Other properties Standard and RNG k- ∗ 0.0035–0.0140 N/A N/A
(Constant or temperature-
dependent)
[93] 2.64 R141b 0.42–1.81; 0.10; 0.13 353–423; 305; 313 Density (Ideal gas) Other properties Realizable k- 6.05–11.11 9.08–36.85 0.26–1.04
(Refprop)
[97] 8.0 Air 0.4–0.6; 0.04–0.078; 0.1 300; 300; 300 Density (Ideal gas) Other properties Standard, Realizable, RNG k- , Standard 0–2 3–16 0–0.3
(Constant) and SST k- ∗ and RSM
[92] N/A R142b 1; 0.173; 0.4681 348; 283; 326.1 All properties (Refprop) Standard, RNG k- , SST k- N/A N/A 0.076–0.4
[132,133] 1.50, Water 0.198–0.360; 9-17e-4; 25- 393–413; 278–288; Density (Ideal gas) Other properties Realizable k- N/A N/A 0.17–0.53
1.75, 2.00 50e-4 294–306 (Constant)
[50] 2.8 R141b 0.45–0.6; 43–47e-3; 356–368; 283–285; Density (Ideal gas) Other properties RNG k- N/A N/A 0.2–0.9
0.09–0.10 302–305 (Refprop)

11
[53] 3.3 Air 0.3–0.6; 0.1; 0.11–0.2 Ambient temperatures Density (Ideal gas) Other properties Standard k- ∗(on-design) SST k- ∗(off- N/A N/A N/A
(Constant) design)
[134] 36.5 Water 0.22–0.267; 124–146e-4; 396–403; 324–327; Density (Ideal gas) Other properties Standard and Realizable k- 78–97 76–100 0.78–1.29
155-205e-4 340–348 (Constant)
[135] 2.96 R245fa 0.786–1.922;33-82e-3;92- 353–393; 273–293; Density (Refprop) Other properties Standard k- N/A N/A 0.11–0.61
298e-3 286–318 (Constant)
[136] N/A Air 0.1–0.20; 0.013; 0.10 N/A N/A Standard, RNG, Realizable k- and RSM N/A N/A N/A
[48] 2.85 N2/He 1.0–3.0; 0.25; 0.25–0.31 N/A Density (Ideal gas) Other properties Standard k- N/A N/A 1–13.25
(Refprop)
[137] 0–1.7 R152a 2.10; 0.79; 0.37 353; 308; 283 Refprop RNG k- N/A N/A 0–0.45
[137] 0–2.8 R600a 1.35; 0.46; 0.22 353; 310; 283 Refprop RNG k- N/A 5–28 0–0.65
[88] 10.65 Air 0.2–0.5; 995-1003e-4; 295; 295; 295 Ideal gas Standard, Realizable k- ; SST k- ; RSM 37–204 132–343 0.24–1.50
0.1–0.1875
[98,138] 2.0 R134a 2.60–3.19; 0.41; 0.76–0.90 352–362; 283; 302–308 Density (Ideal gas, RKS EoS, Standard, RNG, Realizable k- ; SST k- ∗ N/A N/A 0.34–0.50
Refprop) Other properties
(Constant, Refprop)
[99] 2 Water 1.99–2.70; 872.5–1228.7e- 393–403; 278–288; Ideal gas Spalart–Allmaras, Standard, RNG, N/A N/A 0.17–0.53
6;3–4.5e-3 297–313 Realizable k- ; Standard, SST k- ; RSM
[139] 2 Water 1.99–2.70; 87-123e- 393–403; 278–288; Ideal gas Spalart–Allmaras, Standard, RNG, N/A N/A N/A
5;3–4.5e-3 297–313 Realizable k- ; Standard, SST k- ; RSM
Applied Thermal Engineering 167 (2020) 114783
H.S. Cao and H.J.M. ter Brake Applied Thermal Engineering 167 (2020) 114783

Table 6
Two-phase flow CFD models of ejectors. Dt refers to nozzle throat diameter.
Reference Dt Working fluid Gas property model Two-phase flow model Mass transfer mechanism Turbulence model
(mm) (–) (–) (–) (–) (–)

[101] 1 CO2 Refprop Mixture model HEM, Rayleigh-Plesset equation RNG k-

[140] 6.10 Water Constant Eulerian model Shan condensation model Realizable k-
[141] 0.54 CO2 Refprop Mixture model HRM Standard k-
[100] N/A CO2 Refprop Mixture model Cavitation, Boiling phase change model SST k-
[142] 1 CO2 Refprop Mixture model HEM RNG k-
[143] 1 CO2 TEMO-Media Mixture model HEM SST k-
[144] 1.41–2.00 CO2 Refprop Mixture model HEM Realizable k-
[145] 1.41 CO2 Refprop Mixture-Model HRM SST k-
[146] 1.07–1.12 CO2 Refprop Mixture-Model HEM Realizable k-
[107] 0.24 CO2 Refprop Mixture-Model Lee evaporation model SST k-
[147] 3 Liquefied natural gas/Boiling off gas Constant Mixture-Model Sideman condensation model Standard k-

compressibility effects and thus predict critical flow condition. A few
studies summarized in Table 6 discussed the speed of sound in the two-
phase flow. Bulinski et al.[101] calculated the speed of sound in the
two-phase flow simply as a weighted average with respect to the mass
fraction of each phase. Yazdani et al. [100] adopted the speed of sound
in the two-phase flow derived by Brennen [106], which included both
the homogeneous equilibrium (instantaneous heat transfer between the
phases) and homogeneous frozen (no heat transfer between the phases)
models. Yazdani et al. emphasized that the speed of sound depends on
both the physical properties and the interface area of the two-phase
flow. To estimate the speed of sound, a precise tracking of the liquid/
vapor interface is required and that the problem is still challenging.
Giacomelli et al. [107] compared two different equations for the speed
of sound in the two-phase flow including the Brennen equation [106]
and the Wallis equation that was the default equation implemented in
ANSYS Fluent and cannot be modified by the user [108]. Giacomelli
et al. found that the Brennen equation was in better agreement with the
behavior of the flashing flow in nozzles. More details regarding speed of
sound in two-phase flow can be found in papers presented by Flatten
and Lund, and Lorenzo et al. [109,84].
performance. However, ejectors with the same macroscopic fea-
tures may have very different local flow characteristics. To under-
stand the physical mechanism of the ejector performance, the local
flow characteristics of an ejector need to be investigated through
experiments such as local pressure measurements and flow visua-
lization. It is still challenging to relate the two macroscopic features
with local flow characteristics.
