Journal Pre-proof

Development of Ejector Performance Map for Predicting

Fixed-geometry Two-phase Ejector Performance for Wide Range of
Operating Conditions

Muhammad Haider , Stefan Elbel

PII: S0140-7007(21)00121-3
DOI: https://doi.org/10.1016/j.ijrefrig.2021.03.022
Reference: JIJR 5090

To appear in: International Journal of Refrigeration

Received date: 22 February 2021

Revised date: 26 March 2021
Accepted date: 29 March 2021

Please cite this article as: Muhammad Haider , Stefan Elbel , Development of Ejector Performance
Map for Predicting Fixed-geometry Two-phase Ejector Performance for Wide Range of Operating Con-
ditions, International Journal of Refrigeration (2021), doi: https://doi.org/10.1016/j.ijrefrig.2021.03.022

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.

© 2021 Published by Elsevier Ltd.

HIGHLIGHTS
 Novel ejector performance map for a fixed-geometry ejector from experimental data
 New ejector efficiency for performance, vol. entrainment ratio for characterization
 Comparing efficiencies for pressure recovery and recirculation mechanism operation
 Predicts CO2 ejector performance accurately for wide range of operating conditions
 Applicable to other ejector datasets of different refrigerants and cycle layouts
Development of Ejector Performance Map for Predicting Fixed-geometry

Two-phase Ejector Performance for Wide Range of Operating Conditions

Muhammad Haider, Stefan Elbel*

Air Conditioning and Refrigeration Center,

Department of Mechanical Science and Engineering,

University of Illinois at Urbana-Champaign,

1206 West Green Street, Urbana, IL, 61801, USA

Phone: +1 (217) 244-1531; Fax: +1 (217) 333-1942

Email: elbel@illinois.edu

* Corresponding Author

1
Abstract

Recent studies have expanded the operational envelope of ejector by utilizing both pressure

recovery and liquid recirculation mechanisms. This necessitates finding a consistent

representation of fixed-geometry ejector performance data using only operational variables for

predicting ejector performance. Taking inspiration from the compressor performance maps, the

study explores the appropriate motive, performance, and characterization variable for

representing ejector performance while conducting experiments on a transcritical CO2 ejector

system. The existing ejector efficiency is revisited by conducting ejector power analysis. Two

possible motive variables, along with a new ejector efficiency is introduced. The trends of

existing and new ejector efficiencies are studied in controlled motive variable experiments while

changing suction flow conditions. The ejector performance can be represented with the help of a

single curve, termed as ejector performance map, using the new ejector efficiency as

performance variable and the volumetric entrainment ratio as characterization variable. The

prediction accuracy for both the single-phase and the two-phase suction inlet conditions are

evaluated while developing performance map using only the single-phase suction inlet data

points. The methodology can predict 89.3% of the data within 20% accuracy. The applicability

of the methodology is assessed for other datasets involving different refrigerants, cycle

architectures, and applications. The proposed ejector performance map can be utilized in

numerical ejector system analysis for investigating new cycle architectures based on ejector

experimental data, thus, improving system model fidelity. It can also help system design

engineers in making ejector system selection decisions after thorough system performance

analysis.

Keywords: two-phase ejector; ejector performance map; ejector performance prediction; ejector

2
efficiency; ejector system analysis

Nomenclature

Symbols:

APD absolute percentage deviation [%]

COP coefficient of performance [-]

COS condenser outlet split

d throat diameter of the motive nozzle [mm]

D diameter of the mixing section [mm]

h specific enthalpy [

IHX internal heat exchanger

L length of the mixing section [mm]

̇ mass flow rate [ ]

n number of data points

P pressure [kPa]

ph phase at the suction inlet

s specific entropy [ ]

T temperature [°C]

v specific volume [

̇ volumetric flow rate [

̇ power [W]

x vapor quality [-]

Greek symbols:

3
 measured efficiency [-]

̂ predicted efficiency [-]

mass entrainment ratio [-]

volumetric entrainment ratio [-]

Subscripts:

d referring to diffuser

ejec referring to ejector efficiency

i index for representing relevant expansion/compression process

lift referring to pressure lift

map referring to proposed ejector efficiency

mn motive nozzle

ms mixing section

r referring to recovered power

rm referring to maximum recovered power potential

SH referring to level of superheat

sn referring to suction nozzle

th motive nozzle throat

x motive stream compression process

y suction stream compression process

4
1. Introduction

A two-phase ejector, acting as a work recovery device in a vapor compression system,

improves the system efficiency by reducing the exergy losses in a throttling valve. The

experimental studies of the two-phase ejector systems show that the system performance can be

improved for a variety of refrigerants using different cycle architectures involving heating,

refrigeration, and air conditioning applications (Zhang et al., 2020).

