Applied Thermal Engineering 229 (2023) 120576

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Research Paper

Vapor-liquid equilibrium measurement and heating performance modeling

on eco-friendly zeotropic blends of CO2/R1234ze(Z) and CO2/R1336mzz(E)
Jielin Luo a, b, Zhiming Ye a, Zhen Zhao a, Kaiyin Yang a, Shaozhi Zhang a, *, Qin Wang a
a
Key Laboratory of Refrigeration and Cryogenic Technology of Zhejiang Province, Institute of Refrigeration and Cryogenics, Zhejiang University, Hangzhou, China
b
Renewable Energy Research Group (RERG), Department of Building Environment and Energy Engineering, The Hong Kong Polytechnic University, Hong Kong, China

A R T I C L E I N F O A B S T R A C T

Keywords: In the context of sustainable development, CO2 and hydrofluoroolefins (HFOs) are regarded as promising
Vapor liquid equilibrium refrigerant candidates. Despite recognition of application potential, the knowledge of their blends is insufficient.
CO2 In this paper, vapor–liquid equilibrium (VLE) properties of two eco-friendly zeotropic CO2/HFO blends, namely
HFO
CO2/R1234ze(Z) and CO2/R1336mzz(E), were measured at temperature from 253.15 K to 333.15 K at CO2
Mixture
concentration from 0 to 1. Peng-Robinson (PR) model and Soave-Redlich-Kwong (SRK) model are identified as
Heat pump
Model suitable models for describing VLE of CO2/HFO with deviation less than 4.5%, whereas Perturbed-Chain Sta­
tistical Associating Fluid Theory (PC-SAFT) model is shown to have larger deviation. Furthermore, a heat pump
water heater is modelled to demonstrate the working potential of CO2/HFO in terms of thermodynamics and
safety, while the newly fitted binary interaction parameter kij being used. Both CO2/R1234ze(Z) and CO2/
R1336mzz(E) show more than 150% advantage in volumetric heating capacity and comparable heating effi­
ciency compared to R134a. The optimal concentration of CO2 is higher than 0.6, eliminating the flammability of
mixture. Additionally, the operating pressures of CO2/HFO are reduced to lower than 3.2 MPa at typical working
condition due to HFO addition. The results contribute to the accurate VLE description of CO2/HFO and further
investigation on the applications of eco-friendly refrigerants.

extensively studied in recent years. Attributed to their extremely high

1. Introduction flammability (Class A3 in ASHRAE standard), HCs are used with limited
charging amount (e.g., 100 g), which is unfavorable for large-scale
To date, rapid economic development spurs the surging requirement application [6]. As for CO2, its performance has been widely studied
of energy, whereas the increasing consumption of energy brings up the in fields of refrigeration, air-conditioning and heat pump [7–11].
concern of environmental expense. In aspect of thermodynamic systems, Although CO2 is a primary candidate for instantaneous heat pump water
a consensus has been achieved by scientists worldwide that, conven­ heater (i.e., heating water from ambient temperature to temperature
tional working substances are responsible for ozone depletion and the higher than 323.15 K), the applications of pure CO2 are limited by two
greenhouse effect [1–3]. aspects. One is its high operating pressure (above 6 MPa), which results
For either refrigeration, heat pump or organic Rankine cycle, re­ in extra device cost and safety risks in long-term use [7–9]. The other
frigerants are essential working substances. Most of existing refrigerants one is its low critical temperature of 304.13 K, which makes it unsuitable
on the market are hydrofluorocarbons (HFC) with global warming po­ for the case of heat sink with high temperature (e.g., storage-type heat
tential (GWP) higher than 1000 (e.g., R134a, R410A) [3]. With the pump water heater with return water higher than 313.15 K) [8,11].
implementation of the Kigali amendment to the Montreal Protocol, the Synthesized refrigerants, such as HFOs (hydrofluoroolefins), have
phasing out of HFC is put on the agenda in a stepwise manner (e.g., the been gaining attention as a new generation of refrigerants for various
consumption of HFC should be cut down to 20% at the year of 2045 for devices [3,4,12–15]. For instance, R1234yf and R1234ze(E) are pro­
China) [1]. Therefore, the applications of HFC in the future will be posed as alternatives to R134a, and their performances in refrigeration
limited, and its eco-friendly substitutes are being studied by researchers, or air-conditioning have been studied through theoretical demonstra­
which are divided into natural ones and synthesized ones [2–5]. tions, numerical simulations, and experimental evaluations [3]. How­
Natural refrigerants such as CO2 and hydrocarbons (HCs) are ever, it is worth noting that HFOs also have some drawbacks as a

* Corresponding author.
E-mail addresses: jielin.luo@polyu.edu.hk (J. Luo), enezsz@zju.edu.cn (S. Zhang).

https://doi.org/10.1016/j.applthermaleng.2023.120576
Received 1 December 2022; Received in revised form 7 April 2023; Accepted 10 April 2023
Available online 17 April 2023
1359-4311/© 2023 Elsevier Ltd. All rights reserved.
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

Nomenclature VLE Vapor-liquid equilibrium

Abbreviations English letters

AAD Average absolute deviation kij Interaction parameter
COP Coefficient of performance P Vapor pressure, MPa
GWP Global warming potential T Temperature, K
HC Hydrocarbon x Liquid mole fraction
HFC Hydrofluorocarbon y Vapor mole fraction
HFO Hydrofluoroolefin Subscripts
LFL Lower flammable limit a Ambient
MAD Maximum absolute deviation cal Calculated
ODP Ozone depletion potential exp Experiment
PC-SAFT Perturbed-Chain Statistical Associating Fluid Theory ref Reference
PR Peng-Robinson tot Total
SRK Soave-Redlich-Kwong wi Water inlet
UFL Upper flammable limit wo Water outlet
vdW Van del Waals
VHC Volumetric heating capacity, kJ/m3

Table 1
Existing studies and this work on VLE of CO2/HFO (*: not experimental data but outcome of molecular simulation).
Type Temperature range (K) Pressure range (MPa) Datapoints Reference Year

CO2/R1234yf 223.00–373.00 0.062–2.292 96 [21] 2012

273.15–310.92 0.325–4.5 * [22] 2013
283.15–353.15 0.815–7.210 65 [23] 2014
293.13 0.82–3.48 5 [24] 2020
283.15–313.15 0.437–6.938 32 [25] 2022
CO2/R1234ze(E) 233.00–363.00 0.104–1.909 112 [26] 2013
273.15–293.15 0.212–5.684 * [22] 2013
283.32–353.02 0.310–7.595 83 [27] 2019
CO2/R1243zf 288.50–348.50 0.680–7.693 47 [28] 2021
273.15–313.15 0.269–7.391 69 [25] 2022
CO2/R1216 233.18–273.20 0.331–2.595 64 [29] 2016
CO2/R1234ze(Z) 252.51–333.27 0.029–7.171 74 This work
CO2/R1336mzz(E) 251.80–333.20 0.030–6.958 74 This work

