Beyond Temperature Glide: The Compressor Is Key To Realizing Bene Fits of Zeotropic Mixtures in Heat Pumps
Beyond Temperature Glide: The Compressor Is Key To Realizing Bene Fits of Zeotropic Mixtures in Heat Pumps
Beyond Temperature Glide: The Compressor Is Key To Realizing Bene Fits of Zeotropic Mixtures in Heat Pumps
www.entechnol.de
Energy Technol. 2021, 2000955 2000955 (1 of 10) © 2021 The Authors. Energy Technology published by Wiley-VCH GmbH
www.advancedsciencenews.com www.entechnol.de
Economic Assessment Panel to the Montreal Protocol,[3] about rig that was operated in cooling and heating mode. Propane
80% of the climate impact of cooling cycles is from indirect emis- reached the highest COPs of all pure components. For the cooling
sions and 20% from direct emissions. mode, the COPs are increased for mixtures. The highest COPs are
Both direct and indirect emissions of cooling appliances and about 16% higher than those of the pure components isobutane or
heat pumps can be reduced by zeotropic mixtures: on the one n-butane but only increased by 7% compared with propane. For
hand, the GWP and other essential criteria such as flammability the heating mode, no benefit of a mixture was obtained; here, pro-
can be adjusted by the targeted selection of fluid pairs.[4] On the pane had always the highest COP. Only small benefits of mixtures
other hand, zeotropic mixtures have a high potential for coeffi- compared with the better pure component were also reported by
cient of performance (COP) increase due to their nonisothermal Park et al.[14] for ethane/propane mixtures, and by Venzik et al. for
phase change.[5,6] This temperature change during isobaric evap- isobutane/propene mixtures[5] and isobutane/propane mix-
oration and condensation is known as temperature glide. The tures.[15] The main results of these studies can be summarized
extent of the temperature glide generally depends on the constit- as follows: the COPs of both pure components differ, and the max-
uents of a mixture and its composition x.[5,6] The fundamental imum COP is often observed for a specific composition. However,
thermodynamic idea of using zeotropic mixtures is that the non- the COP of the best mixture composition is just a few percentage
isothermal phase change better matches the temperature change points higher compared with the better performing pure fluid. In
in typical heat sources and sinks.[6] Matching of the temperature particular, the COP maximum is generally not located at the com-
profiles minimizes exergy losses in the heat exchangers, and thus position with the best matching temperature glide but is shifted to
increases the COP. higher fractions of the better performing pure fluid.
Mixtures have already been proposed that achieve the desired The significant differences between simulations and experi-
thermodynamic properties such as low GWP (e.g., R404A, ments suggest that at least one crucial parameter has been lack-
R452A),[4] but these mixtures mostly have a negligible tempera- ing in the simulation models used. The resulting fundamental
ture glide. In turn, the targeted use of a temperature glide is not questions are: is it generally impossible to achieve the theoreti-
yet industrial practice, even though it has been discussed in sci- cally predicted COP benefits of mixtures in real plants? Or did we
ence for many decades.[7–9] In particular, the targeted selection of just considered the wrong mixture components so far? These
suitable fluid pairs for a specific application is still challenging. questions lead to the following approach in this work.
