Вы находитесь на странице: 1из 37

P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL

November 13, 1997 13:18 Annual Reviews AR049-13

Annu. Rev. Fluid Mech. 1998. 30:365–401


Copyright c 1998 by Annual Reviews Inc. All rights reserved

BOILING HEAT TRANSFER


V. K. Dhir
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

Mechanical and Aerospace Engineering Department, University of California,


Los Angeles, California 90095-1597; e-mail: vdhir@seas.ucla.edu

KEY WORDS: nucleate boiling, maximum heat flux, transition boiling, film boiling, minimum
by University of Nottingham on 08/04/10. For personal use only.

heat flux, pool boiling, flow boiling

ABSTRACT
This review examines recent advances made in predicting boiling heat fluxes,
including some key results from the past. The topics covered are nucleate boiling,
maximum heat flux, transition boiling, and film boiling. The review focuses on
pool boiling of pure liquids, but flow boiling is also discussed briefly.

INTRODUCTION
Boiling is a phase change process in which vapor bubbles are formed either on a
heated surface or in a superheated liquid layer adjacent to the heated surface. It
differs from evaporation at predetermined vapor/gas-liquid interfaces because it
also involves creation of these interfaces at discrete sites on the heated surface.
Nucleate boiling is a very efficient mode of heat transfer. It is used in various
energy conversion and heat exchange systems and in cooling of high-energy–
density electronic components. Pool boiling refers to boiling under natural
convection conditions, whereas in forced flow boiling, liquid flow over the
heater surface is imposed by external means. Forced flow boiling includes
external and internal flow boiling. In external boiling, liquid flow occurs over
unconfined heated surfaces, whereas internal flow boiling refers to flow inside
tubes.
This review is a follow-up to Rohsenow’s (1971) similar review in this series.
Reviews covering different aspects of boiling have appeared elsewhere (e.g.
Kenning 1977, Dhir 1991, Fujita 1992). This review focuses on pool boiling
of pure liquids. Flow boiling is described only briefly.
Boiling is a complex and elusive process. As such, we often rely on dimen-
sionless groups and empirical constants when correlating data. Concurrent with
the development of correlations useful for engineering applications, progress
365
0066-4189/98/0115-0365$08.00
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

366 DHIR

continues to be made in understanding the physics of the boiling process. Be-


cause the process is so complex and because so many heater and fluid variables
interact, completely theoretical models have not been developed to predict the
boiling heat fluxes as a function of heater surface superheat. In many cases, a
consensus is lacking in the technical community with respect to the dominant
mechanisms of heat transfer (in nucleate and transition boiling) and the degree
to which the contribution of various mechanisms to total heat flux changes with
wall superheat and heater geometry.
Figure 1 shows, qualitatively, the boiling curve (i.e. dependence of the wall
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

heat flux, q, on the wall superheat on a surface submerged in a pool of saturated


liquid). The wall superheat, 1T, is defined as the difference between the wall
temperature and the saturation temperature of the liquid at the system pressure.
by University of Nottingham on 08/04/10. For personal use only.

The plotted curve is for a flat plate or a horizontal wire to which the heat
input rate is controlled. As the rate of heat input to the surface is increased,
natural convection is the first mode of heat transfer to appear in a gravitational
field.
At a certain value of the wall superheat (Point A), vapor bubbles appear on
the heater surface. This is the onset of nucleate boiling. The bubbles form on
cavities or scratches on the surface that contain preexisting gas/vapor nuclei.
In liquids that wet the surface well, the onset of nucleation may be delayed.
For these liquids, a sudden inception of a large number of cavities at a certain
wall superheat causes a reduction in the surface temperature, while the heat
flux remains constant. This behavior is not observed when the boiling curve
is obtained by reducing the heat flux, and hysteresis results. After inception,
a dramatic increase in the slope of the boiling curve is observed. In partial
nucleate boiling, corresponding to region II (curve AB) in Figure 1, discrete
bubbles are released from randomly located active sites on the heater surface.
The density of active sites and the frequency of bubble release increases with
wall superheat.
The transition from isolated bubbles to fully developed nucleate boiling (re-
gion III) occurs when bubbles at a given site begin to merge in the vertical
direction. Vapor appears to leave the heater in the form of jets. The condition
of jet formation also approximately coincides with the merger of vapor bubbles
at the neighboring sites. After lateral merger, vapor structures appear that look
like mushrooms with several stems (Gaertner 1965). Figure 2 shows a photo-
graph of a large vapor structure supported by several smaller bubbles (stems).
A small change in the slope of the boiling curve can occur upon transition from
partial to fully developed nucleate boiling. The heat flux on polished surfaces
varies with wall superheat roughly as
q ∼ 1T m , (1)
where m has a value between 3 and 4.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 367


Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org
by University of Nottingham on 08/04/10. For personal use only.

Figure 1 Typical boiling curve, showing qualitatively the dependence of the wall heat flux, q,
on the wall superheat, 1T, defined as the difference between the wall temperature, Tw, and the
saturation temperature, Tsat, of the liquid. Schematic drawings show the boiling process in regions
I–V. These regions and the transition points A–E are discussed in the text.

The maximum or critical heat flux, qmax, sets the upper limit of fully developed
nucleate boiling for safe operation of equipment. After maximum heat flux is
reached, most of the surface is rapidly covered with vapor. The surface is nearly
insulated, and the surface temperature rises very rapidly. When the rate of heat
input is controlled, the heater surface passes quickly through regions IV and
V (see Figure 1) and stabilizes at point E. If the temperature at E exceeds the
melting temperature of the heater material, the heater will fail (burn out). The
curve ED (region V) represents stable film boiling, and the system can be made
to follow this curve by reducing the heat flux.
In stable film boiling, the surface is covered with vapor film, and liquid
does not contact the solid. On a horizontal surface the vapor release pattern is
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

368 DHIR
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org
by University of Nottingham on 08/04/10. For personal use only.

Figure 2 Photographic observation of vapor phase structure.

governed by Taylor instability of the vapor-liquid interface. With reduction of


heat flux in film boiling, a condition is reached when a stable vapor film on the
heater can no longer be sustained. Heat flux and wall superheat corresponding
to the condition at which vapor film collapse occurs are referred to as the
minimum heat flux qmin, and the minimum wall superheat 1Tmin, respectively.
Upon collapse of the vapor film, the surface goes through regions IV, III,
and II very rapidly and settles in nucleate boiling. Region IV, falling between
nucleate and film boiling, is called transition boiling, which is a mixed mode of
boiling that has features of both nucleate and film boiling. Transition boiling
is very unstable, since it is accompanied by a reduction in the heat flux with
an increase in the wall superheat. As a result, it is difficult to obtain steady-
state data in transition boiling, except when the heater surface temperature is
controlled. Transient transition boiling data can be obtained either by quenching
or by accessing from the nucleate boiling side when heat input to the heater is
controlled.