substitute for R134a, including flammability and price concerns (vapor–liquid equilibrium) property is an essential property in regards
[13,14]. Additionally, researchers have raised concerns about the of operating pressure and phase-changing temperature for working fluid
breakdown of HFOs, which may generate flammable or acidic by- used in thermodynamic systems. However, the studies on VLE of CO2/
products under certain conditions (e.g., hydrosphere, accumulation), HFO are limited, as shown in Table 1.
leading to safety concerns [14]. Although the mechanisms behind the It can be found that for VLE of CO2/HFO, most of existing studies
breakdown of HFOs have not yet been fully understood, the risk of such were focused on CO2/R1234yf and CO2/R1234ze(E). However,
breakdown presents a potential drawback in the practical application of R1234yf, R1234ze(E) and R1243zf have normal boiling points below
HFOs [14]. 255 K, whereas R1234ze(Z) and R1336mzz(E) to be studied here have
As shown above, despite its applications in specific systems, pure boiling points above 280 K [4]. Therefore, their mixture with CO2 ex­
refrigerant suffers from its unique problem (e.g., operating pressure for hibits a larger temperature glide, which is suitable for the working
CO2, price for HFOs). Therefore, the idea of mixed refrigerant is pro­ condition with larger temperature difference between heat sink and heat
posed. Despite facing challenges such as concentration control and source. Previous studies have indicated that CO2/R1234ze(Z) and CO2/
phase accumulation, mixed refrigerants have been confirmed effective R1336mzz(E) delivered application prospect in storage-type heat pump
in alleviating the drawbacks of pure refrigerants [16]. For example, the water heater [18,19]. Therefore, the investigation on their thermo­
mixture of CO2/HC has been extensively studied and demonstrated to physical properties is of importance. In 2021, Bell et al. [30] conducted a
have satisfactory performance in heat pump applications [17–19]. thorough survey on the experimental data and models for mixtures
Compared with HCs, the flammability of HFOs can be better restricted containing HFOs. More than 100 datapoints could be found for VLE of
by flame retardant [13]. The mixture of CO2/HFO may be an even better CO2/R1234yf and CO2/R1234ze(E). However, VLE of other kinds of
candidate for application. Zhao et al. [20] theoretically evaluated the CO2/HFO mixture was lack of experimental data or even had not yet
performance of CO2/R1234ze(Z) in a refrigerator under a large tem­ been studied.
perature span (i.e., a large temperature difference between heat sink and The determination of the binary interaction parameter kij is crucial
heat source). However, accurate systematic investigation of CO2/HFO is for the VLE model of mixtures, as it reflects the degree of component
limited due to the lack of fundamental data on thermophysical combination between the mixture components. Various equations of
properties. state and mixing rules have been employed to determine kij for mixtures
The knowledge of thermophysical properties of a new refrigerant with HFOs, but a large amount of experimental data is required
will lay good foundation for further exploration on its systematic per­ [12,31–36]. Although molecular simulation (MS) can also be used in
formance by either theoretical or experimental means. VLE VLE modeling, its accuracy is dependent on temperature and molecular

2
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

Table 2 then used to regress the data. Finally, a heat pump water heater was
Information of refrigerants used in this work. modeled and analyzed using the obtained data, with a comparison made
Abbreviation CO2 R1234ze(Z) R1336mzz(E) between the new blends and existing refrigerants.
(R744) Accordingly, the significance of this work can be drawn. The results
CAS number 124–38-9 29118–25-0 66711–86–2 in this work: 1) supplement reliable VLE data for new kinds of CO2/HFO
Chemical name carbon cis-1,3,3,3- trans-1,1,1,4,4,4- mixture, 2) contribute to the establishment of accurate VLE calculation
dioxide tetrafluoro-1- hexafluoro-2-butene of CO2/HFO, and 3) validate the potential applications of eco-friendly
propene CO2/HFO. With the pressing challenge of achieving carbon neutrality,
Chemical formula CO2 C3H2F4 C4H2F6
Molar mass (g/ 44.01 114.04 164.05
this paper offers valuable insights for future research on new-generation
mol) refrigerants.
Normal boiling 194.69 282.88 280.58
point (K) 2. Experimental system for VLE measurement
Critical 304.13 423.27 403.37
temperature (K)
Critical pressure 7.377 3.531 2.766 2.1. Materials
(MPa)
Acentric factor 0.2239 0.3274 0.4053 In this work, CO2, R1234ze(Z) and R1336mzz(E) were adopted for
ODP 0 0 0 VLE measurement, and the detailed information of these pure compo­
GWP 1 ~1 7
Safety group A1 A2L A1
nents is shown in Table 2 [4,5,12,13,38,39]. The suppliers of CO2,
Mass purity >99.9% >99.9% >99.0% R1234ze(Z) and R1336mzz(E) were Hangzhou Minxing Chemical
Technology Co., Ltd., Zhejiang Research Institute of Chemical Industry
Co., Ltd. and Dupont China Holding Co., Ltd. respectively.
structure, limiting its universality [27,37]. Therefore, experimental
determination of kij for VLE modeling is currently preferred. 2.2. Experimental apparatus
From the above literature survey, the research gap of existing studies
can be indicated. Even if the recognition is reached that CO2/R1234ze An experimental apparatus for VLE measurement was customized
(Z) and CO2/R1336mzz(E) show promising application potential, the and built, whose schematic diagram is shown in Fig. 1. The experimental
investigation on their VLE is missed. The experiment for VLE measure­ apparatus included three subparts, i.e., charging, controlling and
ment is beneficial for obtaining kij for VLE models and thus promoting measuring.
accurate modeling on its performance in specific applications. The charging subpart was designed for the purpose of refrigerant
In order to evaluate the potential of CO2/R1234ze(Z) and CO2/ charging. Two gas cylinders were connected to the container, with
R1336mzz(E) in storage-type heat pump water heaters, VLE data needs several switch valves installed in the pipeline to control charge amount.
to be obtained up to a temperature limit of 333.15 K for the system’s hot Meanwhile, a vacuum pump was also connected to the container to
side, and down to a temperature limit of 253.15 K for the system’s cold extract air or gas before experiment. The container, made of 316
side when utilizing air source. Therefore, an experimental setup was stainless steel, had a volume of 180 ml, and its gas tightness was
constructed in this study, which allowed for VLE measurements of the examined at a pressure of 10 MPa. Five pipes were connected to the
two mixtures at concentrations ranging from 0 to 1 and temperatures container for charging, temperature measurement, pressure measure­
ranging from 253.15 K to 333.15 K. Three different VLE models were ment, gas sampling, and liquid sampling.

Fig. 1. Schematic diagram of experimental rig for VLE measurement (1. Vacuum pump; 2. PID controller; 3. Heater; 4. Temperature sensor; 5. Circulating pump; 6.
Refrigerator; 7. Platinum resistance thermometer (PRT); 8. Pressure transducer; 9. Temperature datalogger; 10. Pressure datalogger; 11. Magnetic stirrer; 12.
Container; 13. Thermostatic bath; 14. Gas sampling valve; 15. Gas buffer tube; 16. Pressure gauge; 17. Liquid sampling valve; 18. Liquid buffer tube; 19. Six-way
valve; 20. Chromatograph; 21/22/23. Thermostatic hood).