(4) Compared to analytical modelling, CFD modelling provides insights
into the local flow physics along ejectors, that are more reliable and
accurate. Averaged Navier–Stokes based turbulence models have
been adopted to predict the ejector performance. These models can
reasonably well estimate the performance in terms of entrainment
ratio and critical back pressure, and the prediction differences be-
tween turbulence models are small. In terms of local flow phe-
nomena, the SST k- model is recommended.
(5) At cryogenic temperatures, the flow in ejectors may undergo phase
changes. Two-phase flow models should take into account possible
nucleation, growing of condensation droplets, and metastable
states, when the thermodynamic states of the flow in the ejector
cross the equilibrium saturation line.

6. Summary and outlook for future work
This paper presents a review of studies on the use of ejectors for
cryogenic cooling. The ejector performance at the component-scale
determines the overall performance of cryogenic cooling systems at the
system-scale. The design of a well performing ejector at different op-
erating conditions is becoming a bottleneck. To understand the physical
mechanism of the ejector at the local-scale, studies have been carried
out, both experimentally and theoretically. This paper reviews the
progress of the experimental investigation and the computational
modelling of ejectors for cryogenic cooling and discusses the future
research challenges.
The following conclusions can be drawn from the reviewed in-
vestigations:
(1) Differing from the extensive use of ejectors in refrigeration systems,
not much research has been carried out on the use of ejectors at
cryogenic temperatures. Moreover, the relatively little research
done focuses on concept design, experimental investigation and
thermodynamic analysis of cooling systems with ejectors, in which
the physical mechanism of ejectors at the local-scale is not well
explained.
(2) Ejectors with various geometric features have been proposed and
investigated, but only limited research is done on ejectors with a
nozzle throat diameter less than 1 mm. In contrast to the progress of
modeling on ejectors, the innovation in the design of ejectors is less
marked.
(3) The entrainment ratio and the pressure lift ratio are two macro-
scopic features of an ejector, which quantify the ejector
Regarding the direction of future research in the utilization of
ejectors for cryogenic cooling, focus should be on the physical me-
chanism of two-phase flow ejectors at cryogenic temperatures. To ex-
plore the physical mechanism, both macroscopic features and local flow
characteristics of ejectors need to be investigated. Since phase transi-
tions and turbulence are usually involved in ejectors at cryogenic
temperatures, the challenges of modelling turbulent and non-equili-
brium two-phase flows should be further investigated. Moreover, fur-
ther efforts are still needed to study the utilization potential of small-
scale ejectors with a nozzle throat diameter less than 1 mm in miniature
cooling systems.
Declaration of Competing Interest
The authors declared that there is no conflict of interest.
Acknowledgments
This work is supported by the Networking/Partnering Initiative of
European Space Agency (ESA) under Contract No. 4000114493/15/
NL/KML/fg.
References
[1] P. Chaiwongsa, S. Wongwises, Effect of throat diameters of the ejector on the
performance of the refrigeration cycle using a two-phase ejector as an expansion
device, Int. J. Refrig. 30 (2007) 601–608, https://doi.org/10.1016/j.ijrefrig.2006.
11.006.
[2] R.C. Kozinski, Vehicle air conditioning system utilizing refrigerant recirculation
within the evaporator/accumulator circuit, Patent US5493875A, 1996.

12
H.S. Cao and H.J.M. ter Brake
[3] J.M. Abdulateef, K. Sopian, M.A. Alghoul, M.Y. Sulaiman, Review on solar-driven
ejector refrigeration technologies, Renew. Sust. Energ. Rev. 13 (2009) 1338–1349,
https://doi.org/10.1016/j.rser.2008.08.012.
[4] M.J. Bergander, D. Butrymowicz, K. Smierciew, J. Wojciechowski, Refrigeration
cycle with ejector for second step compression, International Refrigeration and Air
Conditioning Conference, 2010, p. 2211.
[5] E.F. Kranakis, The french connection: Giffard’s injector and the nature of heat,
Technol. Cult. 23 (1982) 3–38, https://doi.org/10.2307/3104441.
[6] C.A. Parsons, M.G.G. Stoney, Trials of steam turbines for driving dynamos, Proc.
Inst. Mech. Eng. 61 (1901) 797–816, https://doi.org/10.1243/PIME_PROC_1901_
061_016_02.
[7] M. Leblanc, Refrigerating-machine, Patent US1029201A, 1912.
[8] E.A. Groll, Ejector technology, Int. J. Refrig. 34 (2011) 1543–1544, https://doi.
org/10.1016/j.ijrefrig.2011.08.001.
[9] S. Elbel, Historical and present developments of ejector refrigeration systems with
emphasis on transcritical carbon dioxide air-conditioning applications, Int. J.
Refrig. 34 (2011) 1545–1561, https://doi.org/10.1016/j.ijrefrig.2010.11.011.
[10] J. Fischer, On ejector technology, Int. J. Refrig. 36 (2013) 1399–1400, https://doi.
org/10.1016/j.ijrefrig.2012.10.002.
[11] D. Buyadgie, O. Buyadgie, S. Artemenko, A. Chamchine, O. Drakhnia, Conceptual
design of binary/multicomponent fluid ejector refrigeration systems, Int. J. Low
Carbon Technol. 7 (2012) 120–127, https://doi.org/10.1093/ijlct/cts038.
[12] J.E. Carrillo, F.S. de La Flor, J.S. Lissen, Thermodynamic comparison of ejector
cooling cycles. ejector characterisation by means of entrainment ratio and com-
pression efficiency, Int. J. Refrig. 74 (2017) 371–384, https://doi.org/10.1016/j.
ijrefrig.2016.11.006 URL: http://www.sciencedirect.com/science/article/pii/
S0140700716303723.
[13] M. Hamzaoui, H. Nesreddine, Z. Aidoun, M. Balistrou, Experimental study of a low
grade heat driven ejector cooling system using the working fluid R245fa, Int. J.