NUCLEATE BOILING
Preexisting Nuclei
Vapor/gas trapped in imperfections such as cavities and scratches on the heated
surface serve as nuclei for bubbles. Bankoff (1958) was the first to provide
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 369

a criterion for entrapment of gas in a wedge by an advancing liquid front.


According to this criterion, a wedge-shaped imperfection on a surface will trap
vapor/gas as long as the contact angle1 is greater than the wedge angle.
Wang & Dhir (1993a) developed a vapor/gas entrapment criterion by mini-
mizing the Helmholtz free energy of a system involving a liquid-gas interface
in a cavity. According to this criterion, a cavity will trap vapor/gas if

φ > ψmin , (2)


Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

where ψ min is the minimum cavity-side angle of a spherical, conical, or sinu-


soidal cavity. For the spherical and conical cavities, ψ min occurs at the mouth
of the cavity and is equal to the cavity-mouth angle, ψ m, as measured from the
heater surface. Ward & Forest (1976), while analyzing the relation between
by University of Nottingham on 08/04/10. For personal use only.

platelet adhesion and roughness of a synthetic material, obtained the same cri-
terion for stability of a vapor/gas nucleus in a long narrow fissure. Although
Bankoff’s criterion provides a necessary condition for vapor/gas entrapment in
a wedge, Equation 2 provides a sufficient condition.

Inception
Several approaches have been proposed for determining the incipient wall su-
perheat for boiling from preexisting nuclei. This review discusses two of the
most commonly used approaches. In the first approach, as originally proposed
by Hsu (1962), an embryo will become a bubble if the temperature of the
liquid at the tip of the embryo (the farthest point from the heated wall) is at
least equal to the saturation temperature corresponding to vapor pressure in the
bubble. Thus, Hsu’s criterion requires that the embryo should be surrounded
everywhere by superheated liquid.
In the second approach, boiling incipience is proposed to correspond to
a critical point of instability of the vapor-liquid interface. The interface is
considered to be stable or quasi-stable if the curvature of the interface increases
with an increase in vapor volume (see e.g. Mizukami 1977, Forest 1982, Nishio
1985). Wang & Dhir (1993a) studied the instability of the vapor-liquid interface
in a spherical cavity and showed that nucleation occurs when nondimensional
curvature of the interface attains a maximum value. They obtained the following
relation between wall superheat and diameter, Dc, of a nucleating cavity:
4σ Tsat
1T = K max , (3)
ρv h f g Dc

1 A distinction must be made between an advancing and a receding contact angle. Generally, the
advancing contact angle is greater than the receding contact angle. Because of the large uncertainties
associated with determination of advancing and receding contact angles, a static contact angle, φ,
was used in this work.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

370 DHIR

where
K max = 1 for φ ≤ 90◦
= sin φ for φ > 90◦
In Equation 3, σ is the interfacial tension, Tsat is the saturation temperature,
ρv is the density of vapor, hfg is the latent heat of vaporization, and Dc is the
cavity-mouth diameter. The implicit assumption made in arriving at Equation 3
is that the interface temperature is the same as the wall temperature. Through
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

carefully conducted experiments, Wang (1992) validated Equation 3.


According to Equation 3, few preexisting vapor/gas nuclei are found for well
wetting liquids such as R-113 and FC-72. For these liquids, the expected wall
superheat at nucleation should approach the homogeneous nucleation temper-
by University of Nottingham on 08/04/10. For personal use only.

ature ('90% of critical temperature). The observed inception superheats for


these liquids, although much higher than those observed for partially wetting
liquids, are much smaller than those corresponding to homogeneous nucleation
temperature (see Barthau 1992). Gases dissolved in these liquids may initi-
ate the nucleation, and as a result, the observed superheat is smaller than that
corresponding to homogeneous nucleation. In many instances gas is added by
external means to the wetting liquids to reduce the inception temperature and
to minimize the hysteresis.

Nucleation Site Density


The number density of sites that become active increases as wall heat flux or
superheat increases. Because addition of new nucleation sites influences the
rate of heat transfer from the surface, a knowledge of nucleation site density as
a function of wall superheat is needed in order to develop a credible model for
prediction of nucleate-boiling heat flux. Several other parameters also affect
the site density, including the procedure used in preparing the heater surface,
surface finish, surface wettability, heater material thermophysical properties,
and heater thickness. Until recently, little attention had been given to the effect
of these parameters on the density of active sites. Kocamustafaogullari & Ishii
(1983) correlated the cumulative nucleation site density reported by various
investigators for water boiling on a variety of surfaces at pressures of 1–198
atm. In developing the correlation, heater surface characteristics were not
considered. In general, active site density is correlated as

Na ∼ 1T m 1 , (4)

where m1 varies between 4 and 6. Cornwell & Brown (1978) found that the
proportionality constant in Equation 4 increased with surface roughness, but
the exponent was independent of surface roughness. Bier et al (1978) obtained
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 371

distinctly different values of m1 during boiling on an etched copper surface and


on a turned surface. This discrepancy resulted because in neither case were the
observations tied to the shape and size distribution of cavities.
Wang & Dhir (1993a,b) provided a mechanistic approach for relating the
cavities present on the surface to those that actually nucleate. Their approach
also includes the effect of surface wettability. They first determined the size,
shape, and mouth angle of activities present on a polished copper surface and
then used Equation 2 to determine the fraction of those cavities that will trap
vapor/gas. Most of the cavities that could trap vapor/gas were of reservoir type.
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

The data, consistent with their model, showed a 20-fold reduction in number
density of active sites as the contact angle was decreased from 90◦ to 18◦ .
Although Wang & Dhir showed how the developed criterion could be used to
by University of Nottingham on 08/04/10. For personal use only.