3
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

Table 3 be maintained at 0.1 kPa for 24 h. Then the container and the
Specifications of measuring instruments. pipes were vacuumed. The container was charged with refrig­
Parameter Equipment Model Range Accuracy erant and vacuumed for three times, to fully exclude impurity.
After the final vacuuming, the container can be considered as
Temperature Standard platinum F500 (WIKA 75–1000 ±0.005
precision Instruments K K clean and empty.
thermometer Ltd.) (2) Charging. HFO was charged firstly into the container which was
Pressure Pressure transducer UNIK 5000 0–1 MPa ±0.4 kPa cooled in the thermostatic bath at 253.15 K. Then CO2 was
(General Electric 0–7.5 ±3 kPa charged. The charge amounts were controlled with a balance
Company) MPa
Composition Chromatograph GC1690 0–1 –
with accuracy of 0.1 g.
(Hangzhou (3) First setpoint. The circulating pump and the magnetic stirrer were
Kexiao switched on, to achieve uniform temperature distribution in the
Company) bath and the container. Meanwhile, the thermostatic hoods were
operated. Thermo-equilibrium was achieved when the tempera­
ture fluctuation was less than 10 mK and the pressure fluctuation
Table 4 was less than 0.2% of the measuring value.
Results for validation experiment. (4) Measuring. Once thermo-equilibrium was achieved, the temper­
Vapor pressure of pure CO2 ature and pressure were recorded within the following 10 min. As
for composition measurement, the refrigerants of liquid or gas
T (K) Pexp (MPa) Pref (MPa) Relative deviation (%)
phase in the container were charged into the buffer tubes which
262.63 2.6805 2.6097 2.71% were vacuumed in advance, so that the uncertainty caused by the
273.51 3.5871 3.5185 1.95%
283.17 4.5934 4.5044 1.98%
residue in buffer tubes could be minimized. Then they were dis­
293.16 5.8162 5.7304 1.50% charged, and another piece of refrigerants of liquid or gas phase
Vapor pressure of pure R1234ze(Z) was charged and injected into chromatograph to determine the
T (K) Pexp (MPa) Pref (MPa) Relative deviation (%) composition. Since the mixture in container had reached vapor­
275.03 0.07371 0.07388 − 0.23%
–liquid equilibrium, the measured composition by chromato­
307.92 0.24615 0.2458 0.14%
312.9 0.28775 0.28766 0.03% graph denoted the equilibrium composition at vapor or liquid
317.87 0.33449 0.33471 − 0.07% phase at specific temperature and pressure. The composition
322.84 0.38396 0.38745 − 0.90% measurement was repeated three times for each measuring point.
332.82 0.50749 0.51234 − 0.95% (5) Successive setpoints and measuring. After the measurement of a
342.79 0.65676 0.66554 − 1.32%
Composition of CO2(1)/n-butane(2)
state point, the temperature setpoint of thermostatic bath was
T (K) y1,exp y1,ref Relative deviation (%) changed according to the experimental plan. Once a new equi­
332.97 0.8472 0.8364 1.29% librium state was reached, VLE data were recorded following the
312.28 0.8957 0.9013 − 0.62% same procedure as described in step 4). This process was repeated
294.38 0.9326 0.9371 − 0.48%
until the desired temperature range was covered and sufficient
275.36 0.9687 0.959 1.01%
255.63 0.9773 0.9742 0.32% datapoints were obtained for VLE analysis.
T (K) x1,exp x1,ref Relative deviation (%)
332.97 0.6482 0.651 − 0.43% 3. Experimental results and discussions
312.29 0.6674 0.66 1.12%
294.38 0.6821 0.688 − 0.86%
275.38 0.6867 0.693 − 0.91%
3.1. Validation experiment
255.64 0.7049 0.711 − 0.86%
The vapor pressures of pure CO2 and R1234ze(Z) were measured for
validation, and the obtained data were compared with those calculated
The controlling subpart aimed to create a stable experimental con­ by Refprop V10.0 [40]. The comparison results are shown in Table 4.
dition. A thermostatic bath was equipped with a refrigerator to achieve In addition, the composition measurement accuracy with chro­
test temperatures ranging from 253.15 K to 333.15 K. Temperature matograph was also examined with the mixture of CO2/n-butane at
sensors were used for detection, and a circulation pump was employed equilibrium state, while the data were compared with those given by
for liquid recycling to ensure the uniformity of liquid temperature in the Refprop V10.0 [40]. The comparison results are shown in Table 4 as
thermostatic bath. A magnetic stirrer was placed in the container to well.
ensure uniform temperature of the tested refrigerant. Additionally, The accuracy of pressure measurement was validated at pressure
thermostatic hoods were installed around sampling pipes or devices, to ranging from 0.07 MPa to 5.82 MPa. The experimental data exhibited an
prevent unexpected condensation of sampling gas at experimental average relative deviation of 0.44% against the reference data. For
temperatures higher than ambient. composition measurement, the average relative deviation was 0.04%
The measuring subpart was designed to measure and record the against the reference data. Consequently, the reliability of experiment
temperature, pressure, and composition of the refrigerant, while the was confirmed [27,41,42].
specifications of measuring instruments are shown in Table 3. The detail
of data uncertainty calculation can be found in Appendix A1, and the 3.2. VLE data for CO2/HFO
uncertainty of each datapoint is provided in the Supporting Information.
VLE data of CO2/R1234ze(Z) and CO2/R1336mzz(E) were collected
2.3. Experimental procedure over a temperature range of 253.15 K to 333.15 K, with varying liquid
concentrations of CO2 ranging from 0 to 1.
The experiment was carried out as follows, whose procedure has 74 datapoints were obtained for each mixture, and the results are
been well established in existing studies [12,23,28]: presented in Fig. 2 for CO2/R1234ze(Z) and Fig. 3 for CO2/R1336mzz
(E). The details of data and uncertainty are listed in Table S1 and
(1) Pretreatment. Before experiment, air tightness inspection was Table S2 in the Supporting Information. Overall, the obtained data
conducted, which was examined by a helium mass spectrometer demonstrate the VLE behavior of the two mixtures and serve as a reliable
leak detector. In this experiment, the pressure in container could source for future thermodynamic modeling and analysis.

4
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

Fig. 2. Experimental data and correlations for VLE of CO2/R1234ze(Z).

Fig. 3. Experimental data and correlations for VLE of CO2/R1336mzz(E).

successfully attempted for description on VLE of CO2/HFO [43–45]. In

Table 5
addition, Soave-Redlich-Kwong (SRK) model was proposed in describing
kij of different models for CO2/R1234ze(Z) and CO2/R1336mzz(E).
VLE of mixture composed by components with large difference in
PR SRK PC-SAFT normal boiling point [46,47]. Thus, with van der Waals (vdW) mixing
CO2/R1234ze(Z) 0.05566 0.05915 0.01849 rule, these three models (i.e., PR, SRK and PC-SAFT) are used in this
CO2/R1336mzz(E) − 0.02970 − 0.02789 − 0.09670 work, and further details on these models can be found in Appendix A2
or in relevant literature [43,45,47].
The best fitting results of the interaction parameter, kij, for these
3.3. Data correlations and model comparisons
three models are given in Table 5, while the deviations between fitting
results and experimental datapoints are shown in Table 6 and Table 7.
In this section, different VLE models are adopted to fit the experi­
To present the deviation more visually, the fitting deviations of
mental data in section 3.2. Peng-Robinson (PR) model and Perturbed-
composition for CO2/R1234ze(Z) and CO2/R1336mzz(E) are drawn in
Chain Statistical Associating Fluid Theory (PC-SAFT) model have been

5
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

Table 6
Deviation of different models at different temperatures.
T (K) CO2/R1234ze(Z) CO2/R1336mzz(E)

PR SRK PC-SAFT PR SRK PC-SAFT

AAD(x) AAD(y) AAD(x) AAD(y) AAD(x) AAD(y) AAD(x) AAD(y) AAD(x) AAD(y) AAD(x) AAD(y)

253.15 0.0205 0.0018 0.0189 0.0014 0.0269 0.0041 0.0160 0.0082 0.0159 0.0089 0.0221 0.0154
263.15 0.0207 0.0051 0.0196 0.0052 0.0237 0.0091 0.0157 0.0077 0.0120 0.0084 0.0176 0.0171
273.15 0.0172 0.0088 0.0133 0.0094 0.0204 0.0147 0.0129 0.0076 0.0112 0.0083 0.0227 0.0199
283.15 0.0148 0.0070 0.0122 0.0077 0.0107 0.0147 0.0157 0.0093 0.0139 0.0100 0.0293 0.0243
293.15 0.0142 0.0051 0.0128 0.0057 0.0120 0.0147 0.0136 0.0076 0.0126 0.0085 0.0373 0.0282
303.15 0.0130 0.0050 0.0127 0.0053 0.0068 0.0137 0.0113 0.0058 0.0111 0.0062 0.0408 0.0278
313.15 0.0123 0.0107 0.0151 0.0086 0.0142 0.0076 0.0104 0.0075 0.0109 0.0062 0.0561 0.0293
323.15 0.0093 0.0060 0.0113 0.0071 0.0296 0.0234 0.0106 0.0075 0.0116 0.0054 0.0694 0.0416
333.15 0.0114 0.0062 0.0144 0.0064 0.0338 0.0301 0.0084 0.0126 0.0094 0.0140 0.0634 0.0696

Fig. 4 and Fig. 5. Further information on the fitting results can be found
Table 7
in Table S1 and Table S2 in the Supporting Information.
Deviation of different models for all datapoints.
In Table 6 and Table 7, AAD represents average absolute deviation
CO2/R1234ze(Z) CO2/R1336mzz(E) and MAD represents maximum absolute deviation, as calculated in Eqs.
PR SRK PC-SAFT PR SRK PC-SAFT (1–4). N represents the number of datapoint.
AAD(x) 0.0148 0.0142 0.0199 0.0129 0.0121 0.0384 N (⃒ ⃒ )
1 ∑
MAD(x) 0.0449 0.0274 0.0559 0.0329 0.0261 0.1187 AAD(x) = ⃒x1,cal − x1,exp ⃒
i
(1)
AAD(y) 0.0063 0.0064 0.0143 0.0079 0.0081 0.0284 N i=1
MAD(y) 0.0203 0.0214 0.0399 0.0248 0.0249 0.0859
N (⃒ ⃒ )
1 ∑
AAD(y) = ⃒y1,cal − y1,exp ⃒
i
(2)
N i=1

Fig. 4. Composition deviation of different models against experimental data for CO2/R1234ze(Z) (a. deviation of liquid mole fraction; b. deviation of vapor
mole fraction).

Fig. 5. Composition deviation of different models against experimental data for CO2/R1336mzz(E) (a. deviation of liquid mole fraction; b. deviation of vapor
mole fraction).

