Ecole Polytechnique Federale de Lausanne: Laboratoire de Transfert de Chaleur Et de Masse
Ecole Polytechnique Federale de Lausanne: Laboratoire de Transfert de Chaleur Et de Masse
Ecole Polytechnique Federale de Lausanne: Laboratoire de Transfert de Chaleur Et de Masse
Pool boiling curve. Heat transfer mechanisms in pool boiling. Nucleate pool boiling correlations. Departure from nucleate boiling (DNB). Nucleate boiling of zeotropic mixtures. Enhanced nucleate pool boiling. Boiling on horizontal tube bundles.
Figure 9.5
1000
dbfig 9.5
Figure 9.6
IB - Incipience of Boiling DNB - Departure from Nucleate Boiling MFB - Minimum Film Boiling DNB
10 MFB
IB
Natural convection 1 1 10 Wall Superheat (K) 100 1000
Figure 9.7
Intense Bubble agitation mechanism. convection in the liquid adjacent to the heated wall is induced by the lateral pumping motion of growing and departing bubbles, which transforms the natural convection process into a localized forced convection process. Vapor-liquid exchange mechanism. Transient conduction from the hot wall into the liquid forms a superheated layer on the surface and its removal by departing bubbles gives rise to a cyclic thermal boundary layer stripping phenomenon. Evaporation mechanism. Vapor bubbles grow in the superheated layer formed on the heated surface. Macro-evaporation occurs over the top of the bubble surrounded by the thermal boundary layer while micro-evaporation occurs underneath the bubble across the thin layer of liquid trapped between a rapidly growing bubble and the surface, referred to as microlayer evaporation.
Rohsenow (1962) proposed one of the first nucleate pool boiling correlations, based on the bubble agitation mechanism and formulated as a single phase forced convection correlation:
Nu = C1 Re x Pr y
[9.3.1]
The Nusselt number for boiling was defined as follows where the bracketed term is the bubble departure diameter, designated as the characteristic length:
Nu =
nb ( k L g L G )
1/ 2
[9.3.2]
In this expression, nb is the nucleate pool boiling heat transfer coefficient. His Reynolds number was defined in terms of the superficial velocity of the liquid as
Re =
q h LG L g( L G )
1/ 2
L L
[9.3.3]
where hLG is the latent heat and L is the liquid dynamic viscosity.
Laboratoire de Transfert de Chaleur et de Masse
An empirical constant Csf was introduced to account for the influence of the liquid-surface combination observed in his database as follows:
Nu =
1 Re(1 n ) Pr m Csf
[9.3.4]
[9.3.5]
The specific heat of the liquid is cpL and PrL is the liquid Prandtl number; nb is obtained from the definition of the heat transfer coefficient, that is nb = q/T. The values of the exponents are m=0.7 and n=0.33 (equivalent to q T3) for all fluids except for water for which Rohsenow recommended setting m=0. Values of Csf for various surface-fluid combinations of Rohsenow and additional values proposed by Vachon, Nix and Tangor (1967) are listed in Table 9.1. This method is now mostly of historical value, pointing to the importance of the micro-topology of the boiling surface on nucleate boiling heat transfer.
Laboratoire de Transfert de Chaleur et de Masse
Mostinski (1963) ignored surface effects and applied the principle of corresponding states to nucleate pool boiling heat transfer, correlating data as a function of the reduced pressure of the fluid pr and its critical pressure pcrit. His dimensional reduced pressure correlation gives nb in W/m2K as:
[9.3.6]
The correlation must be used with q in W/m2 and pcrit in kN/m2 (i.e. in kPa, not in N/m2). FP is a non-dimensional pressure correction factor that characterizes pressure effects on nucleate boiling as
[9.3.7]
This correlation gives reasonable results for a wide range of fluids and reduced pressures.
Stephan and Abdelsalam (1980) proposed four specific correlations applying a statistical multiple regression technique to the following fluid classes: water, organics, refrigerants and cryogens. Their organic fluid correlation is the most widely quoted and it is given as:
nb d bub = 0.0546 G kL L
1/ 2
qd bub k T L sat
0.67
h LG d 2 bub a2 L
0.248
L G L
4.33
[9.3.8]
The expression to the left of the equal sign is a Nusselt number and their bubble departure diameter dbub is obtained from
2 d bub = 0.0146 g( L G )
1/ 2
[9.3.9]
The contact angle is assigned a fixed value of 35 irrespective of the fluid, Tsat is the saturation temperature of the fluid in K, and aL is the liquid thermal diffusivity.
