Heat transfer performance for HFO-1234yf

Until now, only very limited information associated with the two-phase heat transfer characteristics as far as HFO-1234yf are concerned. The only published literatures regarding to the two-phase flow characteristics are from Moreno et al.[6]and Park and Jung[7]concerning nucleate boiling, Park et al.[8]for external condensation, Saitoh et al.[9], Li et al.[10]and Mortada et al.[11]

regarding to in-tube evaporation, Padilla et al.[12,13] in associa-tion with two-phase flow pattern, and Col et al. [14]and Wang et al. [15] for in-tube condensation. The following is a brief summary and discussion from the aforementioned results asso-ciated with R-134a counterpart.

2.1. Results for nucleate boiling

Moreno et al.[6]conducted pool boiling experiments for HFO-1234yf and R-134a at system pressures ranging from 0.7 to 1.7 MPa using horizontally oriented 1 cm²heat sources. The test surfaces include a plain and microporous surfaces. Test results for Ts¼40 and 60 1C are depicted in Fig. 2. And it shows that the boiling heat transfer coefficients of HFO-1234yf and R-134a are nearly identical at lower heat fluxes (qo200 kW m^2) while HFO-1234yf yielded lower heat transfer coefficients at higher heat fluxes and lower critical heat flux (CHF) as compared with R-134a.

It is often recognized that three mechanisms, namely bubble agitation, vapor-liquid change phenomenon, and evaporation are associated with basic mechanisms of the nucleate boiling heat transfer[16]. As shown inTable 1a [2,17], the HFO-1234yf has a higher reduced pressure at the same saturation temperature. This is because its critical pressure is about 17% lower than of R-134a.

In fact, at a saturation temperature of 40 1C the reduced pressure pⁿ is approximately 20% higher than that of R-134a, thereby Nomenclature g gravitational acceleration (m²s^1) H heat transfer coefficient (W m^2K^1)

Ps saturation pressure (Pa) Pr Prandtl number, dimensionless Pⁿ reduced pressure

Q heat transfer rate (W) q heat flux (W m^2) T temperature (1C)

Ts saturation temperature (1C) x vapor quality, dimensionless

X Lockhart Martinelli parameter, dimensionless z axial direction (m)

LG difference between liquid phase and vapor phase

i inside

220 240 260 280 300 320 340 360

P (kPa)

10 100 1000 10000

Carbon dioxide (CO₂) R-410A

R-407C R-22 HFO-1234yf R-134a

Fig. 1. Relationship between saturation pressure and saturation temperature for various refrigerants.

leading to a larger activation sites that would boost the heat transfer coefficient.

On the other hand, the smaller bubble departure diameter

 ^s

gðrLrGÞ

 0:5!

of HFO-1234yf implies a lower bubble agita-tion, and a smaller vapor–liquid change contribution which offset the positive contribution from the higher reduced pressure. As a result, an almost identical heat transfer coefficient amid R-134a and HFO-1234yf is seen when qo200 kW m^2. On the other hand, it appears that the heat transfer coefficient for R-134a gradually surpass those of HFO-1234yf when q is above 200 kW m^2. Moreno et al. [6] reported that the CHF for HFO-1234yf is appreciably lower than that of R-134a as shown in Fig. 2(c). Therefore they argued that at a higher heat flux (e.g., q4200 kW m^2) it is likely that the departure of HTC between R-134a and HFO-1234yf is mainly due to the local dryout of HFO-1234yf.

Analogous results are also applicable in microporous surface (Fig. 2(b)). However, the effect of saturation temperature on HTC for the baseline surface (smooth) and microporous surface is opposite. For the baseline surface (smooth surface), higher satura-tion temperature brings about higher heat transfer coefficient due to larger activation sites. On the other hand, the activation sites for microporous surface are mainly controlled by the artificial cavity, thereby lifting the positive contribution of cavity activa-tion, and a reversed influence of saturation temperature.

Park and Jung[7]also reported nucleate boiling heat transfer coefficients (HTCs) of R-134a and HFO-1234yf on a flat plain and low fin surfaces. All data were taken at the liquid pool tempera-ture of 7 1C on a small horizontal square copper plate (9.53  9.53 mm) at heat fluxes from 10 to 200 kW m^2with an interval of 10 kW m^2. Test results show that the nucleate boiling HTCs of HFO-1234yf are very similar to those of R-134a for two surfaces tested as depicted inFig. 3(a). And Park and Jung[7]also found that the conventional boiling correlations can be used for the design of evaporators and boilers with HFO-1234yf. Notice that the maximum test range for their experiments fall within the

‘‘identical range’’ where no appreciable distinction is observed as reported from Moreno et al.[6]. The present study had compared the nucleate boiling HTC for R-134a and HFO-1234yf using the well-known Cooper correlation[18]as shown in Fig. 3(b). The results suggest that the nucleate boiling HTC of HFO-1234yf is marginally higher than that of R-134a.

