Boiling: Novak Zuber Myron Tribus
Boiling: Novak Zuber Myron Tribus
Boiling: Novak Zuber Myron Tribus
11 Novak Zuber
Myron Tribus
II Department of Engineering
University of California
Los Angeles, California .
This report has been reproduced directly from the best available copy.
Printed in USA. Price $2.00. Available from the Office of Technical Services,
Department of Commerce, Washington 25, D. C.
. . . . . ..
: . . : . ., . . ... .. . .
. . .
AECU-3631 .
. . . . .. . . . . . . ....
: .FURT.HER'REMARKS ON.THE. ,. . ; :
. . . : . . . .
. . .. .. .. . .. .. . . . .. .. . .. . . ... . ." r . . ,
.. .. . . . . .: .. .. .. ., .. . . . . . . .. . ' .: ...
i . . . . .. . .. . . : . .
. . , .. . . .
' 1 ;
. . . . , .. '
. .,,' " ;,. :Novak ,Zuber
:f .. .. . .
,
Myron T r i b u s
.
Department of Engineering
University of C alif ornia
L o s Angeles
iii
ACKNOWLEDGMENT
Page
REGIMES OF BOILING HEAT TRANSFER,.............................1
. .
the Kutateladze and Borishanskii crit6ria f o r the " burnout" .heat fl& a r e
derived from the theory. An analytical expression for the.empirical func -
tioli.which appears in the correlation of Griffith is also derived.,
. .
vi
idealized system, .for uniform and nonuniform
.. heat transfer rates i s dis-
:. .... '.
,! '
, ., s. w p. l
The detailed and general features of this .e!..,idealized system
. . ,,,.
,
.,
T
;:
vii
~,
. . . . . .. . . .
--
. . . -
a :therfial diffusivity . . . .. i : :' . . . .
.:
. . .
. ,
. . ..
c = : specifict'heat'atconstant pressure. '.' '
. . . .. .. ... .. . . . , ? . "
.
f . - -'frequ'en'cy',
: :.., . ' .
' . . . . . . .) .
; . , .
. . . . .
. ..
. >. ..
, .
, . . :
g = acceleration due to gravity
h = enthalpy
k =. thermal conductivity
L = latent heat of vaporization
m . = wave number
n = angular frequency
P = pressure
AP = vapor pressure difference corresponding to the liquid superheat
temperature difference (T, T,) -
P = critical thermodynamic pressure
' q = heat flux
R = bubble radius
0
= . 2 o / A P = s a d i u s of critical bubble
Tw = temperature of the heating surface and, consequently, of the
superheated liquid in contact with it
Ts = saturation 'temperature corresponding to pressure on system
T~ = temperature of the bulk liquid
uv -
- . velocity of vapor
UL = velocity of liquid
P = mass density
A. =. wave length
u = surface tension
T = period '
Cr = viscosity
Subscripts
L = liquid .-
v = vapor
viii
.REGIMES OF BOILING HEAT TRANSFER
bubbles which induce a strong agitation',of the liquid'near the heating sur-
face. . This. mechanism which was f i r s t proposed by' Jakob2 was further
investigated and confirmed' by Gunther a d Kreith'= and. by Rohsenow and
Clark. The important factors to consider; therefore, a r e the bubble
population,
. .
the r a t e of bubble formation and the dynamics. The last two
f a c t o r s a r e ifunction of t h e liquid iuperheit temperature (T', T. 1. How- -
ever, for given
' a. .'
heat flux this temperature difference is a function of the
agitatick
. . .
which turn depends.upon the bubble populati6n. ~ h i aspect
I
k'
, of
the pkoblem was excellently discussed a by Courty ahd FousteB
They point out that if i t is assumed that each bubble makes its own
11
quantized" contribution to the agitation of the liquid and, therefore, to
the heat flux, then for any given system the boiling curve in the nucleate
region must be an expression .. .
for the total number
. ,
of bubbles o n the. surface
i t a given superheat. consequently, the slope of the boiling curve in the
nucleate region represents the change in the liquid superheat temperature
5
(Tw -
T,) necessary to alter the number of bubbles in such a way a s to
accommodate a new heat flux. In other words, it appears that for a -
given heat flux the liquid superheat temperature is not an independent
variable but is a sensitive'function of surface nucleating condition which
governs the bubble population.
The variations of the liquid superheat temperature caused by a
change of surface conditions may be small compared to the temperature
of the heating surface. Therefore, in many design problems these varia-
tions can be neglected when the surface temperature is estimated. How-
ever, a s all proposed correlations of heat transfer rates in nucleate
boiling a r e of the form
-
these variations of the liquid superheat temperature (Tw Ts) become
very important when the heat flux is to be computed. This result is a
consequence of the fact that the exponent n is very large ; values between
58 6
3 and 24 have been reported in the literature. It should be noted
also, that in the above equation, both the constant and the exponent n
depend on the condition of the surface. Indeed, it is possible to change
their values by just changing the surface polishing procedure. '
*
To complicate the problem further, the value of the exponent changes with
time because surface characteristics change with prolonged bbiling, i. e. ,
.
the.s h f a c e exhibits the effect of 'I aging"
Zuber. lo Some limited success has. been had and we shall now briefly dis-
cuss this approach in order to point out why when a l l things a r e considered
this line of attack cannot be expected to provide any 'information concerning
the peak nucleate heat flux, i. e., the information in which designers of
apparatus a r e most interested.
