You are on page 1of 8

D. D.

WITTKE Collapse of Vapor Bubbles With


Translator Motion
Assistant Professor o f
Mechanical Engineering,
University o f N e b r a s k a , Lincoln, N e b . ;
F o r m e r l y a t University o f Illinois,
U r b a n a , ill.
Assoc. M e m . A S M E The collapse of vapor bubbles in a subcooled liquid is studied under conditions that
heat transfer is the controlling mechanism. Emphasis is placed on the effect of transla-
B. T. CHAO tory motion and the presence of noncondensable gas on the collapse rate. The governing
Professor o f M e c h a n i c a l conservation equations are cast into appropriate dimensionless forms, and numerical
a n d Nuclear Engineering, solutions are presented for the bubble history with Jakob number, Pcclet number, and
University o f Illinois, U r b a n a , III. dimensionless persistent bubble radius as parameters. The predicted collapse behavior
Mem. ASME
agrees satisfactorily with experimental data• which cover ranges of Jakob number from
7 to 45, Peclet number from 100 to 7700, and persistent bubble radius from 7.5 to 40
percent of the initial radius. The analysis also yields detailed information on the tem-
perature field surrounding a collapsing bubble, with or without translatory motion. Re-
sults for the latter vindicated the anticipated inadequacy of the thin thermal boundary-
layer assumption used in published analyses.

Introduction and collapse of a gas bubble in a liquid-gas solution due to over


or undersaturation. Readers are referred to a recent disserta-
THE mechanics of vapor bubble collapse under tion [8] by one of the authors for further information. Suffice it
spherically symmetrical conditions was recently examined by to say that the effect of translational motion on the collapse rate
Florschuetz and Chao [l]. 1 It was shown that the collapse of vapor bubbles has been recognized but it appears that there
mode could be controlled by liquid inertia or heat transfer or both, has been no concerted effort to obtain analytical and experi-
depending on system conditions. For the case of heat transfer mental information. In practice, translating bubbles are of usual
controlled collapse, their analysis made use of (lie Plesset-Zwick occurrence rather than the exception. This is then the major
temperature integral [2], thus involving the assumption of a thin motivation which prompts the present investigation.
thermal boundary layer. It is intuitively evident that, for the In this paper, attention is directed to the case where heat trans-
relatively slow collapse experienced under such conditions, the fer dominates the collapse mode since the translatory motion is
thin boundary-layer approximation could become increasingly not expected to be a significant factor if liquid inertia is the con-
poorer as the collapse proceeds and eventually invalid. This trolling mechanism.2 The governing equations are solved
limitation was recognized in [1] and held partially responsible for numerically without using the thin boundary-layer simplification
the discrepancy between the analytical and experimental results. and, with this improvement, the case of the stationary bubble is
Another factor cited for the discrepancy was the fact that, in spite reexamined. In addition, bubbles containing noncondensable
of the "gravity-free" condition to which the experimental bubbles gas are considered. Experimental data are presented and com-
were subjected, they oftentimes exhibited small translatory pared with theoretical predictions.
motion. When this occurred, the heat transfer would be en-
hanced, resulting in higher collapse rates.
Plesset and Zwick [2, 3, 4] and Forster and Zuber [5] analyzed Analysis
in great detail the growth of stationary vapor bubbles in super- The growth or collapse of a vapor bubble in a liquid is described
heated liquids. The possible effect of translational motion was by the conservation equations of mass, momentum, and energy
acknowledged but not studied. Ruckenstein [6] examined the for both the liquid and vapor, together with the appropriate
mechanism of heat transfer between a boiling liquid and vapor thermodynamic relation as well as the initial and boundary con-
bubbles moving in it. The analysis was made as if the bubble ditions. A detailed exposition of the problem and possible
radius were unaltered. Ruckenstein's main result was identical simplifications for the spherically symmetrical collapse of a
to that given by Boussinesq [7] more than half a century ago. stationary bubble have been given in [ 1 ]. A similar development
In the literature, there are numerous publications describing and for a traversing bubble collapsing under heat transfer controlled
discussing the influence of translational velocity on the growth conditions shows.that many of the simplifying assumptions in-
troduced there remain valid. For clarity and definiteness, we
1 N u m b e r s in brackets designate References at end of paper.
list the major assumptions used in the present analysis as follows:
C o n t r i b u t e d b y the H e a t Transfer Division and presented at the
Winter A n n u a l Meeting, N e w Y o r k , N . Y . , N o v e m b e r 2 7 - D e c e m b e r 1 Constant translational velocity during collapse. This as-
1, 1966, of THE A M E R I C A N SOCIETY OF M E C H A N I C A L ENGINEERS.
Manuscript received by H e a t Transfer Division, February 21, 1966; 2Here we think of the usual translatory m o t i o n such as that due t o
final draft, June 7, 1966. Paper N o . 6 6 — W A / H T - 1 2 . buoyancy.