Refrig. 86 (2018) 388–400, https://doi.org/10.1016/j.ijrefrig.2017.11.018 URL:
http://www.sciencedirect.com/science/article/pii/S0140700717304619.
[14] Y. Li, J. Yu, H. Qin, Z. Sheng, Q. Wang, An experimental investigation on a
modified cascade refrigeration system with an ejector, Int. J. Refrig. 96 (2018)
63–69, https://doi.org/10.1016/j.ijrefrig.2018.09.015 URL: http://www.
sciencedirect.com/science/article/pii/S0140700718303372.
[15] A.T.A.M. de Waele, Basics of Joule-Thomson liquefaction and JT cooling, J. Low
Temp. Phys. 186 (2017) 385–403, https://doi.org/10.1007/s10909-016-1733-3.
[16] M.N. Sarevski, V.N. Sarevski, Preliminary study of a novel R718 refrigeration cycle
with single stage centrifugal compressor and two-phase ejector, Int. J. Refrig. 40
(2014) 435–449, https://doi.org/10.1016/j.ijrefrig.2013.12.005 URL: http://
www.sciencedirect.com/science/article/pii/S0140700713003848.
[17] N. Lawrence, S. Elbel, Theoretical and practical comparison of two-phase ejector
refrigeration cycles including first and second law analysis, Int. J. Refrig. 36
(2013) 1220–1232, https://doi.org/10.1016/j.ijrefrig.2013.03.007.
[18] N. Lawrence, S. Elbel, Experimental investigation of a two-phase ejector cycle
suitable for use with low-pressure refrigerants R134a and R1234yf, Int. J. Refrig.
38 (2014) 310–322, https://doi.org/10.1016/j.ijrefrig.2013.08.009 URL: http://
www.sciencedirect.com/science/article/pii/S0140700713002132.
[19] N. Lawrence, S. Elbel, Analysis of two-phase ejector performance metrics and
comparison of R134a and CO_2 ejector performance, Sci. Technol. Built En. 21
(2015) 515–525, https://doi.org/10.1080/23744731.2015.1030327.
[20] R.D. Zucker, O. Biblarz, Fundamentals of Gas Dynamics, 2nd ed., John Wiley &
Sons, New Jersey, 2002.
[21] D.L. Johnson, D.L. Daggett, A cold ejector for closed-cycle helium refrigerators,
The Telecommunications and Data Acquisition Report, 1987, pp. 102–111.
[22] R. Routledge, Discoveries and Inventions of the Nineteenth Century, Bracken
Books, London, 1989.
[23] W.A. Little, R.L. Paugh, Development of a fast cooldown Joule-Thomson micro-
miniature refrigerator and vacuum package for infrared focal plane array at 70 K,
Cryocoolers 6 (1991) 161–169.
[24] K.C. Wu, D.P. Brown, A.P. Schlafke, J.H. Sondericker, Helium Refrigerator with
Features for Operation at Supercritical Pressure, Springer US, Boston, MA, 1984,
pp. 495–501. https://doi.org/10.1007/978-1-4613-9865-3_57.
[25] F. Kong, H. Kim, Analytical and computational studies on the performance of a
two-stage ejectorcdiffuser system, Int. J. He at Mass Transf. 85 (2015) 71–87,
https://doi.org/10.1016/j.ijheatmasstransfer.2015.01.117 URL: http://www.
sciencedirect.com/science/article/pii/S0017931015001350.
[26] G. Grazzini, A. Rocchetti, Numerical optimisation of a two-stage ejector re-
frigeration plant, Int. J. Refrig. 25 (2002) 621–633, https://doi.org/10.1016/
S0140-7007(01)00063-9 URL: http://www.sciencedirect.com/science/article/
pii/S0140700701000639.
[27] H.B. Callen, Thermodynamics and an Introduction to Thermostatistics, Wiley, New
York, 1985.
[28] E.W. Lemmon, M.L. Huber, M.O. McLinden, Refprop version 8.0, NIST standard
reference database 23, 2007.
[29] J.A. Rietdijk, The expansion-ejector, a new device for liquefaction and refrigera-
tion at 4 K and lower, Pure Appl. Cryogen. 6 (1966) 241–249, https://doi.org/10.
1016/B978-0-08-012409-4.50023-1.
[30] J.A. Rietdijk, Device for producing cold and/or liquefying gases, Patent
US3427817DA, 1969.
[31] J.A. Rietdijk, Apparatus and ejector for producing cold, Patent US3434298DA,
1969.
[32] K.E. Nicholds, Cooling apparatus of the Joule-Thomson type, Patent US3645113,
1972.
[33] D.E. Daney, P.M. McConnell, T.R. Strobridge, Low-temperature nitrogen ejector
performance, in: K.D. Timmerhaus (Ed.), Advances in Cryogenic Engineering,
13
Applied Thermal Engineering 167 (2020) 114783
Springer, US, Boston, MA, 1973, pp. 476–485.
[34] P.V. Rohr, C. Trepp, Experimental investigation of an ejector, Cryogenics 25
(1985) 684–686, https://doi.org/10.1016/0011-2275(85)90188-2.
[35] J. Yu, Improving the performance of small Joule-Thomson cryocooler, Cryogenics
48 (2008) 426–431, https://doi.org/10.1016/j.cryogenics.2008.05.006.
[36] J. Yu, G. Tian, Z. Xu, Exergy analysis of Joule-Thomson cryogenic refrigeration
cycle with an ejector, Energy 34 (2009) 1864–1869, https://doi.org/10.1016/j.
energy.2009.07.034.
[37] J. Yu, J. Chen, Y. Li, Theoretical study on an innovative ejector enhanced Joule-
Thomson cycle, Int. J. Energy. Res. 34 (2009) 46–53, https://doi.org/10.1002/er.
1555.
[38] M.M. Rashidi, O.A. Beg, A. Habibzadeh, First and second law analysis of an ejector
expansion Joule-Thomson cryogenic refrigeration cycle, Int. J. Energ. Res. 36
(2012) 231–240, https://doi.org/10.1002/er.1800.
[39] H. Cao, S. Vanapalli, H.J. Holland, C.H. Vermeer, T. Tirolien, H.J.M. ter Brake,
High-efficiency Joule-Thomson cryocoolers incorporating an ejector, Cryocoolers
19 (2016) 419–426.