Cooper (1984) proposed a new reduced pressure form of pool boiling correlation including the surface roughness of the boiling surface as a variable:
nb = 55p r
0.12 0.4343 ln R p
[9.3.10]
This is a dimensional correlation in which nb is in W/m2K, the heat flux q is in W/m2, M is the molecular weight of the fluid and Rp is the surface roughness in m. When Rp is unknown, it is set to 1.0 m. He recommended multiplying the above heat transfer coefficient by 1.7 for horizontal copper cylinders; however, the correlation seems to be more accurate for boiling of refrigerants on copper tubes without this correction and that is the approach recommended here. The Cooper correlation covers reduced pressures from about 0.001 to 0.9 and molecular weights from 2 to 200.
Gorenflo (1993) proposed a fluid specific reduced pressure correlation and included the effect of surface roughness. His method uses a reference heat transfer coefficient, o, specified for each fluid at the following fixed reference conditions of pro=0.1, Rpo=0.4 m and qo=20,000 W/m2. His values of o are listed in Table 9.2 for selected fluids. The nucleate boiling heat transfer coefficient nb at other conditions of pressure, heat flux and roughness is then calculated relative to the reference heat transfer coefficient using the following expression:
nb = o FPF (q / q 0 )nf R p / R po
His pressure correction factor FPF is
)0.133
FPF = 1.2p 0.27 + 2.5p r + r pr 1 pr
[9.3.11]
[9.3.12]
The effect of reduced pressure on his exponent nf for the heat flux term is given by:
[9.3.13]
Its value decreases with increasing reduced pressure, which is typical of experimental data. The surface roughness is Rp in m and is set to 0.4 m when unknown.
Laboratoire de Transfert de Chaleur et de Masse
The above method is for all fluids except water and helium; for water the corresponding equations are:
[9.3.14]
[9.3.15]
This method is applicable over the reduced pressure range from about 0.0005 to 0.95. For fluids not listed, experimental values can be input at the reference conditions, or another correlation can be used to estimate o. For fluids on the list, this method gives accurate results over a very wide range of heat flux and pressure and is probably to most reliable of those presented.
Fluid
Methane Ethane Propane n-Butane n-Pentane i-Pentane n-Hexane n-Heptane Benzene Toluene Diphenyl Ethanol n-Propanol i-propanol n-Butanol i-Butanol Acetone R-11 R-12 R-13 R-13B1 R-22 R-23 R-113 R-114 R-115 R-123 R-134a R-152a R-226 R-227 RC318 R-502 Chloromethane Tetrafluoromethane Hydrogen (on Cu) Neon (on Cu) Nitrogen (on Cu) Nitrogen (on Pt) Argon (on Cu) Argon (on Pt) Oxygen (on Cu) Oxygen (on Pt) Water Ammonia Carbon Dioxide * Sulfur Hexafluoride * At triple point.
pcrit
46.0 48.8 42.4 38.0 33.7 33.3 29.7 27.3 48.9 41.1 38.5 63.8 51.7 47.6 49.6 43.0 47.0 44.0 41.6 38.6 39.8 49.9 48.7 34.1 32.6 31.3 36.7 40.6 45.2 30.6 29.3 28.0 40.8 66.8 37.4 12.97 26.5 34.0 34.0 49.0 49.0 50.5 50.5 220.6 113.0 73.8 37.6
M
16.04 30.07 44.10 58.12 72.15 72.15 86.18 100.2 78.11 92.14 154.2 46.07 60.10 60.10 74.12 74.12 58.08 137.4 120.9 104.5 148.9 86.47 70.02 187.4 170.9 154.5 152.9 102.0 66.05 186.5 170.0 200.0 111.6 50.49 88.00 2.02 20.18 28.02 28.02 39.95 39.95 32.00 32.00 18.02 17.03 44.01 146.1
o
7000 4500 4000 3600 3400 2500 3300 3200 2750 2650 2100 4400 3800 3000 2600 4500 3950 2800 4000 3900 3500 3900 4400 2650 2800 4200 2600 4500 4000 3700 3800 4200 3300 4400 4750 24000 20000 10000 7000 8200 6700 9500 7200 5600 7000 5100 3700
Table 9.2. References values of Gorenflo (1993) with o in W/m2K at pro = 0.1, qo = 20,000 W/m2 and Rpo = 0.4 m with pcrit in bar.