The available publications on zeotropic mixtures are mostly We compare experimental results from a water/water heat
theoretical studies and a few experimental studies. The theoreti- pump test rig with simple simulations and explain the origin
cal studies aim either to show the general potential of zeotropic of the deviations between usual models and experiments. The
mixtures or to identify potentially suitable fluid pairs for a spe- experimental results of three zeotropic mixtures are then ana-
cific application. McLinden and Radermacher[10] compared pure lyzed in more detail. We figure out why zeotropic mixtures do
and mixed refrigerants. Considering an ideal vapor-compression not achieve the expected COP benefits in experiments, even if
cycle by assuming isobaric heat exchangers and an isentropic the temperature glide matches the temperature changes in the
compressor, the mixtures R22/R114 and R22/R11 were investi- heat transfer fluids. The fluid-dependent compressor efficiency,
gated over the whole composition range (from pure to pure which is usually neglected in theoretical studies, is shown to be
fluid). For both mixtures, the authors found an increased particularly important here (Section 3). Based on these findings,
COP for a specific composition compared with the respective we propose a compressor-based criterion for the targeted selec-
pure components. The COP increase reached up to 18% for a tion of mixture components, in addition to the usual criterion to
large temperature difference of 25 K for the secondary heat trans- find mixtures with well-matching temperature glides. The com-
fer fluid. Based on their results, the authors concluded that the pressor criterion is solely based on the compressor efficiencies of
better the temperature glide of the mixture would match the tem- the pure components (Section 4), enabling a preselection without
perature change in the secondary heat transfer fluids, the higher the knowledge of mixture properties. By utilizing a physical com-
would be the COP increase in zeotropic mixtures. Similar results pressor model to calculate fluid-dependent compressor efficien-
are also found by Yan et al.[11] for propane/isobutane mixtures in cies,[16] the criterion is applied to fluid selection from a database.
a vapor-compression refrigeration cycle with separation conden- One mixture is selected that fulfills the criterion and has a well-
sation, by Stoecker and Walukas[7] for R12/R114 mixtures in a matching temperature glide. The selected mixture is then tested
domestic refrigerator, and by Zühlsdorf et al.[12] for different in a process simulation over the whole composition range. The
mixtures of hydrocarbons and hydrofluoroolefins in booster heat results show that zeotropic mixtures can achieve a large COP
pumps. In another study, Zühlsdorf et al.[6] investigated the tem- increase if the pure components fulfill our criterion of high
perature glide matching in more detail. For 14 natural refriger- and similar compressor efficiencies.
ants, the authors conducted process simulations for several
mixtures under various compositions. They found large COP
increases up to 27% for zeotropic mixtures if the temperature 2. Experimental Section
change in the heat source is significant. All these theoretical stud-
ies show that a zeotropic mixture improves the COP significantly The vapor-compression cycle considered in our study is a basic
if only the temperature glide well matches the temperature water/water heat pump (Figure 1). The schematic represents
change in the heat transfer fluids. both the test rig in simplified form and the simulation model.
However, these theoretical studies mostly contradict the avail- Figure 1 also defines thermodynamic states. The evaporation
able experimental studies. Chang et al.[13] investigated propane/ temperature is defined as the temperature at the evaporator inlet
isobutane and propane/n-butane mixtures in a heat pump test Tev ¼ T4; the condensation temperature refers to the dew point
Energy Technol. 2021, 2000955 2000955 (2 of 10) © 2021 The Authors. Energy Technology published by Wiley-VCH GmbH
www.advancedsciencenews.com www.entechnol.de
Energy Technol. 2021, 2000955 2000955 (3 of 10) © 2021 The Authors. Energy Technology published by Wiley-VCH GmbH
www.advancedsciencenews.com www.entechnol.de
The condensation temperature is the only remaining degree of equation of state (EOS)[22] combined with a polynomial correlat-
freedom of the simulations. The standard approach in the litera- ing the temperature dependence of the isobaric ideal gas heat
ture for heat pump simulations calculates the condensation tem- capacity. The fluid database used for the Peng-Robinson EOS
perature by a minimum temperature approach. Usually, the property model is composed of different sources[21,23,24] and is
minimum approach temperature is estimated. Here, we use given in Section SI3, Supporting Information. The fluid property
the minimum approach temperature observed in our experi- model used for calculation is indicated in the results section.