Nomenclature 1
c = specific heat of liquid 7'mt = saturation temperature corre- Subscripts

k = thermal conductivity of liquid sponding to final system pres- 0 = refers to initial value
L = latent heat sure, p = refers to persistent value of bubble
t = time radius
Vv = equilibrium vapor pressure U a — bubble translational velocity w = refers to bubble surface
>co*= final system pressure 8 = spherical polar coordinate oo = refers to system value, or value at
R = bubble radius K = thermal diffusivity of liquid large distance from the bubble
/' = radial coordinate p = liquid density
T = temperature p„ = vapor density (Other quantities are defined in the text.)

Journal of Heat Transfer FEBRUARY 19 6 7 / 1 7

Copyright © 1967 by ASME

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


sumption was originated from the experimental evidence re-
vealed by the motion picture records of more than 50 bubbles.
Within the range of experimental conditions used in the present
investigation, it was found that the bubble moved with a nearly
constant velocit}' over a major portion of the collapse. Haber-
man and Morton [9] reported results for the terminal velocity of
air bubbles in various liquids under normal gravitational condi-
tions. Their graph for filtered or distilled water is reproduced in
Fig. 1. Inasmuch as the initial radius of most bubbles studied
ranged from 0.15 to 0.30 cm and the persistent radius was
usually 0.04 cm or larger, it is not surprising that the measured
velocity showed no significant changes although a majority of
our experiments were conducted under "reduced-gravity" condi-
tions. Furthermore, the experimental bubbles traveled at a
steady velocity prior to their collapse. Even if large forces
urging the change of velocity existed upon the initiation of col- O.OI 0.02 0.05 O.I 0.2 0.5 1.0 2.0
lapse, time would be required for its manifestation. While little EQUIVALENT RADIUS (CM.)
information is available on the subject, some estimate may be
inferred from Miyagi's paper [10]. I t was there stated that air Fig. 1 Terminal velocity of air bubbles in filtered or distilled w a t e r as a
function of bubble size (due to H a b e r m a n a n d M o r t o n [9])
bubbles rising from rest in still water required a distance of 3 to 4
cm to attain the terminal velocity. For the vapor bubbles
selected in this study, the collapse was essentially complete within
a travel distance of l 1 /* c m ° r less- Finally, it should be noted
that the analytical formulation which follows does not require
this assumption. However, it does provide conveniences in the
interpretation and presentation of results.
2 Bubble remains spherical at all times. The viscous and
pressure forces acting on a traversing bubble tend to produce dis-
tortion which is opposed by the surface tension effects. For the
magnitude of the translation velocity involved and the size of
bubble considered, motion picture records showed that this as-
sumption was approximately valid although a few exceptions
have been noted. A discussion of the collapse behavior asso-
ciated with nonspherical bubbles will be given later.
3 Irrotational outer flow. Levich [11] and Moore [12] both
demonstrated that, for gas bubbles rising in water at suffi-
ciently large Reynolds numbers, the irrotational solution gave a
Fig. 2 ( a ) Coordinates in physical system
valid approximation. The physical basis of such approximation
rests with the large difference between the gas (or vapor) and
liquid viscosities and hence the propensity of the bubble to de-
velop internal circulation [13]. While the use of ideal fluid solu-
tion does lead to errors in the details of stress distribution, it is
quite permissible insofar as the calculation of convective heat
transport around the bubble is concerned.
4 Liquid is incompressible, infinite in extent, and has con-
stant thermodynamic and transport properties. Thermodynamic
equilibrium exists between the liquid and the vapor. These are
the usual assumptions and are known to be valid.
5 For bubbles containing noncondensable gas, vapor diffusion
Fig. 2(b) Transformed coordinates
is sufficiently rapid that the collapse process remains dominated
by heat transfer in the liquid. This assumption is partly based
on the experimental finding reported in [1], When a heavy gas, velocity components used in the ensuing analysis. A t the instant
like xenon, and a very light gas, like helium, were separately in- under consideration, the bubble radius is R. In view of the
troduced into the vapor of stationary bubbles, only very minor irrotational assumption, there exists a velocity potential 4> which
difference in their collapse behavior was observed. In view of the satisfies the Laplace equation. For the axisymmetical flow under
anticipated presence of internal circulation for a traversing consideration, it is
bubble thus promoting a more uniform condition inside, the
, 1 b
possible retardation of the collapse due to insufficiently rapid = 0. (1)
+ shTo al lsin e So
vapor-gas diffusion is even less likelj'. In [8], an estimate was
made for the buildup of gas concentration adjacent to the
The appropriate boundary conditions are
bubble surface during the collapse of a vapor-gas bubble. The
result also lends support to the assumption. a <>
f
= R at r = R (2a)
dr
V e l o c i t y Field in the Liquid
and
Consider a spherical bubble of variable radius rising with a
constant velocity XJm in a liquid of infinite extent. Alternately,
one may picture that the center of mass of the bubble is at rest + _L bd J
Y =
= C/co as r (2b)
while the unbound liquid flows downward with a uniform and
constant velocity, U„, at a large distance away from it. Fig. In (2a), the dot denotes time derivative. W e note that (1) does
2(a) displays this view, together with the coordinate system and not explicitly contain t, which enters the solution via the boundary