determine theoretically the number density of active sites, the procedure used
in determining the size, shape, and mouth angle of cavities is tedious and time
consuming and cannot be used readily in a practical application.
Wang & Dhir did not consider the thermal interference between sites or
the seeding and deactivation of sites in the neighborhood of an active cavity.
Kenning (1989) noted that thermal and flow conditions in the vicinity of a
heated surface can lead to activation of inactive sites and deactivation of active
sites. Sultan & Judd (1983) studied the bubble growth pattern at neighboring
sites during nucleate pool boiling of water on a copper surface. They found that
elapsed time between the start of bubble growth at two neighboring active sites
increased as the distance separating the two sites increased. They proposed
that thermal diffusion in the substrate in the immediate vicinity of the boiling
surface may be responsible for this behavior. Their work suggests that some
relation may exist between distribution of active nucleation sites and bubble
nucleation phenomenon.
Judd & Chopra (1993) reported results of interactions between neighboring
sites that lead to activation of inactive sites and deactivation of active sites. They
noted that for separation distances between nucleation sites less than one bubble
diameter at departure, the formation of a bubble at the initiating site promotes
the formation of bubbles at the adjacent sites (site seeding). For separation
distances between one and three bubble diameters at departure, formation of
a bubble at the initiating site inhibits the formation of bubbles at the adjacent
site (deactivation of sites). For distances greater than three bubble diameters at
departure, nucleation at one location is not influenced by activation at another
site. Kenning’s and Judd’s studies indicate that thermal interference and site
seeding or deactivation can alter the local active-site density and distribution
at low heat fluxes or in partial nucleate boiling. However, the significance of
these processes with respect to heat transfer during well-established nucleate
boiling on thick heaters is expected to be small.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

372 DHIR

Bubble Dynamics
After inception, a bubble continues to grow (in a saturated liquid) until forces
causing it to detach from the surface exceed those pushing the bubble against
the wall. After departure, cooler liquid from the bulk fills the space vacated
by the bubble, and the thermal layer at and around the nucleation site reforms
(transient conduction). When the required superheat is attained at the tip of
the vapor bubble embryo or the interface instability criterion is met, a new
bubble starts to form at the same nucleation site, and the bubble growth process
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

repeats. Wall heat transfer in nucleate boiling results from natural convection on
the heater surface areas not occupied by bubbles and from transient conduction
and evaporation at and around nucleation sites. Bubble dynamics include the
processes of bubble growth, bubble departure, and bubble release frequency,
by University of Nottingham on 08/04/10. For personal use only.

which includes time for reformation of the thermal layer (waiting period). The
following sections describe each of these processes.

BUBBLE GROWTH The literature highlights two points of view with respect to
bubble growth on a heated surface. One group of investigators has proposed
that the growth of a vapor bubble occurs as a result of evaporation all around the
bubble interface. The energy for evaporation is supplied from the superheated
liquid layer that surrounds the bubble after its inception. Some bubble growth
models are similar to that proposed for growth of a vapor bubble in a sea of
superheated liquid (see Plesset & Zwick 1954). The bubble growth process on
a heater surface, however, is more complex because the bubble shape changes
continuously during the growth process, and superheated liquid is confined to a
thin region around the bubble. Mikic et al (1970) obtained an analytical solution
for the bubble growth rate by using a geometric factor to relate the shape of a
bubble growing on the heater surface to a perfect sphere and accounting for the
thermal energy stored in the superheated liquid layer prior to bubble inception.
Since the initial energy content of the superheated liquid layer surrounding the
bubble depends on the waiting time, the model shows the dependence of bubble
growth rate on waiting time.
The second point of view is that most of the evaporation occurs at the base
of the bubble in that the microlayer between the vapor-liquid interface and
the heater surface plays an important role. Snyder & Edwards (1956) were
the first to propose this mechanism for evaporation. Moore & Mesler (1961)
deduced the existence of a microlayer under the bubble from the oscillations in
the temperature measured at the bubble release site. Cooper & Lloyd (1969)
not only confirmed the existence of a microlayer underneath isolated bubbles
formed on glass or ceramic surfaces but also deduced the thickness of the
microlayer from the observed response of the heater surface thermocouple.
They noted that an expression for local thickness, δ, of the microlayer could be
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 373

written as
p
δ ∼ νl tg , (5)

where ν l is the kinematic viscosity of the liquid, and tg is the bubble growth time.
It was further demonstrated that bubble growth was mostly due to evaporation
from the microlayer. Although Cooper & Lloyd’s work proved the importance
of microlayer evaporation at low pressures, their work was limited in scope. Lee
& Nydahl (1989) calculated the growth of spherical bubbles with a microlayer.
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

For microlayer thickness, they used Cooper & Lloyd’s formulation. They came
to the same conclusion as Cooper & Lloyd, that microlayer evaporation is a
significant contibutor to the heat transfer during bubble growth. However,
Plesset & Prosperetti (1977) concluded that in subcooled boiling, evaporation
by University of Nottingham on 08/04/10. For personal use only.

at the microlayer accounts for only 20% of the total heat flux. After more
than three decades of research, we still do not have an effective, consistent
model for bubble growth on a heated surface that appropriately includes the
microlayer contribution and time-varying temperature and flow field around the
bubble.
BUBBLE DEPARTURE The diameter to which a bubble grows before departing
is dictated by the balance of forces that act on the bubble. These forces are
associated with the inertia of the liquid and vapor, the liquid drag on the bubble,
buoyancy, and the surface tension. Fritz (1935) correlated the bubble departure
diameter by balancing, on a static bubble, the buoyancy with surface tension
force. Although significant deviations of the bubble diameter at departure with
respect to Fritz’s expression have been reported in the literature, especially at
high system pressures, his correlation did provide a correct length scale for the
boiling process.
Several other expressions have been reported for bubble diameter at depar-
ture, obtained either empirically or analytically by involving various forces
acting on a bubble. These expressions (see e.g. Hsu & Graham 1976), how-
ever, are not always consistent with each other. Cole & Rohsenow (1969)
correlated bubble diameter at departure with fluid properties but found it to be
independent of wall superheat. Gorenflow et al (1986) proposed an expression
for bubble diameter at departure that indicates the bubble diameter increases
weakly with wall superheat.
Some investigators disagree about the role of surface tension. Generally,
surface tension tends to push the bubble against the wall and thus inhibits bubble
departure. However, Cooper et al (1978) found that in some cases surface
tension assisted bubble departure by making the bubble spherical. Buyevich &
Webber (1996) also made the same argument. Issues regarding the forces that
act on a growing bubble can be put to rest only through complete numerical
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

374 DHIR

simulation of both bubble growth and departure while properly accounting for
the adhesion forces and interfacial tension.

BUBBLE RELEASE FREQUENCY A theoretical evaluation of the bubble release


frequency f can be made from the expressions for the waiting time tw and
the growth time tg. The waiting time corresponds to the time it takes for the
thermal layer to redevelop to allow nucleation of a bubble. Predictions of bubble
release frequency based on waiting and growth times, however, do not match
well with the data because many simplifications are made in obtaining tw and
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

tg. Thus, correlations have been reported in the literature that include both the
bubble diameter at departure and bubble release frequency. One of the most
comprehensive correlations of this type is given by Malenkov (1971).
by University of Nottingham on 08/04/10. For personal use only.