These recent studies show that the ejector can be employed either in pressure recovery or

liquid recirculation mechanism in improving system performance, thereby expanding the

operational envelop of a two-phase ejector. It has been found that the single-phase vapor inlet

conditions, typically encountered in a standard ejector cycle (Elbel and Hrnjak, 2008) maximizes

pressure recovery mechanism. Similarly, the two-phase suction inlet conditions in flooded

evaporator operation (Pottker and Hrnjak, 2015) optimizes both pressure recovery and liquid

recirculation mechanisms. However, the near saturated liquid flow at the ejector suction port in

the liquid recirculation cycle (Lawrence and Elbel, 2016) maximizes the liquid recirculation

mechanism of the ejector.

The expanded operational envelop necessitates to predict ejector performance for wide

range of operating conditions. This, however, is not possible without having consistent ejector

performance representation for a fixed-geometry ejector. For example, systematic representation

of compressor experimental performance data using multiple compressor speed (motive variable)

curves, where each speed curve relates relevant efficiency (performance variable) with pressure

ratio (characterization variable), is important in predicting compressor performance. Thus, a

consistent ejector performance representation which can be called ejector performance map, may

be developed if appropriate motive, characterization, and performance variables can be

5
identified.

However, the choice of these variables is not straightforward. It is difficult to determine

what could be an appropriate motive variable for an ejector. Also, the recent research points that

the ejector efficiency, which is typically used as performance variable may not be

appropriate for representing ejector performance for different operating conditions. The widely

accepted (Elbel and Hrnjak, 2008) is reported to have very low magnitude of less than 1%

in R410A liquid recirculation cycle (Lawrence and Elbel, 2016). Similarly, decreases as the

flow conditions changes from single-phase to two-phase at the suction port of an ejector, even

though the system performance improves. These observations suggest that the study of consistent

representation of ejector performance may have to involve revisiting the ejector efficiency

definition so that its magnitude should be a good reflection of ejector usefulness for its newly

explored operational envelop of two-phase suction flow conditions.

The proposed ejector performance map is different from an evolving research area of

physics-based ejector modeling. Aidoun et al., (2019) have summarized the latest developments

in ejector physics-based models. The ejector design models may be advantageous in designing

efficient ejectors, but they are computationally expensive. Also, detailed modeling is not a

guarantee for accuracy in prediction for wide range of operating condition. These design models

are not practical for the ejector system analysis that require computationally efficient yet

accurate models. The proposed ejector performance map can find its utility in analyzing different

ejector cycle architectures while predicting ejector performance from experimental data. This can

greatly improve the reliability of such system models that typically assume efficiencies for

different ejector parts. Also, the performance map can help in making ejector system selection

decision after analyzing several component combinations (Haider and Elbel, 2020).

6
 Several studies have developed correlations for predicting ejector performance. Lucas et

al. (2013) developed the ejector performance map using empirical correlation involving the

ejector efficiency, the geometry features, and the operating conditions. Likewise, Liu and Groll

(2013) developed correlations for predicting efficiencies for different parts of the ejector using

geometrical properties and flow conditions. Similarly, Banasiak et al. (2015) developed ejector

performance map for engaging appropriate ejector as the system load is varied in a multi-ejector

rack system using the flow conditions and the ejector geometry parameters. Even though these

correlations are computationally efficient, they are limited to single-phase suction inlet

conditions. Moreover, ejector geometry parameters are used in these correlations for predicting

ejector performance which makes them suitable for ejector design application (Barta et al.,

2021). For predicting performance of a fixed-geometry ejector, the geometry parameters become

redundant. Thus, the proposed ejector performance map is also different from these existing

correlations in literature.

In this study, the main objective is to find a consistent ejector performance representation

involving both single and two-phase suction inlet conditions for a fixed-geometry ejector for

predicting ejector performance accurately. First, an ejector power analysis is presented for

identifying two possible motive variables, along with defining a new ejector efficiency. Then,

controlled motive variable experiments are conducted for observing the trends in ejector

performance, while changing suction inlet conditions from single-phase to two-phase for a

standard ejector cycle. A consistent ejector performance representation methodology, also the

ejector performance map has been developed after carefully observing the performance data.

Later, the prediction accuracy of proposed ejector performance map methodology is evaluated

for large dataset. The applicability of the proposed methodology is tested by utilizing the data

7
from different studies involving various refrigerants and their relevant cycle architectures.