Example Calculation: Determine the nucleate pool boiling heat transfer coefficient for n-pentane boiling on a polished copper surface (1.0 m roughness) at a pressure of 1.01 bar and a heat flux of 30 kW/m2 using the Gorenflo method. Solution: From Table 9.2, we see that o = 3400 W/m2K, pcrit = 33.7 bar and M = 72.15. Thus, pr = psat/pcrit= 1.01/33.7 = 0.030 and thus
The maximum heat flux attainable in the nucleate pool boiling regime of the pool boiling curve illustrated in Figure 9.5 is the DNB point. This heat flux is called qDNB and is also often referred to as the critical heat flux. The maximum in heat transfer rate occurs at the point of onset of a hydrodynamic instability occurring close to the heater surface, which was first explained by Zuber (1959) to be governed by the Taylor and Helmholtz instabilities. His model has since been refined by Lienhard and Dhir (1973) for an infinite surface. The Taylor instability governs the collapse of an infinite, horizontal planar interface of liquid above a vapor or gas. The Taylor wavelength is that which predominates at the interface during such a collapse. In the present case at the DNB, vapor jets formed above a large, flat horizontal heater surface occur at spacings corresponding to the wavelength of the Taylor instability. The Helmholtz instability, instead, describes the point at which a planar liquid interface goes unstable when the velocity of a vapor or gas flowing parallel to their interface reaches some critical value. Essentially, a small perturbation of the interface creates a low pressure zone on the convex side and a high pressure zone on the concave side. The imbalance of the opposing pressure forces acting on the vapor-liquid interface induces an instability, which is attenuated by interfacial surface tension.
According to Zuber, it is the liquid interface of the rising vapor jets that go unstable. The resulting expression for qDNB from such an analysis is:
q DNB = G h LG
2 1 G 2 3
g( L G ) 16
[9.3.16]
which is equivalent to
[9.3.17]
This equation is valid for flat infinite heaters facing upwards, by Lienhard and Dhir (1973), who observed good agreement as long as the diameters or widths of the heaters were sufficiently large. Kutateladze (1948) arrived at nearly the same expression using dimensional analysis:
q DNB = C1/ 2 h LG 4 g( L G ) G
[9.3.18]
His empirical factor C was set to 0.131 based on comparison to experimental data. Zubers analysis yielded a value of C = /24 = 0.1309, which is nearly identical to Kutateladzes value. The Lienhard and Dhir solution for an infinite flat surface facing upwards gives C = 0.149, which is 15% higher. The recommended expression for large surfaces is C = 0.1309 while for tubes this value is corrected by a factor of 0.9.
Laboratoire de Transfert de Chaleur et de Masse
Example Calculation: Determine the value of the heat flux at the DNB for water at 1.01 bar on a horizontal tube where the required fluid physical properties are: L = 958.25 kg/m3; G = 0.6 kg/m3; hLG = 2256120 J/kg; = 0.05878 N/m. Solution: The value of qDNB is first calculated and then that for the tube applying a correction of 0.9 as follows:
1/ 2
Therefore, the heat flux at DNB for the tube is 998100 W/m2 (or 998.1 kW/m2).
Nucleate pool boiling of zeotropic mixtures is similar to boiling of pure fluids and azeotropic mixtures, except for two additional complications: mass transfer effects and estimating the mixture physical properties or the mixture critical pressure. Mass transfer occurs in evaporation of zeotropic mixtures since they have different compositions in their liquid and vapor phases, but not azeotropic mixtures since they have the same composition in each phase. Mass transfer tends to reduce nucleate boiling heat transfer coefficients and, in some cases, may reduce the value of the heat transfer coefficient by up to 90%. Detailed reviews of mixture boiling are given by Thome and Shock (1984) and by Collier and Thome (1994). The mass transfer effect on bubble growth can be explained in simple terms as follows. Since the equilibrium composition of the more volatile component is larger in the vapor phase than in the liquid phase, the more volatile component preferentially evaporates at the bubble interface, which in turn reduces its composition there and induces the formation of a diffusion layer in the liquid surrounding the bubble. The partial depletion of the more volatile component at the interface increases that of the less volatile component, which increases the bubble point temperature at the interface. This incremental rise in the local bubble point temperature can be denoted as . Hence, to evaporate at the same rate as in a pure fluid, a larger superheat is required for a mixture.