Heat Transfer Mechanisms


In partial nucleate boiling, or in the isolated bubble regime, transient conduction
into liquid adjacent to the wall is an important mechanism for heat transfer
from an upward-facing horizontal surface (Forster & Greif 1959). After bubble
inception, the superheated liquid layer is pushed outward and mixes with the
bulk liquid. The bubble acts like a pump in removing hot liquid from the surface
and replacing it with cold liquid. Combining the contribution of transient
conduction on and around nucleation sites, microlayer evaporation underneath
the bubbles, and natural convection on inactive areas of the heater, one can
write an expression for partial nucleate-boiling heat flux as
µ ¶
K2p K2
q= π(kρc p )l f Dd2 Na 1T + 1 − Na π Dd2 h̄ nc 1T
2 2
π 2
+ h̄ ev 1T Na Dd . (6)
4
In the above equation kl is the thermal conductivity of liquid, ρ l is the density
of liquid, and cρl is the specific heat of liquid.
Only the first two terms in Equation 6 were included in Mikic & Rohsenow’s
(1969) original model. The evaporation at the bubble boundary is included
in the first term that represents the transient conduction in the liquid. Judd &
Hwang (1976) suggested the addition of the last term on the right-hand side of
Equation 6. This term accounts for the microlayer evaporation at the base of
bubbles. For Equation 6 to serve as a predictive tool, several variables must
be known: the bubble diameter at departure, Dd; bubble release frequency, f,
the proportionality constant, K, for the bubble diameter of influence; number
density, Na, of active sites; and average heat transfer coefficients, h̄ nc and h̄ ev , for
natural convection and microlayer evaporation, respectively. Using empirical
correlations for several of these parameters, Mikic & Rohsenow (1969) justified
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 375

the validity of Equation 6 when the third term on the right-hand side of the
equation was not included.
Judd & Hwang (1976) matched the heat fluxes predicted from Equation 6
with those observed in the experiments in which dichloromethane was boiled
on a glass surface. In doing so, they relied on the measured values of microlayer
thickness to evaluate h̄ ev and on the assumption that K2 was 1.8. Experimentally
measured values of active nucleation site density and bubble release frequency
were used in the model. Figure 3 shows Judd & Hwang’s data and predictions.
At the total measured heat flux of 6 w/cm2, about one third of the energy is
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

dissipated through evaporation at the bubble base. The data plotted in Figure 3
by University of Nottingham on 08/04/10. For personal use only.

Figure 3 Relative contribution of various mechanisms to nucleate-boiling heat flux (Judd &
Hwang 1976). qM, measured heat flux; qP, predicted heat flux; qME, microlayer evaporation heat
flux; qNC, natural convection heat flux.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

376 DHIR

show that at high heat fluxes or in fully developed nucleate boiling, most of
the energy from the heater is removed by evaporation. This observation is in
general agreement with Gaertner’s (1965) finding that after the first transition
(partial to fully developed nucleate boiling), evaporation is the dominant mode
of heat transfer.
At low heat fluxes, or in partial nucleate boiling, the relative contribution
of various mechanisms depends on the geometry of the heater. In fact, the
details of the heat transfer mechanisms may be altered as heater geometry or
the angular position of the surface with respect to the direction of gravitational
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

acceleration is varied. For example, on a downward-facing surface, the bubbles


may slide along the heater surface for some distance after leaving the nucleation
site but before moving away from the heater surface. During the movement
by University of Nottingham on 08/04/10. For personal use only.

of the bubbles along the heater surface, cyclic disruption and reformation of
the thermal layer will occur and, in turn, will result in a higher heat transfer
rate. Figure 4 shows the nucleate boiling data obtained by Nishikawa et al
(1974) for water on flat plates inclined at different angles with the horizontal. In
partial nucleate boiling, the downward-facing surfaces accommodate heat fluxes
that are higher than those on an upward-facing horizontal surface or a vertical
surface. However, at high heat fluxes or in fully developed nucleate boiling,
the data for all of the surfaces fall on a single line. This observation indicates
that when evaporation is the dominant mode of heat transfer, the orientation of
the plate has little effect on dependence of heat flux on wall superheat.
In fully developed nucleate boiling, mushroom-type bubbles supported by
several vapor stems attached to the heater may be observed (Gaertner 1965;
see also Figure 2). Most evaporation occurs at the periphery of these stems
(smaller bubbles supporting large vapor masses). Energy for the phase change
is supplied by the superheated liquid layer in which the stems are implanted.
Thus, the boiling heat flux can be calculated if the fractional area occupied by
the vapor stems and the thickness of the thermal layer are known. The heater
area fraction occupied by the vapor stems is equal to the product of the number
density of stems and the wall area occupied by one stem.
Alternatively, the heat flux can be calculated if the vaporization rate per stem
and number density of active sites are known. Lay & Dhir (1994) used the latter
approach to predict fully developed nucleate-boiling heat flux. By assuming
that the duration for which vapor stems exist on the heater is much larger than
the time needed to form the stems, Lay & Dhir (1995a) carried out a quasi-static
analysis to determine the maximum diameter of vapor stems as a function of
wall superheat. The shape of the vapor stem depends on the value chosen for
the Hamaker constant.
Lay & Dhir’s analysis also showed that locally, in the ultra-thin film, heat
fluxes as high as 1.54 × 108 W/m2 could exist at a wall superheat of 20 degrees
K for water at 1 atmospheric pressure. This observation is worthy of further
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 377


Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org
by University of Nottingham on 08/04/10. For personal use only.

Figure 4 Nucleate boiling data of Nishikawa et al (1974) on plates oriented at different angles to
the horizontal.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

378 DHIR

exploration in our pursuit to accommodate very high heat fluxes at relatively


low wall superheats. From the analysis, the vaporization rate, ṁ s , per stem can
be calculated as a function of wall superheat. Using Wang & Dhir’s (1993a,b)
model for density of active sites, the heat flux in fully developed nucleate boiling
was calculated from
q = Na ṁ s h f g . (7)
Heat fluxes predicted from Equation 7 agreed well with the data. In fact,
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

good agreement with Gaertner & Westwater’s (1960) data was also seen when
number density of active sites reported by Gaertner & Westwater was used.
This approach needs to be verified further with data from other sources.
by University of Nottingham on 08/04/10. For personal use only.