[40] J. Lee, C. Lee, S. Baek, S. Jeong, Investigation of ejector-equipped Joule-Thomson
refrigerator operating below 77 K, Int. J. Refrig. 78 (2017) 93–107, https://doi.
org/10.1016/j.ijrefrig.2017.03.016.
[41] J. Haisma, Ejector in refrigerating device, Patent US3496735A, 1970.
[42] A. Hemidi, F. Henry, S. Leclaire, J.M. Seynhaeve, Y. Bartosiewicz, CFD analysis of
a supersonic air ejector. Part II: Relation between global operation and local flow
features, Appl. Therm. Eng. 29 (2009) 2990–2998, https://doi.org/10.1016/j.
applthermaleng.2009.03.019.
[43] D.W. Sun, Variable geometry ejectors and their applications in ejector refrigera-
tion systems, Energy 21 (1996) 919–929, https://doi.org/10.1016/0360-5442(96)
00038-2.
[44] P. Chaiwongsa, S. Wongwises, Experimental study on R134a refrigeration system
using a two-phase ejector as an expansion device, Appl. Therm. Eng. 28 (2008)
467–477, https://doi.org/10.1016/j.applthermaleng.2007.05.005.
[45] A. Selvaraju, A. Mani, Experimental investigation on R134a vapour ejector re-
frigeration system, Int. J. Refrig. 29 (2006) 1160–1166, https://doi.org/10.1016/
j.ijrefrig.2006.01.004.
[46] R. Yapici, H.K. Ersoy, A. Aktoprakoglu, H.S. Halkaci, O. Yigit, Experimental de-
termination of the optimum performance of ejector refrigeration system de-
pending on ejector area ratio, Int. J. Refrig. 31 (2008) 1183–1189, https://doi.
org/10.1016/j.ijrefrig.2008.02.010.
[47] K. Chunnanond, S. Aphornratana, An experimental investigation of a steam ejector
refrigerator: the analysis of the pressure profile along the ejector, Appl. Therm.
Eng. 24 (2004) 311–322, https://doi.org/10.1016/j.applthermaleng.2003.07.003.
[48] C. Li, Y. Li, L. Wang, Configuration dependence and optimization of the entrain-
ment performance for gas-gas and gas-liquid ejectors, Appl. Therm. Eng. 48 (2012)
237–248, https://doi.org/10.1016/j.applthermaleng.2011.11.041.
[49] G.J. Hammink, Research towards the application of an ejector in a Joule-Thomson
cooler, Master’s thesis University of Twente, Enschede, the Netherlands, 2017.
[50] Y. Zhu, W. Cai, C. Wen, Y. Li, Numerical investigation of geometry parameters for
design of high performance ejectors, Appl. Therm. Eng. 29 (2009) 898–905,
https://doi.org/10.1016/j.applthermaleng.2008.04.025.
[51] M. Nakagawa, A.R. Marasigan, T. Matsukawa, A. Kurashina, Experimental in-
vestigation on the effect of mixing length on the performance of two-phase ejector
for CO2 refrigeration cycle with and without heat exchanger, Int. J. Refrig. 34
(2011) 1604–1613, https://doi.org/10.1016/j.ijrefrig.2010.07.021.
[52] J. Yan, W. Cai, Area ratio effects to the performance of air-cooled ejector re-
frigeration cycle with R134a refrigerant, Energy Convers. Manag. 53 (2012)
240–246, https://doi.org/10.1016/j.enconman.2011.09.002.
[53] A. Hemidi, F. Henry, S. Leclaire, J.M. Seynhaeve, Y. Bartosiewicz, CFD analysis of
a supersonic air ejector. Part I: Experimental validation of single-phase and two-
phase operation, Appl. Therm. Eng. 29 (2009) 1523–1531, https://doi.org/10.
1016/j.applthermaleng.2008.07.003.
[54] W. Merzkirch, Flow Visualization, Academic Press, New York, 1987.
[55] G. Olden, D. Lineberry, C. Linn, Landrum, C. Brian Hawk, Low-cost flow visuali-
zation for a supersonic ejector, in: 41st AIAA/ASME/SAE/ASEE Joint Propulsion
Conference & Exhibit, 2005, pp. 2990-2998, https://doi.org/10.2514/6.2005-
3523.
[56] A.B. Little, Y. Bartosiewicz, S. Garimella, Visualization and validation of ejector
flow field with computational and first-principles analysis, J. Fluids Eng. 37
(2015), https://doi.org/10.1115/1.4029534 051107.
[57] J.H. Keenan, E.P. Neumann, F. Lustwerk, An investigation of ejector design by
analysis and experiment, J. Appl. Mech. T. ASME 17 (1950) 299–309.
[58] K. Matsuo, K. Sasaguchi, Y. Kiyotoki, H. Mochizuki, Investigation of supersonic air
ejectors: Part 2, effects of throat-area-ratio on ejector performance, Bull. JSME 25
(1982) 1898–1905, https://doi.org/10.1299/jsme1958.25.1898.
[59] M. Chandrasekhara, A. Krothapalli, D. Baganoff, Mixing characteristics of a su-
personic multiple jet ejector, in: 25th AIAA Aerospace Sciences Meeting, AIAA-87-
0248, 1987, https://doi.org/10.2514/6.1987-248.
[60] S. Kim, H. Kim, S. Kwon, Transitional behavior of a supersonic flow in a two-
dimensional diffuser, KSME Int. J. 15 (2001) 1816–1821, https://doi.org/10.
1007/BF03185139.
[61] V. Dvorak, P. Safarik, Transonic instability in entrance part of mixing chamber of
high-speed ejector, J. Therm. Sci. 14 (2005) 258–264, https://doi.org/10.1007/
s11630-005-0011-5.
[62] H. Sugiyama, Y. Tsujiguchi, T. Honma, Structure and oscillation phenomena of
pseudo-shock waves in a straight square duct at Mach 2 and 4, in: 15th AIAA
International Space Planes and Hypersonic Systems and Technologies Conference,
AIAA, 2008, pp. 2008–2646. https://doi.org/10.2514/6.2008-2646.
H.S. Cao and H.J.M. ter Brake
[63] H. Zare-Behtash, N. Gongora-Orozco, K. Kontis, S.J. Holder, Application of novel
pressure-sensitive paint formulations for the surface flow mapping of high-speed
jets, Exp. Therm Fluid Sci. 33 (2009) 852–864, https://doi.org/10.1016/j.
expthermflusci.2009.03.002.