6
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

Table 8 Therefore, PR model and SRK model are suitable for fitting VLE data
Heat pump configurations compared in this paper. of CO2/HFO investigated in this paper. Considering its simplicity, PR
Configuration #1 #2 #3 #4 #5 equation with vdW mixing rule is selected for the following simulation,
while kij in Table 5 is adopted.
Refrigerant R134a CO2/ CO2/ CO2/ CO2/
R600a R600 R1234ze R1336mzz
(Z) (E) 4. Performance modeling on a heat pump water heater using
Drawback GWP is Flammability – CO2/HFO
higher than cannot be
1000 eliminated at
xCO2 <0.9 In our previous studies, theoretical and experimental investigations
VLE Property [52] This work were carried out for air-source heat pump water heater in cold regions,
parameter package while CO2/HC exhibited excellent performance [17–19]. However, the
Cycle pattern Vapor- Single-stage recuperative flammability of HC is an inevitable drawback of such mixture. In this
injection
Model [18,19,53] [19]
section, a storage-type water heater is simulated, while the flammability
reference of CO2/HFO is also discussed to illustrate its potential in safety.

(⃒ 4.1. Model establishment

⃒ )
MAD(x) = max ⃒x1,cal − x1,exp ⃒i (3)
For water heaters producing hot water in cold regions, two modifi­
(⃒ ⃒ ) cations to heat pump water heater are commonly adopted. One is to
MAD(y) = max ⃒y1,cal − y1,exp ⃒i (4)
increase the component of refrigerant (i.e., using mixed refrigerant),
From Table 6, Table 7, Fig. 4 and Fig. 5, some results of model fitting which has been detailly evaluated in our previous studies [17–19]. The
can be found.

(1) Regarding the CO2/HFO mixture studied here, both PR model

and SRK model demonstrate similar performance in fitting the
datapoints with regard to AAD. However, PC-SAFT model ex­
hibits larger deviations. On the other hand, in terms of MAD, SRK
model is more accurate in fitting the liquid concentration
compared to PR model.
(2) For PR model and SRK model, with the increase of temperature,
the fitting deviation of liquid concentration decreases while that
of gas concentration increases. Nevertheless, such variation of
deviation still falls in acceptable range, with the largest deviation
less than 4.5%.
(3) PC-SAFT model exhibits deviation higher than 11% (i.e., MAD(x)
for CO2/R1336mzz(E) in Table 7). From Fig. 4 and Fig. 5, it can
be found that PC-SAFT model prediction gives significantly larger
deviation with temperature increasing. The liquid concentration
exhibits negative deviation while the gas concentration exhibits
positive deviation. The reason for such large deviation can be
ascribed by the unsuitability of PC-SAFT model for calculating Fig. 7. Temperature-entropy diagram of single-stage recuperative cycle
VLE near critical temperature [48]. using mixture.

Fig. 6. Schematic diagram of single-stage recuperative cycle using mixture.

7
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

Fig. 8. Comparisons of heating performances of different configurations at different working conditions.

other one is to increase the stage of compression (e.g., multi-stage heater in cold regions, despite the phasing out of R134a in the process.
compression heat pump), as commercially adopted at present [49–51]. Configurations #2 and #3 are potential transition schemes in our pre­
In this section, both modifications are taken for comparison, to further vious studies with satisfactory energy efficiency, but the involvement of
explore the potential of CO2/HFO. The configuration (i.e., the combi­ HC results in a nonnegligible flammability risk [17–19]. Configurations
nation of heat pump cycle pattern and refrigerant) is described in #4 and #5 adopt CO2/HFO studied in this paper, which are the focus of
Table 8, in which the detailed introduction on each cycle pattern (e.g., this paper.
assumption, operating principle, calculation method of performance) In order to simulate the different configurations described in Table 8,
can be retrieved in corresponding literature. The source of mixture Aspen HYSYS software was utilized. This software is a widely-used tool
property for simulation in this section is also listed in Table 8. in chemical engineering research, and is capable of simulating thermo­
As shown in Table 8, five configurations are compared in this section. dynamic systems such as heat pumps and refrigerators [18,20]. The
Configuration #1 is a common-used two-stage vapor-injection cycle, properties of R134a are pre-coded in Aspen Properties and were used
which is a representative scheme for commercial heat pump water directly in the simulation. For CO2/HC, PR model with vdW mixing rule

8
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

Table 9 (2) CO2/R1234ze(Z) (configuration #4) has slightly lower COP than
Comparison of refrigerants for different configurations in Table 8 (Twi = 323.15 CO2/R600 (configuration #3) at low ambient temperature, while
K; Two = 348.15 K; Ta = 263.15 K). the inferiority becomes obvious with the increase of ambient
Configuration #1 #2 #3 #4 #5 temperature. As evidenced by both the COP and VHC curves, the
Refrigerant R134a CO2/ CO2/ CO2/ CO2/
performance of CO2/R1234ze(Z) was found to be only slightly
R600a R600 R1234ze(Z) R1336mzz(E) inferior to that of CO2/R600. Therefore, CO2/R1234ze(Z) can be
COP 2.277 2.254 2.299 2.294 2.150 considered a promising alternative to CO2/R600, as it offers only
VHC (kJ/m3) 1265 3653 4665 4015 3409 marginal performance deterioration while providing substantial
P1 (MPa) 0.133 1.258 1.496 1.077 0.874
safety benefits.
P2 (MPa) 2.277 3.059 3.614 3.154 2.816
T2 (K) 370.66 374.84 373.96 383.11 360.91 (3) CO2/R1336mzz(E) (configuration #5) shows similar trends of
xCO2,tot 0 0.601 0.677 0.685 0.685 COP and VHC with CO2/R600a (configuration #2). In our pre­
GWP 1300 10 2 ~1 5 vious studies, CO2/R600a was found to exhibit the best seasonal
LFL (vol %) – 1.74 1.68 – – performance among CO2/HC [18]. In Fig. 8, the inferiority of
UFL (vol %) 6.55 6.22
CO2/R1336mzz(E) is not such significant. Taking the case of
– – –
Safety group A1 A3 A3 A1 A1
water inlet/outlet temperatures of 323.15/348.15 K as an
example, at ambient temperature of 263.15 K, COP and VHC of
CO2/R1336mzz(E) are 2.150 and 3409 kJ/m3 respectively,
Table A1 comparable with 2.254 and 3653 kJ/m3 of CO2/R600a.
Governing equations for single-stage recuperative heat pump water heater.
Device Mass Energy conservation Auxiliary In addition, it should be noted that in most of existing engineering
conservation constraint
software, a default value of kij = 0 is used when the VLE data of a mixture
Compressor m1 = m2 Wcomp = Assumption 6) is not available. In our calculations, it was found that using kij = 0 for
(m2 h2 − m1 h1 )/0.85 CO2/HFO resulted in operating parameters (such as operating pressures)
Condenser m2 = m3 , mwi = Qcon = m2 h2 − m3 h3 = Assumption 3)
with a deviation of up to 12.0% compared to the results obtained using
mwo mwo hwo − mwi hwi
Recuperator m3 = m4 , m6 = Qrec = m4 h4 − m3 h3 = Assumption 3) accurate kij. Therefore, an inadequate kij setting in the properties would
m1 m 6 h 6 − m 1 h1 result in inaccurate designs in practical applications [18]. This high­
Throttle m4 = m5 h4 = h5 – lights the importance of experimental VLE results for CO2/HFO in this
Evaporator m5 = m6 , mai = Qeva = m5 h5 − m6 h6 = Assumption 3) study.
mao mao hao − mai hai
From the results in this section, it can be found that CO2/HFO ex­
hibits satisfactory thermodynamic performance in applications of heat
was used, and kij was determined using the method presented in pump water heater, while an accurate model for VLE property is of great
Ref. [52]. For CO2/HFO, its kij was selected based on the results in significance in modeling and design of thermodynamic systems.
Table 5 of this paper.
The single-stage recuperative cycle is illustrated in Fig. 6, and its 4.3. Discussions
operational principles are detailed in Ref. [19]. The thermodynamic
model is briefly described in Appendix A3. Compared to the conven­ The potential in thermodynamics for CO2/HFO has been demon­
tional vapor-injection cycle, the single-stage recuperative cycle has a strated in section 4.2, while the environmental issue is also of great
simpler structure, resulting in lower initial costs and an easier control concern for applications. Compared with existing candidates in ther­
strategy [19]. In addition, the temperature-entropy diagram of refrig­ modynamic system, the advantages of CO2/HFO in terms of environ­
erant at each state in Fig. 6 is shown in Fig. 7, in which the temperature mental characteristics mainly manifest in GWP and flammability.
variations of water and air are also superimposed. As shown in Table 2, R1234ze(Z) and R1336mzz(E) are proposed as
The variables of such air-source heat pump water heaters include new-generation refrigerants with GWP lower than 10. On the contrary,
water inlet temperature, water outlet temperature and ambient tem­ R134a and R410A are refrigerants to be phased out, with high GWP of
perature. To account for the requirement of radiator heating, three pairs 1300 and 1924 respectively [13].
of water inlet/outlet temperatures are considered, i.e., 323.15/348.15 In terms of flammability, Nair [3] suggested that HFOs are a safe
K, 313.15/348.15 K, 313.15/338.15 K. The ambient temperature of cold alternative to HCs in applications with a charge amount up to 900 g. Wu
regions is selected from 243.15 K to 283.15 K, based on the data of et al. [54] conducted a review of existing studies on the flammability of
typical meteorological year of northern China retrieved through Ener­ HFOs. Pure R1336mzz(E) has been confirmed as non-flammable due to
gyPlus [18]. its low ratio of hydrogen atoms to hydrogen plus fluorine atoms. Pure
R1234ze(Z) has been classified as A2L in ASHRAE standard, with lower/
upper flammable limits (LFL/UFL) of 7.5/16.4 vol%, indicating similar
4.2. Thermodynamic performance comparison but slightly lower flammability than common-used R1234 isomers, e.g.,
R1234yf, R1234ze(E) [39,54].
The variations of heating performance with the change of working The flammability of binary blends is also discussed. CO2 is a widely-
condition are shown in this section. The operating composition and recognized combustion retardant, but the flammability of CO2/R1234ze
operating pressure are optimized through a genetic algorithm in MAT­ (Z) has not yet been studied. However, the critical suppression ratio of
LAB as introduced in Refs. [18,19]. Heating performance is evaluated in CO2/R1234yf was reported as 1.398:1 (CO2 mole fraction of 58.3%), so
terms of COP (coefficient of performance) and VHC (volumetric heating less CO2 is anticipated to eliminate the flammability of CO2/R1234ze(Z)
capacity). The heating performances of different configurations in [54]. In section 4.2, the suitable concentration of CO2 is higher than 0.6,
Table 8 are shown in Fig. 8. thus ensuring the safety of such mixture. As for CO2/R1336mzz(E), it is
From Fig. 8, some interesting points can be found as follows: non-flammable at random concentration. On the contrary, due to the
flammability of HC, the binary mixture of CO2/HC remains flammable
(1) CO2/HFO in this paper exhibits comparable performance against even at a CO2 concentration higher than 0.9 [55,56]. To this end, the
existing configurations. For configuration #1 (i.e., R134a), the safety of proposed CO2/HFO in this paper is illustrated.
noticeable drawback of low VHC is observed, attributed to its To show the advantages of CO2/HFO, a comparison of refrigerants
specific volume higher than mixture with CO2. and configurations is conducted, as shown in Table 9. Thermodynamic