18 / FEBRUARY 1967 Transactions of the ASP/IE

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


condition (2a). Furthermore, the second boundary condition
(2b) is nonlinear. One may readily verify that the solution of the B = sin Odd (8)
2p„L dr
boundary value problem is
in which p„ is the vapor density corresponding to If a linear
m R2R
= U„ r + cos 0 H , r > R. (3) density-temperature relation is assumed, one has
2 r> r
T Ta Pv ~ Pr.0
In (3), the first term on the right-hand side is the velocity po- (9)
Ta Pi .flat — Pi'.O
tential associated with the inviscid flow of an infinite fluid past a
sphere of fixed radius R, the undisturbed velocity being Ua.
It may be worth mentioning that, if one chooses to use the
The second term is that due to a radial expansion or contraction
Clausius-Clapeyron relation, namely
of a spherical cavity in an infinite fluid, otherwise at rest. From
(3), one finds the radial and tangential velocity components as
dPv Lpvp ^ pj,
(10)

M )
R2R dT T(p - p.) ~ T
W T, cos 9 -| — (4a)
v, = - — = - Uc
dr
and introduces the approximation that dpv/dT = constant, the
and error turns out to be somewhat larger for the experimental condi-
tions used.
1
(46)
vo =
" 7 ¥ = iJ
" (1 +
f»)sin*- Dimensionless F o r m u l a t i o n
W e define the following quantities to facilitate further discus-
Energy Equation for the Liquid
sion and analysis:
If one ignores viscous dissipation, the energy equation written
in terms of liquid temperature is (i) The dimensionless bubble radius, 7 = R/Rq
(ii) The dimensionless radial coordinate, r* = r/RO
dT dT dT (iii) The dimensionless time, r = nt/Ra2
— + v. —
dt ' dr r (iv) The dimensionless temperature, 0 = {T — Tm)/(T,al —
Tm)
fd*T— 2_ a?' JL^ jJ^T cot 0 57'
= «
r > R. (5) (v) The Peclet number, Pe = 2UJ{q/k
V&r2 r ar + r1 a0 2 + r2 (vi) The Jakob number, Ja = pc(T e „t — TJ)/p^L
(vii) Coordinate transformation according to, y = r* — 7 ,
The initial and boundary conditions appropriate for heat transfer
p. = — cos 6
controlled collapse are

T{r, 9, 0) = T„ (6a) With these, (5), (6a, 6), and (8), respectively, become

ae 1 ae
e, t) = and t(r, e, t) = t„. (66)
ar by
Reference [1] gives detailed arguments leading to the second con- 1 + *
7
dition in (66). For a pure vapor bubble, Tw = Tm and, thus,
is a constant. For vapor containing noncondensable gas, Tw con- Pe ae
1 -
tinually decreases as collapse proceeds. If one assumes a linear £>2/
variation of vapor pressure with temperature, and denotes the 1+*
7
persistent bubble radius by RP, then
Pe 1 - p2 1 ae
+ 1 +
T — T1 CC ( r „ t -
J- 111 r.) ( I - (7) y_ dp
T 1 + 2 1 +
T

as has been shown in [1]. a26 ae 1 - p2 a2e


+ ap2'
a y2
Energy Equation for the Bubble Surface 7(1 + ^
A close examination of the complete energy equation for the
moving surface of a spherically symmetrical collapsing bubble in- 2p ae
dicates that the kinetic energy of both liquid and vapor, viscous y> 0 (11)
V 2 dp'
dissipation, and heat conduction into vapor are insignificant com- 1 + -
pared to latent heat release and heat conduction into the liquid 7
[1], In addition, for heat transfer controlled collapse, the vapor
e {y, p, 0) = 0 (12a)
velocity at the bubble wall is small relative to R and may also be
ignored.
6( CO, p , r ) = 0, 6 ( 0 , p, r ) = 1 - (126)
For a translating bubble, internal circulation usually exists.
While it tends to facilitate the transport of heat, released at the
interface, into the vapor, the external circumferential convective and
motion would likewise enhance heat transport into the liquid.
Hence it appears reasonable to ignore again the heat flowing into ja r+1 ae
7 = (0, p, r)dp (13)
the vapor region. 3 Another consequence of the translatory 2e,
motion is that the local flux at the bubble surface varies with 6,
implying a nonuniform condensation rate. However, the inter- w h e r e 7 p = Rp/Ro, the dimensionless persistent bubble radius and
facial tension may effectively keep the bubble nearly spherical, as
P-.Q \ / I p N
In (11) and (13), the dot de-
has been evidenced in our experiments, and the internal circula- e„ = 1 - 1 -
pv.sat/ \ 7 ,
tory motion would promote a uniform condition within the bub-
ble. Thus it is adequate to use an integral energy relation: notes differentiation with respect to r. For a pure vapor bubble,
7 P = 0 and hence ec = 1.
3 System pressure close to critical is excluded in this discussion. The transformed coordinate system is illustrated in Fig. 2(6).

