Download as pdf or txt
Download as pdf or txt
You are on page 1of 36

0082-06.

fm Page 307 Wednesday, August 23, 2000 3:54 PM

6 Heat Transfer with Phase


Change

6.1 INTRODUCTION
Heat transfer of a phase change coolant is much more complex than the previous
modes of heat transfer that we have studied. By phase change we denote the following
processes:

a. Solid changing to a liquid—fusion, or melting,


b. Liquid changing to a vapor—evaporation, also boiling,
c. Vapor changing to a liquid—condensation,
e. Liquid changing to a solid—crystallization, or freezing,
f. Solid changing to a vapor—sublimation,
g. Vapor changing to a solid—deposition.

Because these processes involve a fluid medium, we generally classify them as


convection processes. Beyond all of the parameters that we previously examined in
convection, we must now add the variables of surface tension, ; ambient pressure,
P; and either the enthalpy or latent heat of evaporation, hfg; the latent heat of
solidification, hfs; or the latent heat of sublimation, hsg. The latent heat is the amount
of heat required to convert a unit mass of a substance from one phase to another
phase. There is also a difference in density between the liquid phase and the vapor
phase that induces a buoyancy force,   l  v , when bubbles are present.
Also, we will note that heat transfer during phase change does not always occur
with a change in temperature of the media. In fact, a very large rate of heat transfer
can be achieved with very little change in temperature. This is one of the attractions
of phase change heat transfer. Furthermore, in contrast to natural or forced convec-
tion, increasing the T may result in a decrease in the heat transfer coefficient.
Because of the number of variables, there are no accurate general equations or
correlations to use. Of the usable equations, most have an empirical value that
changes with the surface characteristics and must be evaluated by experimentation.
The accuracy of these correlations without experimental verification may be
50%.
Although heat transfer by phase change is not yet widely used in electronics
cooling, as the component heat flux rises the laws of physics dictate that high-end
cooling technologies will progress from air-cooled to liquid-cooled to phase change.
Our studies in this section will be largely theoretical.

© 2001 by CRC PRESS LLC


0082-06.fm Page 308 Wednesday, August 23, 2000 3:54 PM

6.1.1 DEFINITIONS OF PHASE CHANGE PARAMETERS


Vapor pressure—At any temperature above absolute zero, the molecules in a liquid
are in constant motion. Some of these molecules will have a higher velocity than
the average. If the energy in one of these high-speed molecules is greater than the
cohesive forces, the molecule can “escape” through the surface of the liquid. We
know this as evaporation. Since these higher-speed molecules contain more energy
than the remaining molecules, the reduction in energy causes the bulk liquid to cool.
If the liquid is in an airtight enclosure, the escaped molecules will fill the airspace.
Some “escaped” liquid molecules will even reenter the liquid. Eventually, the number
of molecules escaping the liquid equals the number of molecules reentering the liquid.
When this occurs, we call the air space a saturated vapor. Since the molecules exert
a pressure within this enclosure, we call this the saturated vapor pressure. If we increase
the pressure within the airspace, or decrease the volume of the airspace, the vapor will
contain more fluid particles than it can hold. We can then say that we have supersat-
urated the vapor. Supersaturation conditions can occur only temporarily.
Phase change—In the previous discussion, the liquid molecules changed phase,
from a liquid to a vapor. If the walls of the enclosure (from the previous discussion)
are suddenly cooled, we would again supersaturate the vapor. Randomly moving liquid
molecules that strike the cool enclosure walls would leave their vapor state and return
to the liquid state, forming condensation. This is also a change in phase—from vapor
phase to liquid phase. We see an identical process when water, in the solid form of
ice, melts and becomes a liquid.
Another process, known as sublimation, occurs when a material changes from
a solid phase directly to a vapor phase. The opposite of sublimation is deposition,
where the vapor phase changes directly to a solid.
The triple point of a material is the combination of temperature and pressure
conditions at which the material can exist in a solid, liquid, and vapor state simul-
taneously. Water has a triple point of 0.01°C at 4.58 torr. The sublimation point is
the temperature and pressure at which the material can exist as solid and vapor. The
critical point is the temperature at which no amount of pressure will cause the vapor
to liquefy. This occurs when the molecules are moving so fast that the internal
cohesive forces are not strong enough to form a surface.
Figure 6.1 shows the phase diagram for H2O. A is the triple point. B is the critical
point. Curve AB shows the points at which H2O can exist in both the liquid and vapor
phase. Line AC shows where H2O can exist as a solid and a liquid, and AD shows
where it can exist as both a solid and a vapor. Standard atmospheric pressure is 760
torr (760 mmHg). Moving along the 760 torr line we see that H2O has a phase change
from solid to liquid at 0°C, and changes from a liquid to a gas at 100°C. Sublimation
occurs along line efg, where the solid may change directly into a vapor without turning
into a liquid first. At point B, temperatures of 374°C and above, H2O cannot liquefy
at any pressure. This is called the critical temperature and occurs when the gas
molecules are moving so fast that cohesive forces are unable to form a surface.
Technically, a vapor is called a gas when it exceeds the critical temperature.
As previously noted, a liquid will evaporate if the vapor pressure is lower than
saturation pressure. Saturation pressure increases with temperature. If the saturation

© 2001 by CRC PRESS LLC


0082-06.fm Page 309 Wednesday, August 23, 2000 3:54 PM

FIGURE 6.1 The phase diagram for H2O.

pressure is equal to the surrounding atmospheric pressure, the liquid will form vapor
bubbles throughout and boil.
Latent heat—As an object absorbs energy, the object will display a temperature
increase, defined by the specific heat of the object. We call this temperature increase
sensible heat. Another form of heat is latent heat, by which an object will absorb
energy but will not increase in temperature. The latent heat of vaporization is higher
than the latent heat of fusion because the molecules are spaced farther apart in a
gas than in a liquid.

6.2 DIMENSIONLESS PARAMETERS IN BOILING


AND CONDENSATION
For boiling, the Jakob (Ja) number is the ratio of maximum sensible heat absorbed
by the liquid to the latent heat absorbed by the vapor, or the reverse for condensation.
The Jakob number can also be described in terms of temperature difference as

c p ( T s  T sat )
Ja  -------------------------------
-
h fg

This dimensionless group characterizes the heat transfer during a phase change.
The Jakob number is usually quite small. For example, we can let Ts equal the
temperature of the liquid surface of an ice block Ti equal the internal temperature
of the ice block. If we say the temperature difference between an ice block and a

© 2001 by CRC PRESS LLC


0082-06.fm Page 310 Wednesday, August 23, 2000 3:54 PM

liquid surface is 10°C, then Ja  0.058. The Bond number is the ratio of the
gravitational buoyancy force to the surface tension.

6.3 MODES OF BOILING LIQUIDS


Boiling occurs when the temperature of a surface wall, Tw, exceeds that of the liquid
saturation temperature, Tsat. The heat transferred from the heated wall to the liquid
can be found by

qs  h ( T w  T sat )  h  T e

where qs is the rate of heat generation per unit of surface (W/m2) and Te is the excess
temperature. As we know from experience, bubbles within the liquid characterize
the boiling process. The dynamics of the bubbles reaching the surface affect the
fluid motion and, therefore, the convection.
The boiling heat transfer process consists of two basic types: pool boiling, which
occurs in an initially stagnant liquid, and flow boiling, which occurs in the presence
of liquid velocity. Boiling may occur in both process when no bubbles are visible.
We call this subcooled boiling. In this process, the temperature of the liquid is below
the saturation temperature. In subcooled boiling, bubbles form in the superheated
liquid at the wall but are condensed when they grow large enough to extend into the
subcooled bulk liquid. When the vapor extends into the subcooled liquid, it loses its
heat and collapses. Figure 6.2 shows the phases in subcooled boiling. Saturated boiling

FIGURE 6.2 Fluid flow pattern induced by a bubble in a subcooled liquid in different stages
of development collapse.