The effect of mass transfer on nucleate pool boiling heat transfer can therefore be explained by introducing the parameter , which represents the increase in the bubble point temperature at the surface due to preferential evaporation of the more volatile component. At a given heat flux, the boiling superheat of the mixture is TI+ while that for an ideal fluid with the same physical properties as the mixture is TI. Thus, the ratio of the mixture boiling heat transfer coefficient nb to that of the ideal heat transfer coefficient nb,I at the same heat flux is:
TI nb = nb , I TI +
[9.4.1]
The value of TI is the wall superheat that corresponds to nb,I, which is determined for instance using the Cooper correlation with the molecular weight and critical pressure of the mixture. Hence, as the value of increases, the ratio nb/nb,I decreases, which means that a larger wall superheat is required in a mixture to transfer the same heat flux. As exploited in an early mixture boiling prediction by Thome (1983), the maximum value of is the boiling range of the mixture bp, which is equal to the difference between the dew point and the bubble point temperatures at the composition of the liquid [bp is also referred to as the temperature glide for refrigerant mixtures]. In fact, the actual value of varies from zero at the inception of boiling, since no mass transfer occurs until evaporation takes place, up to bp at the DNB, where all the liquid is assumed to be converted to vapor.
Laboratoire de Transfert de Chaleur et de Masse
Starting from a mass transfer balance around an evaporating bubble and simplifying with an approximate slope of the bubble point curve, the following expression was obtained to predict heat transfer in the boiling of mixtures
nb nb ,I q = 1 + bp 1 exp nb ,I L h LG mL q
[9.4.2]
where mL is the mass transfer coefficient in the liquid (set to a fixed value of 0.0003 m/s). The value of nb,I is determined with one of the pure fluid correlations presented earlier (with the exception of the Gorenflo method that is not adapted to this). This method was proposed in 1985 and later published by Thome (1989). It is applicable to organic, refrigerant, aqueous, hydrocarbon, and cryogenic multi-component mixtures (i.e. with two or more components) for boiling ranges up to 30 K.
10
Example Calculation: Assuming an ideal heat transfer coefficient of 3000 W/m2K, determine the mixture boiling coefficient at 50 kW/m2 for a mixture with a boiling range of 15 K, a liquid density of 700 kg/m3 and a latent heat of 300000 J/kg. Solution: Substituting values into the above expression,
nb 3000 50000 = 0.670 = 1 + (15)1 exp 3000 50000 700(300000)(0.0003) nb = 0.67(3000 ) = 2009 W m 2 K
Thus, nb for the mixture is 2009 W/m2K, which is 33% lower than that of the equivalent pure fluid.
Figure 9.1
Laboratoire de Transfert de Chaleur et de Masse
11
Figure 9.2 depicts a photograph of a Turbo-Chil low finned tube made by Wolverine Tube Inc.
Figure 9.2
By mechanically deforming low finned tubes, it is possible for create a high density of reentrant channels and pores on a tubes surface, which increases heat transfer performance significantly. Figure 9.3 shows several diagrams of the Turbo-B tube made by Wolverine Tube Inc. that is widely used in refrigerant flooded evaporators, showing its external enhanced boiling geometry and its internal helical fins for augmenting heat transfer to chilled water.
Figure 9.3
12
Figure 9.8
Heat Transfer Mechanisms Compared to a plain tube, enhanced nucleate boiling surfaces have significant performance advantages. For instance, the enhancement ratio at the same wall superheat relative to that of a comparable plain tube range from about 2-4 for low finned tubes but increase up to 15 times for mechanically deformed low finned tubes such as for the Turbo-B tubes of Wolverine Tube Inc. or Gewa-B tube of Wieland. Evaporation and convection occur both on the outside surface of an enhanced boiling surface and inside its re-entrant passageways. There are four possible paths by which heat can leave an enhanced surface: 1. As latent heat in vapor formed within the enhancement passageways (primary importance); 2. As latent heat in bubbles growing on the exterior of tube or while they are emerging from reentrant channels (secondary importance); 3. As sensible heat to liquid pumped through the re-entrant passageways (primary importance); 4. As sensible heat to liquid on exterior of tube (secondary importance).