2. Experimental facility and uncertainty analysis

2.1 Test facility

The experiments for measuring ejector performance are conducted using an air conditioning

system working on a transcritical CO2 standard ejector cycle with internal heat exchanger (IHX).

The schematic of the lab facility is shown in Figure 1. The detailed system layout and description

can be found in Lawrence and Elbel (2019). The microchannel gas cooler and the evaporator are

housed inside two separate closed loop wind tunnels. In the evaporator side wind tunnel, an

electric heater is installed for controlling inlet air temperature, and thus, the cooling capacity of

the evaporator. Similarly, the gas cooler wind tunnel is equipped with a chilled water heat

exchanger which helps in controlling inlet air temperature and heat transfer rate. The

compressor used in the study is a variable speed radial piston compressor. The IHX is a

microchannel heat exchanger with a high and low-pressure line. A bypass valve is installed on

the suction line of the IHX which allows changing the IHX effectiveness in the range of 0.4-

0.97. The dimensions of the ejector used in the study are given in Lawrence and Elbel (2019),

with the only exception of motive nozzle divergent section length of 10mm instead of 5mm. For

optimizing system COP, an additional control of the high side pressure is often implemented in a

transcritical system. In this study, the needle control or the expansion valve is not used.

8
 Figure 1 Schematic of lab facility working on a standard ejector transcritical CO2 cycle
with IHX

The air conditioning system can operate the ejector in wide range of conditions. The

motive inlet conditions are in the supercritical range, making both the temperature and the

pressure independent of each other. The motive inlet pressure and temperature can be controlled

by either varying the compressor speed, or changing the gas cooler air inlet conditions, or

changing the IHX effectiveness. The ejector suction port inlet conditions are the same as the exit

conditions of the evaporator, which can be either single-phase or two-phase. The single-phase

suction inlet conditions are achieved by operating system with a superheated evaporator exit

conditions. For single-phase suction flow both the temperature and the pressure are independent

of each other. The pressure, however, is dictated by the evaporator coil temperature, which can

be controlled by changing the cooling load of the system. The two-phase suction inlet conditions

are achieved by feeding high mass flow rate to the evaporator, resulting in two-phase conditions

at the evaporator exit. The expansion valve is used to achieve the desired flow conditions at the

evaporator exit.

2.2 Uncertainty analysis

9
 The detailed uncertainties in temperature, pressure and mass flow rate sensors installed in

the experimental facility can be found in Zhu et al. (2018). The uncertainty in important

variables, calculated using uncertainty propagation (Moffat, 1998) is summarized in Table 1. The

suction inlet temperature can only be useful for the superheated single-phase suction inlet

conditions. The uncertainty is higher for two-phase flow condition as the suction quality is

estimated by an energy balance from the air side heat transfer measurements. Table 2 reports the

difference in uncertainty in the maximum work recovery potential, ̇ work recovered, ̇ and

the ejector efficiencies ( , ) for both single and two-phase suction inlet conditions.

Table 1 Uncertainty in the key ejector performance variables

Variable 1ph 2ph 1ph/2ph

̇ (W) 5.5% 9% -

̇ (W) - - 3%

(-) 0.01 0.013 -

(-) 0.004 0.008 -

(kPa) - - ±17

(-) 2% 5% -

3. Ejector Performance Representation

The problem of consistent ejector performance representation has been linked with the

ejector efficiency definition. It is assumed that the maximum recovered power potential ̇ is a

possible motive variable. In this section, a power analysis is introduced for analyzing ejector

performance which leads to identifying two possible definitions for motive variable, ̇ , and a

10
new ejector efficiency definition . Then, the experimental data from the controlled ̇

experiments are presented in finding the ̇ definition that works better as a motive variable.

The suitable characterization and performance variables are chosen while observing the

performance data.

3.1 Ejector power analysis

The power analysis introduced here is based on the analysis framework provided by Elbel

and Hrnjak (2008) in which they divided the ejector flow into four isentropic expansions and

four isentropic compressions (pressure recovery) processes. The analysis had been limited to the

pressure differences only. In this study, the power is considered using relevant entropy through

which an isentropic process occurs.

Figure 2 shows that the expansion process within the motive nozzle can be subdivided

into three power terms as flow is expanded from to , from to , and ultimately from

to through isentropic processes along . Similarly, the suction flow is expanded from

to through an isentropic process along . Each expansion power term is labelled as

̇ . The mixing process occurs at the and both the motive and suction streams are

compressed from to , and ultimately from to through pressure recovery processes,

termed as ̇ . It is assumed that motive and suction stream follow isentropic processes and

, respectively. Table 2 gives all the expansion and compression power terms in the form of

products of the mass flow rates and the specific enthalpy differences.