© 2001 by CRC PRESS LLC


0082-06.fm Page 311 Wednesday, August 23, 2000 3:54 PM

is the type of boiling with which we are most familiar: the temperature of the liquid
exceeds the saturation temperature, bubbles form, and they rise to the liquid surface.

6.3.1 BUBBLE PHENOMENON


Bubbles are an important characteristic of saturated boiling. Bubbles form individ-
ually in microscopic surface imperfections on the heated body. The bubbles grow
until they separate from the heated surface and reach the surface of the liquid. At
higher heat fluxes bubbles form, separate, and reform so quickly that continuous
streams or vapor columns are seen. At still higher heat flux levels, the process occurs
so quickly that the liquid cannot reach the surface of the heated body. In this case
the heated body is continually blanketed by a vapor film.
A vapor bubble forms and grows in a liquid as long as the bubble vapor pressure
is higher than the ambient liquid pressure, that is, p pl. Stability is reached when

2
p  p l  -------
Rb

where Rb is the radius of the vapor bubble. If we assume that the bubble and the
liquid are at identical temperatures, the  p between the vapor and the liquid can be
translated into a temperature by the Clausius-Clapeyron relation. Therefore, from
Hetsroni1 we see that

2  fg 
T  Tl  Ts 1 ------- ------ -----------------
R b h fg  l  

While this equation is for an isothermal condition, Bergles and Rohsenow2 found
the theoretical temperature difference between the wall surface and the liquid satu-
ration temperature that will form the first bubble, or the incipience of boiling

8qT 
 T b  T w  T sat  -----------------
 v h fg k f

This equation assumes that temperature decreases linearly with increasing distance
from the heated wall. Since this is not strictly true in actual occurrences of wall
superheat, the equation underpredicts the wall superheat, Tb.
The familiar boiling process is actually called nucleate boiling. When a bubble
becomes large enough to detach from a heated surface, the bubble is said to have
nucleated. In nucleate boiling the characteristic length is the size of the bubble when
it separates from the surface, Lb. To find this dimension, we balance the bubble surface
tension to the bubble buoyancy force, as shown in Figure 6.3; therefore,

Surface Tension  Buoyancy Force


2 3
2 R b   --- R b (  l   )g
3

where Rb is the bubble radius.

© 2001 by CRC PRESS LLC


0082-06.fm Page 312 Wednesday, August 23, 2000 3:54 PM

FIGURE 6.3 Force balance on vapor bubbles in a fluid: (a) unattached forces, (b) attached
vapor bubble forces.

The bubble radius is therefore found by

3  gc 0.5
R b  ------------------------
-
(  l   )g

and the characteristic length is then related to the radius by

 gc 0.5
L b  ------------------------
-
(  l   )g

Cole and Rohsenow3 correlated vapor bubble departure diameters in a saturated


water pool by a dimensionless Eustis number

4 1.25 2
E 0  [ 1.5 10 ( Ja ) ]

and in saturated pools of other liquids by

4 2
10 ( Ja ) ]
1.25
E 0  [ 4.65

where:
g (  l   )D b
2

E0  Eustis number, --------------------------------


gc 
 cT s
Ja*  modified Jakob number ------------
v h f s
-

© 2001 by CRC PRESS LLC


0082-06.fm Page 313 Wednesday, August 23, 2000 3:54 PM

Now that we have correlated the bubble departure diameter, we can find the
frequency of departure. Although current experimental data are irregular, Hetsroni1
found that bubbles that grow very slowly depart at a frequency of about

0.5625  Ja
2

f  -------------------------------
2
-
Db

Cole4 roughly correlated the frequency of departure for rapidly growing bubbles as

0.5
g ( l   )
4
---
f  ---------------------------
3
C D l Db

where CD is the dimensionless drag coefficient.


Theoretically, now that we can calculate the diameter of the bubbles, Db, and
the frequency of bubble departure, f, we can determine the bubble, or vapor flux,
Qv, if we can find the number and size of nucleation sites per unit area. Since this
value is usually unknown, another way to find the vapor flux is by the equation

q
Q   ------------
h fg 

where  is the dimensionless fraction of the heat flux that will result in a net generation
of bubbles. Unfortunately,  is not a constant. Graham and Hendricks5 found that 
varies in a complex manner from about 0.01 to 0.02 at low heat fluxes, to 0.5 at
about 20% of the critical heat flux, and approaches 1.0 at the critical heat flux.
If the vapor flux can be determined, then we can find the amount of agitation
caused by nucleate boiling, which is described as the bubble Reynolds number

Db Q
Re b  -------------
l

A good estimate of the maximum bubble velocity can be found by balancing


the vapor kinetic energy against the buoyancy force of the bubble in terms of the
characteristic length. Therefore,

Vapor Kinetic Energy  Buoyancy


1
---  V max  g (  l   )L
2
2

Since we know the characteristic length is

 gc 0.5
L b  ------------------------
-
(  l   )g

© 2001 by CRC PRESS LLC


0082-06.fm Page 314 Wednesday, August 23, 2000 3:54 PM

the maximum bubble velocity can be described as

 (  l   )gg c 0.25
V max ~  ---------------------------------
-
 
2 

The maximum bubble velocity can also be described using what is known as
the Helmholtz* theory of instability. That is, the bubble vapor column is disrupted
by a wavelength disturbance of Lb, the characteristic length, and becomes unstable
at VH, or

2  g 0.5
V H ~  ----------------c
  Lb 

If the vapor bubble diameter is limited to an experimental range found by


Bromley et al.,6 that is

gc  gc 
-  D b  5.45 ------------------------
3.14 ------------------------ -
g ( l   ) g ( l   )

then we may use a quick estimate of the velocity of an undisturbed bubble given
by Zuber et al.7

1
V b  --- 4gr
3

Similar to bubble growth in nucleate boiling, vapor bubbles collapse when


subcooled. Florschuetz and Chao8 used a temperature integral developed by Plesset
and Zwick9 to determine the rate of collapse, H, of a bubble:

 ˙ 
4 2  1  -----------
2 Rb  2
 ------- 
H -t  ---   R˙ -
 ---- Ja ------- -  3
2
R b, i 3   ---------
b  R b, i 

  R b, i 

where Rb is the bubble radius and Rb,i is the initial bubble radius.
If we define the bubble collapse period as the time when the bubble volume is
1% of the departure bubble volume, then the ratio Rb/Rb,i  0.2. We now define the
collapse period, c, as

 
1  -----------
2 Rb  2
 ------- 
 c  ---   R˙ - -  3
 R b, i Rb
 2.32
3  ---------
b 
  ----------  0.2
  R b, i 
R b, i

* Hermann von Helmholtz (1821–1894) described the theory of instability and also provided a physical
proof of Fourier’s Theorem by producing complex musical tones using individual tuning forks.

© 2001 by CRC PRESS LLC


0082-06.fm Page 315 Wednesday, August 23, 2000 3:54 PM

and substituting and rearranging, we see that

2 2
2.32 R b, i R b, i
 c  ---------- ------------
-  0.580 ------------
-
4 Ja  2
Ja 
2

We can now determine the relationship among the rise velocity, Vb, the rate of
collapse, c, and the departure diameter, Db , to the collapse length, Lc, as

c

Lc  0 V b dt

Substituting the equations for the collapse rate, H, and the undisturbed bubble
velocity, Vb, and letting the ratio of the bubble radius, Rb /Rb,i, equal the ratio of the
bubble diameter, Db /Db,i, we have an equation for the collapse length, Lc, as

2.5 2.5
D b 2g D b 2g
L c  1.4 -------------------
- ------  0.0292 -----------------------
-
Ja  Ja 
2 48 2

6.3.2 POOL BOILING


There are two different types of boiling: pool, and flow. The type of boiling known
as saturated pool boiling is depicted in Figure 6.4. Note that the temperature of the
boiling liquid is nearly constant except at the surface of the heated wall. At the
heated surface the liquid temperature increases sharply.