In nucleate boiling heat transfer rates appear,to be independent of the
geometry of the system (provided the system is not too'small) because the
bubbles induce strong localized agitation in the superheated liquid film ad-
jacent to the heating surface. Consequently, all proposed correlations a r e
formulated by considering dimensionless ratios which utilize bubble dimen-
sions rather than the dimensions of the apparatus. In this way there have
been proposed a Nusselt modulus, a bubble Reynolds modulus and a Prandtl
'
modulus for the liquid. In pool boiling ~ l l i o n has shown that the bubble
radius and radial velocity (which he .determined from experiments) furnish
a suitable characteristic.length and velocity for the Reynolds number of the
flow adjacent to the heating surface. EiQon, for example,, substituted these
measured values of bubble radius and velocity in the usual equations for
I1
pipe flow" with the results shown in Figure 3.
Inasmuch '.as the problem of the growth of a bubble in a superheated
. . . .. .
7
T H E DEGREE OF
AGITATION
OF THE LIQUID THE HEAT
FLUX
1
cn
+
0
THE SUPERHEAT
W AT THE WALL
Lr.
Lr.
4 .
THE BUBBLE
POPULATION AND
DYNAMICS
-.
1
t < n
A +
5 Ow
< WLr.
a ~ar .
~3
L
THE SURFACE
CONDITIONS
b
_I
0
The product of the bubble radius R and velocity R is then time independent
These two dimensionless groups include aspects of two .of the important
phenomena: the bubble dynamics and nucleation.
.
When Equations (4) and (3) a r e substituted for the characteristic
length and velocity in the bubble Nusselt and Reynolds moduli a relation
between the heat flux and the liquid superheat temperature (Tw - T 6 ) is
obtained in t e r m s of the thermodynamic properties of the vapor and the
liquid:
:;.
Recently Perkins and Westwater17 have reported that the heat transfer rates .
of the problem, :no account is taken, of bubble interaction, but rather the
dimensionless groups a r e derived from consideration of a single bubble. Such
a restriction applies
. . not:... only to
..
Equation
. . , . . .
(5) but:also
: .._:
to all .,...correlations. which
, ' . .- . ., . .
a r e derived by considering the effect of a single bubble. It-seems reasonable
that if a quantitative,understanding of the change from nucleate to transition
. . .
Figure 5. An Empirical Fit of Data in the Nucleate Regime as
Obtained by Rohsenow, Griffith and Berenson. l 7
boiling is desired, and if the critical heat flux is to be predicted from an
equation for the nucleate portion of the curve, the bubble population has to
be taken into account. This is a consequence of the fact that limitation
on the maximum nucleate heat flux is brought about because of the mutual
interference of many bubbles. It has been noted that no information con-
cerning bubble population is now available either.from experiments or
theory.
We shall show in thefollowing sections that even if the bubble
. .
population were known it would.'not permit one to solve the 11 burnout"
problem. ''>'ASwe shall see, the I' burnout" is caused by interactions which'
occur away from the surface. This point of view originated with
Kutateladze and .has been adopted by many (but.not all).Russian workers.
In the next section the matter will be taken up in detail.
-
PREVIOUS EMPIRICAL CORRELATIONS OF PEAK NUCLEATE HEAT FLUX
. .
~ a n ~ h - i v e s t i ~ a thave
o r s adopted the view that the peak nucleate
. . heat
flux marks a transition from one regime to another (just a s the critical
. .
Reynolds modulus marks the transition from laminar to turbulent flow)
and have proposed equations which do not derive from the behavior of the
nucleate region alone.
Such contributions a r e to be found in the work by ~utateladze",
' 2l2 and by Rohsenow and Griffith.
Sterman2O , ~ o r i s h a n s k i i ~ The analy-
sis of?Kutateladze will be discussed in greater detail because he was appar-
ently the f i r s t to propose that the peak nucleate heat flux was caused by the
hydrodynamic instability of two phase flow. It will be seen, also, why
Kutateladze' s and Borishanskii' s method, which consists of analyzing the
maximum heat flux by considering the nucleate region, necessarily leads to
dimensional analysis and a t least one empirical constant. The advantage of
an analysis of transition boiling a s opposed to nucleate boiling will be:.seen
in the next section, for such an analysis yields an analytical expression for
this heat flux without recourse to experimentally determined constants.
Kutateladze noted in 1951 that " the essential feature of the theory of
the phenomenon can be derived if one assumes that the c r i s i s in the boiling
process is purely a hydrodynamic phenomenon: the destruction of stability
of two-phase flow existing close to the heating surface. " According to this
view the change from nucleate to transition boiling occurs when the velocity
in the vapor phase reaches a critical value. Starting from the nonlinear
Euler equation of motion and the energy equation he derived, by dimensional
analysis, the following equation for the peak nucleate heat flux in pool boiling
of liquids at saturation,
The constant K was, determined from experiments and i t s value was found to be
The same equation, using dimensional analysis but a different thought
. . . .
model, 'was rederived by Sterman.20
'In two recent papers Borishanskii2" 2 2 extended Kutateladzel s
dimensional analysis to include the effects of viscosity. In the discussion
he points .out several interesting aspects of nucleate boiling 'close to the
maximum heat flux. These comments a r e worth repeating here. ~ c c o r d -
ing to ~ o r i s h & s k i ithe continuous kxistenc e of steady heat transfer rates
in the neighborhood of the peak nucleate heat flux leads to the conclusion
that there exists a direct steady movement of liquid toward the heating
surface and of vapor away f r o m it. Because of the density difference be-
tween the two phases, Borishanskii reasoned that a larger part of a c r o s s
sectional a r e a close to the heating surface is occupied by vapor than by
liquid. Therefore, near the heating surface one can consider a two-phase
boundary region whose thickness is of the order of a disengaged bubble.