9
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

performance, safety factor and environmental index are compared, at which significantly hinders accurate modeling of its performance for
the working condition of water inlet/outlet temperatures of 323.15/ specific applications. Therefore, further research in this aspect is
348.15 K and ambient temperature of 263.15 K. necessary to fully explore the potential of CO2/HFO and optimize its use
In Table 9, the working potential of CO2/HFO in this paper can be as eco-friendly refrigerants.
clearly illustrated. In this paper, an experimental rig for VLE measurement was built,
In terms of thermodynamic performance, these five configurations and two kinds of new CO2/HFO (i.e., CO2/R1234ze(Z) and CO2/
exhibit COP deviation less than 6.5%. Except for CO2/R1336mzz(E), the R1336mzz(E)) were measured at temperature from 253.15 K to 333.15
other configurations yield COP deviation less than 2%. Compared to K at CO2 concentration from 0 to 1. 148 datapoints were obtained and
R134a, the most significant advantage of using a mixed refrigerant is the fitted by three typical VLE models. PR model and SRK model are selected
VHC. In Fig. 8 and Table 9, the VHC of configurations #2~5 is more than as suitable models for describing VLE of CO2/HFO with deviation less
150% higher than that of configuration #1, which is attributed to the than 4.5%, whereas PC-SAFT model exhibits larger deviation. There­
low specific volume of CO2. As for operating parameters, it can be found after, a simulation of heat pump water heater is conducted with the new
that CO2/HFO has lower operating pressure and higher CO2 concen­ fitted kij, along with the comparison with existing refrigerants. At typical
tration than CO2/HC. The operating pressure of CO2 system is effectively working condition of radiator heating (i.e., providing hot water up to
reduced by the addition of HFO. In Table 9, the discharge pressures of 348.15 K at ambient temperature down to 253.15 K), CO2/R1234ze(Z)
CO2/HFO do not exceed 3.2 MPa, which are comparable to common- and CO2/R1336mzz(E) show similar COP with common-used R134a and
used refrigerant R410A. CO2/HC. VHC of CO2/HFO is more than 150% higher than that of
In terms of environmental impact, CO2/HFOs are clearly superior R134a. The addition of HFO reduces the operating pressures of CO2
among five configurations. Although the breakdown effect of HFOs system, while the addition of CO2 eliminates the flammability of HFO in
needs further study, the environmental advantages of using HFOs are mixture. The studied CO2/HFOs possess noteworthy environmental
evident in terms of flammability and GWP. The GWP of CO2/HFO is characteristics, with an ODP of 0 and a GWP lower than 10.
lower than 10, much lower than 1300 of R134a. The flammability of The results in this work supplement reliable VLE data for CO2/HFO
CO2/HFOs are much lower than those of CO2/HCs. Even though CO2 is mixtures and demonstrate the potential of CO2/R1234ze(Z) and CO2/
included with a mole fraction of more than 0.6, the flammability of HC R1336mzz(E) as refrigerant for heat pump water heater.
makes CO2/HC flammable. According to the ASHRAE handbook, re­
frigerants with LFL lower than 3.5% are classified as A3 (i.e., highly Declaration of Competing Interest
flammable), so configurations #2 and #3 present severe safety concerns
in practical application, especially under large-scale circumstance. For The authors declare that they have no known competing financial
CO2/HFO, both CO2/R1234ze(Z) and CO2/R1336mzz(E) are non- interests or personal relationships that could have appeared to influence
flammable, indicating promising working prospects. the work reported in this paper.

5. Conclusions Data availability

In the context of sustainable development, the exploration of eco- Data will be made available on request.
friendly refrigerants becomes a popular research topic, while CO2 and
HFOs are regarded as promising candidates. With conventional re­ Acknowledgement
frigerants being gradually phased out, CO2/HFO is being emerged as a
viable alternative with significant potential for widespread applications. This work is supported by the National Natural Science Foundation
However, the investigation on the VLE of CO2/HFO is currently limited, of China (No.51976179 and No.51876185).

Appendix A1. Uncertainty analysis

The uncertainty of pressure is caused by the accuracy of pressure sensor up1 (listed in Table 3) and the accuracy of datalogger up2. In the
experiment, Agilent 34970A is used to collect the voltage signal output from pressure sensor and transfer it into pressure signal. According to the
manual of datalogger, up2 is found as (0.002% measuring value + 0.0005% FS). Therefore, the expanded uncertainty for pressure measurement (k = 2)
is calculated as follows.
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
up = 2 up1 2 + up2 2 = 0.8 kPa (for p from 0 − 1 MPa) or 6 kPa (for p from 1 − 7.5 MPa) (A1)

The uncertainty of temperature is caused by the accuracy of platinum precision thermometer uT1 of 5 mK, the accuracy of datalogger uT2 of 1 mK
and the temperature control fluctuation of thermostatic bath uT3 of 50 mK. Therefore, the expanded uncertainty for temperature measurement (k = 2)
is calculated as follows.
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
uT = 2 uT,1 2 + uT,2 2 + uT,3 2 = 0.1 K (A2)

The uncertainty of composition is caused by the response ratio error during calibration of chromatograph and the peak area ratio error in
experiment. As introduced in out studies, the maximum expanded uncertainty for composition is 0.009 [57,58].
The detailed uncertainty for each datapoint can be found in Supporting Information.