11
 Figure 2 Power analysis schematics showing ideal expansion and compression processes
between different pressures within an ejector

Table 2 Power terms as product of mass flow rate and specific enthalpy differences

Ideal Expansion Ideal Compression

̇ ̇

MN ̇ ̇ ̇ ̇

̇ ̇ ̇ ̇

̇ ̇ ̇ ̇ ( ( ) ( ))
SN
̇ ̇ ( ( ) ( ))

The ejector performance can be given by the ratio of power recovered, ̇ to the

maximum recoverable power potential, ̇ . The processes involving can be ignored while

characterizing ejector performance at the system level. Thus, only the processes between ,

, and are considered in writing power ratio as given in Equation (1).

12
 ̇ ̇ ̇
̇ ̇ ̇
(1)

The efficiency is the ratio of the useful output work to the input work. In an ejector, the

useful work is the power imparted to the suction flow ̇ , whereas the total input work can be

referred to the maximum power recovery potential ̇ . Elbel and Hrnjak (2008) defined the

ratio between ̇ and ̇ as ejector efficiency, by canceling ̇ and ̇ on the

assumption that the power expanded by the motive stream in the motive nozzle, ̇ between

to is equivalent to the power recovered, ̇ by the motive stream from to . Thus,

the ejector efficiency, (Elbel and Hrnjak, 2008) can be written as in Equation 2.

̇ ̇ ̇ ̇
̇ ̇ ̇ ̇ ( )
(2)

However, ̇ and ̇ can only be equal if it is assumed that the pressure recovery of

motive stream occurs along . In Equation (2), the isentropic work recovered ̇

̇ is calculated assuming . Thus, does not account for any mixing process

inside ejector and assumes that both the motive and suction streams are compressed through

independent isentropic processes. However, realistically both the motive and the suction streams

are compressed such that and . This leads that ̇ and ̇ are not equal.

Furthermore, mathematically ̇ and ̇ cannot be canceled in Equation (1) even if they are

equal, as it will affect the magnitude of the ratio. A new ejector efficiency labelled as

can be defined which retains ̇ in ̇ while only considering ̇ as recovered work. It is

given in Equation (3)

̇ ̇ ̇ ̇ ( )
̇ ̇ ̇ ( )
(3)

In Equation (3), it is assumed that the suction stream is compressed through an isentropic

13
work recovery, ̇ ̇ such that . The ejector exit entropy, can be found using

and the measured . The ejector exit enthalpy, can be found by applying energy balance

across the ejector using Equation (4).

̇ ̇
̇ ̇
(4)

The different processes involved in defining and are shown on a Ph

diagram in Figure 3 for a two-phase suction inlet condition. The Equation (2) and Equation (3)

shows that ̇ can be defined in two possible ways, i.e., is the ̇ for , whereas

̇ is ̇ for . In the next subsection, controlled experiments for and

will be conducted to evaluate which one of these definitions for ̇ is a better motive

variable.

Figure 3 Ph diagram of a transcritical standard ejector CO2 cycle with IHX showing processes
involved in the definition of ejector efficiencies and

3.2 Ejector performance map

The motive variables and ̇ are dependent on the motive inlet temperature

, pressure , mass flow rate ̇ , and either or . Thus, and ̇ can have

same magnitude from different operating conditions. A suitable ejector motive variable though

14
obtained by different operating conditions must relate the characterization and the performance

variable using a single curve for wide range of operating conditions. For example, in compressor

performance map, a single curve of a particular compressor speed (motive variable) defines

relationship between the characterization and the performance variable.

For finding suitable motive variable, two series of experiments are conducted - ̇ is

controlled in the first, while ̇ in the second. In these experiments, the ejector suction flow

is changed from single-phase to two-phase for comparing trends of the ejector efficiency ,

with the newly proposed and finding which of the two motive variables can give a

consistent ejector performance representation. The needle control, which is usually employed for

maximizing system COP of a transcritical CO2 system for increasing the high side pressure, is

not used in these experiments.

3.2.1. Controlled motive variable experiments

From experimental standpoint, the difference between the two ̇ definitions is choice

of considering either of the two low-side measurable ejector pressures as the reference pressure,

i.e., for ̇ and for ̇ . Figure 4 shows the measured ejector efficiencies against

for the controlled ̇ and ̇ experiments. The data points with higher than 0.7

are for two-phase suction flow. For each data point, the flow conditions ( ̇ , , , either

or ) are adjusted as is changed such that the magnitude of ̇ and ̇ remains

same. Figure 4(a) compares trends of and for ̇ W experiments.

has lower magnitude than and variation in both the efficiencies is not significant.