FIGURE 6.4 Temperature distribution of the heated solid surface, Ts, and the boiling liquid,
Tsat , during saturated pool boiling.

© 2001 by CRC PRESS LLC


0082-06.fm Page 316 Wednesday, August 23, 2000 3:54 PM

FIGURE 6.5 The boiling curve generated by Nukiyama10 for saturated water. Adapted from
Incropera, F. P., and DeWitt, D. P., Fundamentals of Heat and Mass Transfer, 3rd ed., John
Wiley & Sons, New York, 1995, 722, using the data of Nukiyama, S., “The Maximum and
Minimum Values of Heat Transmitted from Metal to Boiling Water Under Atmospheric
Pressure,” J. Japan Soc. Mech. Eng., 37, 367, 1934. (Translation: Int. J. Heat Mass Transfer,
1419, 1966).

6.3.2.1 Pool Boiling Curve

In 1934, Nukiyama10 published experiments defining the different regimes of pool


boiling. Nukiyama used a nichrome wire immersed in a bath of water. As he
increased the power to the wire, he noticed bubbles forming at an excess temperature,
 Te , of about 5°C. As he increased power further, he noticed that the power supplied
to the wire could be increased greatly without a large increase in temperature. As
he increased the power further still, the temperature suddenly increased dramatically,
and the wire reached its melting point. After experimenting with a platinum wire
with a higher melting point, he noted that the heat flux and temperature were related,
as shown in Figure 6.5. Nukiyama believed that a method of temperature control of
Te, instead of a power-controlled method, would yield a better curve. In 1937, Drew
and Mueller11 performed the experiment using a steam pipe and obtained a curve
similar to that shown in Figure 6.6. This figure shows the relationship of heat flux
and temperature for water at sea-level atmosphere. The relationship is similar for
most other liquids.
We call the range when  Te   Te, A, free convection boiling. Bubbles do not
form in this region because there is not enough vapor in contact with the liquid. Fluid
motion in this range is caused by natural convection. Nucleate boiling occurs when
individual bubbles form on the heated surface and rise to the liquid surface. This point
on the curve is designated A. The portion of the curve designated AB is characterized
by isolated bubbles. As the temperature of the heated surface is increased further, the
bubbles are generated faster and sometimes merge to form vapor columns. This region
is shown by the portion of the curve designated BC. Nucleate boiling occurs in the

© 2001 by CRC PRESS LLC


0082-06.fm Page 317 Wednesday, August 23, 2000 3:54 PM

Boiling
regimes
Free convection Nucleate transition Film

{
{

{
{
Isolated Jets and
bubbles columns
107


qmax
 max
qmax
106 C Critical heat flux, q"
P
q"s (W / m2 )

B
105


qmin
D  min
Leindenfrost point, qq"min
A
10 4
ONB
∆Te,A ∆Te,B ∆Te,C ∆Te,D

10 3
1 5 10 30 120 1000

∆Te = Ts Tsat( C)
_ o

FIGURE 6.6 Characteristics of the boiling curve for a heated horizontal surface in water.

temperature range of  Te,A   Te   Te,C where  Te,C is about 30°C. This is the
range of operation for most heat transfer work. High levels of power can be dissipated
without a large increase in temperature. Point P indicates the point of the maximum
heat transfer coefficient. Ideally, equipment should operate at this point. In water, the
convection coefficient in this region can exceed 104 W/m2 K.
The region when  Te,C   Te   Te,D, where  Te,D is about 120°C, is called
the transition boiling, partial film boiling, or unstable film boiling range. In the
previous range, as each bubble left the surface, liquid covered the surface until a
new bubble formed. In the transition boiling range, new bubbles are formed before
the liquid can reach the surface. A continuous vapor film forms on the surface. The
entire heated surface oscillates between a liquid and vapor blanket. As the temper-
ature differential  Te increases, the entire surface is more often covered by a vapor
layer than the liquid layer. Also, as  Te increases, hc and therefore qs decrease,
because the thermal conductivity of the vapor layer is much lower than when the
adjacent layer was a liquid.
The region when  Te  Te,D is called film boiling. Point D of the boiling curve
is called the Leidenfrost point. In 1756, Leidenfrost noticed that when water droplets
are placed on a hot surface, the droplets dart about the surface, supported by a vapor
layer. We know that during the transition phase, as  Te increases, a higher percentage
of the surface is covered by vapor at any point in time. The Leidenfrost point occurs
when the entire surface is covered by a vapor layer, and the heat flux reaches a minimum,
qs,D  qmin. In the film boiling range, heat transfer can only occur by conduction through
the vapor layer. After the Leidenfrost point, radiation heat transfer through the vapor
layer becomes more important, and the heat flux increases with increasing  Te.

© 2001 by CRC PRESS LLC


0082-06.fm Page 318 Wednesday, August 23, 2000 3:54 PM

Researchers have experimented with the region after point C, but in actual
engineering applications this region is difficult to control. Any increase in the heat
flux after point C creates a marked increase in temperature. The size of this increase
may cause destruction of the heat flux surface. For this reason, point C is often
called the burnout point or, more commonly, the Critical Heat Flux (CHF).

6.3.2.2 Pool Boiling Correlations

Researchers have calculated a number of correlations to explain the actions of pool


boiling, based on the pool boiling curve. For Nusselt numbers up to the Te, A point,
standard free convection correlations can be used. In the region of nucleate boiling,
where  Te,A   Te   Te,C and where  Te,C is about 30°C, the Nusselt number is
highly dependent upon the rate of bubble formation. Although no exact models are
available to describe this phenomenon, Yamagata et al.12 correlated the influence of
nucleation sites on the heat flux by

qs  C  T e n
a b

where:

qs  surface heat flux (W/m2)


C  constant dependent upon liquid/surface combination
Te  excess temperature (Ts  Tsat) (°C)
a  1.2
n  bubble site density (N/m2) (   T e )
6

b  1/3

Currently, the Rohsenow13 correlation is the most useful:

g ( l   ) 0.5
c p, l  T e  3
 ----------------------
qs   l h fg ------------------------
-
 gc  C h Pr n
sf fg l

where:

  absolute viscosity (N s/m2) cp  specific heat (J/kg K)


hfg  latent heat of vaporization Cs f  coefficient of liquid/surface
(J/kg) combination
g  gravitational acceleration Pr  Prandtl number, cp/k
(9.806 m/s2) n  exponent for liquid
  mass density (kg/m3) l  liquid phase
  surface tension (N/m)  vapor phase
gc  gravitational constant
(1.0 kg m/N s2)

Figure 6.7 shows pool boiling data points for water that were correlated by the
Rohsenow method. Collier14 recommends the following correlation as being simpler

© 2001 by CRC PRESS LLC


0082-06.fm Page 319 Wednesday, August 23, 2000 3:54 PM

FIGURE 6.7 Pool boiling data for water correlated by the Rohsenow method. (Adapted from
References 13 and 36.)

to use than the Rohsenow correlation:


3.33
p 0.17 p 1.2 p 10
qs  0.000481  T 1.8  ------ 4  ------ 10  ------
3.33 2.3
p
e cr
 p cr  p cr  p cr

where:

Te  excess temperature, °C


pcr  critical pressure, atmosphere (101,325 N/m2)
p  operating pressure, atmosphere (101,325 N/m2)

The Rohsenow correlation13 can be manipulated to find the nucleate boiling


Nusselt number in terms of the Jakob number
2
c p, l ( T w  T sat )
------------------------------------- 2
h fg Ja
3 m
-  ---------------
Nu  ----------------------------------- 3
-
m
C sf Pr l C sf Pr l

where the exponent m is 2.0 for water and 4.1 for other liquids.
Danielson et al.15 report a number of values for Csf . The value of Csf can be
assumed to equal ~0.013 when the experimental value is unknown. Hetsroni1 reports
that this value correlates a wide spectrum of experimental data to within 20%.
The most important variables that affect Csf are the surface roughness of the heated