Appendix A2. Description of VLE models used in this study

In this study, three VLE models are adopted, and they are briefly introduced here.

A2.1 SRK model

10
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

The basic expression of SRK equation of state is shown as follows [47].

RT a*α(T)
p= − (A3)
v − b v(v + b)
In Eq. (A3), p, T and v are pressure, temperature and molar volume respectively. R is the universal gas constant of 8.314 J/K/mol. a, b and α(T) are
determined as follows.
a = 0.4278R2 Tc2 /pc (A4)

b = 0.08664RTc /pc (A5)

[ ]2
α(T) = 1 + (0.480 + 1.574ω − 0.176ω2 )(1 − (T/Tc )0.5 ) (A6)

In the above equations, the subscript ‘c’ represents critical properties, and ω is acentric factor.
To expand the applicability of above equations to mixture, the mixing rule should be involved. Here vdW mixing rule is used as follows [45,47].
∑∑
am = xi xj aij (A7)
i j

∑
bm = xi bi (A8)
i

√̅̅̅̅̅̅̅̅
aij = ai aj (1 − kij ) (A9)

In the above equations, the subscript ‘m’ represents mixture, and × is mole fraction. kij is interaction parameter. Specifically, kii = kjj = 0 and kij =
kji.
SRK model has been successfully used for VLE description of binary system with HFOs [47].

A2.2 PR model

The basic expression of PR equation of state is shown as follows [45].

RT a*α(T)
p= − (A10)
v − b v(v + b) + b(v − b)
In Eq. (A10), p, T and v are pressure, temperature and molar volume respectively. R is the universal gas constant of 8.314 J/K/mol. a, b and α(T) are
determined as follows.
a = 0.45724R2 Tc2 /pc (A11)

b = 0.0778RTc /pc (A12)

[ ]2
α(T) = 1 + (0.3746 + 1.54226ω − 0.26992ω2 )(1 − (T/Tc )0.5 ) (A13)

In the above equations, the subscript ‘c’ represents critical properties, and ω is acentric factor.
To expand the applicability of above equations to mixture, vdW mixing rule is used as introduced in Eqs. (A7-A9). PR model has been successfully
used for VLE description of binary system with HFOs or CO2 [45].

A2.3 PC-SAFT model

The detailed derivation for PC-SAFT model can be found in Ref. [59], and here a brief introduction for binary system is shown. The basic expression
of PR equation of state is shown as follows. For this model, the residual Helmholtz free energy is composed with two terms of hard-chain reference
contribution and dispersion contribution, as shown below.
ares = ̃
̃ ahc + ̃
adisp (A14)
res
In Eq. (A14), ã is residual Helmholtz free energy, and the superscripts of ‘hc’ and ‘disp’ denote hard-chain and dispersion respectively.
For ãhc, it is determined as follows.
∑
ahc = m̃
̃ ahs − xi (mi − 1)lnghs
ii (σ ii ) (A15)
i

∑
m= xi mi (A16)
i

[ ( ) ]
1 3ξ1 ξ2 ξ32 ξ32
ahs =
̃ + + − ξ 0 ln(1 − ξ 3 ) (A17)
ξ0 (1 − ξ3 ) ξ3 (1 − ξ3 )2 ξ23

( ) ( )2
1 di dj 3ξ2 di dj 2ξ22
ghs
ij = + 2
+ (A18)
(1 − ξ3 ) di + dj (1 − ξ3 ) di + dj (1 − ξ3 )3

11
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

π ∑
ξn = ρ xi mi din , n ∈ {0, 1, 2, 3} (A19)
6 i
[ ( )]
3∊i
di (T) = σi 1 − 0.12exp − (A20)
kT
In Eqs. (A15)–(A20), m is the number of segments per chain, and σ is the segment diameter. x is mole fraction. ρ is density. k is the Boltzmann
constant. ∊ is the depth of pair potential. Through Eqs. (A15-A20), if three pure component parameters (i.e., m, σ , ∊) are given, ãhc can be determined
for pure substance.
For ãdisp, it is determined as follows.
adisp = − 2πρI1 (η, m)m2 ∊σ3 − πρmC1 I2 (η, m)m2 ∊2 σ3
̃ (A21)

∑
6
I1 (η, m) = ai (m)ηi (A22)
i=0

∑∑ (∊ )
(A23)
ij
m 2 ∊σ 3 = xi xj mi mj σ3
i j
kT ij
( )− 1
8η − 2η2 ( 20η − 27η2 + 12η3 − 2η4
C1 = 1+m 4
+ 1 − m) (A24)
(1 − η) [(1 − η)(2 − η)]2

∑
6
I2 (η, m) = bi (m)ηi (A25)
i=0

∑∑ ( ∊ )2
(A26)
ij
m 2 ∊2 σ 3 = xi xj mi mj σ 3ij
i j
kT

m− 1 m− 1 m− 2
ai (m) = a0i + a1i + a2i (A27)
m m m
m− 1 m− 1 m− 2
bi (m) = b0i + b1i + b2i (A28)
m m m
In Eqs. (A21-A28), η is packing fraction, whose value is equal to ξ3 [59]. aki and bki are universal model constants whose values refer to Ref. [59].
The meanings of other symbols are the same with those in Eqs. (A15-A20). Through Eqs. (A21)–(A28), if three pure component parameters (i.e., m, σ,
∊) are given, ãdisp can be determined for pure substance.
In this study, pure component parameters (i.e., m, σ, ∊) are obtained by fitting the vapor pressure of pure refrigerant reported in literature
[12,40,42,60]. To calculate the property of mixture, the combining rules are employed as follows.
1( )
σ ij = σi + σj (A29)
2
√̅̅̅̅̅̅̅̅( )
∊ij = ∊i ∊j 1 − kij (A30)

In Eq. (A30), kij is binary interaction parameter to be fitted in this study. PC-SAFT model has been successfully used for VLE description of binary
system with HFOs [43].

Appendix A3. Description of heat pump water heater model in this study

In this study, an air-source heat pump heater is simulated to illustrate the application potential of CO2/HFO.
The model of single-stage recuperative heat pump was detailly introduced and experimentally validated in our previous studies, at the working
condition same as that in this study [17–19]. It is briefly explained as follows.
The following assumption is performed for modeling [18,19]: 1) heat pump operates at steady conditions; 2) pressure drops and heat losses are
neglected along pipeline; 3) the pinch points of condenser and recuperator are not less than 2 K, and that of evaporator is not less than 5 K; 4) the
isentropic efficiency of compressor is a function of pressure ratio (i.e., ηi = 0.9343–0.04478P2/P1), and the motor efficiency is 0.85; 5) adiabatic
expansion is undergoing in throttle; 6) the upper limit of discharge temperature for normal operation is 393.15 K.
The temperatures of inlet and outlet water are decided by user requirement, and the inlet air temperature is ambient temperature. The property
source of refrigerant is introduced in Table 8. Using mass and energy conservations, the governing equations can be established in Table A1.
In Table A1, all of subscripts denote the state point in Fig. 6. h represents specific enthalpy, and Wcomp is the power consumption of compressor. The
calculation flowchart can be found in Ref. [19]. Two performance indexes are used in this study, i.e., COP and VHC, which are calculated as follows.
COP = Qcon /Wcomp (A31)

VHC = Qcon /(m1 /ρ1 ) (A32)

COP is an index representing the amount of heating capacity output for per unit of power input. VHC is an index representing the amount of heating
capacity output for per unit of suction volumetric flow rate of compressor. Lower VHC means that larger-scale compressor is needed for given heating
requirement.
As for the model of vapor-injection cycle using R134a, it is a common choice for such application, and it is also used for comparison in the studies

12
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

on heat pump advancement. The detailed information for the model can refer to Refs. [18,19], and it is omitted here. It should be noted that this model
is also validated by experiment under working condition of condensing temperature of 313.15 K and evaporation temperature down to 253.15 K,
whose deviation was less than 5.2% [61]. Therefore, the model of vapor-injection cycle using R134a can be validated.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.applthermaleng.2023.120576.