However, increases as suction flow changes from single-phase vapor to two-phase,

whereas decreases slightly. The variation in is obvious in ̇ W

15
experiments as shown in Figure 4(b). The decreases considerably when the flow changes

from single-phase vapor to two-phase suction inlet conditions for all operating conditions. On the

contrary, the either increases or remains almost same as the suction flow conditions are

changed.

(a) (b)

Figure 4 Ejector performance in terms of efficiency and mass entrainment ratio for
controlled (a) ̇ (b) ̇

Figure 5 shows the experimental data of Figure 4 in terms of the pressure lift, and

the mass entrainment ratio, . A real-time system operation is closer to the controlled ̇

experiment since when is increased decreases as shown in Figure 5. This decrease is

largely reflected by decrease in , while does not change significantly as it is related with

the evaporator coil temperature. The drop in in Figure 4(b) for controlled ̇

experiment shows that has a bias towards pressure recovery mechanism as it reflects drop

in the ejector usefulness when liquid recirculation mechanism is preferred over pressure recovery

mechanism. On the other hand, tends to adjust the drop in pressure recovery

16
mechanism with the rise in liquid recirculation mechanism much better as it either almost

remains constant or increases slightly.

For ̇ W experiments as shown in Figure 5(a), for different operating

conditions are almost identical for corresponding . The trend in Figure 5(b) however, is

somewhat different for ̇ W experiments. The tends to converge for single-

phase suction flow, whereas for two-phase flow it tends to diverge, suggesting that is not

accounting for suction inlet quality.

(a) (b)

Figure 5 Ejector performance in terms of pressure lift and mass entrainment ratio for
controlled (a) ̇ (b) ̇

3.2.2. Characterization using volumetric entrainment ratio

For accounting effects of suction inlet quality, the volumetric entrainment ratio,

instead of given by Equation (5) is used.

̇ ̇
̇ ̇
(5)

Figure 6 shows the experimental data of Figure 4 with new choice of variable for both

̇ and ̇ experiments. data points appear to be scattered in both experiments.

17
However, drop in becomes visible for small change in in ̇ controlled

experiments. The data points for tend to have less scatter for both the experiments,

ideal for representing ejector performance in consistent form using a correlation based on .

(a) (b)

Figure 6 Ejector performance in terms of efficiency and volumetric entrainment ratio for
controlled (a) ̇ (b) ̇

Further experiments for different values of ̇ are conducted while changing suction

inlet conditions. Figure 7(a) shows the recovered power, ̇ for different ̇ as function of

. The data shows that for each magnitude of ̇ , an independent curve can relate work

recovered, ̇ with , suggesting that ̇ is a suitable motive variable. However, using

as in Figure 7(b) allows curve-fitting different operating conditions into a second order

polynomial curve ̂ with an APD of 5.5% as a function of . This allows representing

ejector performance using only two variables, i.e., the ejector efficiency, as the

performance variable and the volumetric entrainment ratio, as the characterization parameter.

There is no motive variable meaning the data for developing ejector performance map can be

collected with much ease. The curve ̂ is the proposed ejector performance map.

18
 (a) (b)

Figure 7 Different ̇ experiment data against (a) ̇ (b) ejector performance

map, ̂ and the measured

4. Analysis and Discussion

4.1 Prediction accuracy of and

The performance of an ejector needs to be predicted accurately in terms of for both

pressure recovery and liquid recirculation mechanisms. Prediction accuracy of the methodology

is evaluated while considering two performance curves each based on and as

shown in the Figure 8 using single-phase suction flow only. There are multiple reasons for

choosing single-phase suction flow for developing performance map. Firstly, can represent

single-phase suction flow consistently. Secondly, single-phase data points have less uncertainty

than two-phase suction flow. Thirdly, most of the research reports the ejector performance for

the single-phase suction flow. If the proposed methodology can predict the ejector performance

for the two-phase suction flow while utilizing single-phase flow conditions, then it can be helpful

in analyzing the ejector system performance for extended operating conditions using the existing

experimental data.

19
 A total of 149 data points for the steady state ejector operation are collected for the

ejector used in this study. Among these 149 data points, 62 of them are for the single-phase

vapor suction inlet data. The ejector performance data is collected which accounts for the

variation of corresponding to the expected coil temperature (1-15oC). The change in the

motive inlet temperature is in the range of 20-44oC, whereas the variation in the motive inlet

pressure is in the range of 7800-10500kPa. The dataset includes high ambient temperature (up to

50oC), data with modified performance of other system components, such as changing IHX

effectiveness from 0.6 to 0.97, and altering compressor performance appreciably by changing

cooling mechanism from natural cooling to forced cooling by fan (Haider and Elbel, 2020). A

formal procedure for developing the performance maps is listed in the Appendix A.