[64] H. Zare-Behtash, N. Gongora-Orozco, K. Kontis, Effect of primary jet geometry on
ejector performance: A cold-flow investigation, Int. J. Heat Fluid Fl. 32 (2011)
596–607, https://doi.org/10.1016/j.ijheatfluidflow.2011.02.013.
[65] M.V.S. Rao, G. Jagadeesh, Visualization and image processing of compressible
flow in a supersonic gaseous ejector, J. Indian Inst. Sci. 93 (2013) 57–66.
[66] S.K. Karthick, S.M.V. Rao, G. Jagadeesh, K.P.J. Reddy, Experimental parametric
studies on the performance and mixing characteristics of a low area ratio rec-
tangular supersonic gaseous ejector by varying the secondary flow rate, Energy
161 (2018) 832–845, https://doi.org/10.1016/j.energy.2018.07.135.
[67] N.T. Clemens, M.G. Mungal, A planar Mie scattering technique for visualizing
supersonic mixing flows, Exp. Fluids 11 (1991) 175–185, https://doi.org/10.
1007/BF00190296.
[68] P. Desevaux, J.P. Prenel, G. Hostache, An optical analysis of an induced flow
ejector using light polarization properties, Exp. Fluids 16 (1994) 165–170, https://
doi.org/10.1007/BF00206535.
[69] P. Desevaux, A method for visualizing the mixing zone between two co-axial flows
in an ejector, Opt. Lasers Eng. 35 (2001) 317–323, https://doi.org/10.1016/
S0143-8166(01)00020-3.
[70] A. Bouhanguel, P. Desevaux, Y. Bailly, L. Girardot, Flow velocity investigation by
particle image velocimetry in supersonic air ejector, Appl. Mech. Mater. 232
(2012) 256–260, https://doi.org/10.4028/www.scientific.net/AMM.232.256.
[71] S. He, Y. Li, R.Z. Wang, Progress of mathematical modeling on ejectors, Renew.
Sust. Energ. Rev. 13 (2009) 1760–1780, https://doi.org/10.1016/j.rser.2008.09.
032.
[72] S. Elbel, N. Lawrence, Review of recent developments in advanced ejector tech-
nology, Int. J. Refrig. 62 (2016) 1–18, https://doi.org/10.1016/j.ijrefrig.2015.10.
031.
[73] B.J. Huang, J.M. Chang, C.P. Wang, V.A. Petrenko, A 1-D analysis of ejector
performance, Int. J. Refrig. 22 (1999) 354–364, https://doi.org/10.1016/S0140-
7007(99)00004-3.
[74] M. Ouzzane, Z. Aidoun, Model development and numerical procedure for detailed
ejector analysis and design, Appl. Therm. Eng. 23 (2003) 2337–2351, https://doi.
org/10.1016/S1359-4311(03)00208-4.
[75] J.G. del Valle, J.M.S. Jabardo, F.C. Ruiz, J.S.J. Alonso, A one dimensional model
for the determination of an ejector entrainment ratio, Int. J. Refrig. 35 (2012)
772–784, https://doi.org/10.1016/j.ijrefrig.2011.11.020 URL: http://www.
sciencedirect.com/science/article/pii/S0140700711003008.
[76] A.B. Little, S. Garimella, A review of ejector technology for refrigeration appli-
cations, Int. J. Air-Cond Refrig. 19 (2011) 1–15, https://doi.org/10.1142/
S2010132511000351.
[77] J.D. Anderson Jr., Fundamentals of Aerodynamics, 3rd ed., McGraw-Hill, New
York, 2001.
[78] E.D. Rogdakis, G.K. Alexis, Design and parametric investigation of an ejector in an
air-conditioning system, Appl. Therm. Eng. 20 (2000) 213–226, https://doi.org/
10.1016/S1359-4311(99)00013-7.
[79] J. Yu, Y. Ren, H. Chen, Y. Li, Applying mechanical subcooling to ejector re-
frigeration cycle for improving the coefficient of performance, Energy Convers.
Manag. 48 (2007) 1193–1199, https://doi.org/10.1016/j.enconman.2006.10.009.
[80] F. Liu, E.A. Groli, Analysis of a two phase flow ejector for the transcritical CO2
cycle, International Refrigeration and Air Conditioning Conference, 2008, p. 2131.
[81] A. Attou, J.M. Seynhaeve, Steady-state critical two-phase flashing flow with pos-
sible multiple choking phenomenon: Part 1: Physical modelling and numerical
procedure, J. Loss Prevent. Proc. 12 (1999) 335–345, https://doi.org/10.1016/
S0950-4230(98)00017-5.
[82] V.P. Carey, Liquid Vapor Phase Change Phenomena: An Introduction to the
Thermophysics of Vaporization and Condensation Processes in Heat Transfer
Equipment, 2nd ed., Taylor & Francis Inc, New York, NY, 2007.
[83] V. Feburie, M. Giot, S. Granger, J. Seynhaeve, A model for choked flow through
cracks with inlet subcooling, Int. J. Multiph. Flow 19 (1993) 541–562, https://doi.
org/10.1016/0301-9322(93)90087-B.
[84] M.D. Lorenzo, P. Lafon, J.M. Seynhaeve, Y. Bartosiewicz, Benchmark of delayed
equilibrium model (DEM) and classic two-phase critical flow models against ex-
perimental data, Int. J. Multiph. Flow 92 (2017) 112–130, https://doi.org/10.
1016/j.ijmultiphaseflow.2017.03.004 URL: http://www.sciencedirect.com/
science/article/pii/S0301932216304475.
[85] Wojciech Angielczyk, D.B.J.M.S. Yann Bartosiewicz, 1-D modeling of supersonic
carbon dioxide two-phase flow through ejector motive nozzle, International
Refrigeration and Air Conditioning Conference, 2010 pp. 2362, URL: http://docs.
lib.purdue.edu/iracc.
[86] W.E. Lear, G.M. Parker, S.A. Sherif, Analysis of two-phase ejectors with fabri
choking. Proc. Inst, Mech. Eng. C J. Mech. Eng. Sci. 216 (2002) 607–621, https://
doi.org/10.1243/0954406021525287.