ments. All experiments showed a similar minimum approach
temperature of about ΔTmin ¼ 0.2 K. By using the minimum
approach temperature observed in the experiments, we aim to 3. Zeotropic Mixtures: Theory and Real Life
reproduce the behavior of the test rig in the simulations. 3.1. Temperature Glide
Thus, the condensation temperature Tcon is always calculated
as the minimum value while fulfilling the minimum approach In this work, we used three zeotropic mixtures of alkanes and
temperature ΔTmin. This calculation is numerically implemented alkenes: isobutane (R600a)/propane (R290), isobutane
by an optimization (cf. Section SI1, Supporting Information). (R600a)/propene (R1270), and n-butane (R600)/propene
The overall compressor efficiency ηcomp is either set to a fixed (R1270). These pure fluids and their mixtures have frequently
value or calculated by a compressor model, as indicated in the been discussed in the literature for heat pumps and chill-
results section. As compressor model, we implemented a semi- ers.[13,14,25,26] Mixtures of alkanes and alkenes usually form zeo-
physical compressor model taken from our earlier work.[16] This tropic mixtures, while their vapor pressures and temperature
model for reciprocating compressors predicts overall and volu- glides match well with the requirements of typical heat pump
metric efficiencies depending on the inlet state, the outlet pres- and chiller applications. C3 and C4 alkanes/alkenes are particu-
sure, and the refrigerant. The model was specially developed for larly suited for the temperature range of the considered case
fluid extrapolation and can handle both pure fluids and mixtures. study on domestic heat pumps. With the chosen mixtures,
The fitting to a specific compressor is straightforward: just the our analysis thus covers a set of promising candidate mixtures
geometric specifications (e.g., bore and stroke) of the compressor for heat pumps. Several compositions in the entire range
and one set of overall and volumetric efficiencies for one opera- between the pure fluid states were investigated for the mix-
tion point with one refrigerant are needed. In previous work,[16] tures isobutane/propane and isobutane/propene. The mixture
the compressor model was validated with numerous experimen- n-butane/propene could only be investigated up to an n-butane
tal results obtained for different pure fluids and mixtures at vari- fraction of 80 mol% because higher fractions of n-butane lead to
ous operation points. Mean deviations of 3.0% for the pressures below the ambient pressure. We avoided subatmo-
predicted overall compressor efficiencies were calculated. The spheric pressures in the test rig due to the risk of forming an
model fit to the compressor installed in our test rig led to a rela- ignitable mixture by the intake of ambient air.
tive clearance volume of ccl ¼ 0.1 and a friction pressure of Figure 2a shows the temperature glide for the considered
pfr ¼ 80.12 kPa, which are the two fitting parameters of the com- mixtures as a function of mole fraction. The values refer to an
pressor model. The compressor model only requires usual ther- isobaric state change from dew to bubble point at the respective
mal and caloric state variables, which are either available from condensation temperatures (defined at dew point). The selected
fluid databases or can be calculated by equations of state for mixtures cover a wide range of temperature glides, which are
numerous substances. close to the temperature change in the condenser water flow
In this study, fluid properties are calculated either by using (10 K). However, the mixtures differ in the composition
Refprop Version 10[21] or by applying the Peng-Robinson dependence of the temperature glide. In particular, the
Figure 2. a) Theoretical temperature glide for the condenser and b) COP as function of mole fraction. Lines with symbols (b): measured results. Symbols refer
to measured points; estimated errors are below the symbol size. Solid line (b): model results for isobutane/propane, ηcomp ¼ ηcomp,propane, using Refprop.[21]
Energy Technol. 2021, 2000955 2000955 (4 of 10) © 2021 The Authors. Energy Technology published by Wiley-VCH GmbH
www.advancedsciencenews.com www.entechnol.de
mixture isobutane/propane has a broad range with similar achieve the predicted large COP increase by zeotropic mixtures
temperature glides. in practice?