Heat Transfer Correlations


Because mechanistic models are lacking for several parameters (e.g. Na, Dd,
f, h̄ ev ), prediction of heat flux from Equation 6 requires adjustment of several
empirical constants embedded in these parameters. As a result, Equation 6, pre-
sumably obtained on mechanistic arguments, cannot be readily used to predict
the dependence of nucleate-boiling heat flux on wall superheat. Most often,
correlations reported in the literature have been used for this purpose. These
correlations generally are valid for both partial and fully developed nucleate
boiling. For example, Rohsenow’s (1952) correlation has been used widely,
even though it is not based on correct physics. Stephan & Abdelsalam (1980)
developed a comprehensive correlation for saturated nucleate pool boiling of
different liquids. Their correlation is based on both fluid and solid properties,
but no consideration is given to heater geometry.
Cooper (1984a,b) proposed a simple correlation for saturated nucleate pool
boiling. His correlation uses reduced pressure, molecular weight, and surface
roughness as the correlating parameters. His correlation for a flat plate can be
written as
µ ¶0.12−0.21 log10 R p µ ¶
(q)1/3 p p −0.55
= 55.0 · − log10 · M −0.50 . (8)
1T pc pc
In Equation 8, the roughness, Rp, is measured in microns, M is the molecular
weight, p is the system pressure, pc is the critical pressure, 1T is measured
in degrees K, and q is given in W/m2. Cooper suggested that for application
of the correlation to horizontal cylinders, the lead constant on the right-hand
side should be increased to 95. Equation 8 accounts for roughness but does not
account for variations in degree of surface wettability. These correlations should
be used with caution, as large deviations between actual data and predictions
can occur when the conditions under which the correlation was developed are
not duplicated.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 379

Effect of System Variables


Several system variables such as surface finish, surface wettability, surface con-
tamination, heater geometry, liquid subcooling, flow velocity, gravity, system
pressure, thermal properties of the solid, and the mode in which the tests are
performed influence the dependence of nucleate-boiling heat flux on wall super-
heat. For example, surface roughness pushes the boiling curve to the left. Im-
proved wettability suppresses nucleation and, as a result, shifts the boiling curve
to the right. Physicochemical changes on the surface can take place because of
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

deposition of inert matter contained in the host liquid, slow chemical reaction
of the surface with the gases dissolved in the liquid or with the vapor, and strong
chemical reaction of the metal with the concentrated solutions of electrolytes.
Generally, the effect of surface contamination is to enhance the wettability and
by University of Nottingham on 08/04/10. For personal use only.

thereby reduce the nucleate-boiling heat flux for a given wall superheat.
As noted from Figure 4, partial nucleate-boiling heat fluxes generally are
higher on a downward-facing surface, but in fully developed nucleate boiling
the surface orientation has little effect. Thus, the geometry of the surface can
have an effect on partial nucleate-boiling heat fluxes. The rate of convective
heat transfer increases with liquid subcooling. As a result, liquid subcool-
ing influences the inception and partial nucleate boiling regions of the boiling
curve. On the wall heat flux vs wall superheat plots, convective and partial
nucleate-boiling heat fluxes for subcooled liquids lie higher than those for sat-
urated boiling. However, at high nucleate-boiling heat fluxes, the subcooled
and saturated boiling curves almost overlap. Similarly, flow velocity enhances
convective and partial nucleate-boiling heat fluxes but has little effect on fully
developed nucleate boiling.
The magnitude and direction of gravitational acceleration with respect to
the heater surface influences the hydrodynamic and thermal boundary layers
and bubble trajectory. In partial nucleate boiling, heat transfer by convection
represents a major fraction of the total heat transfer rate. Thus, gravity plays
an important role in this mode of boiling. However, Merte’s (1988) centrifuge
data and Zell et al’s (1989) low-gravity data showed that the magnitude of
gravity has little effect on fully developed nucleate boiling. With an increase
in system pressure, the incipience superheat decreases and the nucleate boiling
curve shifts to the left.
The nucleate-boiling heat transfer data collected by Stephan & Abdelsalam
(1980) suggested that thermophysical properties of the solid can have a weak
effect on nucleate-boiling heat fluxes. The boiling curve can be affected by
the manner in which the heat flux is imposed on the surface—steady state or
transient. Sakurai & Shiotsu’s (1977a,b) experiments on platinum wires sub-
merged in a pool of saturated water showed that for exponential heating periods
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

380 DHIR

varying from 5 ms to 1 s, the incipient heat flux increases as the exponential


time decreases. In nucleate boiling, the transient heat transfer coefficients gen-
erally are lower than those obtained under steady-state conditions. The ratio
of transient and steady-state heat fluxes depends on the magnitude of the heat
flux, but this ratio can be as low as 0.5.
Heat Transfer Enhancement
The size of equipment needed for a given heat load can be reduced if already
high nucleate-boiling heat fluxes can be further enhanced. Many research ef-
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

forts have been made in this direction. Recent advances in enhancing nucleate-
boiling heat flux have included the development of heater surfaces with high-
density interconnected artificial cavities of the re-entrant type. Figure 5 shows
by University of Nottingham on 08/04/10. For personal use only.

a reentrant-type cavity and the structures of two of the commercially available


surfaces. The cavities on these surfaces nucleate at very low superheats. The
nucleate-boiling heat fluxes are enhanced not only by the high density of active
nucleation sites at low superheats but also by the evaporation of a thin liquid
film formed on the cavity walls. The film results from the liquid that is pushed
into the cavity after a bubble leaves. The enhanced surfaces have led to an
order of magnitude increase in already high nucleate-boiling heat transfer co-
efficients. However, enhancement is much less at high wall superheats or near
the maximum heat flux condition.

MAXIMUM HEAT FLUX


The maximum or critical heat flux represents the upper limit of nucleate-boiling
heat flux and marks the termination of efficient cooling conditions on the sur-
face. Several experimental and theoretical studies have been reported that
delineate the physics of onset of the critical heat flux condition in pool boiling.
However, no clear consensus exists in the technical community as to the actual
mechanism of critical heat flux.

Figure 5 Reentrant cavity and commercially available enhanced surfaces. (a) Cross-section of
reentrant-type cavity. (b) Thermal excel-E surface. (c) Gewa-T surface.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 381

Mechanisms
Kutateladze (1948) and Zuber (1959) proposed two early models for prediction
of maximum heat flux on large horizontal surfaces. Both models are based
on the hydrodynamics of vapor outflow. Kutateladze developed dimensionless
groups from the equations governing the flow of vapor and liquid. Zuber, in
contrast, proposed that the maximum heat flux occurs when velocity in the vapor
jets issuing from the surface reaches a critical velocity. The critical velocity is
the velocity at which vapor jets become Kelvin-Helmholtz unstable. Zuber also
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

assumed that the jet diameter was half of the jet spacing that was bounded by
the “critical” and the “most dangerous” two-dimensional Taylor wavelengths.
For inviscid liquids, Zuber obtained an expression for the maximum heat flux
on infinite flat plates as
by University of Nottingham on 08/04/10. For personal use only.