2.2. Results of outside condensation

Park et al. [8] conducted experiments concerning external condensation experiments for plain, low fin, and Turbo-C tubes at a saturated vapor temperature of 39 1C with the wall subcool-ing rangsubcool-ing from 3 to 8 1C. The geometry of the test tubes is shown inTable 2. Test results show that the condensation HTCs of HFO-1234yf are very similar to those of R-134a for all three surfaces tested as shown inFig. 4(a). At first glance, it seems that the condensation HTCs for HFO-1234yf is also identical to that of R-134 as those shown in nucleate boiling. However, the authors Fig. 2. Nucleate boiling HTCs of HFC-134a and HFO-1234yf for smooth and

microporous surfaces (from Moreno et al.[6]). (a) Nucleate boiling HTC vs. q for smooth surface. (b) Nucleate boiling HTC vs. q for microporous surface. (c) Ratio of critical heat flux for smooth and microporous surface.

Table 1a

Fundamental constants of HFO-1234yf.

Molecular weight ( g mol^1)

Critical temperature (K)

Critical pressure (Mpa)

R-134a 102 374.13 4.07

R-1234yf 114.042 367.85 3.382

also made comparisons (smooth tube) with their condensation HTCs against the well accepted Nusselt’s equation:

hc¼0:728 

r

r

r

m

" #1=4

ð1Þ

Park et al. [8] had slightly modified the original Nusselt equation based on their previous study[19], as

hpredict¼0:79 

r

r

r

m

" #1=4

ð2Þ

Their comparison (using Eq.(2)) revealed that the measured data for R-134a and HFO-1234yf were 9.0% and 27.1% larger than the predicted values. They argued that the relatively large devia-tion associated with HFO-1234yf were from the large uncertain-ties of various properuncertain-ties of HFO-1234yf. A similar calculation was made by the present author using Eq. (1) for plain tube with do¼19.05 mm and Tw¼20 1C. The calculated results inFig. 4(b) show that the condensation HTCs for R-134a are much higher than that of HFO-1234yf (around 30–60%). It is not totally clear why the tested condensation HTC for R-134a and HFO-1234yf[8]

are comparable for all the test tubes at the same condensation temperature. One of the possible reasons may be associated the large uncertainty of their measurements. This is because their test tube is quite short and the acquired heat transfer rate (tempera-ture difference subtracted from the inlet and outlet of cooling water) and temperature difference between the surface and saturation temperature (Ts–Tw) are comparatively small. This is especially pronounced when enhanced tubes (low fin and turbo C) were used. In addition to the uncertainty, another possible explana-tion may be attributed to their relative short test length (L¼ 290 mm) which may cause some end effect (lateral conduction from the test tube to the flange) that inevitably promotes condensation. Note that most of properties influencing the condensation HTC suggest a lower condensation HTC of HFO-1234yf.

2.3. Results of in-tube evaporation

Saitoh et al.[9]conducted study for boiling heat transfer of the refrigerant HFO-1234yf flowing in a smooth small-diameter Table 1b

Thermodynamic and transport properties of HFC-1234yf.