Journal of Heat Transfer FEBRUARY 1 9 6 7 / 1 9

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


The line y = 0 corresponds to the bubble surface. The dimensional image of the bubble was more elliptical than circular,
boundaries at p. = —1 and + 1 are the axes of symmetry corre- four peripheral points corresponding to the major and minor
sponding to 6 = 0 and w and, thus, are zero flux boundaries. diameters were used to determine the equivalent bubble radius
The equation set (11) to (13) was solved numerically using the as well as the location of its mass center relative to a fixed ref-
Peaceman-Rachford alternating direction technique [14]. A erence point inside the test chamber. A least-square curve fitting
brief description of that procedure and the computer program is of the displacement data was executed by the University of Il-
given in the Appendix. A study of the effects of various parame- linois I B M 7094 computer and the resulting equation differen-
ters on the bubble collapse behavior is reserved until after the tiated to determine the translational velocity. It was found that
next section in which the measured collapse data are presented first-order fitting was entirely adequate, implying that the veloc-
and compared with the theoretical results. ity was a constant. The persistent bubble radius was obtained
from the machine plotted radius-time curve at sufficiently long
time.
Experimental Investigation and Results
The experimental apparatus and procedure were selected to Bulk Liquid M o t i o n
closely approximate the conditions used in the foregoing analysis. The procedure of determining the bubble translational velocity,
The test facility described in [1] was used but was modified so as herein described, would clearly be in error if a bulk sloshing
that various translatory velocities could be imparted to the motion of the liquid existed after dropping of the platform. The
bubble. This was achieved by attaching variable counterweights relevant quantity sought is the relative motion between the bubble
to a falling platform on which the test chamber and a high-speed and its surrounding liquid.
motion picture camera were mounted. The 'effective gravita- In an effort to ascertain if bulk liquid motion of significant
tional field experienced by the bubble ranged from approximately extent existed, the following test was performed. An assortment
zero to nearly normal. The experimental procedure and data of small plastic pellets (Dow-Pelaspan 8, heavier than water as
reduction technique deviated somewhat from those reported in supplied) was first suitably expanded by heating, transferred to
[1] and are briefly described subsequently. room-temperature water and then stirred to form a suspension.
Triple distilled water was heated to boiling in the test chamber Those of density very close to that of water would remain sus-
at atmospheric (room) pressure. The heat source was removed pended for hours and were sorted out. Such selected pellets were
and the spring-loaded cover plate of a bubble holder was closed. introduced into the bubble collapse chamber and tests carried out
The latter was located at the bottom of the test chamber and was under simulated conditions except that the water was at room
essentially a small glass tube in which the test bubble was incu- temperature and air bubbles were used instead of vapor bubbles.
bated and held.4 As the water in the test chamber slowly cooled Of the 28 pellets studied, 2 had abnormal velocities close to 1
to the desired temperature, the convective currents became dis- cm/sec but the average was around 0.25 cm/sec. These are at
sipated. When ths intended temperature was finally approached, least an order of magnitude less than the bubble velocities ex-
the chamber pressure was reduced and then a swit ch was actuated, perienced in the majority of tests conducted during the present
bringing forth the following sequence of events: investigation.