s µ ¶ · ¸
σ g(ρl − ρv ) ρl + ρv 1/2 ρl (16 − π )
qmax F = Cρv h f g 4 . (9)
ρv2 ρl ρl (16 − π ) + ρv π

At low system pressures, Equation 9 and the expression obtained by Kutate-


ladze are nearly identical. The value obtained by Zuber for constant C was
π/24, whereas Kutateladze correlated the data available at that time and found
the constant C to have a value of 0.168. Subsequently, Lienhard & Dhir (1973)
obtained data with a variety of fluids at different accelerations normal to the
heaters and concluded that for large horizontal plates constant, C should have
a value of 0.15. From the data they also deduced that for a plate to be called a
large plate it should at least accommodate three Taylor wavelengths. Neither
model accounted for the surface wettability, and presumably the underlying
assumption in these models was that liquids wetted the heater surface well.
Equation 9 has also been extended to predict maximum heat flux on heaters
of different geometry, size, and orientation (Lienhard & Dhir 1973). For heaters
of other geometries, the maximum heat flux is written as

qmax = f (l 0 )qmax F , (10)

where f (l 0 ) is a function of dimensionless characteristics width, l 0 , of the heater,


which is defined as
l
l0 = q . (11)
σ
g(ρl −ρv )

For small heaters, l 0 <


e 1, the function f (l 0 ) increases as l 0 decreases. For
large heaters, the function f (l 0 ) becomes independent of l 0 and attains a value
slightly less than unity. However, the exact value of f (l 0 ) depends on the
heater geometry. The prediction of maximum heat flux on heaters of different
geometries requires a knowledge of the ratio of vapor jet to heater area and of
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

382 DHIR

critical velocity of vapor in the jets. Lienhard & Dhir (1973) summarized the
methodology for evaluating f (l 0 ) for various heater geometries. Predictions
from Equation 10 agree with a large set of maximum heat flux data obtained
with different liquids and heater geometries.
Haramura & Katto (1983) have questioned the validity of the assumption of
instability of large vapor jets used in the hydrodynamic theory as originally pro-
posed by Zuber and its subsequent augmentation by Lienhard & Dhir (1973).
This questioning is based on the fact that visual observations show the presence
of large vapor mushroom-type bubbles on the heater surface rather than tall
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

vapor jets. Haramura & Katto (1983) suggested an alternative hydrodynamic


model for prediction of maximum heat flux under pool boiling conditions.
In their model, the vapor stems supporting mushroom-type bubbles become
by University of Nottingham on 08/04/10. For personal use only.

Helmholtz unstable. Maximum heat flux is proposed to occur when the liquid
film trapped between the base of the mushroom-type bubble and the wall dries
out prior to departure of the bubble (hovering period). The thickness of the
liquid film is assumed to be one fourth of the Helmholtz unstable wavelength.
By comparing the maximum heat flux predicted from their model with the pre-
diction from Equation 9 with C = π/24, Haramura & Katto found the ratio
of vapor stem to heater area to be a function of vapor to liquid density ratio
only. However, the predicted magnitude of the area ratio is not borne out by
experiments, and the model is unable to describe the observed effect of surface
wettability on the maximum heat flux. Nevertheless, several investigators have
extended the Haramura & Katto model to flow boiling and to jet impingement
cooling.
From the early 1960s to the late 1970s, the hydrodynamic theory was well
accepted as a model of the maximum heat flux mechanism under pool boiling
conditions. However, during that time period, questions persisted regarding the
theory’s ability to predict maximum heat fluxes on surfaces that were not well
wetted. The observed maximum heat fluxes on partially wetted surfaces are
lower than those predicted by the hydrodynamic theory.
Only recently have several studies documented, unambiguously, the effect
of surface wettability. Liaw & Dhir (1986) studied boiling of saturated water at
1 atm on a vertical surface. A prescribed procedure was followed for oxidation
of the surface, and the static contact angle was used as the measure of the
degree of wettability. Maracy & Winterton (1988) obtained similar data on a
horizontal plate, whereas Hahne & Disselhorst (1978) used horizontal cylinders
of different materials. These investigators showed that the maximum heat flux
decreases with increase in contact angle. However, in comparison to the data
of Liaw & Dhir and Maracy & Winterton, the data of Hahne & Disselhorst
obtained on cylinders showed a much stronger dependence of maximum heat
flux on contact angle.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 383


Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org
by University of Nottingham on 08/04/10. For personal use only.

Figure 6 Dependence of peak heat flux on contact angle.

Figure 6 shows the steady-state peak heat flux data obtained by Liaw & Dhir
(1986). Saturated water at 1 atm pressure was the test liquid. The data are
plotted as a function of contact angle and were taken on a 6.3-cm wide and
10.3-cm high copper plate. Also plotted are the data obtained with R-113,
which wets the polished copper surface well. The dotted lines in Figure 6 show
the predictions obtained from Equation 9 using the value of C suggested by
Zuber (for an infinite horizontal plate) and that suggested by Lienhard & Dhir
for a vertical plate. The data obtained with R-113 and with water at a contact
angle of 18◦ are within a few percent of the prediction based on hydrodynamic
theory. However, water data covering a range of contact angles from 27◦ to
107◦ are much lower.
For a contact angle of 90◦ (polished copper, distilled water), the observed
maximum heat flux is only about 55% of that given by Lienhard & Dhir (1973).
The reduction of maximum heat flux noted by Costello & Frea (1965) when
distilled water was used instead of tap water can thus be attributed to the reduced
wettability. Lienhard & Dhir explained these data by considering the number
of vapor jets that the heater accommodated. Since the available buoyancy
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

384 DHIR

force can sustain a vapor removal rate corresponding to the maximum heat
flux for an 18◦ contact angle, the hydrodynamics of the vapor outflow cannot
determine the maximum heat flux on partially wetted surfaces. In contrast,
because the maximum heat flux data appear to be correlated with the surface
wettability (surface property), the upper limit of heat removal is likely set by the
surface.
Further evidence that hydrodynamics does not control the maximum heat flux
on partially wetted surfaces is obtained from the void fraction profiles. The void
fraction on partially wetted surfaces at maximum heat flux is less than unity
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

everywhere normal to the surface; thus, flow paths are available for the liquid
to reach the heater surface. Dhir & Liaw (1989) explained the occurrence of
maximum heat flux on partially wetted surfaces on the basis of evaporation area
by University of Nottingham on 08/04/10. For personal use only.