T

0 R-134a 292.8 1295 14.43 271.1 10.73 0.092 0.01151 198.6 0.01156 1.341 0.0897

R-1234yf 315 1175 17.17 220 11.44 0.0746 0.0091 162.3 0.0093 1.259 0.933

5 R-134a 350 1278 17.14 254.4 10.94 0.0898 0.01195 194.8 0.01085 1.355 0.921

R-1234yf 372 1160 20.8 206 11.67 0.073 0.0094 159 0.00868 1.275 0.957

10 R-134a 414.6 1261 20.23 238.8 11.15 0.0876 0.0124 190.7 0.01014 1.37 0.946

R-1234yf 436 1144 24.4 194 11.9 0.0713 0.0098 155.6 0.0081 1.293 0.983

20 R-134a 571.7 1225 27.78 210.7 11.58 0.0833 0.01333 182.2 0.00876 1.405 1.001

R-1234yf 590 1111 33 171 12.36 0.0672 0.0106 148.3 0.0067 1.332 1.041

30 R-134a 770.2 1187 37.54 185.8 12.04 0.079 0.01433 173.1 0.00742 1.446 1.065

R-1234yf 782 1075 44 152 12.86 0.0631 0.01143 140.1 0.00563 1.379 1.11

40 R-134a 1017 1147 50.09 163.4 12.55 0.0747 0.01544 163 0.0061 1.498 1.145

R-1234yf 1017 1037 58.3 134 13.49 0.0586 0.0123 131.1 0.00462 1.437 1.196

50 R-134a 1318 1102 66.27 143.1 13.12 0.0704 0.01672 151.8 0.0048 1.566 1.246

R-1234yf 1301 993.3 76.7 118 14.12 0.054 0.01326 120.9 0.0035 1.515 1.31

5 10 20 50 100 200 500

Fig. 3. Nucleate boiling HTCs of HFC-134a and HFO-1234yf on flat surfaces.

(a) Nucleate boiling HTCs of HFC-134a and HFO-1234yf on two flat copper surfaces simulating a plain or low fin tube, respectively,[7]. (b) Comparison of the calculated nucleate boiling HTCs of HFC-134a and HFO-1234yf on flat plate using the Cooper correlation.

Table 2

Specification of the enhanced tube tested[8].

Tube type Outside

Low fin 18.90 1.214 0.252 0.576 1024

Turbo-C 18.90 0.760 0.250 0.350 1654

horizontal tube (inner diameter (ID): 2 mm) and Li et al.[10]used similar test facility and identical test tube for comparing the HTC amid HFC-32 and HFO-1234yf. The test tube was heated by direct electrification using a DC power supply connected to two electro-des soldered at the flanges of the two ends of the test tube. Their experimental conditions are Ts¼15 1C, q¼6–24 kW m^2, and G ¼100–400 kg m^2s^1.Fig. 5(a) shows the variation in the heat transfer coefficient against the vapor quality. The mass flux was kept at 200 kg m^2s^1; the measured results are for three different heat fluxes: 6, 12, and 24 kW m^2, respectively. At the lowest heat flux of 6 kW m^2, the measured heat transfer coefficient increased with the vapor quality, showing that the convective heat transfer intensifies with increasing quality. The dryout quality was about 0.8 and did not change with heat flux.

Increasing the heat flux from 6 to 12 and 24 kW m^2showed that the heat transfer coefficient increases with heat flux at low vapor quality; thus, nucleate boiling is the dominant heat transfer coefficient mechanism at low vapor quality. On the other hand, the detectable rise of HTC (heat transfer coefficient) vs. vapor quality for a low heat flux of 6 kW m^2 is associated with the change of flow pattern since annular flow may prevail at high

quality region. However, as claimed by the authors who argues that nucleate boiling is dominant heat transfer process when q¼12 and 24 kW m^2, thereby showing a moderate change of HTC as vapor quality is increased. This seems feasible but the relative effect of heat flux, based on the test results of Saitoh et al.

[9], is in fact much lower. A rough estimation of the heat flux dependency is about h q^0.42which is generally much lower than the pure nucleate boiling where h q^06–0.7. In this sense, it is expected that convective evaporation still plays certain role rather than pure nucleate boiling.

Fig. 5(b) shows the effect of mass flux on the boiling heat transfer at a heat flux of 12 kW m^-2. The dryout occurs at a vapor quality of 0.8 for all the conditions. In the high quality region (40.4), the heat transfer coefficients at both mass fluxes (200 and 400 kg m^2s^1) increased with the increasing vapor quality, and the heat transfer coefficient was higher at 400 kg m^2s^1than at 200 kg m^2s^1. At a mass flux of 100 kg m^2s^1, the effect of vapor quality on the heat transfer coefficient was weak. The results suggest that in the high vapor quality region, forced convective evaporation is dominant. In the lower quality region, xo0.4, the HTC is rather insensitive to change of mass flux, indicating a nucleate boiling dominant regime.Fig. 5(c) depicts a comparison between the boiling heat transfer performances of HFO-1234yf and R-134a at a mass flux of 300 kg m^2s^1and a heat flux of 12 kW m^2. The figure shows that in the wide vapor quality region, the difference between the heat transfer coeffi-cients of HFO-1234yf and R-134a is small, Saitoh et al. [9]

attributed this to the small differences in their thermodynamic properties. In addition to this possible explanation, as explained earlier, the contribution of nucleate boiling for both refrigerants is about the same when q is small. Moreover, it will be shown in subsequent section that the flow patterns[12]for both fluids are virtually similar, thereby resulting in a comparable convective boiling performance. The present author also made a calculation of the convective boiling HTC between R-134a and HFO-1234yf at a saturation temperature of 10 1C with d_i¼10 mm, q¼20–