Figure 2b shows COPs of the heat pump cycle as a function of The first step is to go back to the reason for using the temper-
the mole fraction of isobutane or n-butane. Both experimental ature glide of zeotropic mixtures in heat pumps: the temperature
results and simulations are shown. The simulation refers to glide can reduce the exergy losses in the heat exchangers and the
the model given in Section 2.3 while assuming a constant overall gap between the thermodynamic mean temperatures of evapora-
compressor efficiency of ηcomp ¼ 0.56 for all compositions. This tion TL and condensation TH. Both effects increase COP, in par-
overall compressor efficiency corresponds to the measured com- ticular, if the temperature glide of the mixture matches the
pressor efficiency of pure propane. temperature change in the heat transfer fluids. Figure 3 shows
The model results show the typical COP composition depen- the specific exergy losses in the evaporator (Figure 3a) and the
dency for zeotropic mixtures of simple thermodynamic cycle cal- ratio of both thermodynamic mean temperatures TH/TL
culations:[6,12] both pure fluids have similar COPs, while a (Figure 3b) derived from the experiments. The trends for the con-
specific composition shows a distinctive COP maximum. Such denser are similar to those of the evaporator. The experimental
results are often shown in the literature and are the basis of results show that zeotropic mixtures actually reduce exergy
the great expectations in using zeotropic mixtures. However, losses. As predicted by theory, the minima of the exergy losses
reality shows a different picture: the experimental results as well as of the temperature ratios are located at compositions
strongly differ from the simulation. All investigated mixtures with temperature glides that well match the heat source and sink
show a similar behavior: the COPs of the respective pure com- temperature regimes (Figure 2a). Thus, zeotropic mixtures
ponents of the mixture differ, while the maximum COP is still indeed reduce exergy losses and the gap between the thermody-
achieved for a certain composition. This maximum is only sig- namic mean temperatures. Still, the COP maxima are found for
nificant compared with the pure component with the lower COP, higher mole fractions of propane or propene (Figure 2b). Thus,
here isobutane or n-butane. In comparison, the maximum COP the COP composition dependency cannot solely depend on the
is just slightly better than the other pure component. Other pre- match of the temperature glide and the thermodynamic mean
vious studies observed similar findings.[13,14] Still, the experi- temperature ratios: there has to be another crucial physical effect.
mental results show that the COP increase due to the use of a
zeotropic mixture can be improved if the temperature glide 3.2. The Compressor is the Heart of the Matter
matches the temperature change in the heat transfer fluids.
As shown in Figure 2a, the temperature glides of the mixtures The compressor is the central part of a vapor-compression heat
isobutane/propane and isobutane/propene are smaller than the pump from an exergetic point of view. Although the compressor
temperature change in the condenser water flow of 10 K for all efficiency is known to strongly depend on fluid properties and on
compositions. The mixture n-butane/propene shows tempera- the operation point,[5,16] theoretical studies usually use constant
ture glides between 0 and 17.5 K. Due to the higher and, thus, efficiencies for all mixtures and operation points. Thus, the fluid
better fitting glide, n-butane/propene reaches the highest COP dependence of the compressor efficiencies is a promising candi-
and shows the largest COP increase in the mixture compared date to explain the large deviations between simulations and
with both pure fluids. experiments, as investigated later.
In summary, there is a great discrepancy between the Figure 4 shows the experimental results of the overall com-
expectations from theoretical studies and what is actually pressor efficiencies (Figure 4a) and the specific exergy losses
observed in experiments. The critical question is: how can we in the compressor and the evaporator (Figure 4b) for the three
Figure 3. a) Measured specific exergy loss in the evaporator and b) ratio of thermodynamic mean temperatures as function of mole fraction. Symbols
refer to measured points; estimated errors are below the symbol size.
Energy Technol. 2021, 2000955 2000955 (5 of 10) © 2021 The Authors. Energy Technology published by Wiley-VCH GmbH
www.advancedsciencenews.com www.entechnol.de
Figure 4. a) Overall compressor efficiency ηcomp and b) specific exergy loss in the compressor and the evaporator as function of mole fraction. Symbols
refer to measured points; estimated errors are below the symbol size.
investigated mixtures. The overall compressor efficiency is quite remain nearly constant over the composition range or should
similar for all mixtures but strongly depends on the composition. even increase for the mixtures. We expect that the exergy losses
The efficiencies of isobutane and n-butane are significantly in the heat exchangers then have a higher impact on the exergy
smaller than for propane and propene. The exergy losses in loss of the whole cycle leading to a stronger exergy loss minimum
the compressor show a reversed trend: compositions with small and COP maximum.
compressor efficiencies have high exergy losses and vice versa.