1 A photospot light was turned on and the camera started. Experimental Results a n d C o m p a r i s o n With Theory
2 After a short delay to allow the film drum to accelerate to Experimental results of four test runs are shown plotted in Figs.
the set speed, the cover plate of the bubble holder was flipped 3 through 6.5 The solid line represents the theoretical prediction
open, thus releasing the bubble. according to the numerical solution of the governing equations.
3 After another short delay, the platform was dropped. The abscissa in the plots is a dimensionless time, Tlr, appropriate
4 When the platform fell through a sufficient distance for the for heat transfer controlled collapse. It is defined by
bubble to attain thermodynamic equilibrium with the liquid, the
test chamber pressure was rapidly increased to atmospheric by 4 4 T Kt
t„ = - JaV = - Ja — . (14)
opening a port at the top of the test chamber, thus initiating the 7T 7T IXo
collapse. The pressure rise time was monitored by a quartz
transducer, the output of which was displayed on an oscilloscope Pertinent data and parameters for these test runs are listed in
and photographed. It usually ranged from 3 - 4 millisec. Table 1. The B-number is a parameter characterizing the rela-
5 Continuous photographing of the bubble. tive importance of heat transfer and liquid inertia in the spheri-
6 The platform was decelerated and eventually brought to cally symmetrical collapse of vapor bubbles. It is defined by
rest. After a preset time, the camera was turned off.
B = Ja2 - —-, Ap = p — p„,o • (15)
The total duration from camera start to shutoff was about 3 ffo p'
sec but the time from bubble release to the beginning of plat-
The use of Ttl and B was first introduced in [1]. It was there
form deceleration was only about 6 /io sec. Further discussion on
proposed that, for B < 0.05, the collapse would be dominated by
the selection of suitable time intervals between various events
heat transfer effects. The direct use of the foregoing discriminat-
as influenced by the test conditions can be found in [8].
ing value for translating bubbles is subject to scrutiny but,
After the film was processed and edited, a Benson-Lehner A / D
judging from the extremely low values encountered in these experi-
Converter with a magnification of approximately 10 was used to
ments, it appears likely that heat transfer remains the controlling
obtain and convert analog information to digital data which were
then punched on cards for computer processing. Since the two- 5 Similar results of 15 additional test runs were presented in [8].

T h e discussion and conclusion drawn here reflect the results of all test
' See Fig. 7 in reference [1 ] for details. runs.

Table 1 Data pertinent to the four e x p e r i m e n t a l bubbles s h o w n in Figs. 3 through 6

Average
Run It 0 To, AT *
Ap Ja B aspect
no. cm cm/sec °C °C cm °Hg cm Hg 7p (at r» a t ) Pe (at Tnt) ratio
1680 0 .143 5.5 95.06 4.47 74. 75 11.20 0.39 13.6 936 5.5 X 10~3 1.1
1800 0 .263 15.4 96.47 2.63 73. 61 6.69 0.2 8.1 4820 1.4 X 10" 3 1.5
1870 0 .393 2.2 93.95 5.49 74. 57 13.67 0.075 16.7 1029 2.7 X 10" 3 1.1
1560 0 .354 18.2 S8.99 10.54 74. 75 24.15 0.125 32.0 7664 8.4 X 10- 3 1.3
A21 = r „ t - AV = Va* ~ pv,o

20 / FEBRUARY 1 9 6 7 Transactions of the ASP/IE

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


). 1680
R0 = O.I43 cm
U„ = 5.5 cm/sec —
AT=4.47 °C

0 .5 1.0 1.5 2.0


T„
r„
Fig. 3 C o m p a r i s o n b e t w e e n experimental data a n d theory
Fig. 4 C o m p a r i s o n b e t w e e n e x p e r i m e n t a l data a n d theory

factor. The average aspect ratio is given as an indication of 1 1


k
!
the departure from spherical shape during collapse. RUN NO. 1870
In Figs. 3, 4, and 6, the theoretical curves seem to fit the data
quite well in spite of the fact that the bubble in Fig. 4 deviates
X Ja= 16.7
pfe=i02y
Ro=0.393 cm
U„ = 2.2 cm/sec
considerably from being spherical. This is reflected in the some- Xp =0.075 AT=5.49 °C
what larger scatter of data. On the other hand, the data of Run
1870 fell below the theoretical prediction by a large amount.
Upon reexamination of the motion picture records, it was noted
that the bubble experienced localized, irregular, oscillatory sur-
face motion during collapse. It was of relatively higher fre-
quency although the amplitude was small. It should be stated
that, almost without exception, the bubbles oscillated somewhat
during collapse. It began at the time when they left the glass
incubating tube but was damped after the dropping of the plat-
form. Another possible explanation for the rather large dis-
crepancy between theory and data is that the very low translating 0 .1 .2 .3 .4 .5
velocity used in that particular run could involve large measure- r„
ment errors.
Fig. 5 C o m p a r i s o n b e t w e e n experimental d a t a and theory (bubble
The theoretical curves in Figs. 4 and 6 mildly exhibited two in- exhibited localized, irregular, oscillatory surface motion during collapse)
flections. This occurs under suitable combinations of ypl Pe, and
Ja. Similar observations may be made if one refers to Figs. 10(a),
10(6), and 11. Clearly, a nonzero y p is a necessary but insuf-
ficient condition for the double inflection to occur. Unfortu-
nately, the scatter of experimental data in Runs 1800 and 1560
(and many others reported in [8]) makes it impossible to obtain
either experimental support or the lack of it at the present time.
In general, the theoretical prediction of the radius-time curves
compares reasonably well with the experimental data except when
irregular, oscillatory surface motion takes place.