40 kW m^2 using the well-known Chen correlation [20]. The calculated results indicated that the difference between R-134a and HFO-1234yf is very small. Mortada et al.[11]performed an experiment for HFO-1234yf and R-134a in a 1.1 mm rectangular channel with rather small mass flux of 20–100 kg m^2s^1 and heat flux from 2–15 kW m^2. However, their results showed that the HTC for HFO-1234yf is lower than that of R-134a as much as 40% and convective boiling is the major heat transfer mechanism even at this mini-size tube. Notice that nucleate boiling is the major heat transfer mechanism in most published works in mini-size or micro-mini-size tubes. The results are contradictory to previous results. It is not totally clear why the test results of Mortada et al.

[11]showed a different trend.

Saitoh et al.[9]also measured two-phase pressure drop (DP).

The pressure drops were compared with the Lockhart Martinelli correlation, which defines the pressure drop as:

 dP

wherefLandfGare the two-phase multipliers in the liquid and gas phases, respectively. The multipliers are defined asfL¼(1þC/

Xþ1/X²)^0.5 and fG¼(1þ CXþ X²)^0.5, where X is the Lockhart Martinelli parameter, which is the square root of the ratio between the pressure drop assuming liquid flow alone and assuming gas flow alone. When the liquid and gas phases are turbulent, C ¼20; when the liquid phase is laminar and the gas phase is turbulent, C ¼12.Fig. 5(d) shows the measured pressure drops and those predictions using the Lockhart Martinelli correla-tion. The measured pressure drops agreed well with the Lockhart Martinelli correlation.

Fig. 4. Condensation HTC for R-134a and HFO-1234yf. (a) External condensation HTCs of R-134a and HFO-1234yf on various tubes[8]. (b) Comparison of the calculated condensation HTCs of R-134a and HFO-1234yf using Nusselt equation for plain tube.

Despite the Lockhart Martinnlli correlation seems to give very excellent prediction of the pressure drop upon Saitoh et al.’s mea-surement. It should be mentioned that the results are applicable only for a 2-mm tube. Padilla et al.[12]performed flow visualization as well as pressure drop measurement for three working fluids—

R-410A, R-134a, and HFO-1234yf. The tube diameter (D) varies from 7.90 to 10.85 mm. The mass velocity ranges from 187 to 1702 kg m^2s^1 and the saturation temperatures from 4.8 to 20.7 1C. For a saturation temperature of 10 1C and D¼6.7 mm, the corresponding flow pattern for G¼ 300 kg m^2s^1 and G¼ 500 kg m^2s^1 subject to vapor quality for R-134a and HFO-1234yf are depicted in Figs. 6 and 7. The different flow regimes observed are: slug, intermittent and annular flows. The observed flow patterns, for the same vapor quality, saturation temperature, mass flux, and tube diameter, are virtually the same amid R-134a and HFO-1234yf. This is desirable since HFO-1234yf was designed so that not only its properties would be close to those of R-134a but also equipped with the same flow phenomena. Analogous flow patterns amid R-134a and HFO-1234yf in a horizontal return bend was also reported by Padilla et al.[13].

Their measured results of pressure drop are also compared against 10 well-known two-phase frictional pressure drop prediction meth-ods. However, unlike that of Saitoh et al. [9] who reported the Lockhart Martinnlli correlation shows excellent predictive ability, Padilla et al.[13] shows that the predictive ability of the Lockhart Martinnlli correlation is actually among the poorest one. Padilla et al.

[12] showed that the M ¨uller-Steinhagen and Heck correlation[21]

gives the best predictive capability. The mean absolute error is near 19% and the mean relative error is around73%. Typical comparison is shown in Fig. 8. Although the method proposed by M ¨uller-Steinhagen and Heck[21] is a method which has been developed without considering flow pattern effects on the process, this method gives the best prediction for intermittent and annular flows, but also for their entire database.

2.4. In-tube condensation

Col et al.[14]conducted experiments for measurement of local heat transfer coefficients during condensation of HFO-1234yf within a single circular 0.96 mm diameter microchannel at Fig. 5. In-tube convective boiling heat transfer coefficients and pressure drop data of Saitoh et al.[9]. (a) Effect of heat flux on local heat transfer coefficient for HFO-1234yf. (b) Effect of mass flux on local heat transfer coefficient for HFO-HFO-1234yf. (c) Comparison of heat transfer coefficients between HFO-1234yf and R-134a at mass flux of 300 kg m^2s and at a heat flux of 12 kW m^2. (d) Comparison of pressure drops between measured values and values calculated by the Lockhart Martinelli correlation for HFO-1234yf.