© 2001 by CRC PRESS LLC


0082-06.fm Page 320 Wednesday, August 23, 2000 3:54 PM

TABLE 6.1
Values of the Surface/Liquid Coefficients
Liquid/Surface Combination Csf

Benzene–Chromium 0.010
Carbon Tetrachloride–Copper 0.013
Carbon Tetrachloride–Polished Copper 0.007
Ethyl Alcohol–Chromium 0.0027
Isopropyl Alcohol–Copper 0.0023
n-Pentane–Chromium 0.015
n-Pentane–Polished Copper 0.0154
n-Pentane–Lapped Copper 0.0049
Water–Brass 0.006
Water–Copper 0.013
Water–Scored Copper 0.0068
Water–Polished Copper 0.013
Water–Nickel 0.006
Water–Chemically Etched Stainless Steel 0.0133
Water–Mechanically Polished Stainless Steel 0.0132
Water–Ground and Polished Stainless Steel 0.008

surface and the angle of contact between the vapor bubble and the heated surface.
The surface roughness affects the number of nucleation sites, and the angle of contact
is a measure of the wettability of the surface. Smaller contact angles represent greater
wettability. A totally wetted surface has the least amount of vapor and represents
the greatest heat transfer coefficient. Values of the coefficients of liquid/surface
combinations are shown in Table 6.1.
Table 6.2 presents important boiling point thermophysical data for coolants
commonly used in electronic cooling.

6.3.2.3 Pool Boiling Critical Heat Flux Correlations

The critical heat flux, point C in Figure 6.6, is determined by the maximum rate
that vapor bubbles can leave the heated surface. This can also be described as the
maximum speed that the liquid can re-wet the heated surface after a bubble leaves
the wall. We can now relate the bubble velocity to the critical heat flux by

qmax   V max h fg

or

qmax
-  C max
----------------------
 V max h fg

© 2001 by CRC PRESS LLC


0082-06.fm Page 321 Wednesday, August 23, 2000 3:54 PM
© 2001 by CRC PRESS LLC

TABLE 6.2
Fluid Properties at Respective Boiling Points (1.0 atm)

Coolant
Property FC-72 FC-77 FC-84 FC-87 L-1402 R-12 R-113 Water

Boiling Point, °C 52.0 100.0 83.0 30.0 51.0 30.0 48.0 100.0
Density, Liquid, l, kg/m3 1592 1590 1575 1633 1635 1487 1511 958.0
Density, Vapor,  , kg/m3 12.68 14.31 13.28 11.58 11.25 6.34 7.40 0.59
Absolute Viscosity, , N s/m2 0.00045 0.00045 0.00042 0.00042 0.00052 0.00036 0.00050 0.00027
Specific Heat, cp, J/kg K 1088 1172 1130 1088 1059 – 979 4184
Heat Vaporization, hfg, J/kg 87927 83740 79553 87927 104675 165065 146824 2257044
Thermal Conductivity, k, W/m K 0.0545 0.0570 0.0535 0.0551 0.0596 0.0900 0.0702 0.683
Surface Tension, , N/m 0.0085 0.0080 0.0077 0.0089 0.0109 0.0118 0.0147 0.0589
Coefficient of Expansion, , 1/K 0.0016 0.0014 0.0015 0.0016 0.0016 – 0.0017 0.0002
0082-06.fm Page 322 Wednesday, August 23, 2000 3:54 PM

where:

qmax  critical heat flux (W/m2 )


v  vapor density (kg/m3)
Vmax  maximum bubble velocity (m/s)
hfg  latent heat of vaporization (J/kg)
Cmax  coefficient of critical heat flux geometry

Kutateladze17 defined the equation for the critical heat flux in 1948 by dimen-
sional analysis. Later, in 1959, Zuber18 found the equation for critical heat flux by
a hydrodynamic stability analysis. The critical heat flux can be found by

0.25
qmax  C max h fg [  v (  l   v )gg c ]
2

where Cmax is  /24, also called the Zuber constant. Table 6.3 presents the exper-
imental data of Lienhard and Dhir19 for the value of Cmax using a dimensionless
parameter called L*. This parameter is the ratio of the characteristic length of the
heated surface, Ls, to the characteristic bubble dimension, Lb.
Pressure affects the maximum heat flux because of the influence on vapor density
and the boiling point of the liquid. As the boiling point changes, so does the heat
of vaporization and the surface tension. Therefore, each fluid has a specific pressure
that will yield a maximum heat flux. Cichelli and Bonilla20 have experimentally
demonstrated that the critical heat flux increases with pressure up to 1/3 of the critical
pressure. After this peak, the critical heat flux falls to zero at the critical pressure, as
shown in Figure 6.8.

TABLE 6.3
Values of Cmax (Zuber Constant) for Critical Heat Flux
Surface Geometry Cmax Characteristic Length Range of Applicability

Infinite flat heater 0.15 Width or Diameter L* 27


Small flat heater 12 L
2
Width or Diameter 9  L*  20
0.15 ----------------b
As
Large horizontal 0.12 Radius L 1.2
cylinder
Small horizontal 0.12L* 0.25 Radius 0.15  L*  1.2
cylinder
Large sphere 0.11 Radius 4.26  L*
Small sphere 0.227L* 0.5 Radius 0.15  L*  4.26
Large arbitrary body 0.12 — —
Ls Ls
Note: L*  ----
Lb
-  -------------------------------------
 gc 0.5
-
--------------------------
(  l   )g

© 2001 by CRC PRESS LLC


0082-06.fm Page 323 Wednesday, August 23, 2000 3:54 PM

FIGURE 6.8 Critical heat flux in nucleate boiling as a function of pressure. (From Cichelli,
M. T. and Bonilla, C. F., Trans. AIChE, 41, 755, 1945. With permission.)

When the liquid is subcooled, Zuber et al.7 have found that the critical heat flux
is correlated to a reasonable estimation by

 2k l ( T sat  T l ) 24 
2

0.25
qmax  qmax sat  1 ---------------------------------
- ---------------- ---------------------------------- 
 l  h fg  v g  g c ( l   ) 

where:

qmax sat  critical heat flux in saturated pool boiling (W/m2)


kl  thermal conductivity of the liquid (W/m K)
Tl  temperature of the bulk liquid (C)
l  thermal diffusivity of the liquid, k/cp

 gc 
0.5 2 0.25
  ---- 2 ------------------------
- ----------------------------------
3 g ( l   )  gg c (  l   v )

The effect of subcooling on the critical heat flux was studied by Ellion.21 All of
the previous correlations are for clean horizontal surfaces. Bernath22 has shown that
a vertical surface may have a critical heat flux as much as 25% less than a horizontal
surface.

© 2001 by CRC PRESS LLC


0082-06.fm Page 324 Wednesday, August 23, 2000 3:54 PM

6.3.2.4 Pool Boiling Minimum Heat Flux Correlations

There are no adequate theories that describe the transition regime between critical heat
flux and vapor film boiling. This region corresponds to qs , D of Figure 6.6 and is called
the minimum heat flux, or the Leidenfrost point. Fortunately, this area of the boiling
curve has little practical value in electronic cooling. At lower temperatures within this
region, fluid motion is characterized by periodic, unstable liquid and heater contact. At
the higher-temperature region, a stable vapor film forms and the heat flux reaches a
minimum. If the heat flux drops below the minimum required to initiate a stable vapor
film, the liquid will again contact the heater surface and nucleate boiling will be rees-
tablished. Zuber23 used stability theory to derive an equation for minimum heat flux, qmin
gc  g ( l   ) 0.25
qmin  C  h fg ----------------------------------
( l v )
2

Kovalev24 notes that this correlation is accurate to only about 50% at moderate
pressures and has even less accuracy at high pressures. Nevertheless, Berenson,25
for a large flat horizontal heated surface, and Lienhard and Wong,26 for a horizontal
heated cylinder, found that C  ~0.09.