Laboratoire de Transfert de Chaleur et de Masse
The principal factors contributing to the high thermal performance of enhanced surfaces have been identified by Thome (1990) as follows: Nucleation superheat. Enhanced surfaces have re-entrant nucleation cavities (except for low finned tubes) that are able to nucleate at very low wall superheats with respect to plain surfaces, see Figure 9.9. Wetted surface area. Low finned tubes have from 2 to 4 times the surface area of a plain tube while complex enhancements have area ratios from 4 to 10 times those of plain tubes. Thin film evaporation. Thin evaporating liquid films form on the extensive inner surfaces of the re-entrant passageways. Capillary evaporation. In corners, liquid menisci evaporate as heat is conducted into the liquid. Internal convection. Liquid is pumped in and then back out of the re-entrant channels by the action of the departing bubbles. External convection. The high density and departure frequency of the bubbles emerging from the re-entrant channels accentuates the external convection mechanisms, i.e. bubble agitation and thermal boundary layer stripping.
13
With an end (cross-sectional) view of the process in re-entrant channels of circular, triangular and rectangular shape, Arshad and Thome (1983) used a motor driven still camera to observe the formation and dryout of the thin liquid films. Figure 9.10 shows their experimental setup and Figure 9.11 shows some of their photographs of the process. Figure 9.12 illustrates a schematic sequence of nucleation, formation and dryout of a liquid film in a triangular re-entrant channel upon application of a high heating rate.
Figure 9.13 Figure 9.13 illustrates a thin film evaporation model hypothesized by Xin and Chao (1985) for rectangular re-entrant channels with a slit opening. Hence, thin evaporating films were seen to be an important heat transfer mechanism, analogous to the microlayers formed underneath rapidly growing bubbles on plain surfaces, and are active over a large surface area. Refer to Thome (1990) for a complete treatment of this subject.
14
100
Figure 9.14
Heat transfer coefficient (kW/m
2
K)
10
30
20
15
dbfig.9.16
100000
Turbo B Plain
Heat transfer coefficient (W/m2K)
10000
1000
R-123
100 1000 10000 Heat flux (W/m 2) 100000
dbfig 9.17
200 R-123 at 4 C 180
160
120
100
80
60
40
20
16
dbFig 9.18
40000 Tsat = 40F (4.4 C) 35000
25000
20000
15000
10000
5000
THERMOCOUPLES
REFRIGERANT INLET
Figure 9.20
Laboratoire de Transfert de Chaleur et de Masse
17
Figure 9.21
dbFig 9.24
6000 Mass velocity = 14.5 kg /m2s Pressure = 1.3 bar 5000
4000
34 kW / m2 25 kW / m2
3000
9 kW / m2 3 kW / m2
2000
1000
18
dbFig 9.22
12000
11000
10000
45 kW / m2
Heat Transfer Coefficient (W/m 2K) 9000
35 kW / m2 25 kW / m2 15 kW / m2
8000
7000
45 kW / m2 35 kW / m2
6000
5000
25 kW / m
4000
15 kW / m2
3000
2000 0,0 0,1 0,2 0,3 Vapor quality 0,4 0,5 0,6
dbFig 9.25
11000 R - 123 at 4.4 C Turbo - B tube 19-25 kg/m2s 10000 Heat Transfer Coefficient (W/m 2K) 36 kW / m2
44 kW / m2
23 kW / m2 8000 44 kW / m2 36 kW / m2 7000 23 kW / m2 14 kW / m2 6000 0,0 0,1 0,2 Vapor quality 0,3 0,4 0,5 14 kW / m2
19
For R-134a at 4C (psat = 337.7 kPa), the physical properties are: M = 102.03, L = 1281 kg/m3; G = 16.56 kg/m3; L = 0.0002576 Ns/m2; G = 0.0000109 Ns/m2; hLG = 195500 J/kg; kL = 0.0902 W/m K; cpL = 1352 J/kg K; = 0.011 N/m; PrL = cpLL/kL =3.861; pr = 0.08320. 1. Determine the nucleate pool boiling heat transfer coefficient for R-134a at 4C using the methods of Mostinski, Cooper, Stephan-Abdelsalam and Gorenflo at a heat flux of 10 kW m-2 and surface roughness of 0.5 micron. 2. Calculate the heat flux at DNB for R-134a at 4C for a horizontal cylinder of 5 mm diameter.
20