References [24] A. Arami-Niya, X. Xiao, S.Z.S. Al Ghafri, F. Jiao, M. Khamphasith, E. Sadeghi

Pouya, M.S. Sadaghiani, X. Yang, T. Tsuji, Y. Tanaka, Y. Seiki, E.F. May,
Measurement and modelling of the thermodynamic properties of carbon dioxide
[1] L. Höglund-Isaksson, P. Purohit, M. Amann, I. Bertok, P. Rafaj, W. Schöpp,
mixtures with HFO-1234yf, HFC-125, HFC-134a, and HFC-32: vapour-liquid
J. Borken-Kleefeld, Cost estimates of the Kigali Amendment to phase-down
equilibrium, density, and heat capacity, Int. J. Refrig. 118 (2020) 514–528.
hydrofluorocarbons, Environ. Sci. Policy 75 (2017) 138–147.
[25] S. Li, S. Peng, Z. Yang, Y. Duan, Vapor-liquid equilibrium measurements for binary
[2] Í.F. Guilherme, D.F. Marcucci Pico, D.D.O. Dos Santos, E.P. Bandarra Filho,
mixtures of carbon dioxide (CO2) + 2,3,3,3-Tetrafluoroprop-1-ene (R-1234yf) and
A review on the performance and environmental assessment of R-410A alternative
carbon dioxide (CO2) + 3,3,3-Trifluoropropene (R-1243zf), Fluid Phase Equilib.
refrigerants, J. Build. Eng. 47 (2022), 103847.
561 (2022), 113542.
[3] V. Nair, HFO refrigerants: a review of present status and future prospects, Int. J.
[26] G. Di Nicola, G. Passerini, F. Polonara, R. Stryjek, PVTx measurements of the
Refrig. 122 (2021) 156–170.
carbon dioxide+trans-1,3,3,3-tetrafluoroprop-1-ene binary system, Fluid Phase
[4] S. Bobbo, G.D. Nicola, C. Zilio, J.S. Brown, L. Fedele, Low GWP halocarbon
Equilib. 360 (2013) 124–128.
refrigerants: a review of thermophysical properties, Int. J. Refrig. 90 (2018)
[27] S. Wang, R. Fauve, C. Coquelet, A. Valtz, C. Houriez, P. Artola, E.E. Ahmar,
181–201.
B. Rousseau, H. Hu, Vapor–liquid equilibrium and molecular simulation data for
[5] C. Mateu-Royo, C. Arpagaus, A. Mota-Babiloni, J. Navarro-Esbrí, S.S. Bertsch,
carbon dioxide (CO2) + trans-1,3,3,3-tetrafluoroprop-1-ene (R-1234ze(E)) mixture
Advanced high temperature heat pump configurations using low GWP refrigerants
at temperatures from 283.32 to 353.02 K and pressures up to 7.6 MPa, Int. J.
for industrial waste heat recovery: a comprehensive study, Energ. Conver. Manage.
Refrig. 98 (2019) 362–371.
229 (2021), 113752.
[28] M.S. Sadaghiani, A. Arami-Niya, B. Marsh, S.Z.S. Al Ghafri, E.F. May, Vapor-liquid
[6] A. Jones, A. Wolf, S.M. Kwark, Refrigeration system development with limited
equilibria for carbon dioxide + 3,3,3-trifluoropropene binary mixtures at
charge of flammable Refrigerant, R-290, Therm. Sci. Eng. Progr. 34 (2022),
temperatures between (288 and 348) K, J. Chem. Eng. Data 66 (2021) 4044–4055.
101392.
[29] W.M. Nelson, T.P. Benecke, D. Ramjugernath, Phase equilibrium data for binary
[7] J. Wang, M. Belusko, M. Evans, M. Liu, C. Zhao, F. Bruno, A comprehensive review
mixtures of carbon dioxide with 1,1,2,3,3,3-hexafluoro-1-propene or 2,2,3-tri­
and analysis on CO2 heat pump water heaters, Energy Convers. Manage.: X 15
fluoro-3-(trifluoromethyl)oxirane at temperatures between (233 and 273) K, Fluid
(2022), 100277.
Phase Equilib. 425 (2016) 114–119.
[8] Y. Song, C. Cui, X. Yin, F. Cao, Advanced development and application of
[30] I.H. Bell, D. Riccardi, A. Bazyleva, M.O. McLinden, Survey of data and models for
transcritical CO2 refrigeration and heat pump technology—a review, Energy Rep. 8
refrigerant mixtures containing halogenated olefins, J. Chem. Eng. Data 66 (2021)
(2022) 7840–7869.
2335–2354.
[9] R.B. Barta, E.A. Groll, D. Ziviani, Review of stationary and transport CO2
[31] X. Yao, L. Ding, X. Dong, Y. Zhao, X. Wang, J. Shen, M. Gong, Experimental
refrigeration and air conditioning technologies, Appl. Therm. Eng. 185 (2021),
measurement of vapor-liquid equilibrium for 3,3,3-trifluoropropene(R1243zf) +
116422.
1,1,1,2-tetrafluoroethane(R134a) at temperatures from 243.150 to 293.150 K, Int.
[10] B. Yu, J. Yang, D. Wang, J. Shi, J. Chen, An updated review of recent advances on
J. Refrig. 120 (2020) 97–103.
modified technologies in transcritical CO2 refrigeration cycle, Energy 189 (2019),
[32] Z. Yang, X. Tang, J. Wu, J. Lu, Experimental measurements of saturated vapor
116147.
pressure and isothermal vapor-liquid equilibria for 1,1,1,2-Tetrafluoroethane
[11] F. Cao, Y. Song, M. Li, Review on development of air source transcritical CO2 heat
(HFC-134a) + 3,3,3-trifluoropropene (HFO-1243zf) binary system, Fluid Phase
pump systems using direct-heated type and recirculating-heated type, Int. J. Refrig.
Equilib. 498 (2019) 86–93.
104 (2019) 455–475.
[33] X. Zhang, X. Dong, H. Guo, M. Gong, J. Shen, J. Wu, Measurements and
[12] E. Boonaert, A. Valtz, J. Brocus, C. Coquelet, Y. Beucher, F. De Carlan,
correlations of isothermal (vapour + liquid) equilibrium for the isobutane (R600a)
J. Fourmigué, Vapor-liquid equilibrium measurements for 5 binary mixtures
+ cis-1,3,3,3-tetrafluoropropene (R1234ze(Z)) system at temperatures from
involving HFO-1336mzz(E) at temperatures from 313 to 353 K and pressures up to
(303.150 to 353.150) K, J. Chem. Thermodyn. 103 (2016) 349–354.
2.735 MPa, Int. J. Refrig. 114 (2020) 210–220.
[34] X. Zhang, X. Dong, H. Guo, Y. Zhao, H. Zhang, M. Gong, J. Shen, Measurements of
[13] Z. Yang, B. Feng, H. Ma, L. Zhang, C. Duan, B. Liu, Y. Zhang, S. Chen, Z. Yang,
isothermal (vapour + liquid) equilibrium for the 1,1,2,2–1,1,2,2-tetrafluoroethane
Analysis of lower GWP and flammable alternative refrigerants, Int. J. Refrig. 126
(R134) + cis -1,3,3,3-tetrafluoropropene (R1234ze(Z)) system at temperatures
(2021) 12–22.
from (303.150 to 343.150) K, J. Chem. Thermodyn. 111 (2017) 20–26.
[14] D.C. Savitha, P.K. Ranjith, B. Talawar, R.P. Reddy, N., Refrigerants for sustainable
[35] D. Mambo-Lomba, P. Paricaud, Predictions of thermodynamic properties and phase
environment - a literature review, Int. J. Sustain. Energ. 41 (2022) 235–256.
equilibria of refrigerant systems with COSMO approaches, Int. J. Refrig. 124
[15] Z. Roy, G. Halder, Replacement of halogenated refrigerants towards sustainable
(2021) 52–63.
cooling system: a review, Chem. Eng. J. Adv. 3 (2020), 100027.
[36] S. Tomassetti, G. Di Nicola, M. Pierantozzi, J.S. Brown, Two-phase and vapor phase
[16] M. Bai, L. Zhao, R. Zhao, Review on applications of zeotropic mixtures, J. Therm.
PvTx properties of the difluoromethane +cis-1,3,3,3-tetrafluoroprop-1-ene binary
Sci. 31 (2022) 285–307.
system, J. Chem. Eng. Data 65 (2020) 4326–4334.
[17] J. Luo, K. Yang, Z. Zhao, G. Chen, Q. Wang, Experimental investigations on the
[37] X. Nie, L. Zhao, S. Deng, W. Su, Y. Zhang, A review of molecular simulation applied
performance of a single-stage compound air-source heat pump using CO2/R600a in
in vapor-liquid equilibria (VLE) estimation of thermodynamic cycles, J. Mol. Liq.
cold regions, Appl. Therm. Eng. 205 (2022), 118050.
264 (2018) 652–674.
[18] J. Luo, Q. Wang, K. Yang, Z. Zhao, G. Chen, S. Zhang, Performance investigation on
[38] T. Katsuyuki, I. Junichi, K.K. Konstantinos, Thermodynamic properties of HFO-
a novel air-source heat pump using CO2/HC for recirculated water heater in cold
1336mzz(E) (trans-1,1,1,4,4,4-hexafluoro-2-butene) at saturation conditions, Int.
regions, Sustainable Energy Technol. Assess. 53 (2022), 102496.
J. Refrig. 82 (2017) 283–287.
[19] J. Luo, Q. Wang, Z. Zhao, K. Yang, G. Chen, S. Zhang, Theoretical investigation on
[39] J. Yang, Z. Ye, B. Yu, H. Ouyang, J. Chen, Simultaneous experimental comparison
the heating performance of a recuperative heat pump water heater using CO2/HCs
of low-GWP refrigerants as drop-in replacements to R245fa for Organic Rankine
with large temperature glides in cold regions, J. Therm. Sci. Eng. Appl. 14 (2022),
cycle application: R1234ze(Z), R1233zd(E), and R1336mzz(E), Energy 173 (2019)
031012.
721–731.
[20] Z. Zhao, J. Luo, Q. Song, K. Yang, Q. Wang, G. Chen, Theoretical investigation and
[40] Refprop V10. National Institute Of Standards (NIST).
comparative analysis of the Linde-Hampson refrigeration system using eco-friendly
[41] J. He, X. Xu, X. Gui, New vapor-liquid phase equilibrium data of CO2 in several
zeotropic refrigerants based on R744/R1234ze(Z) for freezing process applications,
heavy n-alkanes at high pressures, J. Chem. Eng. Data 66 (2021) 1600–1610.
Int. J. Refrig. 145 (2023) 30–39.
[42] K. Tanaka, Y. Higashi, R. Akasaka, Y. Kayukawa, K. Fujii, Measurements of the
[21] G. Di Nicola, C. Di Nicola, A. Arteconi, R. Stryjek, PVTx measurements of the
vapor− liquid equilibrium for the CO2 + R290 mixture, J. Chem. Eng. Data 54
carbon dioxide + 2,3,3,3-tetrafluoroprop-1-ene binary system, J. Chem. Eng. Data
(2009) 1029–1033.
57 (2012) 450–455.
[43] A. Barati-Harooni, A. Najafi-Marghmaleki, Prediction of vapor–liquid equilibrium
[22] G. Raabe, Molecular simulation studies on the vapor-liquid phase equilibria of
for binary mixtures containing R1234yf or R1234ze (E), Int. J. Refrig. 88 (2018)
binary mixtures of R-1234yf and R-1234ze(E) with R-32 and CO2, J. Chem. Eng.
239–247.
Data 58 (2013) 1867–1873.
[44] Z. Wu, R. Sun, L. Shi, H. Tian, X. Wang, G. Shu, A comparative and predictive study
[23] N. Juntarachat, A. Valtz, C. Coquelet, R. Privat, J. Jaubert, Experimental
of the mixing rules for the vapor-liquid equilibria of CO2-based mixtures, CIESC J.
measurements and correlation of vapor–liquid equilibrium and critical data for the
73 (2022) 1483–1492.
CO2 + R1234yf and CO2 + R1234ze(E) binary mixtures, Int. J. Refrig. 47 (2014)
[45] Z. Wu, L. Shi, R. Sun, H. Tian, X. Wang, P. Hu, G. Shu, A temperature-independent
141–152.
prediction model predicts the vapor-liquid equilibrium of CO2-based binary
mixtures, Int. J. Refrig. 140 (2022) 125–138.