This two-phase boundary region may be visualized a s consisting of liquid
s t r e a m s flowing toward the surface and surrounded by vapor. The shapes
of the filaments of liquid as they flow towards the wall a r e not well defined
'
because of the inherent randomness in the bubble dynamics and coalescence
in the nucleate regime. (This limitation disappears in transitional boil-
ing a s will be shown. )
At low nucleate boiling heat transfer r a t e s the discrete phase
appears a s a vapor 'bubble surrounded by a mass of liquid;, whereas close
to the peak heat flux i t is rather a liquid stream surrounded by vapor. Con-
sequently, . an analysis of nucleate boiling below the peak flux should con-
s i s t of an analysis of bubble.formation, while close to the maximum flux
the analysis should be directed toward the dynamics of a liquid stream fila-
ment bounded by a group of bubbles. The change from nucleate to transition
'
boiling occurs when the steady flow of the liquid towards.the wall is dis-
rupted, i. e., when at a critical velocity in the vapor phase, the liquid streams
a r e destroyed. Borishanskii noted further that: I I This problem seems
analogous from a theoretical point of view to the disturbance of steady flow
of a liquid stream in gas which is moving coaxially with ,it. The solution of
?I>
3
the problem leads to a relationship between the increment of the oscillation
' and the wave length. F'urther analysis of the equation for the limiting c a s e
of stable flow leads to the conclusion that the critical boiling point corresponds
to the establishment of a definite geometrical structure of the two-phase
boundary layer. 11 Borishanskii, theref ore, considered' that .the phenomenon
was to be explained by analyzing the stability of a liquid jet surrounded by a
moving, coaxial, vapor phase. F r o m the equation which determines the am-
plitude of the w a v e h d from the energy equation he-established, by dimen-
sional analysis, two similarity criteria: K given by Equation ( 6 ) a s previously
found by Kutateladze, and N, given by
P L =/a
. N = (8)
I r2 [gb, - Pv ) ] q2
K = 0 . 1 3 + 4N'Oo4 (9)
In order to establish the above correlation 117 experimental d& points were
used. The experimental data represent the following liquid- solid combination:
water boiling on a graphite surface; ethanol, benzene, n- heptane and n-
pentane boiling on a chromium plated surface; l5 ethanol2"nd water2= boil-
ing from a nichrome surface. The correlation and data a r e shown on
Figure 6 which is reproduced from Borishanskii' s paper. The viscosity
of the liquid appears in Equation ( 9 ) only in the additive correction factor N.
Inasmuch as the deviation from a horizontal line is small it can be seen from
Figure 6 that the effect of the viscosity is also small.
Good agreement with experimental data was achieved, also, by the
correlation- proposed by Rohsenow and Griffith:
1 0 ~ 1 2 16 2 3 4 5 6 78910'12 16 2 3 4 5 6 7 8 9 1 0 ~ 1 2 16 2 3 4 5 6 7 8
N
1 water 6 heptane 11 ethanol
2 benzene 7 propane 12 heptane
3 ethanol 8 propane 13 water
4 pentane 9 water
w
03 5 pentane 10 benzene
ture differences play no role. One would expect that the complexity of the ,. ,..,
heat transfer process would increase a s the bubbles become more dense and -\
interact more; the fact that the opposite is true indicates that a new mechan- *
A S-
ism enters and dominates. The hypothesis that a new mechanism governs -
the phenomenon gains credence when i t is recognized that the critical heat TI>
flux not only marks the end of the nucleate region (upon increasing surface
temperature) but equally well denotes the beginning of the transitional region. (., -l$,
It is an often observed but seldom discussed fact that the decrease in heat
transfer that accompanies an increase in surface temperature is found only
in transitional boiling and not in any other heat transfer system. It is there-
fore reasonable to propose that insight into the new mechanism can be ob-
tained by considering transitional boiling.
It will b,e seen that a well defined geometry is the characteristic of
transition boiling. Therefore, an analytical solution of the problem can be
attained." ' In nucleate boiling the flow configuration is not well defined.
As noted by Borishanskii, 1 1 the form of the liquid s t r e a m s differs considerably
since i t is determined by the not- well regulated order of the combination of
'. vapor bubbles. 1 1 Given a field equation and no geometrical data, Kutateladze
and Borishanskii found i t necessary to use kmensional analysis, which is the
best that can be done under the circumstances and which reduces the problem
to the determination of an empiric a1 constarit.
THIS PAGE
WAS INTENTHONALLY
LEFT BLANK
TRANSITION BOILING OF LIQUIDS AT SATURATION
mum and minimum heat flux in transition boiling from a horizontal surface.
Transition boiling in general will be discussed in a later publication.
To facilitate the presentation, the phenomenon known a s Taylor in-
stability is f i r s t discussed.'