Figure 8 Two ejector performance maps/curves developed using and

can be predicted using the ejector performance maps by following the procedure

mentioned in the Appendix A. Figure 9 shows prediction from both the ̂ and the

̂ prediction curves. The ̂ prediction curve overpredicts as the suction flow

changes from the single to two-phase inlet conditions, whereas the prediction accuracy of

20
 ̂ is not affected by changing the suction flow conditions. Table 3 gives the statistical

comparison between the prediction accuracy of the two performance maps. The ̂ curve

can predict 89.3% of the data within 20% accuracy, whereas ̂ curve can only predict 57.7%

of the data within 20% accuracy. Thus, the proposed ejector performance map ̂ using

as performance variable and volumetric entrainment ratio, can accurately predict

ejector performance for wide range of operating conditions. The coefficients of ̂ is

given in Appendix B.

Figure 9 Pressure lift prediction using ̂ and ̂

Table 3 Data (%) predicted within specified accuracy using ̂ and ̂

Accuracy

30% 20% 10%

̂ 96.6 89.3 61.7

̂ 69.8 57.7 36.2

21
4.2 Applicability to other datasets

The generality of the proposed ejector performance map is evaluated by representing

ejector performance data from various sources using the proposed ejector performance maps. A

total of six ejector datasets are presented here – three are R134a ejectors, one is an R410A

ejector and the remaining two are CO2 ejectors.

Lawrence and Elbel (2014) studied effects of mixing-section dimensions on ejector

performance for low-pressure refrigerants (R134a and R1234yf) in a Condenser Outlet Split

(COS) ejector system. Figure 10 represents three different R134a ejector geometries from the

study – all having the same motive nozzle throat diameter of 1mm, but different mixing section.

The scatter plot shows for each ejector dataset, whereas the respective performance

maps are shown as the curves. The strength of proposed choice of variables becomes obvious

when comparing a large set of ejector performance data visually and finding the ejector geometry

that works best. In this case, the ejector with a mixing section of 5mm diameter and 60mm

length, clearly outperforms other geometries. The range of is large for an R134a ejector,

because the difference in the density of the vapor and the liquid is significant. Interestingly, the

ejector performance map has a peak after which the ejector efficiency drops. This finding can be

utilized in system optimization studies for maximizing the system performance.

22
 Figure 10 Ejector performance map for three different R134a ejector geometries
(Lawrence and Elbel, 2014)

Figure 11 shows the data of the other three ejectors – the first is of the R410A ejector system and

the other two are of the CO2 systems. Lawrence and Elbel (2016) worked on an R410A liquid

recirculation cycle, in which the density of both the motive and the suction stream is similar,

resulting in relatively low . The study by Zhu et al. (2017) is about the performance of the

transcritical CO2 ejector system for the heat pump application. All data points have superheated

suction inlet conditions. The third study by Giacomelli et al. (2019) contains both the

transcritical and the subcritical data points for the CO2 ejector refrigeration cycle. of

the ejectors can be represented using their respective ejector performance map. The APD for

Giacomelli et al. (2019) is 26%, which is relatively higher than other datasets (Table B. 1) This

could be because of needle control while collecting data.

23
 Figure 11 Ejector performance map for R410A and CO2 ejectors

The proposed methodology is also applicable to other refrigerants and systems working

on different cycle layouts and applications. The data of all seven ejectors, i.e., the coefficients of

̂ performance map, APD, and the applicable range of is presented in Appendix B.

4.3 Difference between and

The difference in the ejector efficiency comes from how the ̇ and ̇ are defined in

Equation 1 and Equation 2. The influence of ̇ is significant in determining the magnitude of

the ejector efficiency, but its effect on the trend is relatively less as compared to ̇ as suction

flow conditions are changed from single-phase to two-phase. In controlled ̇ experiments,

is raised by increasing the coil temperature for compensating the drop in as is

increased, suggesting that ̇ decreases slightly in these experiments which can contribute to

the increasing shown in Figure 4(a). On the contrary, in the controlled ̇

experiments, is dropped as is increased, suggesting that ̇ should increase. This

increase in ̇ can contribute to the decrease in in Figure 4(b).

However, the trend in both the efficiencies are primarily dependent on the definition of

24
 ̇ . Figure 12 breakdowns ̇ for both and into the constituent terms of

suction mass flow rate, ̇ and the change in enthalpy during the isentropic compression, .