Theoretical Results
The satisfactory agreement between the observed data and the
•theoretical results for the collapse behavior of translating bubbles
provides evidence of the general validity of the physical model
considered and the assumptions used in the analysis. In addi-
tion, the procedure employed in the numerical solution of the
nonlinear, parabolic differential equation has met with reasona- Fig. 6 Comparison between experimental data and theory

ble success. W e now proceed with a parametric study of the


collapse behavior of vapor bubbles experiencing translatory trolled conditions involved the use of the Plessef^Zwick tempera-
motion. Specifically, the effects of Peclet number, Jakob num- ture integral. As such, it assumes the existence of a thin thermal
ber, and the presence of noncondensable gas in the vapor are in- boundary layer, the validity of which is subject to severe limita-
vestigated. A few selected results for the temperature distribu- tion, particularly when the collapse is slow. Such limitation was
tion in the liquid are given which provide further insight into the recognized by the authors of that reference and, indeed, they
problem. In the following, we first present some additional in- pointed out that their result could only be regarded as an upper
formation for stationary vapor bubbles, with and without the bound. It was thus deemed desirable to obtain the solutions by
presence of noncondensable gas. These supplement the findings a direct numerical integration of the governing equations without
reported in [1], imposing the thin boundary-layer restriction. The results are
summarized in Fig. 7. The significance of the Jakob number
Stationary Bubbles (Spherically Symmetrical Collapse) needs no elaboration. W e note that the curves with smaller
The theoretical result presented in [1] for spherically sym- Ja lie below those with larger Ja, but this should not be taken to
metrical collapse of pure vapor bubbles under heat transfer con- mean faster collapse for bubbles with smaller Jakob number. On

Journal of Heat Transfer FEBRUARY 1 9 6 7 / 2 1

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


0 .02 .04
0 .2 A .6 .8 1.0
Th
TH
(a) For Ja = 1
Fig. 7 Spherically symmetrical collapse of pure v a p o r bubbles (the
curve due to Florschuetz a n d Chao is a n upper-bound solution)

[H (b) For Ja = 10

Fig. 10 Influence of translalory motion on the collapse behavior of


Fig. 8 Temperature profiles surrounding pure v a p o r bubbles—spheri-
v a p o r bubbles
cally symmetric collapse

Pe = 0
J o = 10
Ja = IO
Pfe=3000
—y p = 0 ~
\ \ \ —y P =0.3
\ \ \
\\ \
\\ \
\\ \ — >;=o.5
~ A > \

\>
0.3
\\
\ s v

V
\ 0.775
\ v 0.472 ^ ^ y= 0.372
v
o.o\
0 .4 .8 1.2 1.6
0 .2 .4

[ H r„
Fig. 9 Temperature profiles surrounding v a p o r - g a s b u b b l e s — s p h e r i c a l l y Fig. 11 Influence of the a m o u n t of noncondensable gas on the collapse
symmetric collapse behavior of translating bubbles

the contrary, the actual time required for a pure vapor bubble to 0.775, i.e., the bubble radius is still more than 3 / 4 of its initial size,
collapse to a preassigned fraction of its initial radius increases with the thermal boundary layer has already grown to a thickness
decreasing Ja. The seemingly anomalous result is the conse- comparable to the bubble radius itself. This is quite under-
quence of the association of Ja2 in the definition of rn. standable since, for small Jakob number, the collapse is slow, thus
Computed temperature profiles in the liquid for Ja = 1 and 10 allowing ample time for heat to diffuse into a region remote from
are shown in Fig. 8. The abscissa is the distance from the bubble the source at the bubble surface.
surface normalized with respect to the bubble radius. It is evi- The upper bound solution due to Florschuetz and Chao is
dent that the validity of the thin boundary-layer approximation theoretically valid for large Jakob numbers if conditions for the
is very limited and particularly so when the Jakob number is thin boundary-layer approximation were to be met. On the other
small. Consider, for example, the case of Ja = 1. When 7 = hand, heat transfer controlled collapse would usually ensue when