6.3.2.5 Pool Boiling Vapor Film Correlations

The region of vapor film boiling, like minimum heat flux boiling, is not of great
concern in applications involving electronic cooling. Above the Leidenfrost point
the heat flux increases, but at a dramatic increase in surface temperature. In this
range the heated body is enveloped in a continuous blanket of vapor; therefore, heat
transfer is by conduction through the vapor. At temperature ranges much higher than
those seen in electronics, radiation heat transfer across the vapor film becomes
important. Because the vapor film is thin, the Rayleigh number is low, which indicates
heat transfer by conduction only. Because of this, analysis of the heat flux for vapor
film boiling is relatively simple. The minimum wall superheat that sustains the vapor
film was found by Berenson25 to be

 , f h fg g (  l   ) 23
gc  0.5
 ,f 13
T min  0.127 ---------------
- ------------------------- ----------------- ------------------------------
-
k ,f l  g ( l  ) gc  ( l   )

where the subscript f is that property at the film temperature. For film boiling, the
following correlation for the average heat transfer coefficient is commonly used:

g (  l   )  k ( h fg 0.35c p,  T e )
3 0.25
h c  C fb -------------------------------------------------------------------------------------
-
D T e

where the constant Cfb is 0.62 for a horizontal cylinder (Bromley27), 0.67 for a sphere
(Frederking and Daniels28), and replacing D with L is 0.71 for a plane vertical surface
(Mills29). When the vapor flow rate is high, that is, when

L (  l   )g ( h fg 0.50c p, v  T e )
3
--------------------------------------------------------------------------------
- 5 10
7
k  T e

© 2001 by CRC PRESS LLC


0082-06.fm Page 325 Wednesday, August 23, 2000 3:54 PM

Frederking and Clark30 recommend the following equation for the average film
boiling heat transfer coefficient
13

h  0.15 --------------------------------------------------------------------------------------
(  l   )g ( h fg 0.50c p,  T e )k v
2
-
 T e

6.3.3 FLOW BOILING


Similar to normal convection, boiling has two modes: pool boiling and forced
convection boiling. Pool boiling is dominated by fluid motion caused by buoyancy
forces, both liquid and vapor, within the fluid. In forced convection boiling, fluid
motion is characterized by an external force such as a pump. Also, similar to normal
convection, flow can be divided into two major categories: external flow and internal
flow. In electronics cooling, external flow usually occurs in large enclosures directly
on the surfaces of hot components. Internal flow occurs through tubes and ducts
and requires a complex analysis because the bulk flow is part liquid and part vapor.
In instances of pool boiling, thermal overshoot is a problem. Thermal overshoot is
the initial resistance of a liquid to boil. The heated surface may reach 30°C above
the normal boiling temperature before actual nucleate boiling occurs. This problem
does not normally occur in forced convection boiling.
In two-phase flow, a variable known as the vapor mass quality, , is often used.
The vapor mass quality describes the ratio of the vapor mass flow to the total mass
flow and is found by


  -------------------
ṁ ṁ l
-

where:

ṁ  mass vapor flow rate (kg/s)


ṁ l  mass liquid flow rate (kg/s)

Another term often used in two-phase internal flow is the mass velocity, G,
which is the mass flow per cross-sectional area. The mass velocity is further divided
into vapor mass flow, Gv , and liquid mass flow, Gl. The quantities can be shown as

G  G Gl

where:


G v  -----
Ac
-

ṁ l
G l  -----
Ac

© 2001 by CRC PRESS LLC


0082-06.fm Page 326 Wednesday, August 23, 2000 3:54 PM

6.3.3.1 External Forced Convection Boiling

In electronic cooling external flow is usually a relative statement. The flow may be
external over hot components, but it is still within a conduit. The pressure drop in
this type of flow is not at all well understood. Some correlations are available, but
the range of applicability is so narrow as to be nearly useless for engineering design.
In cases of conduit flow which passes over exposed components, the engineer should
always build a scale model to determine pressure drop.
At temperatures before boiling incipience, standard correlations for forced con-
vection may be used, with attention to temperature-induced property variations within
the single phase. The maximum heat flux can be increased substantially by using
forced convection in the boiling regime. Researchers have recorded heat flux values
of 35 MW/m2. Lienhard and Eichhorn31 developed a correlation for the maximum
heat flux using the Weber number, which is similar to the Reynolds number. Their
correlation is usually accurate to about 20%. The Weber number, We, is the ratio of
inertia to surface tension and can be described as

V D
2

We  ---------------

-

For a heated cylinder in a low-velocity liquid cross flow, the critical heat flux
is estimated by

1
qmax   h fg V ---- 1 4 13
 ----------
-
 We D

where low velocity is described as:

qmax 
 ---------------- 0.275  0.5
 -------------  -----l  1
  h fg V   

For a heated cylinder in a high-velocity liquid cross flow, the critical heat flux
is estimated by

l 
 ----
0.75 l 
 ----
0.5
- -
   
q m ax   h fg V -------------- -----------------------------
-
169 13
19.2 We D

where high velocity is described as:

qmax 
 ---------------- 0.275   l  0.5
  h fg V  ------------- ----- 1
 

© 2001 by CRC PRESS LLC


0082-06.fm Page 327 Wednesday, August 23, 2000 3:54 PM

FIGURE 6.9 The fluid flow regime progression from liquid to vapor for two-phase vertical
internal flow.

6.3.3.2 Internal Forced Convection Boiling

Forced convection boiling in a tube is quite complex. Researchers have identified


six regions of heat transfer in a vertical tube. These regions may blend into each
other and are not well defined. Approximate views of the modes of vertical internal
two-phase flow are shown in Figure 6.9. When a liquid flow enters a vertical tube
having a surface temperature above Tsat, the liquid initially is heated to Tsat and
initiates nucleate boiling. When nucleate boiling occurs, the single-phase liquid
(mode 1) flow becomes a two-phase bubbly flow (mode 2), composed of vapor
bubbles and liquid. As the fluid temperature increases, the vapor bubbles merge and
form large bubbles called slug flow (mode 3). Churn flow (mode 4) is highly irregular
and unsteady, and is composed of large bubbles that continually break apart and
then merge together. Higher in the tube, annular flow (mode 5) occurs. Annular flow
occurs when liquid is only in contact with the tube walls, and a core of vapor and
a liquid mist fills the center of the tube. Still higher in the tube, flow is composed
of pure vapor and the liquid mist called droplet flow (mode 6), until finally the mist
evaporates and only the single-phase vapor (mode 7) is present. Horizontal tubes
show the same modes of heat transfer but have more of a sloshing effect which tends
to blend the modes even more, as shown in Figure 6.10.
The calculation of a two-phase pressure drop within a tube is extremely complex.
Recently, this has been an area of extensive research, with only some success. The
most generally applicable model is based on the assumption of homogeneous flow.
That is, this model ignores the mode of flow and treats the flow as an average of
vapor and liquid. For a simplistic case of flow through a round tube, the total pressure
gradient is the sum of the pressure gradients related to wall friction, gravitational
resistance, and momentum changes:

   ( 1   ) l

© 2001 by CRC PRESS LLC


0082-06.fm Page 328 Wednesday, August 23, 2000 3:54 PM

FIGURE 6.10 The fluid flow regime progression from liquid to vapor for two-phase hori-
zontal internal flow.