When two liquids of different densities and having a common interface
a r e accelerated in a direction perpendicular to the boundary, any small irreg-
ularity of the interface will tend to change in shape. The interface is stable,
i. e., the irregularities will tend to smooth out in time,if the acceleration is
directed from the denser to the lighter fluid. The interface is unstable, i. e.,
* . e..
the irregularities of the interface will grow with time, when the acceleration
is directed from the lighter to the heavier medium. A familiar example of
this is a glass of water turned upside do-. The air-water inter-
face instead of remaining plane a s it falls will tend to form long liquid spikes
which subsequently disintegrate into drops. An analysis of this phenomenon
was developed by Taylor using potential theory and neglecting surface tension
and viscosity. He found that for an unstable interface the initial growth of
the amplitude of the disturbance is exponential in time. The stabilizing effect
exerted by the surface tension was recently demonstrated by Bellman and
Pennington. 30 Their analysis indicates that disturbances of the interface
can be stable o r unstable according to whether the wave length is shorter o r
longer than a critical value. This critical wave length is given by Kelvin' s
equation (3 1):
They show, also, that the "most dangerous" wave length, i. e., the wave
length for which the amplitude of the disturbance grows most rapidly is
. .
given by
Bellman and Pennington further point out that "This phenomenon explains
the hanging .of water droplets on the underside of a horizontal surface, such
a s a ceiling, Such a droplet is undergoing an upward acceleration of 980
2
c m / s e c and w i l l tend to drip because of Taylor instability unless i t s
effective wave length is too small to satisfy Equation (11). F o r water, the
critical wave length is about Xol= 1. 73.cm. Droplets of larger diameter
will tend to drip, while smaller ones will tend to hang. " Another evidence
11
is cited by Lamb and Milne-Thompson who point to experiments in ' .
-
-
-'
, ' , - '-: ---
I '
*"
e% iC! no*+:&
onsider saturated liquid on a 1 [ ~ ~ i z o n t asurface
l in the
nucleate boiling regime. As the heat flux is increased a value is reached
when, because of bubble interaction and of the disruption of the liquid
streams, an instantaneous vapor film is formed. This value corresponds
to the peak nucleate heat flux. The vapor liquid interface of t h i s film Is
hydrodynamically unstable because the acceleration is directed from the
less dense to the more dense medium. It can be expected that, because of
agitation, the interface has random initial perturbations, distributed con-
tinuously over a spectrum of wave lengths. For a two-dimensional system
all perturbations with wave length longer than the critical one, i. e. , than
Xol a r e unstable. In view of the initial, exponential growth it can be
expected that these unstable perturbations will dominate and that those
near Xo2 will be the first to achieve finite amplitude. Therefore, as a .
consequence of Taylor' s instability a definite geometric a1 configuration
in transition boiling can be expected. F o r a two-dimensional system this
geometry should be characterized by disturbances with wave lengths in the
spectrum
24
Figure 8 shows a high speed still photograph of the nucleate regime. *
There is no regularity to the photograph. Figure 9 shows the simultaneous
nucleate and transitional boiling. To the left the system is quite irregular;
in the center and to the right it will be seen that the vapor shoots vertically
upward in three columns, more o r less equidistant.
At the conditions represented by the right half of Figure 0, one should
11
analyze the fall of the I' spikes" of liquid rather than the " rise of bubbles.
In their downward fall the spikes approach the heated surface and rapid
evaporation occurs. The question now arises a s to whether there is a solid-
liquid contact in transition boiling. The similarity between transition boiling
II
and the " spheroidal state, " also known a s the " Leidenfrost phenomenon,
w a s first pointed out and discussed by Drew and Mueller. They concluded
$!
that the collapse of the spheroidal state corresponds to tke change from trans- Y
$
the heated surface in the nucleate region at the instant of formation of the
c ,,
unstable vapor film does not necessarily correspond to the temperature at the # ;
instant of collapse of transition boiling. This important fact was clearly , 1i1
Q.
demonstrated by Perkins and Westwater. l7 They observed that the change t ' ,, ;:
from nucleate to transition boiling occurs over a temperature range instead
of at a definite critical temperature. The peak heat flux is constant throughout
this range of temperature. Therefore, in the q-T plane instead of a critical
point there is a " critical region", i. e. , a plateau. The temperature at the
two extremities of this plateau a r e of the same order of magnitude a s correctly
observed by Drew and Mueller. 3 7 Because of this region a simultaneous
coexistence of nucleate and transition boiling can be expected. It appears,
therefore, that in transition boiling no solid liquid contact exists. This fact
is confirmed by the moving pictures of Westwater and Santangelo.
-
'
= - >I. -'
-
'
-
- 7 -
d 8hultaur-I
Figure 0. P b t ~ g r a P #T 1eah (ldt ~ e Isd
) Trmoitisd
(right side) Boiling. OIIwall A T a BL , q = 171.000 tern
hr
As the liquid evaporates from the spikes the vapor flows in the
region between two spikes. It was noted, already, that this region has
a form of a rising bubble. Because of the Taylor instability a release of
bubbles at regular intervals can be expected. F o r large evaporation
rates the release of a bubble will appear a s a burst. As a row of bubbles
is released an unstable interface is again formed. Because of the down-
ward flow of the liquid a spike will be formed now underneath the re-
leased bubble and the process is renewed. The process fs, therefore,
thermally stable because it can continue indefinitely. However it is
hydrodynamically unstable, indeed, it is this instability which is the cause
~f the phenomenon.
The characteristics of transition boiling will be summarized now
and qualitatively compared to experimental data.