22 / FEBRUARY 1967 Transactions of the ASP/IE

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


1.0 g-_0 0 9=57° 0=90' 0 = 127 ' going observation may be understood if one realizes that the
R U N NO. 1680
relative contribution to heat transfer at the bubble surface due to
Jo = 13.6 the translatory motion is greater when the Jakob number is
P'e =936
smaller. The presence of the noncondensable gas always retards
the collapse, but its effect is manifested earlier when the P^clet
r number is larger. T o demonstrate the highly nonlinear nature of
=0.927
=0.583 the influence of the noncondensable gas as its amount varies, Fig.
\ =0.481 11 is included.
\ \ \ =0.432
11 \\ — =0.397 Temperature profiles calculated for conditions corresponding
w
\
to two of the experimental runs, one with relatively slow transla-
\\
V\ \ V tional velocity and large persistent radius (Run 1680) and another
\\\
\\v Vs with high translational velocity and moderate persistent radius
\\\ \\ (Run 1800), are shown in Figs. 12(a) and 12(6). The profiles are
\\
presented for four angular positions of 0, 57, 90, and 127 deg
LLl
0.2 0 0.2 0 0.2 0.2 0.4 0.6
measured from the front stagnation point. At a given instant,
the thickness of the thermal boundary layer continuously in-
M creases as one moves awa}' from the front stagnation in a meridian
(a) S m a l l t r a n s l a t i o n a l v e l o c i t y a n d l a r g e a m o u n t of n o n c o n d e n s a b l e gas plane. For a given angular location, however, the boundary-
layer thickness first increases with time, reaches a maximum, and
1.0 eventually decreases as collapse approaches completion. This is
RUN NO. 1800
a consequence of the continual drop in bubble surface tempera-
0=0° 9=57" 0 = 90° ture as the partial pressure of the vapor falls.
The appearance of a temperature maximum away from the
bubble surface and the reversal of temperature gradient as col-
lapse proceeds can be seen in Fig. 12(a) for 6 = 127 deg. The
liquid circumferential motion relative to the bubble surface
convects heat from the front stagnation toward the rear. Simul-
taneous with this event is the progressive reduction of bubble
wall temperature as we noted earlier. When the translational
velocity is small and if the bubble surface temperature drops
sufficiently rapidly, a temperature reversal could occur. This is,
of course, purely a local phenomenon. Overall, there is a con-
\
tinual transfer of heat from the bubble surface into the liquid.

.1 0
Acknowledgment
U-'] The research reported herein was supported by a National
(b) Large t r a n s l a t i o n a l v e l o c i t y a n d m o d e r a t e a m o u n t of n o n c o n d e n s a - Science Foundation Grant, GP-2717, for which the authors ex-
ble g a s press their deep appreciation. Thanks are also clue to L . W .
Fig. 1 2 Temperature profiles for collapsing vapor-gas bubbles with Florschuetz and S. A. Zwick for their helpful discussions
translatory motion prior to publication.

References
the Jakob number is small.6 In this sense, its applicability is 1 L. W . Florschuetz and B. T . Chao, " O n the Mechanics of Va-
restricted. p o r B u b b l e C o l l a p s e , " J O U R N A L OF H E A T T R A N S F E R , T R A N S . ASME,
If vapor diffusion plays no significant role, as is oftentimes the Series C, vol. 87, 1965, pp. 209-220.
2 M . S. Plesset and S. A. Zwick, " A Non-Steady Heat Diffusion
case for relatively slow collapse, the presence of the noncon-
Problem With Spherical Symmetry," Journal of Applied Physics, vol.
densable gas is essentially manifested in the continual reduction of 23, 1952, p . 95.
the vapor pressure and the corresponding reduction of bubble 3 M . S. Plesset and S. A. Zwick, " T h e Growth of Vapor Bubbles
wall temperature as collapse proceeds. The temperature in Superheated Liquids," Journal of Applied Physics, vol. 25, 1954,
gradient at the wall is thus reduced with a corresponding re- p. 493.
4 S. A. Zwick, " T h e Growth and Collapse of Vapor Bubbles,"
duction in heat transfer and collapse rate. A secondary effect is Hydrodynamics Lab. Report No. 21-19, California Institute of Tech-
the reduction in vapor density as the bubble shrinks, resulting in nology, Pasadena, Calif., 1954.
a smaller rate of latent heat release. The rather dramatic in- 5 H. Ii. Forstel' and N. Zuber, "Growth of a Vapor Bubble in a
fluence of noncondensable gas on the transient temperature field Superheated Liquid," Journal of Applied Physics, vol. 25, 1954, p.
474.
toward the later stages of collapse is illustrated in Fig. 9.
6 E. Ruekenstein, " O n Heat Transfer Between Vapor Bubbles
in Motion and the Boiling Liquid From Which They Are Gen-
Translaiing Bubbles erated," Chemical Engineering Science, vol. 10, 1959, p. 22.
The influence of translational velocity on the bubble collapse 7 M . J. Boussinesq, "Caleul du pouvoir refroidissant des
courant fluides," Journal of Mathematics, series 6, vol. 70, 1905, p.
behavior is shown in Figs. 10(a) and 10(6), respectively, for Ja = 1 285.
and 10. Both the case of a pure vapor bubble (Y p = 0) and that S D . D . Wittke, "Collapse of Vapor Bubbles With Translatory
containing noncondensable gas (yp = 0.3) are considered. The Motion," P h D thesis, University of Illinois, Urbana, 111., 1966.
translational velocity is expressed in terms of the dimensionless 9 W . L. Haberman and R. K . Morton, " A n Experimental In-
vestigation of the Drag and Shape of Air Bubbles Rising in Various
P6clet number. One may obtain an estimate of the bubble veloci-
Liquids," David Taylor Model Basin Report No. 802, September,
ties corresponding to the several P6clet numbers shown by re- 1953.
ferring to Table 1. From these figures, it is seen that, as ex- 10 O. Miyagi, " T h e Motion of an Ail- Bubble Rising in Water,"
pected, the faster the bubble translates, the more rapidly it col- Philosophical Magazine, series 6, vol. 50, 1925, p. 112.
11 V. G. Levich, " M o t i o n of Gas Bubbles at Large Reynolds
lapses. The translatory motion produces a greater effect for the
Numbers," Zhur. EUsjitl. i Teoret. Fiz., vol. 19, 1949, p. 18. See also,
smaller Jakob number except at the beginning of the collapse Physicochemical Hydrodynamics, Prentice-IIall, Inc., Englewood Cliffs,
when the conduction transient dominates in any case. The fore- N. J . , 1962, p. 436.
12 D . W . Moore, " T h e Rise of Gas Bubbles in a Viscous Liquid,"
6 Small B-number is associated with small Ja. See equation (15). Journal of Fluid Mechanics, vol. 6, 1959, p. 113.