Researchers have proposed many different and quite complex heat transfer
correlations to describe the regimes of internal forced boiling flow. Klimenko32
proposed a methodology which has sufficient accuracy, is not overly complex, and
applies to heat transfer stages before mist flow. To use the Klimenko method, we
must first determine if the heat transfer is dominated by film evaporation or by
nucleate boiling, by evaluating the dimensionless parameter Φ.
l
  -
13
Gh
  ----------fg- 1   ----
-
  1  ----
q l 

If Φ 1.6 104, film evaporation is the dominant mode of heat transfer; if Φ 


1.6 104, nucleate boiling dominates. The next step in the Klimenko method is to
determine the Nusselt number resulting from the two-phase heat transfer coefficient,
hTP , in either film evaporation or by nucleate boiling. The characteristic length for the
Nusselt number is based on bubble size, which is found by
 gc 0.5
L b  ------------------------
-
g ( l   )
If the heat transfer mode is nucleate boiling, Φ  1.6 104, the Nusselt number
is found by
0.15
Nu  7.4
3 0.6
10 q* P* Pr l
0.5 1  3  k----w-
 kl 
where:
qL b
q*  ----------------
h fg   l
-

p pL
p*  --------------------------------------
-
0.5 
---------b
[  g ( l   ) ] 

kw  wall material thermal conductivity (W/m K)


© 2001 by CRC PRESS LLC
0082-06.fm Page 329 Wednesday, August 23, 2000 3:54 PM

If the heat transfer mode is film evaporation , Φ 1.6 104, the Nusselt number
is found by

 -----  k----w-
0.2 0.09
0.6 1  6
Nu  0.087Re Pr l
 l   k l 

where:

 V Lb
Re  ------------
l
-

G l
V  ---- 1   -----  1
pl 

Finally, knowing the Nusselt number for the heat transfer mode, we can calculate
a single-phase heat transfer coefficient, which is based on the Reynolds number of
the liquid portion of the flow. Usually the single-phase heat transfer coefficient, hFC,
is so much smaller than the two-phase coefficient, hTP , that it can be ignored.
However, if it is significant, the overall heat transfer coefficient can be found by

3 3 13
h c  ( h TP h FC )

Klimenko’s correlations,32 shown in Figure 6.11 and 6.12, match experimental


data with a mean absolute deviation of about 13% for most common coolants used
in electronic cooling. The limitations are

6.1 103 N/m2  p  3.04 106 N/m2


50 kg/m2 s  G  2690 kg/m2 s
0.017    1.00
0.00163 m  D  0.0413 m

6.4 EVAPORATION
Liquid may exist on a surface if the surface is below the saturation temperature, Tsat.
Evaporation, like boiling, is considered a convection heat transfer process. Evapora-
tion occurs when the molecules in a liquid gain enough energy to escape the liquid
binding energy and enter the vapor state. Since the molecules leaving the liquid
contain excess energy, the net effect is a loss of energy of the liquid. Since temperature
is a measure of energy, the temperature of the liquid decreases when evaporation
occurs. The convection of vapor from a surface is related to the mass transfer coef-
ficient, hm. The magnitude of evaporative cooling can be expressed as the equation

h
T   T s  h fg  -----m [  A, sat ( T s )   A,  ]
 hc 

© 2001 by CRC PRESS LLC


0082-06.fm Page 330 Wednesday, August 23, 2000 3:54 PM
© 2001 by CRC PRESS LLC

FIGURE 6.11 Klimenko data for the effect of wall conductivity on the Nusselt number for internal forced convection boiling flow. (From Klimenko, V. V.,
Int. J. Heat Mass Transfer, 31, 541, 1988. With permission.)
0082-06.fm Page 331 Wednesday, August 23, 2000 3:54 PM
© 2001 by CRC PRESS LLC

FIGURE 6.12 Klimenko data showing the transition from nucleate boiling to film evaporation for internal forced convection flow. (From Klimenko,
V. V., Int. J. Heat Mass Transfer, 31, 541, 1988. With permission.)
0082-06.fm Page 332 Wednesday, August 23, 2000 3:54 PM

where:

hm  mass transfer convection coefficient (m/s)


hc  convection heat transfer coefficient (W/m2 K)
A,sat(Ts)  saturated vapor density (kg/m 3)
A,∞  infinite vapor density (kg/m3)

The ratio of the heat transfer coefficient to the mass transfer coefficient, hc/hm,
can also be expressed in terms of a dimensionless Lewis number, Le,

h k
-----c  ------------------n   c Le l  n
hm D AB Le p

where:

DAB  binary mass diffusion coefficient (m2/s)


Le  Lewis number, /DAB
n  13

Using this relationship we see that the cooling effect can also be described as

 A h fg p A, sat ( T s ) p A, 
T   T s  --------------------------
23
-  ----------
- ----------------------
 c p Le Ts T

where:

  molecular weight of species (kg/kmol)


  universal gas constant (8.315 kJ/kmol)
pA,sat(Ts)  saturated vapor pressure (N/m2)
pA,  infinite vapor pressure (N/m2)

Gilliland and Sherwood33 correlated the mass transfer coefficient from liquids
to air in a wetted-wall column as

D AB 0.83 0.44


h m  0.023  --------
- Re Sc
 D L

where:

D  tube diameter (m)


Sc  Schmidt number, /DAB 
2000  Re  35,000
0.6  Sc  2.5

© 2001 by CRC PRESS LLC


0082-06.fm Page 333 Wednesday, August 23, 2000 3:54 PM

Rohsenow and Choi34 indicate that by using the Chilton-Colburn factor, j, the
mass transfer coefficient of evaporation from a flat, wetted surface in laminar flow
can be found by

V
h m  0.664 -------------------------
0.67
-
Re L Sc

and for turbulent flow by

V
h m  0.037 -----------------------
0.2 0.67
-
Re L Sc

where V is the velocity in m/s, and the Reynolds number is based on the length of
the surface.
A Reynolds number based on the thickness of a falling evaporative film has also
been proposed. In this case, the falling film Reynolds number, Re, is defined in
terms of the hydraulic diameter, DH, and the mean velocity, um, of the film as


 l  ------
- 4
l um DH  l  4
Re   -----------------
-  ---------------------
-  -------
l l l

where:

4 Ac 4
D H  --------
-
P  P
------


u m  ------
l 
-

  mass flow per unit film width (kg/m s)


g l  ( l  v )
 3

  0 l u dy  ----------------------------------
3 l
-

In evaporation from falling films of water, Chun and Seban35 determined the
following correlations for the Nusselt number

1  3
Nu   --- Re 
3
0  Re   30
4 
 0.22
Nu  0.822Re  30  Re   Re tr
3 0.4 0.65
Nu  3.8 10 Re Pr  l Re tr  Re 

where Retr is the turbulent Reynolds number, which is equal to 5800Prl1.06.

© 2001 by CRC PRESS LLC


0082-06.fm Page 334 Wednesday, August 23, 2000 3:54 PM

6.5 CONDENSATION
Condensation will form on a surface if the vapor contacting the surface is saturated,
and the surface temperature is lower than the temperature of the saturated vapor.
When this occurs, the latent heat within the vapor is transferred to the cooler surface.
Normally, condensation occurs as drops on a surface. If the conditions that caused
the condensation are steady state, and the surface is clean, the droplets will coalesce
and form a condensate film on the surface. In a gravity field, the laminar fluid film
will flow. If the surface is long enough, the laminar flow of condensate may become
wavy and then turbulent.
Contrary to other forms of heat transfer, as T increases the heat transfer coef-
ficient decreases. This is because as the temperature difference increases, more vapor
becomes condensate, which causes a thicker layer of film. Because condensation heat
transfer is a conduction process, the thicker film impedes the transfer of heat.
Because of the heat transfer impedance caused by the thick fluid film, applica-
tions using condensation often have short flow paths or use small horizontal cylin-
ders. The short flow paths do not allow the film to become very thick. Because of
the action of film condensation, droplet condensation provides a higher heat transfer
coefficient. Heat transfer rates of droplet condensation can be more than a magnitude
higher than film condensation. Therefore, most applications use some type of coating
on the condensation surfaces which promotes droplet formation.
The condensation film begins at point x  0 and continues down a surface. The
thickness of the film increases as x increases, as shown in Figure 6.13. Because the

FIGURE 6.13 The geometry and nomenclature used to describe film condensation on a
vertical surface.