[87] N. Lawrence, S. Elbel, Experimental and analytical investigation of automotive
ejector air-conditioning cycles using low-pressure refrigerants, International
Refrigeration and Air Conditioning Conference, 2012, p. 2118.
[88] F. Mazzelli, A.B. Little, S. Garimella, Y. Bartosiewicz, Computational and experi-
mental analysis of supersonic air ejector: Turbulence modeling and assessment of
3D effects, Int. J. Heat Fluid Fl 56 (2015) 305–316, https://doi.org/10.1016/j.
ijheatfluidflow.2015.08.003 URL: http://www.sciencedirect.com/science/article/
pii/S0142727X15001101.
[89] J.A.E. Carrillo, F.J.S. de La Flor, J.M.S. Lissen, Single-phase ejector geometry
optimisation by means of a multi-objective evolutionary algorithm and a surrogate
14
Applied Thermal Engineering 167 (2020) 114783
CFD model, Energy 164 (2018) 46–64, https://doi.org/10.1016/j.energy.2018.08.
176 URL: http://www.sciencedirect.com/science/article/pii/
S0360544218317122.
[90] J.E. Carrillo, F.S. de La Flor, J.S. Lissen, Seasonal performance optimisation of
thermally driven ejector cooling cycles working with R134a, Int. J. Refrig. 104
(2019) 356–366, https://doi.org/10.1016/j.ijrefrig.2019.05.020 URL: http://
www.sciencedirect.com/science/article/pii/S0140700719302166.
[91] K. Pianthong, W. Seehanam, M. Behnia, T. Sriveerakul, S. Aphornratana,
Investigation and improvement of ejector refrigeration system using computa-
tional fluid dynamics technique, Energy Convers. Manag. 48 (2007) 2556–2564,
https://doi.org/10.1016/j.enconman.2007.03.021.
[92] Y. Bartosiewicz, Z. Aidoun, Y. Mercadier, Numerical assessment of ejector op-
eration for refrigeration applications based on CFD, Appl. Therm. Eng. 26 (2006)
604–612, https://doi.org/10.1016/j.applthermaleng.2005.07.003.
[93] E. Rusly, L. Aye, W.W.S. Charters, A. Ooi, CFD analysis of ejector in a combined
ejector cooling system, Int. J. Refrig. 28 (2005) 1092–1101, https://doi.org/10.
1016/j.ijrefrig.2005.02.005.
[94] O. Lamberts, P. Chatelain, Y. Bartosiewicz, New methods for analyzing transport
phenomena in supersonic ejectors, Int. J. Heat Fluid Fl. 64 (2017) 23–40, https://
doi.org/10.1016/j.ijheatfluidflow.2017.01.009 URL: http://www.sciencedirect.
com/science/article/pii/S0142727X16304222.
[95] J. Sierra-Pallares, J.G. del Valle, P.G. Carrascal, F.C. Ruiz, A computational study
about the types of entropy generation in three different R134a ejector mixing
chambers, Int. J. Refrig. 63 (2016) 199–213, https://doi.org/10.1016/j.ijrefrig.
2015.11.007 URL: http://www.sciencedirect.com/science/article/pii/
S0140700715003473.
[96] K. Banasiak, M. Palacz, A. Hafner, Z. Bulinski, J. Smolka, A.J. Nowak, A. Fic, A
CFD-based investigation of the energy performance of two-phase R744 ejectors to
recover the expansion work in refrigeration systems: An irreversibility analysis,
Int. J. Refrig. 40 (2014) 328–337, https://doi.org/10.1016/j.ijrefrig.2013.12.002
URL: http://www.sciencedirect.com/science/article/pii/S0140700713003812.
[97] Y. Bartosiewicz, Z. Aidoun, P. Desevaux, Y. Mercadier, Numerical and experi-
mental investigations on supersonic ejectors, Int. J. Heat Fluid Fl. 26 (2005)
56–70, https://doi.org/10.1016/j.ijheatfluidflow.2004.07.003.
[98] S. Croquer, S. Poncet, Z. Aidoun, Turbulence modeling of a single-phase R134a
supersonic ejector. Part 1: Numerical benchmark, Int. J. Refrig. 61 (2016)
140–152, https://doi.org/10.1016/j.ijrefrig.2015.07.030.
[99] G. Besagni, F. Inzoli, Computational fluid-dynamics modeling of supersonic ejec-
tors: Screening of turbulence modeling approaches, Appl. Therm. Eng. 117 (2017)
122–144, https://doi.org/10.1016/j.applthermaleng.2017.02.011 URL: http://
www.sciencedirect.com/science/article/pii/S1359431116328290.
[100] M. Yazdani, A.A. Alahyari, T.D. Radcliff, Numerical modeling of two-phase su-
personic ejectors for work-recovery applications, Int. J. Heat Mass Transf. 55
(2012) 5744–5753, https://doi.org/10.1016/j.ijheatmasstransfer.2012.05.071
URL: http://www.sciencedirect.com/science/article/pii/S0017931012003924.
[101] Z. Bulinski, J. Smolka, A. Fic, K. Banasiak, A. Nowak, A comparison of hetero-
geneous and homogenous models of two-phase transonic compressible CO_2 flow
through a heat pump ejector, IOP Conference Series Materials Science and
Engineering 10 (2010), https://doi.org/10.1088/1757-899X/10/1/012019
012019.
[102] W. Lee, A pressure iteration scheme for two-phase flow modeling, 2013, pp.
61–82, https://doi.org/10.1142/9789814460286_0004.
[103] A. Shah, I.R. Chughtai, M.H. Inayat, Numerical simulation of direct-contact con-
densation from a supersonic steam jet in subcooled water, Chinese. J. Chem. Eng
18 (2010) 577–587, https://doi.org/10.1016/S1004-9541(10)60261-3 URL:
http://www.sciencedirect.com/science/article/pii/S1004954110602613.
[104] S. Sideman, D. Moalem-Maron, Direct contact condensation, Elsevier. volume 15
of Advances in Heat Transfer, 1982, pp. 227–281. URL: http://www.sciencedirect.
com/science/article/pii/S0065271708701758, https://doi.org/10.1016/S0065-
2717(08)70175-8.