Journal of Heat Transfer FEBRUARY 1967 / 23

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


13 S. Winnikow and B. T . Chao, "Droplet Motion in Purified peratures, 0 , be essentially zero there. Its extent increases as
Systems," Physics of Fluids, vol. 9, 1900, p. 50. computation proceeds and, eventually, it is limited by the
14 D . W. Peaceman and H. H. Rachford, Jr., " T h e Numerical available machine storage capacity.
Solution of Parabolic and Elliptic Differential Equations," Journal of
the Society of Industrial and Applied Mathematics, vol. 3, 1955, p. 28.
When the temperature field at the end of a full time step be-
15 F. B. Hildebrand, Introduction to Numerical Analysis, M c - came known, its radial gradient at the surface was evaluated
Graw-Hill Book Company, Inc., New York, N. Y., 1956, p. S2. using a three-point differentiation formula given by Hildebrand
[15]. Equation (13) was integrated using Simpson's rule and
the dimensionless radius y calculated. T o start the computa-
APPENDIX tion, y for the very first time step was determined from the
theoretical solution of Florschuetz and Chao—equation (26) in
Brief Comments on the N u m e r i c a l Procedure reference [1]—at a time when y = 0.99. Thereafter, it was
The Peaceman-Rachford alternating direction technique sub- evaluated from the numerical solution. It wTas found that this
divides the time step into two half steps. For the first half time convenient procedure produced no adverse effect on the resulting
step, the finite difference quotients are written in implicit form y — TH curve. In order to reduce the computation cost, a
for one space variable but in explicit form for the other. These variable total time step was used. This was achieved by imposing
representations are then interchanged for the second half time the requirement that the change in y from one time step to the
step. This technique offers the likelihood of obtaining stability next should not be less than 0.005 but, in no case, was it allowed
usually inherent in implicit schemes but also has the advantage to be greater than 0.01.
that the resulting set of simultaneous, algebraic equations is tri- The numerical procedure was programmed on the University of
diagonal and can be solved by simple elimination. A list, of the Illinois I B M 7094 computer. In the course of developing the
relevent difference quotients for the derivatives appearing in program, it was noted that, in some instances, small negative
equation (11) is given in [8]. Here, we merely note that a for- values of 0 appeared in regions apparently outside the thermal
ward difference quotient was used for the time derivative, a boundary layer, which is, of course, physically impossible. After
simple mean of the forward and backward difference quotient for several attempts to establish the cause, it was found that insuf-
the first-order space derivatives, and a central difference quotient ficiently fine grid spacing in the p-direction contributed to the
for the second-order space derivatives. difficulty. A suitable size for A p would depend on the Peclet
The alternating direction technique, as described in [14], was number, being smaller for larger Pe. Readers are referred to [8]
intended for finite regions whereas the domain under considera- for additional information and discussion. Included in that
tion is theoretically infinite. However, it appears physically reference are the effect of changing step size and the number of
clear that one would introduce little error by using a finite y [see steps in the ^-direction, a discussion of computation with increas-
Fig. 2(6)] provided that it is sufficiently large. The latter is ing digits, arbitrary replacement of small negative values of 0 by
specified by demanding that the computed dimensionless tem- zero, and so on.

24 / FEBRUARY 1967 Transactions of the ASP/IE

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like