1) As a consequence of Taylor instability a definite geometrical
configuration can be expected in transition boiling. For a
two-dimensional system this geometry is characterized by
disturbances with wave length in the spectrum
-
!, ,=$
8
I,
-
L
The maximum and the minimum heat flux correspond, therefore,
to the maximum and the minimum allowable frequencies
system.
i of the
' I
.
rr
.
i
,
I '
-
I
-
'
-
_ -
- ,, ,; - 8 -
(3t.
and striking photographs in this report. Indeed, they were fortunate to have c 8
the result of these outstanding experiments a s a guide f o r the theoretical
analysis. Westwater and Santangelo give the following description of transition
boiling: 1 1 Most prior workers have failed to realize that this boiling is entire- kk
ly different from both nucleate boiling and film boiling. No active nuclei exist. 9 .$
I
-IF
In fact, no liquid-solid contact exists either. The tube is completely blanketed
by a film of vapor, but the film is n o t smooth nor stable. The film is irregu-
lar and is in violent motion. 11
II
Vapor is formed by sudden bursts at random locations along the film.
Liquid rushes in toward the hot tube, but before the two can touch, a miniature
explosion of vapor occurs and the liquid is thrust back violently. The newly-
formed slug of vapor finally ruptures, and the surrounding liquid surges
11 I1
toward the tube. The process is repeated indefinitely. One observer of
these high speed motion pictures has expressed an opinion that occasional
liquid-solid contact does occur during transition boiling. If so, these contacts
a r e r a t e and of exceedingly short duration. The present writers do not believe
there is a r e a l contac s q a5 . 4ylpIt"~ ,F 7 - -a,
., --#
- ,.-
' TbXhe frequency of the vapor bursts Am**
I 8 \
71 .-3
is surprisingly high. F o r an
--,>
: - , , ',- * s- , _ Y . - lh;&
- - - - -& 8
.= A
- r 3 8 !- - b
- .
1
-I
overall T,- TS = 1 3 3 (and ~ ~ U of 164 Btulhr it2 OF) each inch of the
photographed side of the tube exhibited 84 bursts per second. The bursts
occur so suddenly and unexpectedly that even in slow motion they resemble
explosions. I 1
It is seen that the above observations are'in qualitative agreement
with the previous statements except for our previously mentioned observa-
tion of regularity, A qualitative agreement is obtained also when photo-
graphs, published by Westwater and Santangelol and reproduced in this re-
port, a r e analyzed. We have already remarked on the indeterminate flow
configuration in nucleate boiling even at moderate heat transfer rates a s
seen in Figure 8. This is even more evident at the peak nucleate heat flux
discussed in connection with Figure 9. The coexistence of nucleate on the
left and of transition boiling on the right hand side of the photograph of the
tube permits the comparison. The random spectrum of disturbances is
shown on Figure 0 and Figure 10. Spikes of liquid and rounded regions of
vapor a r e seen at the minimum heat flux in transition boiling which is
shown on Figure 11. The difference between Figure 11 and Figure 10 will
be discussed in the following section when the stabilking effect of the sur-
face tension and the destabilizing effect of large velocities in the vapor
phase, i. e., of Helmholtz instability, a r e considered. The regularity in
space and periodicity in time can be seen in Figure l l a .
In the following section the problem will be formulated mathemat -
ically and the results quantitatively compared with experimental data.
Figure 10. Photograph of TransitiW ~
q = 27,200 BtuIBr fta.
'
g A T* 124 9.
A -
Figure I l k Photograph of 1i.hPctiUg, A T = 181 9,
.
q = 6,47Q B f u / b ft
FORMULATION AND SOLUTION OF THE PROBLEM
and the number of bubbles per unit area per unit time is
where a. b is the area of interest. The vapor mass flow rate per unit area
is obtained from Equations (16) and (171, thus
At a constant pressure because the geometry does not change, Equation (19)
can be written a s
s.
constant . - I -
- (20)
9 = 7
.
,
.--'
-t ,;
I ) ' . . -
- L,, ' 8 -, 8 , - . .
-
I
-
- 7 .'
36 . v.. ,
allowable frequencies which, according to the hypothesis, correspond to
the minimum and maximum heat flux. These two regions are characterized
by a low and a high evaporation rate; consequently a small and a large ve-
locity in the vapor phase can be expected at the minimum and at the maximum
heat flux respectively. The stabilizing effect of the surface tension in c onnec -
tion with Taylor instability was discussed by Bellman and P e n n i n g t ~ n . ~It
was observed, also, in the experiments performed by Allred and Blount. '13
These experiments revealed, also, the effect of the Helmholtz instability.
This instability arises at the interface when there is a relative motion of two
fluids in a direction parallel to the interface. The mushrooming of the inter-
face observed in!,theseexperiments was attributed by Allred and Blount to
Helmholtz instability. It is important to note here that the geometry of the
interface observed in these experiments, i. e., the form of mushrooms, is
identical with the geometry of the interface in transition boiling. This state-
ment can be verified easily by comparing Figures 3.9 and 5.6 from Reference
36 (reproduced here a s Figures 12 and 13) with photographs published by
Westwater and Santangelo. In view of the above considerations it can be
expected that the peak flux in transition boiling is characterized by the com-
bined effects of Taylor and Helmholtz instabilities; whereas the minimum
flux is characterized by Taylor instability only.