© 2001 by CRC PRESS LLC


0082-06.fm Page 335 Wednesday, August 23, 2000 3:54 PM

flow moves at a constant velocity, and because the film is thicker at larger values
of x, the mass flow rate also increases with x.
The following correlations for condensation heat transfer are most accurate when
evaluated at the film temperature recommended by Addoms36:

T eval  T w 0.33 ( T sv  T w )

Nusselt37 obtained the first relationships for the condensation variables in 1916.
Nusselt used simplifying assumptions to arrive at the basic relationship of the
thickness of the film, ,
0.25
4  l k l x ( T sat  T w )
  -------------------------------------------
-
g  l (  l   )hf g

the local heat transfer coefficient, hx,


0.25
 l (  l   )hf g k l
3
h x  g--------------------------------------------
-
4  l x ( T sat  T w )

and the local Nusselt number, Nu,


0.25
h x x g  (    )h x 3
Nu  -------
- l l fg
k l --------------------------------------------
4  l k l ( T sat  T w )
-

For a vertical surface of unit width and a height of L, the average heat transfer
coefficient can be written as
0.25
g  l (  l   )hf g k l sin 
3
h c  0.943 --------------------------------------------------------
-
 l L ( T sat  T w )

In the previous equations by Nusselt, the modified latent heat of vaporization is


hf g  hfg0.375cp,l(Tsat  Tw). Rohsenow38 modified the relationship to match exper-
imental data by equating the hf g  hfg0.68cp,l(Tsat  Tw), for the range when Pr
0.5 and when cp,l(Tsat  Tw) hf g  1.0. Sadasivan and Lienhard39 calculated the
modified latent heat without Nusselt's simplifying assumptions as hf g 
hfg (0.6830.228/Prl)cp,l(Tsat  Tw). Chen40 further modified Nusselt's equation to
account for interfacial shear and momentum forces. Chen corrected the heat transfer
coefficient as

0.25
p, l ( T sat  T w )
 1 0.68 c---------------------------- c p, l ( T sat  T w ) k l ( T sat  T w )
- 0.02 ---------------------------- - --------------------------
 h fg h fg  l h fg 
hc  hc  -----------------------------------------------------------------------------------------------------------------------------
k l ( T sat  T w ) c p, l ( T sat  T w ) k l ( T sat  T w )
-
 1 0.85 --------------------------  0.15 ----------------------------- -------------------------- 
  l h fg h fg  l h fg 

© 2001 by CRC PRESS LLC


0082-06.fm Page 336 Wednesday, August 23, 2000 3:54 PM

where:

c p, l ( T sat  T w )
------------------------------------  2.0
h fg
k l ( T sat  T w )
--------------------------------  20
 l h fg
1  Pr l  10.05

Chun and Seban35 found that the following correlations for the local and the
average Nusselt number are applicable for water condensation flowing down a flat
surface. When the Reynolds number, based on the film thickness, Reδ , represents
laminar film condensation, and 0  Reδ  30,

1  3
Nu   --- Re 
3
4 

When the Reynolds number indicates wavy laminar film condensation, and 30 
Reδ  Retr,
 0.22
Nu  0.822Re 

When the Reynolds number indicates turbulent film condensation, and Retr  Reδ,

3 0.4 0.65
Nu  3.8 10 Re  Pr l

 1.06
where Retr is the turbulent Reynolds number, which is equal to 5800 Pr l .
To find the average Nusselt number, we use an equation based on the distance
the condensate has traveled down the surface. Using the Reynolds number based on
that length, we have the equation for laminar film condensation, when ReL  30,
written as

0.25
2 13
4  -------l 
Nu  --- Pr l  g 
3 --------- ---------------
4Ja l L

When the Reynolds number indicates wavy laminar condensation flow, that is,
when 30  ReL  Retr

0.18
2 13
 l 
Pr l  ------
Nu  -------- g
-
- ---------------
4Ja l L

© 2001 by CRC PRESS LLC


0082-06.fm Page 337 Wednesday, August 23, 2000 3:54 PM

FIGURE 6.14 Local Nusselt number as a function of the Reynolds number for laminar and
turbulent film condensation of water. Adapted from References 29 and 35.

When the Reynolds number indicates turbulent film condensation, that is, when
ReL Retr, then the average Nusselt number is described as

2 13
10  6
3
 -------l  9.12 10 Ja l ( L  x tr )
Pr  g ----------------------------------------------------------
- 0.6
Nu  --------l- --------------- 13 Re tr
4Ja l L  ˙2l  0.35
 ------- Pr l
 g

Figure 6.14 shows the local Nusselt number as a function of the Reynolds number
for laminar and turbulent film condensation data gathered by Chun and Seban.35
The average heat transfer coefficient of a pure saturated vapor condensing on a
horizontal tube can be found by

0.25
g  l (  l   )hf g k l
3
h c  0.725 --------------------------------------------
-
D  l ( T sat  T w )

Some applications have vertical rows of horizontal tubes; that is, the tubes are
arranged so that the film is falling from an upper horizontal tube to a lower horizontal
tube. If the condensation flow is continuous, the tube diameter, D, in the previous
equation can be replaced by DN, where N is the number of tubes. This method gives
conservative results because the condensation flow is rarely continuous.
Chen40 suggested that since the condensation flow is subcooled, additional con-
densation occurs when the condensate is between the horizontal tubes. Assuming
that all of the subcooling is used for additional condensation, Chen obtained the

© 2001 by CRC PRESS LLC


0082-06.fm Page 338 Wednesday, August 23, 2000 3:54 PM

average heat transfer coefficient as

0.25
g  l (  l   )hf g k l
3
c p ( T sv  T w )
h c  0.728 1 0.2 ------------------------------- ( N  1 ) ---------------------------------------------
h fg DN  l ( T sat  T w )

This correlation is very accurate when the following condition is met:

( N  1 )c p ( T sv  T w )  2.0
---------------------------------------------------
-
h fg

6.6 MELTING AND FREEZING


Phase change materials are used in many specialized electronic cooling applications.
Most often, transient power applications such as those used in missiles realize these
benefits. As we have seen, during a phase change a material may absorb great
amounts of power with only a small increase in temperature. For an application such
as a missile, an onboard package of phase change material may absorb the heat
given off by an electronic package, without the need for a dedicated cooling system.
The phase change material absorbs the heat and melts. Depending on the quantity of
heat and the mass of phase change material, the cooling effect will last until all of
the material has melted. The opposite effect, freezing, occurs when the material gives
up its latent heat to the surroundings. Applications for the freezing effect might be
seen in missiles that travel outside the atmosphere for a short time, where the elec-
tronics package may need to be protected from the extreme cold of space until re-entry.
The quantity of heat that a phase change material can absorb can be found by
the equation

q  m am  hm c p, s ( T m  T i ) c p, l ( T 2  T m )

where:

q  quantity of heat stored (W)


m  mass of phase change material (kg/m3)
am  fraction melted (%)
hm  latent heat of fusion (J/kg)
T  absolute temperature (K)
Ti  initial absolute temperature (K)
Tm  melting point absolute temperature (K)
T2  final absolute temperature (K)
cp  specific heat (J/kgK)
cp,s  average specific heat between Ti and Tm (J/kgK)
cp,l  average specific heat between Tm and T2 (J/kgK)

© 2001 by CRC PRESS LLC


0082-06.fm Page 339 Wednesday, August 23, 2000 3:54 PM

FIGURE 6.15 Temperature distribution for ice forming on water, with ambient air as a heat
sink.