[105] I. Karathanassis, P. Koukouvinis, M. Gavaises, Comparative evaluation of phase-
change mechanisms for the prediction of flashing flows, Int. J. Multiph. Flow 95
(2017) 257–270, https://doi.org/10.1016/j.ijmultiphaseflow.2017.06.006 URL:
http://www.sciencedirect.com/science/article/pii/S0301932216306772.
[106] C.E. Brennen, Fundamentals of Multiphase Flow, Cambridge University Press,
2005, https://doi.org/10.1017/CBO9780511807169.
[107] F. Giacomelli, F. Mazzelli, A. Milazzo, A novel CFD approach for the computation
of R744 flashing nozzles in compressible and metastable conditions, Energy 162
(2018) 1092–1105, https://doi.org/10.1016/j.energy.2018.08.050 URL: http://
www.sciencedirect.com/science/article/pii/S0360544218315706.
[108] ANSYS, ANSYS Fluent theory guide, ANSYS, Canonsburg, PA, 2018.
[109] T. Flatten, H. Lund, Relaxation two-phase flow models and the subcharacteristic
condition, Math. Models Methods Appl. Sci. 21 (2011) 2379–2407, https://doi.
org/10.1142/S0218202511005775.
[110] S.M.V. Rao, G. Jagadeesh, Observations on the non-mixed length and unsteady
shock motion in a two dimensional supersonic ejector, Phys. Fluids 26 (2014),
https://doi.org/10.1063/1.4868879 036103.
[111] X. Wang, J. Dong, G. Zhang, Q. Fu, H. Li, Y. Han, J. Tu, The primary pseudo-shock
pattern of steam ejector and its influence on pumping efficiency based on CFD
approach, Energy 167 (2019) 224–234, https://doi.org/10.1016/j.energy.2018.
10.097 URL: http://www.sciencedirect.com/science/article/pii/
S0360544218320875.
[112] D.W. Sun, I. Eames, Recent developments in the design theories and applications
of ejectors, J. Energy Inst. 68 (1995) 65–79.
[113] K. Matsuo, Y. Miyazato, H.D. Kim, Shock train and pseudo-shock phenomena in
internal gas flows, Prog. Aerosp. Sci. 35 (1999) 33–100, https://doi.org/10.1016/
H.S. Cao and H.J.M. ter Brake
S0376-0421(98)00011-6.
[114] K. Chunnanond, S. Aphornratana, Ejectors: applications in refrigeration tech-
nology, Renew. Sust. Energ. Rev. 8 (2004) 129–155, https://doi.org/10.1016/j.
rser.2003.10.001.
[115] S.B. Riffat, L. Jiang, G. Gan, Recent development in ejector technology-a review,
Int. J. Ambient Energy 26 (2005) 13–26, https://doi.org/10.1080/01430750.
2005.9674967.
[116] S. Elbel, P. Hrnjak, Ejector refrigeration: An overview of historical and present
developments with an emphasis on air-conditioning applications, International
Refrigeration and Air Conditioning Conference, 2008, p. 2350.
[117] H.E.G. Bravo, R.D. Rodriguez, J.H. Gutierrez, R.B. y Brown, R.R. Aguila,
H.T. Pena, State of art of simple and hybrid jet compression refrigeration systems
and the working fluid influence, Int. J. Refrig. 35 (2012) 386–396, https://doi.
org/10.1016/j.ijrefrig.2011.10.012.
[118] K. Sumeru, H. Nasution, F.N. Ani, A review on two-phase ejector as an expansion
device in vapor compression refrigeration cycle, Renew. Sust. Energ. Rev. 16
(2012) 4927–4937, https://doi.org/10.1016/j.rser.2012.04.058.
[119] X. Chen, S. Omer, M. Worall, S. Riffat, Recent developments in ejector refrigera-
tion technologies, Renew. Sust. Energ. Rev. 19 (2013) 629–651, https://doi.org/
10.1016/j.rser.2012.11.028.
[120] J. Chen, S. Jarall, H. Havtun, B. Palm, A review on versatile ejector applications in
refrigeration systems, Renew. Sust. Energ. Rev. 49 (2015) 67–90, https://doi.org/
10.1016/j.rser.2015.04.073.
[121] A.B. Little, S. Garimella, A critical review linking ejector flow phenomena with
component- and system-level performance, Int. J. Refrig. 70 (2016) 243–268,
https://doi.org/10.1016/j.ijrefrig.2016.05.015 URL: http://www.sciencedirect.
com/science/article/pii/S0140700716301207.
[122] G. Besagni, R. Mereu, F. Inzoli, Ejector refrigeration: A comprehensive review,
Renew. Sustain. Energy Rev. 53 (2016) 373–407, https://doi.org/10.1016/j.rser.
2015.08.059 URL: http://www.sciencedirect.com/science/article/pii/
S1364032115009223.
[123] G. Besagni, Ejectors on the cutting edge: The past, the present and the perspective,
Energy 170 (2019) 998–1003, https://doi.org/10.1016/j.energy.2018.12.214.
[124] M. Jelinek, A. Levy, I. Borde, Performance of a triple-pressure-level absorption
cycle with R125-N,N’-dimethylethylurea, Appl. Energy 71 (2002) 171–189,
https://doi.org/10.1016/S0306-2619(02)00003-X.
[125] Y. Zhu, P. Jiang, Hybrid vapor compression refrigeration system with an in-
tegrated ejector cooling cycle, Int. J. Refrig. 35 (2012) 68–78, https://doi.org/10.
1016/j.ijrefrig.2011.09.003.
[126] K. Cizungu, M. Groll, Z.G. Ling, Modelling and optimization of two-phase ejectors
for cooling systems, Appl. Therm. Eng. 25 (2005) 1979–1994, https://doi.org/10.
1016/j.applthermaleng.2004.11.014.
[127] M. Sokolov, D. Hershgal, Enhanced ejector refrigeration cycles powered by low
grade heat. Part 1. Systems characterization, Int. J. Refrig. 13 (1990) 351–356,
https://doi.org/10.1016/0140-7007(90)90023-P.
[128] H. Li, F. Cao, X. Bu, L. Wang, X. Wang, Performance characteristics of r1234yf
ejector-expansion refrigeration cycle, Appl. Energy 121 (2014) 96–103, https://
doi.org/10.1016/j.apenergy.2014.01.079 URL: http://www.sciencedirect.com/
science/article/pii/S0306261914001093.