It was noted that at the minimum heat flux the velocities in the vapor
and liquid phases a r e low and that the phenomenon is characterized by the
stabilizing effect of the surface tension. Consequently, it was assumed in
Reference (27) that the motion of the interface is governed by the surface
tension. The angular frequency of such a motion is:
and
Substituting alternately Equation (11)and (24), and Equation (12) and (25)
in Equation (10) it can be expected that, because of the spectrum of unstable
disturbances, the minimum heat flux in transition boiling from a plane,
horizontal surface can be determined within thd limits given by:
It should be emphasized here that the above equation holds for a horf-
zontal plane surface only. As discussed in Reference (2'71, the minimum
heat flux in transition boiling from a(horizontal tube depends on the sta-
bility of two systems of capillary waves. It is, therefore, determined by
the motion with higher energy requirement.
An analysis of the minimum heat flux in transition boiling from a
plane, horizontal surface can be formulated, also, by considering the
asymptotic gro* of Taylor instability. It will be seen in the following that
this formulation leads to an expression for the heat flux which, at low
-
pressure, exhibits the same dependence on liquid properties as Equation (26).
*
Allred and Blount s6 have reported that the asymptotic growth of
Taylor instability can be predicted from the equation, derived by Davies and I
Taylor, S g for.- the
, r i s e of a large, two-dimensional bubble:
..
..
I where r is the radius of curvature of the bubble vertex.
' <
.- .'; If one considers.'that the bubbles follow one another very closely, then
I to a good approximation: . .
. .
: . . . . . , .
I -.
Using Equation (29) for the frequency, t h e m a s s flow r a t e of the vapor associa-
41
It will be seen that the heat transfer rates.predicted by Equation (2 6) and
by Equation (31A) a r e in agreement with experimental data at atmospheric
pressure. This could have been expected inasmuch as both equations ex-
hibit a t low. p r e s s u r e s the s a m e dependence on the density, since $, < < p L
At high pressure, however, this is not the case. Therefore, experimental
data a t high p r e s s u r e a r e needed to show which equation is a better approx-
imation to the phenomenon.
We shall investigate now the peak flux in transition boiling. It was
noted, already, that the peak flux is characterized by the joint effect of
Taylor and Helmholtz instabilities. The problem is to determine in what
way they manifest themselves and interact. It was noted, also, that phenom-
enon b e a r s a similarity to a r e l e a s e of bubbles with variable frequency f r o m
a s e t of regularly spaced orifices of fixed geometry. In accordance with
the hypothesis, at the peak flux the frequency reaches the maximum. In
view of Taylor instability we a r e lead, therefore, to consider a vapor column
i. e . , a jet of diameter X 1 2 . In view of Helmholtz instability we have to
.
investigate the stability of this jet, i. e., to determine what is the maximum
velocity in the vapor phase which permits a periodic behaviour. In accordance
with the previous analysis the stability of a two-dimensional system is again
investigated.
The analysis is most easily made if i t is recalled that the stability
of these s y s t e m s is usually investigated by considering the stability of small
31 32
disturbances. Whether one considers the stability of a plane jet o r a
round jet will make little difference, provided i t is the axial disturbances
which a r e investigated. Since the plane surface is mathematically simpler,
i t is taken here.
- . ' PV
( u V + U L2 )
PL+Pv (P, + P v)
Rayleigh 38 has shown that the critical wave length for. a cylindrical jet':is** .
. .
., ,
where R is the radius of the jet. I£ the jet is to break into spheres, the
wave lengths must be th.e. same axially a s circumferentially, hence
equation (35) is sibstituted in Equation (34) to a relation between
, .
..
* As pointed, out by Birlchoff ' O in .considering t h i i problem, one may always
adopt a coordinate system such that (33) is true.
a* Note that X r e f e r s to the wave lengths of disturbances in the interface
between ascending and descending streams.'. . . .X 01 . . X o 2 r e f e r to wave-
and
lengths in the horizontaf interface."
the velocity and wave length.
F r o m the relation between the wave number and the wave length
and f r o m the assumption concerning the diameter of the jet, i t follows
that the critical wave number in Equation (34) is given by
mo
. . . . . . . . .
. . .
the critical velocity in the vapor phase becomes . .
The relation between the critical velocity and the frequency is obtained
f r o m the equation of continuity, which, because of the double frequency*
, ; : .is given by
I
I . .. ,.
...... . . .
Pi
#
i;
...
. .
(+
3
. = i. .(";) 2
uv
~ftek
cancelling common t e r m s ~ ~ u a U o n yields:
(38)
.._ . . > .. ,
' .a
. .: . ... . . . . .. .. . . .. . . : . . . . . .
.
.
~
.
.. a
.
.. .,, . . . . . . .. . . ..
* i. e. , allowkg for the two t r a i n s of . ,
waves; at, iight. .angles.
~
: I ,.
..........
:%
I The peak flux is obtained by substituting Equation (40) into Equation (19). 4
. . '
and
The algebraic mean is;
. . . .
therefore, .,' .:.
. I-
. .
- ..
, . . . . . .. .. . . .,.'
, I ,
;
. . . . .. . . . .!.
.. " .
A convenient average value for the peak flux can
.
be
.
obtained by replacing
. t . . . .
m, in Equation (41)by the upper limit obtained f i o m ' ~ i u a t i o n(42); arid'
by approximating the numeric aL constant 3 1 f i by' unity. The resulting equa-
. .
tion is then of the form previously derived
. .