When a phase change material freezes, the problem can be considered to be one
of boundary condition values, as shown in Figure 6.15. We can see that the solution
to this problem is quite complex. Researchers have obtained only solutions for simple
cases. We can instead use an approximate solution that was obtained by assuming
that the heat capacity of the subcooled solid phase is negligible relative to the latent
heat of solidification. Another simplifying assumption is that the physical properties
are uniform and that the conductance and the heat sink temperature are constant
throughout the process. To find the time to form a solid phase of a specific thickness
we have:

 -----
L
2
L
 2k- --- U
-
t  T T -
---------------------
 -------------------
s o
-
  h fs 

REFERENCES
1. Hetsroni, G., Ed., Handbook of Multiphase Systems, Hemisphere, Washington, D.C.,
1982.
2. Bergles, A. E. and Rohsenow, W. M., The Determination of Forced Convection
Surface Boiling Heat Transfer, J. Heat Transfer, 86, 365, 1964.
3. Cole, R. and Rohsenow, W. M., Correlation of Bubble Diameters of Saturated Liquids,
Chem. Eng. Prog. Symp. Ser., 65(92), 211, 1969.
4. Cole, R., Photographic Study of Boiling in Region of Critical Heat Flux, AIChE J.,
6, 533, 1960.

© 2001 by CRC PRESS LLC


0082-06.fm Page 340 Wednesday, August 23, 2000 3:54 PM

340 Thermal Design of Electronic Equipment

5. Graham, R. W. and Hendricks, R. C., Assessment of Convection, Conduction and


Evaporation in Nucleate Boiling, NASA TND-3943, National Technical Information
Service, Springfield, VA, 5/1967.
6. Bromley, L. A., Leroy, N. R., and Robbers, A., Heat Transfer in Forced Convection
Film Boiling, Ind. Eng. Chem., 45, 2639, 1953.
7. Zuber, N., Tribus, M., and Westwater, J. W., The Hydrodynamic Crisis in Pool Boiling
of Saturated and Subcooled Liquids, Paper 27, Int. Dev. in Heat Transfer, ASME,
New York, 1961.
8. Florschuetz, L. W. and Chao, B. T., On the Mechanics of Vapor Bubble Collapse, J.
Heat Transfer, 87, 209, 1965.
9. Plesset, M. S. and Zwick, S. A., A Non-Steady Heat Diffusion Problem with Spherical
Symmetry, J. Appl. Phys., 23, 95, 1952.
10. Nukiyama, S., The Maximum and Minimum Values of Heat Transmitted from Metal
to Boiling Water Under Atmospheric Pressure, J. Jpn. Soc. Mech. Eng., 37, 367, 1934.
(Translation: Int. J. Heat Mass Transfer, 9, 1419, 1966.)
11. Drew, T. B. and Mueller, C., Boiling, Trans. AIChE, 33, 449, 1937.
12. Yamagata, K., Kirano, F., Nishiwaka, K., and Matsuoka, H., Nucleate Boiling of Water
on the Horizontal Heating Surface, Mem. Fac. Eng. Kyushu Imp. Univ., 15, 98, 1955.
13. Rohsenow, W. M., A Method of Correlating Heat Transfer Data for Surface Boiling
of Liquids, Trans. ASME, 74, 969, 1952.
14. Collier, J. G., Convective Boiling and Condensation, 2nd ed., McGraw-Hill, New
York, 1981.
15. Danielson, R. D., Tousignant, L., and Bar-Cohen, A., Saturated Pool Boiling of
Commercially Available Perfluorinated Inert Liquids, Proc. ASME-JJME Therm. Eng.
Joint Conf., 1987, 419.
16. Hetsroni, G., Ed., Handbook of Multiphase Systems, Hemisphere, Washington, D.C.,
1982.
17. Kutateladze, S. S., On the Transition to Film Boiling Under Natural Convection,
Kotloturbostroenie, 3, 10, 1948.
18. Zuber, N., Hydrodynamic Aspects of Nucleate Boiling, Ph.D. dissertation, University
of California, Los Angeles, 1959.
19. Lienhard, J. H. and Dhir, V. K., Hydrodynamic Prediction of Peak Pool Boiling Heat
Fluxes from Finite Bodies, J. Heat Transfer, 95, 152, 1973.
20. Cichelli, M. T. and Bonilla, C. F., Heat Transfer to Liquids Boiling Under Pressure,
Trans. AIChE, 41, 755, 1945.
21. Ellion, M. E., A Study of the Mechanism of Boiling Heat Transfer, Mem. 20–88, Jet
Propulsion Laboratory, Pasadena, CA, 3/1954.
22. Bernath, L., A Theory of Local-Boiling Burnout and Its Application to Existing Data,
Chem. Eng. Prog. Symp. Ser., 56(30), 95, 1960.
23. Zuber, N., On the Stability of Boiling Heat Transfer, Trans. ASME, 80, 711, 1958.
24. Kovalev, S. A., An Investigation of Minimum Heat Fluxes in Pool Boiling of Water,
Int. J. Heat Mass Transfer, 9, 1219, 1966.
25. Berenson, P. J., Film Boiling Heat Transfer for a Horizontal Surface, J. Heat Transfer,
83, 351, 1961.
26. Lienhard, J. H. and Wong, P. T. Y., The Dominant Unstable Wavelength and Minimum
Heat Flux During Film Boiling on a Horizontal Cylinder, J. Heat Transfer, 86, 220,
1964.
27. Bromley, L. A., Heat Transfer in Stable Film Boiling, Chem. Eng. Prog., 46, 221,
1950.

© 2001 by CRC PRESS LLC


0082-06.fm Page 341 Wednesday, August 23, 2000 3:54 PM

Heat Transfer with Phase Change 341

28. Frederking, T. H. K. and Daniels, D. J., The Relation Between Bubble Diameter and
Frequency of Removal From a Sphere During Film Boiling, J. Heat Transfer, 88, 87,
1966.
29. Mills, A. F., Heat and Mass Transfer, Irwin, Chicago, 1995, 640.
30. Frederking, T. H. K. and Clark, J. A., Natural Convection Film Boiling on a Sphere,
Adv. Cryog. Eng., 8, 501, 1962.
31. Lienhard, J. H. and Eichhorn, R., Peak Boiling Heat Flux on Cylinders in a Cross
Flow, Int. J. Heat Mass Transfer, 19, 1135, 1976.
32. Klimenko, V. V., A Generalized Correlation For Two-Phase Forced Flow Heat Trans-
fer, Int. J. Heat Mass Transfer, 31, 541, 1988.
33. Gilliland, E. R. and Sherwood, T. K., Diffusion of Vapors into Air Streams, Ind. Eng.
Chem., 26, 516, 1934.
34. Rohsenow, W. M. and Choi, H., Heat, Mass, and Momentum Transfer, Prentice-Hall,
Englewood Cliffs, NJ, 1961.
35. Chun, K. R. and Seban, R. A., Heat Transfer to Evaporating Liquid Films, J. Heat
Transfer, 93, 391, 1971.
36. Addoms, J. N., Heat Transfer at High Rates to Water Boiling Outside Cylinders,
D. Sc. thesis, Massachusetts Institute of Technology, Cambridge, MA, 1948.
37. Nusselt, W., Die Oberflachenkondensation des Wasserdampfes, Z. Vergle. D-Ing., 60,
541, 1916.
38. Rohsenow, W. M., Heat Transfer and Temperature Distribution in Laminar Film
Condensation, Trans. ASME, 78, 1645, 1956.
39. Sadasivan, P. and Lienhard, J. H., Sensible Heat Correction in Laminar Film Boiling
and Condensation, J. Heat Transfer, 109, 545, 1987.
40. Chen, M. M., An Analytical Study of Laminar Film Condensation. I. Flat Plates.
II. Single and Multiple Horizontal Tubes, Trans. ASME, Ser. C., 83, 48, 1961.

© 2001 by CRC PRESS LLC


0082-06.fm Page 342 Wednesday, August 23, 2000 3:54 PM

© 2001 by CRC PRESS LLC

You might also like