40 1C saturation temperature and compares them to the ones of R-134a. Condensation tests are carried out at mass fluxes ranging between 200 and 1000 kg m^2s^1. They reported sufficient heat

transfer deteriorations for the HFO-1234yf as compared to the ones of R-134a. InFig. 9the local heat transfer coefficients and at three different mass velocities are compared to the ones of Fig. 6. Top and side views of the R-134a flow patterns for Ts¼10 1C and D ¼6.70 mm[12]. (a) Slug flow, x ¼0.05, G ¼ 300 kg m^2s^1. (b) Intermittent flow, x¼ 0.05, G ¼300 kg m^2s^1. (c) Annular flow, x ¼0.6, G ¼ 300 kg m^2s^1. (d) Intermittent flow, x ¼ 0.05, G ¼ 500 kg m^2s^1. (e) Intermittent flow, x ¼0.2, G ¼ 500 kg m^2s^1. (f) Annular flow, x¼ 0.6, G ¼ 500 kg m^2s^1.

Fig. 7. Top and side views of the HFO-1234yf flow patterns for Ts¼10 1C and D ¼ 6.70 mm[12]. (a) Slug flow, x¼ 0.05, G ¼300 kg m^2s^1. (b) Intermittent flow, x¼ 0.05, G ¼300 kg m^2s^1. (c) Annular flow, x ¼0.6, G ¼ 300 kg m^2s^1. (d) Intermittent flow, x ¼ 0.05, G ¼ 500 kg m^2s^1. (e) Intermittent flow, x ¼0.2, G ¼ 500 kg m^2s^1. (f) Annular flow, x¼ 0.6, G ¼ 500 kg m^2s^1.

R-134a at the same operating conditions. Except for the lowest values of vapor quality, R-134a displays a heat transfer coefficient higher than HFO-1234yf for all three values of mass velocities. At a mass flux of 200 kg m^2s^1 (Fig. 9(a)), the heat transfer coefficient of HFO-1234yf is lower than that of R-134a by 15%

at 0.4 vapor quality and by 30% at 0.8 vapor quality. One of the explanations about the pronounced difference in HTC between HFO-1234yf and R-134a at a higher vapor quality region is due to film thickness on the periphery. Notice that at higher vapor quality regime the annular flow prevails. On the other hand, the liquid density for R-134a is about 11% higher than that of HFO-1234yf (seeTable 1b). This implies a thinner film thickness of R-134a provided the vapor quality is the same, thereby leading to a higher HTC for R-134a.

A similar trend is found at 400 and 800 kg m^2s^1 mass velocity as shown inFig. 9(b) and (c). When comparing the heat transfer coefficient of HFO-1234yf to the one measured for R-134a, one can see that the latter fluid displays a higher coefficient at the same operating conditions, and this is related to the different properties of the two fluids. Additionally, as aforementioned in the nucleate boiling section, the higher reduced pressure of HFO-1234yf also contributed to decrease the associated condensation heat transfer coefficient, this can be easily seen from the well-known Shah correlation[22]where:

hc¼hL 1þ3:8

A similar calculation for comparing the in-tube condensation HTC for R-134a and HFO-1234yf using the Shah correlation is also made. And it clearly substantiated the measured results of Col et al.

[14]who shows the HTC of HFO-1234yf is inferior to that of R-134a.

Test results by Wang et al.[15]also unveiled similar trend.

Fig. 9(d) shows the total experimental pressure drop measured with HFO-1234yf and R-134a at 40 1C saturation temperature with three different values of mass velocity: 400, 600 and 800 kg m^2s^1. It can be seen that the fluid HFO-1234yf displays a lower pressure drop by 10–12% as compared to that of R-134a, at the same operating conditions. This may be easily understood since the reduced pressure of HFO-1234yf is 20% greater than that of R-134a at 40 1C saturation temperature (seeTable 1a and 1b). A similar result was also reported by Park et al.[23]who condensed R-1234ze(E) within a multi-port MAC condenser with an internal hydraulic diameter of 1.45 mm. They reported that the heat transfer performance of HFO-1234ze(E) was about 15–25% lower than for R-134a. However, experimental flow boiling heat transfer results with R1234ze(E) by Tibirica et al.[24]for horizontally test tubes having 1.0 and 2.2 mm I.D. (internal diameter) stainless steel tubes shows a comparable performance amid HFO-1234ze(E) and R-134a. The reason is analogous to the aforemen-tioned discussion in nucleate boiling. Pressure drop measure-ments for R-134a and HFO-1234yf by Padilla et al. [13] also depicted analogous results.