13
J. Luo et al. Applied Thermal Engineering 229 (2023) 120576

[46] S. Zid, J. Bazile, J. Daridon, A. Piña-Martinez, J. Jaubert, S. Vitu, Fluid phase [54] X. Wu, C. Dang, S. Xu, E. Hihara, State of the art on the flammability of
equilibria for the CO2 + 2,3-dimethylbutane binary system from 291.9 K to 373.1 hydrofluoroolefin (HFO) refrigerants, Int. J. Refrig. 108 (2019) 209–223.
K, J. Supercrit. Fluids 179 (2022), 105387. [55] H. Tian, M. Wu, G. Shu, Y. Liu, X. Wang, Experimental and theoretical study of
[47] Y. Qin, N. Li, B. Liu, H. Zhang, Isothermal measurement and modeling of VLE flammability limits of hydrocarbon–CO2 mixture, Int. J. Hydrogen Energy 42
properties for 2,3,3,3-tetrafluoroprop-1-ene + trifluoromethane + (2017) 29597–29605.
tetrafluoromethane ternary system at temperatures from 253.15 K to 273.15 K, Int. [56] H. Tian, Y. Liu, G. Shu, L. Li, X. Huo, Theoretical and experimental research on the
J. Refrig. 117 (2020) 284–294. influence of initial temperature on the flammability of hydrocarbon-CO2 mixture
[48] N.I. Diamantonis, G.C. Boulougouris, E. Mansoor, D.M. Tsangaris, I.G. Economou, using in organic Rankine cycle, Energy 167 (2019) 939–949.
Evaluation of cubic, SAFT, and PC-SAFT equations of state for the vapor-liquid [57] G. Ye, Y. Fang, Y. Li, Z. Guo, L. Wang, X. Han, Isothermal vapor-liquid equilibria of
equilibrium modeling of CO2 mixtures with other gases, Ind. Eng. Chem. Res. 52 the absorption working pairs (R1234ze(E) + NMP and R1234ze(E) + DMETrEG) at
(2013) 3933–3942. temperatures from 293.15 to 353.15 K, J. Chem. Eng. Data 66 (2021) 3582–3589.
[49] A. Redón, E. Navarro-Peris, M. Pitarch, J. Gonzálvez-Macia, J.M. Corberán, [58] G. Ye, Y. Fang, Y. Zhuang, Z. Guo, H. Ni, X. Wu, X. Han, Vapor-liquid equilibria for
Analysis and optimization of subcritical two-stage vapor injection heat pump the binary systems difluoromethane (R32) +trans-1,1,1,4,4,4-hexafluoro-2-butene
systems, Appl. Energy 124 (2014) 231–240. (R1336mzz(E)) and fluoroethane (R161) +trans-1,1,1,4,4,4-hexafluoro-2-butene
[50] W. Wei, L. Ni, L. Xu, Y. Yang, Y. Yao, Application characteristics of variable (R1336mzz(E)), J. Chem. Eng. Data 66 (2021) 3473–3482.
refrigerant flow heat pump system with vapor injection in severe cold region, [59] J. Gross, G. Sadowski, Perturbed-chain SAFT: an equation of state based on a
Energ. Build. 211 (2020), 109798. perturbation theory for chain molecules, Ind. Eng. Chem. Res. 40 (2001)
[51] L. Xu, E. Li, Y. Xu, N. Mao, X. Shen, X. Wang, An experimental energy performance 1244–1260.
investigation and economic analysis on a cascade heat pump for high-temperature [60] N. Sakoda, Y. Higashi, R. Akasaka, Measurements of PvT properties, vapor
water in cold region, Renew. Energy 152 (2020) 674–683. pressures, saturated densities, and critical parameters for trans-1,1,1,4,4,4-
[52] H. Zhang, M. Gong, H. Li, Y. Zhao, Q. Zhong, X. Dong, J. Shen, J. Wu, A simple hexafluoro-2-butene (R1336mzz(E)), J. Chem. Eng. Data 66 (2021) 734–739.
model for temperature-independent k of the PR-vdW model for mixtures [61] J. Wang, D. Qv, Y. Yao, L. Ni, The difference between vapor injection cycle with
containing HCs, HFCs, PFCs, HFOs, CO2, RE170 and R13I1, Fluid Phase Equilib. flash tank and intermediate heat exchanger for air source heat pump: an
425 (2016) 374–384. experimental and theoretical study, Energy 221 (2021), 119796.
[53] E.A. Rad, S. Maddah, Entropic optimization of the economizer’s pressure in a heat
pump cycle integrated with a flash-tank and vapor-injection system, Int. J. Refrig.
97 (2019) 56–66.

14