The data shown here is from the controlled ̇ experiment, consisting of the three

sub-datasets, each for different . There are four data points in each sub-dataset, in which the

first data point is for a single-phase superheated vapor suction inlet condition (the amount of

superheat in oC is shown in Figure 12), whereas the other three data points are for the changing

suction inlet conditions from to the lowest possible value. The change in in is

greater than it is in as the suction inlet conditions change from the single-phase to the

two-phase flow, resulting in drop of . For , the decrease in and increase in ̇

are adjusted such that the either remains same or improves slightly. Thus,

accounts for the liquid recirculation mechanism better than .

Figure 12 Breakdown of ̇ into ̇ and from controlled ̇ W experiment

The different amount of change in comes from the thermodynamic property of the refrigerant

as the operating conditions are changed. The slope of isentropic lines on the right side of Ph

25
diagram is flatter, whereas it becomes steeper to the left side of the Ph diagram. In a standard

ejector cycle, in is calculated with reference to and which remains to the right-

side of on Ph diagram. The suction inlet enthalpy, moves significantly from right to left

side of the Ph diagram as suction flow conditions are changed, resulting in large changes in the

slope of isentropic lines. On the contrary, in is calculated with respect to and

which remains in the middle of the two-phase dome. The slope of isentropic lines changes

negligibly as the suction flow conditions are changed. This can also explain why on contrary to

the standard ejector cycle operation, has higher magnitude than (which remains

under 1%) for R410A liquid recirculation cycle in Figure 11. In liquid recirculation cycle, is

to the left side of in the Ph diagram, resulting in higher for than for .

5. Conclusions

In this study, an attempt is made to represent fixed-geometry ejector performance data for

an extended ejector operational envelope that involves both pressure recovery and liquid

recirculation mechanisms. The study relied upon how the compressor performance data is

represented. This helped in outlining the need to identify appropriate motive, performance, and

characterization variable for ejector operation.

The study found that the consistent ejector representation is linked with revisiting the

ejector efficiency definition. The ejector power analysis not only helped in identifying two

possible motive variables ̇ and ̇ , but also introduced a new ejector efficiency

definition . The trends of existing efficiency , and new efficiency have

been studied by conducting controlled motive variable experiments on a transcritical CO2 ejector

system. It has been found that favors pressure recovery mechanism over liquid

recirculation mechanism, whereas tends to balance the drop in pressure recovery with

26
an increase in liquid recirculation mechanism. This allowed in consistent ejector performance

representation using ejector efficiency, as the performance variable and the volumetric

entrainment ratio, as characterization variable for both single-phase and two-phase suction

inlet conditions. Significantly different operating conditions were curve-fitted using a second

order polynomial ̂ which has been termed as the ejector performance map.

The prediction accuracy of the ejector performance map methodology is evaluated by

comparing performance curves based on both and . Only single-phase suction

inlet data points were considered for developing these performance curves while predicting

performance for two-phase suction inlet conditions. It was found that the prediction curve based

on performs better than , by predicting 89.3% of the data points within 20%

accuracy.

The generality of the proposed methodology was tested on six different ejector datasets,

involving different refrigerants, systems, and applications. The methodology successfully

represented three R134a ejector datasets having different geometries working on a COS ejector

cycle, an R410A ejector dataset working on a liquid recirculation cycle, and two CO2 ejector

datasets: one working on the heat pump application and the other on an air conditioning

application with both transcritical and subcritical data points.

The proposed ejector performance map can be utilized in system design studies to find

new ejector cycle architectures. The numerical results will be more reliable as the ejector

performance is predicted from experimental data, and not by assuming efficiencies of different

ejector part. The methodology can also aid in the ejector system selection process.

Acknowledgements

The authors would like to thank the member companies of the Air Conditioning and

27
Refrigeration Center at the University of Illinois at Urbana-Champaign for their financial and

technical support, grant number was ACRC #373 and Creative Thermal Solutions, Inc. (CTS) for

their technical support.

References

Aidoun, Z., Ameur, K., Falsafioon, M., Badache, M., 2019. Current advances in ejector

modeling, experimentation and applications for refrigeration and heat pumps. Part 2: Two-phase

ejectors. Inventions, 4(1), 16.

Banasiak, K., Hafner, A., Kriezi, E. E., Madsen, K. B., Birkelund, M., Fredslund, K., Olsson, R.,

2015. Development and performance mapping of a multiejector expansion work recovery pack

for R744 vapour compression units. International Journal of Refrigeration, pp. 265-276.

Barta, R.B., Dhillon, P., Braun, J.E., Ziviani, D., Groll, E.A., 2021. Design and optimization

strategy for ejectors applied in refrigeration cycles, Applied Thermal Engineering.