The value, of the numerical constant in this c a s e is
In order to test the validity of the above analysis the detailed and gross
features of the thought . .
model will..be compared, in ...the following
. . . . 7 ....
section with
experimental data. The experimental results reported by Westwater and
~ a n t a n ~ e l l oin' this case, again, a r e invaluable.
. .
In the following the theoretical predictions a r e compared with experi-
mental data for boiling methanol a t a t i n ~ s ~ h e r i c ' p r e s s ureported
. -
re .by Westwater
.1
and Santangelo.
. , . .
Eq. (24)
Eq. (25)
Eq. (30)
Analysis Experiment
.Btu/hr ft2 Btulhr ft2
,
As a further check of the hypothesis the ratio of the periods com-
' "
puted f r o m the analysis will be compared with the ratio of the maximum ahd
minimum heat fluxdetermineid from experiments. The period at the peak
flux is obtained by substituting .alternately Equation (.l'l.)
and Equaticiq (12)
into Equation (401, thus
Borishanskii:
Eq. (48) . K = --= 0,131 K = 0.13.
. . 24
. .I. .
.
a r e indicated on ~ i ' g u r e6.
* . ..
This uncertainty should i d n o wise be considered a s a defect in the theory
. . . ,
Equations (43): and (48) which predict the peak flux in transition boiling
were determined from stability considerations and an energy balance for
liquids.at saturation temperature. ' To extend the analysis to subcooled liquids
a basic assumption will be made: The change from nucleate to.transition boil-
ing is determined only by the hydrodynamic stability in pool boiling, i. e.,
I
in the absence of a forced flow velocity, the change occurs when the vapor flux
attains a given value. Therefore, if the liquid is at saturation and the viscosity
is neglected, the heat transferred across the solid surface is equal to the
energy required for the generation of that particular vapor mass flow. Since
the stability is the mechanical aspect of the problem i t will be unaffected by
whether the liquid is subcooled o r not, but depend only on the mass transport.
However, as the heat flux across the solid surface is determined by an energy
balance i t will depend on the subcooling. Therefore, all energy requirements
will appear a s additive t e r m s to the energy needed for the generation of the
critical vapor flow. The problem is now reduced to the determination of
energy requirements associated with a boiling, subcooled liquid.
It was already noted by Kutateladzelgthat when the degree of subcooling
is low, ,bubbles will depart from the liquid-vapor interface. The m a s s of vapor
is replaced by an equivalent m a s s of liquid which, a s the bulk temperature is
subcooled, has to be heated f i r s t up to the saturation temperature before
evaporation can occur. It was shown that Equation (48) is a mean value of
,
Consequently, the heat flux density associated'with this enthalpy change of the
liquid is
,
The energy (q ) that is transferred from the film interface (which is a t
saturation temper,ature) to the subcooled liquid will now be es.t@nated.
.. . ..
The problem of t h e transfer of energy between avapor-liquid interface
and the, body of the liquid and the analogous. mass transfer problem,,
i. e., the absorption of a gas into an agitated liquid, were,recently:
analyzed. .' . The~,Mportantsimplificatjon introduced in these . -
4s
papers is the concept of a I' contact time"" :or 1 I exposure time I t
,, ,
& J . ,,
. . . .
where n is the normal to the interface.
. .
. . .
.
. .
. .
I -
from the interface t o the subcooled liquid becomes upon integration
. .
and the
. .
ratio of the wave l'ength,to the period by
The heat transferfed across t&e solid surface is obtained again from
I
an energy balance, i. e., by adding (Equations (48); (55) and (58). Conse-
quently, the critical heat flux in transition boiling of a subcooled liquid is '
! ,
expected to be given by
. , .
where r and X 0 / .T a r e given b y Equations (59). and (60)respectively.
, .
,
. .
;
. .. . , '
. . , .... . ... ;, .
,.:::,,9
.. ,<
.. . . .. . . . .. :, .:., . . . . , : , .'f.,.',
,
%. :,
..
"_
I
Equation ( 61) can be used.as a f i r s t approximation for thes.e su.rf;aces
I -,
als 0. . . . . . .. . ,
reference 47, Leppert, Costello &d Hogiund give data.on heat transfer to '
,
. . .
.
t
S . L
.
Peak
I-
eat Flux .
4. 55-- 5.12 x
..,:
7 13.4 lo6 5.28x105
J.".
.; .:;:,
.., .:
..,.: ...
.
. .
:' ..~
Before concluding this section we shall explore the possibility of
applying to the present problem the concept of "contact time" in a forced . :..
.. -
..;,
flow system. 41' 4.8 Denoting the forced flow velocity by V, the contact time . ,..,
.. .I;
. - ;.:c
. ;. '4 .. :!,<:..
':., .':
:
A0
t = - (62)
v
The energy associated with this forced flow system becomes, therefore
emphasized, however, that Equation ( 6 3 ) does not take into account the
effect of the. forced flow velocity on.the.stability. This effect,, as.wel1 a s
the effect of a restricted geometry. of"the conduit c,an:be expected t o have
an influence :on the peak flux. ,. Thi,s aspect of the problem still ,remains..to
be so1ve.d. . .The agreement shown on Figure 13 shou1.d .be taken, therefore,
only a s an encouraging indication concerning the possibility .of extending
the analysis to forced flow systems in closed conduits. P
Figure 17. Comparison of Theoretical with Experimental Peak '
C
The agreement of the analytical predictions, based on this simple
idealized system, with experimental data was shown in preceding sections.