The compressor exergy losses are significantly larger than the
exergy losses from evaporation and depend more strongly on 4. Selection of Zeotropic Mixtures
composition (Figure 4b). Thus, the compressor efficiency dom-
inates the efficiency of the whole heat pump cycle. The previous analysis showed that fluid pairs are beneficial with
The typical experimental COP composition curves (Figure 2b) high compressor efficiencies over the whole composition range
can be explained by combining the results: if we start in the plots and a well-matching temperature glide. However, these charac-
at pure isobutane or n-butane and go to the COP maxima, the teristics are not helpful as criteria for mixture selection because
exergy losses decrease both in the evaporator and in the compres-
sor (Figure 4b), and thus the COP increases strongly (Figure 2b).
Now, we start at pure propane or propene, with increasing iso-
butane or n-butane fraction; the exergy loss in the evaporator
decreases again. However, the compressor exergy loss decreases
very slightly or remains nearly constant (Figure 4b). Thus, the
COP increases just slightly, leading to a flat maximum
(Figure 2b). Therefore, the behavior of the compressor is respon-
sible for the merely flat maxima and the shifts of the maxima
away from compositions with well-matching temperature glide,
as observed in experiments. To be more precise, compressor
behavior means the very different overall compressor efficiencies
for the pure components and the trends for mixtures.
This finding is also confirmed by simulations if the simple
process model of Figure 2b is enhanced by a compressor model
accounting for fluid-dependent overall compressor efficien-
cies[16] (Figure 5). With the compressor model, the calculated
COPs are all too large because further losses, such as pressure
losses, are neglected by the model. However, these losses only
lead to a constant offset, and thus the mixture behavior is fully
Figure 5. Coefficient of performance (COP) as function of isobutane mole
captured by the model. fraction. Blue circles: experimental results; solid black line: model results
As a result of the analysis, zeotropic mixtures should not only without compressor model; dashed red line: model results with compres-
show a temperature glide that matches the temperature change sor model; and dotted red line: model results with compressor model
in the heat transfer fluids. The compressor efficiency should also shifted to the COP of pure isobutane. Mixture: isobutane/propane.
Energy Technol. 2021, 2000955 2000955 (6 of 10) © 2021 The Authors. Energy Technology published by Wiley-VCH GmbH
www.advancedsciencenews.com www.entechnol.de
the decisive mixture properties are mostly unknown. However, work, we use the physical compressor model, as introduced in
the behavior of zeotropic mixtures is usually close to ideal mix- Section 2.3.
tures facilitating the selection of promising fluid pairs. To test the findings, vapor pressures were calculated for 336
For ideal mixtures, composition-dependent temperature fluids from different substance groups (cf. Section SI3,
glides can be calculated by Raoult’s law just depending on the Supporting Information). For all fluids with vapor pressures
pure component vapor pressures. Calculations with Raoult’s between 0.01 and 1 MPa at 0 C, the overall compressor
law show that the maximum temperature glide mainly depends efficiency was calculated by the physical compressor model
on the ratio of vapor pressures of the pure components (cf. Section 2.3). The selected pressure range refers to the pres-
(cf. Section SI2, Supporting Information). The temperature glide sure limits of the test rig. Following the case study, the overall
is 0 K for a ratio of 1 (equal vapor pressures) but increases with an compressor efficiencies are calculated for an evaporation tem-
increasing ratio of vapor pressures. The desired temperature perature of 0 C, a compressor inlet temperature of 17 C, and
glide in this case study of about 8–10 K, e.g., requires a vapor a condensation temperature of 33 C. The condensation tem-
pressure ratio from 3.0 to 3.5. As another characteristic of ideal perature is estimated to an average value because the conden-
mixtures, most state variables are linear functions of the compo- sation temperatures are, in reality, fluid dependent. Fluid
sition. Accordingly, linear composition dependency of compres- properties are modeled by the Peng-Robinson EOS property
sor efficiency promises to be a reasonable assumption for a first model (cf. Section 2.3). Cubic equations of state have disadvan-
fluid screening. Figure 4a supports this assumption for the mix- tages regarding the accuracy, in particular, for polar mole-
tures studied experimentally. Consequently, we expect that the cules. However, the availability of fluid data for numerous
compressor efficiencies are nearly constant with composition substances is a great advantage for screening. Figure 6 shows
if the compressor efficiencies of the pure fluids are similar. the calculated overall compressor efficiency as a function of
Therefore, the screening for promising fluid pairs just requires vapor pressures at 0 C for the 95 remaining fluids fulfilling
the calculation of compressor efficiencies of pure fluids. In this the pressure limits.