Elbel, S., Hrnjak, P., 2008. Experimental validation of a prototype ejector designed to reduce

throttling losses encountered in transcritical R744 system operation. International Journal of

Refrigeration 31(3), 411-422.

Giacomelli, F., Mazzelli, F., Banasiak, K., Hafner, A., Milazzo, A., 2019. Experimental and

computational analysis of a R744 flashing ejector. International Journal of Refrigeration. 107,

326-343.

Haider, M., Elbel, S., 2020. Development of a new ejector performance map for design of an

automotive air conditioning system. SAE World Congress, Detroit, MI, USA, April 21-23, Paper

2020-01-1244.

Lawrence, N., Elbel, S., 2014. Experimental investigation of a two-phase ejector cycle suitable

for use with low-pressure refrigerants R134a and R1234yf. International Journal of

28
Refrigeration, 38, 310-322.

Lawrence, N., Elbel, S., 2016. Experimental investigation on the effect of evaporator design and

application of work recovery on the performance of two-phase ejector liquid recirculation cycles

with R410A. Applied Thermal Engineering 100, 398–411.

Lawrence, N., Elbel, S., 2019. Experimental investigation on control methods and strategies for

off-design operation of the transcritical R744 two-phase ejector cycle, International Journal of

Refrigeration. 106, 570-582.

Liu, F., Groll, E., A., 2013. Study of ejector efficiencies in refrigeration cycles, Applied Thermal

Engineering 52, 360-370.

Lucas, C., Koehler, J., Schroeder, A., Tischendorf, C., 2013. Experimentally validated CO2

ejector operation characteristic used in a numerical investigation of ejector cycle. International

Journal of Refrigeration 36, 881-891.

Moffat, R., J., 1998. Describing the Uncertainties in Experimental Results. Experimental

Thermal and Fluid Science. 1: 3-17.

Pottker, G., Hrnjak, P., 2015. Ejector in R410A vapor compression systems with experimental

quantification of two major mechanisms of performance improvement: Work recovery and liquid

feeding. International Journal of Refrigeration, 50, 184-192.

Zhang, Z., Feng, X., Tian, D., Yang, J., Chang, Li., 2020. Progress in ejector-expansion vapor

compression refrigeration and heat pump systems. Energy Conversion and Management, 207,

112529.

Zhu, J., Botticella, F., Elbel, S., 2018. Experimental investigation and theoretical analysis of oil

circulation rates in ejector cooling cycles. Energy 157, 718-733.

Zhu, Y., Li, C., Zhang, F., Jiang, P.-X., 2017. Comprehensive experimental study on a

29
transcritical CO2 ejector-expansion refrigeration system. Energy Conversion and Management.

151, 98-106.

Appendix A

Developing ejector performance map

The steps for developing an ejector performance map are as follows.

1: Collect n data points for single-phase suction inlet conditions for wide range of operation

2: Calculate using Equation (3), and using Equation (5)

3: Use curve-fitting tool to estimate the coefficients of the polynomial given in Equation (A.1)

̂ such that (∑ ( ̂ ) ) (A.1)

The same procedure can be used for developing performance map using

Predicting pressure lift

For a given operating condition ( , , ̇ , , ̇ , or ), can be predicted by

following these steps.

1: Calculate using Equation (5) for the ejector inlet conditions

2: Calculate ̂ from Equation (A.1)

3: Calculate ̂̇ using, ̂̇ ̇ ( )

4: Calculate ̂̇ using, ̂̇ ̂ ̇

5: Calculate from Equation (4).

6: Solve for such that, ̂̇ ̇ ( )

7: Calculate, ̂

prediction using ̂ will involve an iterative loop between step 3 and step 6 such that the

change in is reflected in all calculations.

30
Appendix B

Table B. 1 Coefficients of ̂ , APD and range of for the ejectors

Ejector Refrigerant Study/Description APD Coefficient of range

1 CO2 This study 8.9% -0.021

0.052

-0.004

2 R134a Lawrence and Elbel 4.7% -0.0021

(2014), D5xL40 0.0023

-3e-05

3 R134a Lawrence and Elbel 5.8% -0.0032

(2014), D5xL60 0.0041

-5.59e-05

4 R134a Lawrence and Elbel 7.5% -0.0055

(2014), D5xL70 0.0033

-6.65e-05

5 R410A Lawrence and Elbel 7.2% -3.3e-06

(2016) 0.0767

-0.0321

6 CO2 Zhu et al. (2017) 10.7% -0.0313

0.0747

-0.0068

31
7 CO2 Giacomelli et al., 26% -0.0084

(2019) 0.0464

-0.0056

32