available at the stem interface. For a contact angle of 90◦ , the vapor stems merge
at the wall when the wall void fraction attains a value of π /4. For contact angles
less than 90◦ , the merger occurs away from the wall. After merger, the stem
circumference in contact with liquid decreases rapidly. The higher the interface
area available for evaporation at a given superheat, the higher the heat flux (Dhir
& Liaw 1989). Thus, a maximum in the interfacial area corresponds to onset
of the maximum heat flux condition, or the maximum rate of evaporation at the
stems sets the upper limit of nucleate-boiling heat flux. For partially wetted
surfaces, the merger of vapor stems signals a degradation in heat removal rate
or onset of transition boiling. In some respects, this mechanism for maximum
heat flux is similar to Rohsenow & Griffith’s (1956) proposal that onset of
the maximum heat flux condition is due to packing of bubbles, which leads to
coalescence of isolated bubbles at the heater surface.
For surfaces with contact angles less than 90◦ , maximum void fraction occurs
slightly away from the wall. At maximum heat flux on well-wetted surfaces
(maximum heat fluxes approaching those predicted from the hydrodynamic
theory), the void fraction away from the wall reaches a value of unity. At a
void fraction equal to unity, an obstruction to the flow of liquid to the wall can
develop. Since in steady state the vapor production rate must equal the vapor
removal rate, this condition appears to be analogous to what has been assumed
in the past with respect to the hydrodynamically controlled boiling crisis in
pool boiling. The hydrodynamic theory proposed by Zuber (1959) assumed
that the maximum heat flux occurs when vapor escape velocity and vapor flow
area fraction reach their critical values. A void fraction of unity slightly away
from the heater is an alternative form of the same criterion.
Thus, Dhir & Liaw (1989) have proposed different mechanisms for maximum
heat flux on partially wetted and well-wetted surfaces. For partially wetted
surfaces the mechanism depends on the evaporation rate at the surface, whereas
for well-wetted surfaces the mechanism depends on the vapor removal rate or
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 385

occurrence of the void fraction of unity (hydrodynamic limit) slightly away


from the wall.

Unified Approach
Dhir & Liaw also provided a framework for a unified model for nucleate and
transition pool boiling on partially wetted surfaces. In their model the maxi-
mum heat flux condition was not considered as a disjoint process but rather as
a transition point in the continuously evolving q − 1T curve encompassing the
three modes of boiling (i.e. nucleate boiling, transition boiling, and film boil-
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

ing). Since heat removal on both the wet and dry areas contributes to the overall
heat transfer from the surface, an expression for the time- and area-averaged
heat flux may be written as follows:
by University of Nottingham on 08/04/10. For personal use only.

q = q̄l (1 − ᾱw ) + q̄v ᾱw


≡ h̄ l (1 − ᾱw )1T + h̄ v ᾱw 1T. (12)

In Equation 12, q̄l and h̄ l are the time- and area-averaged heat flux and heat
transfer coefficient, respectively, over the liquid-occupied region. Similarly, q̄v
and h̄ v are the time- and area-averaged heat flux and heat transfer coefficient,
respectively, on the dry region.
In Equation 12, the temperature over the dry and wet areas is assumed to be
the same. Such an assumption is true for thick copper plates. However, for
thin heaters made out of low-conductivity materials, a significant difference in
temperature between dry and wet areas may exist. In evaluating the heat flux
from Equation 12, Dhir & Liaw (1989) calculated h̄ l by knowing the energy
removal rate by evaporation at the periphery of vapor stems. The heat transfer
coefficient, h̄ v , over the dry region was obtained from Bui & Dhir’s (1985a)
data for film boiling on a vertical surface, and experimentally measured values
of wall void fraction were used. As discussed earlier, Wang & Dhir (1993a,b)
and Lay & Dhir (1995a) provided a theoretical basis for prediction of number
density of active sites and the diameters of the vapor stems, respectively. If the
number density and the dry area underneath a vapor stem are known, the wall
void fraction can be determined.
Although Dhir & Liaw used the unified model to predict nucleate and transi-
tion boiling on partially wetted surfaces, the approach can be applied to surfaces
on which maximum heat flux is determined by the vapor removal limit (e.g.
well-wetted surfaces). However, implementation of the model requires knowl-
edge of the variation in wall void fraction with wall superheat after the vapor
removal limit has been reached. To have a totally predictive model for post–
critical heat flux, one needs to have models for wall void fraction and the shape
of the interface near the heater surface.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

386 DHIR

Effect of System Variables


Several system variables affect the maximum heat flux: surface wettability;
heater geometry, size, material, and thickness; liquid subcooling; flow velocity;
gravity; system pressure; and the mode in which the surface is heated. The effect
of wettability, geometry, and size of the heater on the maximum heat flux was
discussed earlier. Ample evidence supports the conclusion that for thin heaters
made of low-conductivity materials such as steel or inconel, the maximum heat
flux is lower than that predicted from the hydrodynamic theory (well-wetted
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

surface). In the earlier studies, the maximum heat flux was correlated with the
product of density, specific heat, and thickness of the heater material (Houchin
& Lienhard 1966, Tachibana et al 1967).
Bar-Cohen & McNeil (1992), Carvalho & Bergles (1992), and Golobic &
by University of Nottingham on 08/04/10. For personal use only.

Bergles (1992) analyzed a large body of critical heat flux data on heaters of
different materials and thickness. They correlated the critical heat flux with
conpacitance (the product of the heater thickness and of the square root of the
product of the thermal conductivity, specific heat, and density of the heater
material). For thick heaters, the critical heat flux appeared to asymptotically
approach the hydrodynamic limit. From such a correlation of the data, Carvalho
& Bergles found the thickness of the heater material required to achieve at least
90% of the asymptotic value of the critical heat flux.
The maximum heat flux increases with liquid subcooling. Zuber et al (1961)
extended Equation 9 to a subcooled liquid by accounting for heat lost to the
liquid in a transient manner during growth of a bubble. Elkassabgi & Lienhard
(1988) investigated maximum heat fluxes during subcooled pool boiling on
horizontal cylinders. They identified three subcooling regimes. For low sub-
coolings, the maximum heat flux varies linearly with subcooling in a manner
similar to that observed by Zuber et al. At moderate subcoolings, bubbles sur-
rounded the heater without detaching. For these subcoolings, the maximum
heat flux varied slightly nonlinearly with liquid subcooling and was determined
by natural convection from the outer edge of the bubble boundary layer. At
high subcoolings, the maximum heat flux was independent of liquid subcooling
and was limited by the evaporation rate at the heater surface (i.e. molecular
effusion limit) and not by the rate at which energy could be removed by natural
convection from the outer edge of the bubble boundary layer.
The flow velocity enhances the maximum heat flux and is discussed later.