Figure 6. a) Calculated overall compressor efficiency ηcomp as function of vapor pressure at 0 C. Symbols refer to different substance groups. b) Zoom of
(a) of the interesting range ηcomp ≥ 0.50.
Energy Technol. 2021, 2000955 2000955 (7 of 10) © 2021 The Authors. Energy Technology published by Wiley-VCH GmbH
www.advancedsciencenews.com www.entechnol.de
Figure 6a shows the general trend of higher overall com- range of corresponding vapor pressures to achieve a temperature
pressor efficiencies for larger vapor pressures. The relation glide of 8–10 K (gray background).
between vapor pressure and overall compressor efficiency limits Apart from hydrocarbons, mixtures of propene with the fluo-
the opportunities of finding fluid pairs with different vapor pres- rinated substances within the range of 0.2 MPa ≤ psat ≤ 0.3 MPa
sures but similar overall compressor efficiencies. This trend may be attractive (Figure 6b). Usually, alkanes and alkenes form
applies, in particular, to the group of hydrocarbons (red circles). azeotropic mixtures with fluorinated hydrocarbons (e.g., pro-
As mostly mixtures of alkanes or alkenes were investigated in the pane/R32[27] or propane/R1243zf[28]), which can have higher
literature, hydrocarbons are of particular interest. If unstable temperature glides as expected for zeotropic mixtures with simi-
substances, such as ketene, 1,2-propadiene, and propyne, are lar vapor pressure ratios. Even though azeotropic mixtures might
excluded from the analysis, there is no fluid pair of hydrocarbons be promising, we follow previous studies and focus on zeotropic
with both similar compressor efficiencies and sufficient differ- mixtures.
ence in vapor pressures. The screening results suggest that, at For R32 (difluoromethane), a few fluids are identified
least within the normal operating conditions of heat pumps within the desired range of vapor pressure (magenta back-
and refrigeration cycles, mixtures of hydrocarbons are not ground in Figure 6b), which are mainly also fluorinated
expected to improve the COP significantly. Thus, the analysis hydrocarbons. The most promising fluid pair is R32 and
suggests that mixtures with at least one component from another R152a (1,1-difluoroethane) because both fluids are already
molecular family than hydrocarbons have to be investigated. established refrigerants which form zeotropic mixtures.[29]
Figure 6b shows the range of interest with higher compressor Due to their high GWPs (GWPR152a ¼ 677; GWPR32 ¼ 138)[30],
efficiencies (ηcomp ≥ 0.5), as shown in Figure 6, in more detail. R32, as well as mixtures of R32 and R152a, is not expected having
Mixtures of fluid pairs with overall compressor efficiencies below a great future as refrigerants. However, they are still approved,
this limit can achieve a significant COP increase compared with e.g., on the European continent, for heat pumps and chillers with
the pure components. However, the COP increase in these mix- a refrigerant charge smaller than 3 kg.[31] Mixtures of R32 and
tures resulting from the temperature glide is expected to be too R152a were already tested and shown to be feasible in a
small to beat the pure fluid with the best overall compressor effi- vapor-compression cycle.[32]
ciency. Thus, only mixtures of pure fluids with high overall com- To validate our criterion, we assessed various compositions of
pressor efficiencies are reasonable to achieve a benefit against R152a/R32 with the simulation model, as given in Section 2.3. The
pure fluids. However, the search for appropriate fluid pairs overall compressor efficiency ηcomp is calculated with the compres-
remains challenging, even if other substance groups are included sor model as a function of inlet state, outlet pressure, and mixture
in the analysis (Figure 6b). In particular, the small number of composition[16] (cf. Section 2.3). Fluid properties are calculated
fluids with high vapor pressures limits the possibilities. Here, using Refprop Version 10,[21] showing good agreement with exper-
just propene and R32 are suitable candidates. Ketene and cyano- imental data.[33] The process model has also been shown to repro-
gen fluoride are chemically unstable, carbonyl sulfide is toxic, the duce the composition dependence of the COP well (Figure 5).