How,.ever, more experimental data in transition boiling a r e needed in order
to establish the limitations of the two-dimensional approach. It was noted
already that an extension to three dimensions can be made easily. The
4 authors were unable to find experimental data for pool boiling of subcooled
liquids at high pressures and test Equation (60) in the high pressure range.
It can be expected, however, that for a constant temperature difference
(T, - T, ) the effect of subcoo1ing will decrease :with an increase of pressure
because the term X , / . T decreases while T increases with an increase of
pressure. ,
r
and the treatment of the interface a s though it were .plane was introduced a s
the simplest idealiz'ation. However, i t w i l l follow .that no matter what
geometry is postulated, the form of the resulting solution is unchanged and
only the constant (i. e:, will be affected in Equation ( 61). This change
will alter each of the slopes of Figure 16 in the same way. The group k f i
will still determine the relative slopes for the different liquids and the ratios
of slopes will be unaffected by the constant. The agreement in ~ i ~ u r 16 'e
concerning absolute a s well a s relative slopes serves to indicate that the con-
ception of the process is proper.
Similar remarks apply to the use of . a 11 contact time. 11
The left hand side represents the energy transferred from the surface,
whereas the right hand side is the energy required to bring about the
II
flow instability. It should be kept in mind that the burnout" is a local
phenomenon. I t will occur when the equality sign in ~ ~ u a t i(65)
& is
reached. F o r a uniformly heated conduit of constant c r o s s sectional
characteristics the lowest v a l i e of subcooling occurs at the' downstream
end which is, therefore, the location of the burnout point. F o r a non-
uniform, say cosine heat flux distribution a s occurs in a reactor the
maximum value of q occurs at x = $12 where 4 is the length of the heat-
ing element. At that point, however, the subcooling is still high, there-
fore the inequality sign in Equation (65) may still hold. The subcooling
is minimum at the downstream end, i. e., a t x = 8, but at this .point the
heat flux q is low so that again the inequality may hold. The burnout
point if i t occurs will occur at apoint wher'e the equality f i r s t occurs,
4 / 2 " < x < 8. A quantitative investigation of this aspect of the
problem. will be made. and.'reported a t ' a future. date.
-
we shall conclude this section by deriving from the above theory
the analytical expression of the empirical function which appears in the
correlation of the peak nucleate heat flux recently proposed by Griffith. 5 0
The correlation of Griffith has the following form:
where V is the
. . bGk velocity
. .<
, .
of the,.liquid
. ..
(forced flow). The coeffici,ents, .,
. .
Al ,.A2, A were also foundfrom
. ..'
experiments. In pool boiling
. . .
the bulk. . ,
is
, < ; ,
3
velocity zero (V =*0)and the coefficient A2was determined by
. .choosing:
. . ..
r e a r ranging
,
. .
AS previously mentioned Griffith 5 0 determined the value of the, .
:
,. .
. . . .A2 from experiments . with
constant . water.at i)tmosph&ric, pressure
.:. . . . ,
The.
value
- .
he foundwas
. .
A2 = 0.014. quat ti on( f d g i v e s f o r the same .
'
. . . . ,
11
Westwater, J. W., and J. G. Santangelo, I' Photographic Study of Boiling,
Ind. Eng. Chem. ,,August, 1955, p. 1605. . 1.
Jakob, M., Heat Transfer, Vol. I, John Wiley New York, 1949, Po 642-
10. Forster, H. K. and N. Zuber, " Dynamics of Vapor Bubbles and Boiling
Heat Transfer, " AIChE Journal, December 1955, p. 531.
. ,
11. Zuber, N. , The Rate of Bubble Growth in a Superheated Liquid, "
6
M. S. Thesis, Department of Engineering, University of California,
Los Angeles, January 1954.
Superheated Liquids, l 1 Jour. Appl. Physics, vol. 25, no. 4, pp. 493-500,
April
,. 1954.
14. ~ u b e r ,N. , I ' On the Correlation of Data in Nucleate pool Boiling from
a Horizontal Surface,
II
AIChE Journal, Sept.
. .
.
1957, pp. 9'. s-1 1 s.
15. C i c h e U , M. T. and C . F. Bonilla, " Heat Transfer to Liquids Boil-
ing Under P r e s s u r e , " Trans. AIChE, v01. 41, 1945, pp. 755-7878
27. ~ u b k r ,Novak, " On the Stability of Boiling Heat Transfer, " ASME
P a p e r No. 57-HT-4, Pennsylvania State ASME-AIChE Heat T r a n s f e r
Conference, August, 1957. . . .
.. .
11
28. Zuber, Novak, On the Maximum Heat Flux in Pool Nucleate ~ o i f i n g
of Subcooled Liquids, 11 submitted f o r publication to J e t Propulsion.
,
. .. ,
..
29. Taylor, G. I. , " The Instability of Liquid Surfaces when Accelerated
in a Direction Perpendicular .to their Plane, Proc. Roy: Soc. , London,
A-201, 1950, p. 192.
Z
3.0. Bellrnan, R. and R. H. Penninqon, "Effects of Surface Tension and
Viscosity on Taylor Instability, Quar. Appl. Math. 12, 1'954, p. 151. '
37. Drew, T. B. and A.C. Mueller, oilin in^", Trans. AIChE, vol. .33, 4,. , ,....
.:;
1937, p. 449. ,. . .,...,
,
.. p. 2470. : .. . . ..