[129] M.S. Lee, H. Lee, Y. Hwang, R. Radermacher, H.M. Jeong, Optimization of two-
phase R600a ejector geometries using a non-equilibrium CFD model, Appl. Therm.
Eng. 109 (2016) 272–282, https://doi.org/10.1016/j.applthermaleng.2016.08.
078 URL: http://www.sciencedirect.com/science/article/pii/
S1359431116314375.
[130] D. Li, E.A. Groll, Transcritical CO2 refrigeration cycle with ejector-expansion de-
vice, Int. J. Refrig. 28 (2005) 766–773, https://doi.org/10.1016/j.ijrefrig.2004.
10.008.
[131] H.A.M. Al-Ansary, S.M. Jeter, Numerical and experimental analysis of single-phase
and two-phase flow in ejectors, HVAC&R Res. 10 (2004) 521–538, https://doi.
org/10.1080/10789669.2004.10391118.
[132] T. Sriveerakul, S. Aphornratana, K. Chunnanond, Performance prediction of steam
15
Applied Thermal Engineering 167 (2020) 114783
ejector using computational fluid dynamics: Part 1. Validation of the CFD results,
Int. J. Therm. Sci. 46 (2007) 812–822, https://doi.org/10.1016/j.ijthermalsci.
2006.10.014.
[133] T. Sriveerakul, S. Aphornratana, K. Chunnanond, Performance prediction of steam
ejector using computational fluid dynamics: Part 2. Flow structure of a steam
ejector influenced by operating pressures and geometries, Int. J. Therm. Sci. 46
(2007) 823–833, https://doi.org/10.1016/j.ijthermalsci.2006.10.012.
[134] M. Ji, T. Utomo, J. Woo, Y. Lee, H. Jeong, H. Chung, CFD investigation on the flow
structure inside thermo vapor compressor, Energy 35 (2010) 2694–2702, https://
doi.org/10.1016/j.energy.2009.12.002.
[135] D. Scott, Z. Aidoun, M. Ouzzane, An experimental investigation of an ejector for
validating numerical simulations, Int. J. Refrig. 34 (2011) 1717–1723, https://doi.
org/10.1016/j.ijrefrig.2011.06.001.
[136] J. Fan, J. Eves, H.M. Thompson, V.V. Toropov, N. Kapur, D. Copley, A. Mincher,
Computational fluid dynamic analysis and design optimization of jet pumps,
Comput. Fluids 46 (2011) 212–217, https://doi.org/10.1016/j.compfluid.2010.
10.024.
[137] S. Varga, P.M. Lebre, A.C. Oliveira, CFD study of a variable area ratio ejector using
R600a and R152a refrigerants, Int. J. Refrig. 36 (2013) 157–165, https://doi.org/
10.1016/j.ijrefrig.2012.10.016.
[138] S. Croquer, S. Poncet, Z. Aidoun, Turbulence modeling of a single-phase R134a
supersonic ejector. Part 2: Local flow structure and exergy analysis, Int. J. Refrig.
61 (2016) 153–165, https://doi.org/10.1016/j.ijrefrig.2015.07.029.
[139] S. Varga, J. Soares, R. Lima, A. Oliveira, On the selection of a turbulence model for
the simulation of steam ejectors using CFD, Int. J. Low Carbon Technol. 12 (2017),
https://doi.org/10.1093/ijlct/ctx007.
[140] A. Shah, I.R. Chughtai, M.H. Inayat, Experimental and numerical analysis of steam
jet pump, Int. J. Multiph. Flow 37 (2011) 1305–1314, https://doi.org/10.1016/j.
ijmultiphaseflow.2011.07.008 URL: http://www.sciencedirect.com/science/
article/pii/S0301932211001698.
[141] M. Colarossi, N. Trask, D.P. Schmidt, M.J. Bergander, Multidimensional modeling
of condensing two-phase ejector flow, Int. J. Refrig. 35 (2012) 290–299, https://
doi.org/10.1016/j.ijrefrig.2011.08.013.
[142] J. Smolka, Z. Bulinski, A. Fic, A.J. Nowak, K. Banasiak, A. Hafner, A computational
model of a transcritical R744 ejector based on a homogeneous real fluid approach,
Appl. Math. Model. 37 (2013) 1208–1224, https://doi.org/10.1016/j.apm.2012.
03.044 URL: http://www.sciencedirect.com/science/article/pii/
S0307904X12002077.
[143] C. Lucas, H. Rusche, A. Schroeder, J. Koehler, Numerical investigation of a two-
phase CO2 ejector, Int. J. Refrig. 43 (2014) 154–166, https://doi.org/10.1016/j.
ijrefrig.2014.03.003 URL: http://www.sciencedirect.com/science/article/pii/
S0140700714000504.
[144] M. Palacz, J. Smolka, W. Kus, A. Fic, Z. Bulinski, A.J. Nowak, K. Banasiak,
A. Hafner, CFD-based shape optimisation of a CO2 two-phase ejector mixing sec-
tion, Appl. Therm. Eng. 95 (2016) 62–69, https://doi.org/10.1016/j.
applthermaleng.2015.11.012 URL: http://www.sciencedirect.com/science/
article/pii/S1359431115012478.
[145] M. Haida, J. Smolka, A. Hafner, M. Palacz, K. Banasiak, A.J. Nowak, Modified
homogeneous relaxation model for the R744 trans-critical flow in a two-phase
ejector, Int. J. Refrig. 85 (2018) 314–333, https://doi.org/10.1016/j.ijrefrig.2017.
10.010 URL: http://www.sciencedirect.com/science/article/pii/
S0140700717303948.
[146] M. Palacz, J. Smolka, A.J. Nowak, K. Banasiak, A. Hafner, Shape optimisation of a
two-phase ejector for CO2 refrigeration systems, Int. J. Refrig. 74 (2017) 212–223,
https://doi.org/10.1016/j.ijrefrig.2016.10.013 URL: http://www.sciencedirect.
com/science/article/pii/S0140700716303425.
[147] P. Zheng, B. Li, J. Qin, CFD simulation of two-phase ejector performance influ-
enced by different operation conditions, Energy 155 (2018) 1129–1145, https://
doi.org/10.1016/j.energy.2018.04.066 URL: http://www.sciencedirect.com/
science/article/pii/S036054421830673X.