use of R22 is prohibited, and propane already has a too-small For the considered mixture R152a/R32, the overall compres-
vapor pressure to find a partner leading to a zeotropic mixture sor efficiency changes only slightly with mole fraction of R152a
with the desired temperature glide between 8 and 10 K (Figure 7a). The temperature glide in the condenser, calculated
(psat ≤ 0.14 MPa). For propene, no fluid lies precisely in the for the composition-dependent condensation pressure, shows a
Figure 7. a) Overall compressor efficiency and condenser temperature glide, b) COP, all as function of mole fraction R152a. Mixture: R152a/R32.
Energy Technol. 2021, 2000955 2000955 (8 of 10) © 2021 The Authors. Energy Technology published by Wiley-VCH GmbH
www.advancedsciencenews.com www.entechnol.de
Energy Technol. 2021, 2000955 2000955 (9 of 10) © 2021 The Authors. Energy Technology published by Wiley-VCH GmbH
www.advancedsciencenews.com www.entechnol.de
Database (REFPROP) Version 10 – SRD 23, National Institute of [28] L. Ding, X. Yao, Y. Hou, Y. Zhao, X. Dong, M. Gong, J. Chem.
Standards and Technology, Gaithersburg, MD 2018. Thermodyn. 2020, 144, 106091.
[22] D.-Y. Peng, D. B. Robinson, Ind. Eng. Chem. Fund. 1976, 15, [29] B. G. Lee, J. Y. Park, J. S. Lim, S. Y. Cho, K. Y. Park, J. Chem. Eng. Data
59–64. 1999, 44, 190.
[23] R. C. Reid, J. M. Prausnitz, B. E. Poling, The Properties of Gases and [30] Intergovernmental Panel on Climate Change (IPCC), Working Group I
Liquids, McGraw-Hill, New York 1987. Contribution to the IPCC Fifth Assessment Report Climate Change 2013:
[24] VDI-Wärmeatlas. Mit 320 Tabellen, Springer Vieweg, Berlin 2013. The Physical Science Basis, IPCC, Geneva 2013.
[25] Q. Chen, L. Zhou, G. Yan, J. Yu, Int. J. Refrig. 2019, 104, 282. [31] Official J. Eur. Union, 2014, 150, 195.
[26] C. Fan, G. Yan, J. Yu, Appl. Thermal Eng. 2019, 150, 42. [32] H.-S. Lee, H.-J. Kim, D.-G. Kang, D. Jung, Energy 2012, 40, 100.
[27] S. Bobbo, L. Fedele, R. Camporese, R. Stryjek, Fluid Phase Equilib. [33] Z. Yang, M. Gong, X. Dong, X. Li, J. Wu, Fluid Phase Equilib. 2015,
2002, 199, 175. 394, 93.
Energy Technol. 2021, 2000955 2000955 (10 of 10) © 2021 The Authors. Energy Technology published by Wiley-VCH GmbH