According to Equation 9, the maximum heat flux should scale as 4 g. However,
for very low gravities (µg), the functional dependence of maximum heat flux
on gravity is weaker than that obtained from Zuber’s hydrodynamic analysis.
The reasons for weaker dependence of maximum heat flux on gravity under
microgravity conditions are not clearly understood. Questions also remain
about the stability of boiling. Merte (1994) reported that subcooled boiling
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 387

during long periods of microgravity is unstable and that the surface alternately
wets and dries out prior to occurrence of critical heat flux.
The effect of system pressure on maximum heat flux is built into the hy-
drodynamic model (Equation 9). With increase in system pressure, the critical
heat flux attains a maximum value near a reduced pressure of about 0.35. The
magnitude of the maximum heat flux is affected if the heat input to the heater
is increased rapidly. Sakurai & Shiotsu (1977a,b) found that for exponential
heating periods less than 100 ms, the transient maximum and DNB heat fluxes
increase as the exponential time decreases. The DNB heat flux is defined as
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

the highest nucleate-boiling heat flux at which a linear relationship between


ln q and ln 1T ceases to exist. Expressions for transient maximum heat fluxes
using a steady-state critical heat flux model as the starting point have been de-
by University of Nottingham on 08/04/10. For personal use only.

veloped (Serizawa 1983, Pasamehmetaglou et al 1987). Maximum heat fluxes


observed during quenching of solids are generally lower than their steady-state
values. By carrying out quenching experiments on copper discs in liquid ni-
trogen, Peyayopanakul & Westwater (1978) showed that transient maximum
heat fluxes decrease as disc thickness decreases. However, for discs thicker
than 2.5 cm, the maximum heat flux is independent of thickness. If the time to
traverse the top 10% of the boiling curve is greater than 1 s, the boiling process
can be called quasi-steady. Lin & Westwater (1982) showed that similar to
steady-state experiments, the heater thickness and thermophysical properties
of the heater have some influence on the boiling curve, including maximum
heat flux obtained during quenching.

TRANSITION BOILING
Transition boiling is characterized by a reduction in surface heat flux with an
increase in wall superheat. As a result, the process is inherently unstable. In
transition boiling, periods of liquid-solid and vapor-solid contact occur alterna-
tively at a given location on the heated surface. Conditions similar to nucleate
boiling and film boiling prevail during wet and dry periods respectively. The
variation in heat flux with wall superheat is a result of change in the fraction
of time each boiling mode is present on a given area. Reviews on transition
boiling have been presented (e.g. Dhir 1991, Auracher 1992), and major points
from these reviews and some recent results are included here.
Since Witte & Lienhard (1982) asserted that Berenson’s (1962) transition
boiling data showed hysteresis, several studies on the issue have appeared in
the literature. Bui & Dhir (1985b) showed from their experiments on a vertical
surface that different transition boiling curves were obtained depending on
the side of the curve from which the boiling was accessed (i.e. the nucleate
boiling side or the film boiling side). Bui & Dhir (1985b) and Liaw & Dhir
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

388 DHIR

(1986) showed that the magnitude of hysteresis in transition boiling curves


depends on the static contact angle. The larger the static contact angle, the larger
the hysteresis. In fact, for R-113, which wets the surface well, no hysteresis
was observed when transition boiling was accessed either from the film boiling
side or from the nucleate boiling side. However, Rajab & Winterton (1990)
claim that in their experiments on a horizontal surface, hysteresis persisted even
when liquid wetted the surface well (i.e. nearly zero contact angle).
Ramilison & Lienhard (1987) recreated Berenson’s apparatus in which steam
was condensed on the underside of a copper disc while boiling occurred on
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

the top of the disc. In this experiment, although thermal resistance of the
copper heater was reduced, not all points in transition boiling were accessible.
Ramilison & Lienhard conjectured that the shift from film-transition boiling
by University of Nottingham on 08/04/10. For personal use only.

to nucleate-transition boiling was a result of the change from an advancing


contact angle to a receding contact angle. Thus, they implied that hysteresis
was a result of differences in advancing and receding contact angles.
Haramura’s (1991) data, also obtained on an apparatus similar to that of
Berenson, showed that for R-113, some hysteresis existed even though the data
were obtained under steady-state conditions. Auracher (1992) developed a
feedback system so that in electrically heated systems, steady-state transition
boiling data could be obtained. He found that no hysteresis existed during tran-
sition boiling of R-113 when data were obtained under steady-state conditions
under either increasing or decreasing temperature conditions. From this obser-
vation, Auracher concluded that hysteresis observed in transition boiling was
due to the transient nature of the data obtained in previous studies and that no
hysteresis existed when the data were obtained under steady-state conditions.
The issue is far from being resolved since Auracher did not provide any steady-
state transition boiling data with liquids having contact angles vastly different
from zero, and a majority of the transition boiling data showing hysteresis were
obtained by other investigators only under relatively slow transient conditions.
Equation 12 has been used to correlate the dependence of wall heat flux
on wall superheat in transition boiling. Generally, empirical expressions for
dependence of q̄l , q̄v , and ᾱw on 1T are used so that predicted heat fluxes match
the transition boiling data and the maximum and minimum heat fluxes at the
upper and lower end of the transition boiling data. A few semi-mechanistic
approaches (e.g. Shoji 1992) have been proposed for prediction of transition
boiling heat fluxes. Although these models agree with the data reasonably well,
independent verification of different submodels that contribute to the overall
model is lacking.
To facilitate mechanistic modeling of transition boiling, several experimental
studies have measured the wet area fraction during nucleate boiling (see Dhir
1991) and heat transfer associated with liquid contacts (see Chen & Hsu 1995).
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 389

Using a microthermocouple probe, Chen & Hsu (1995) measured the transient
surface heat fluxes when a liquid droplet lands on a hot, initially dry, surface.
They noted that the process of heat transfer during a short period of liquid
contact is very complex but found that the average heat flux increases with wall
superheat. During contact periods the droplets that initially had a subcooling of
80◦ C yielded average heat fluxes as high as 107 W/m2 at high wall superheats.
More systematic studies similar to that of Chen & Hsu (1995) are needed
to understand the physics of heat transfer during transient liquid-solid contacts
before credible models for transition boiling can be developed. These studies
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

need to be supplemented with measurement of position- and time-dependent


temperature and wall void fractions and with visual observations of the structure
of the vapor-liquid interface.
by University of Nottingham on 08/04/10. For personal use only.

FILM BOILING
Film boiling is amenable to straightforward analysis; however, in carrying out
the analysis, many simplifications with respect to the matching of interfacial
conditions and the shape of the interface are made. As a result, empirical
constants must be used to match predictions with the data.

Semi-Mechanistic Models
Sakurai et al (1990a,b) developed comprehensive correlations for saturated and
subcooled film boiling on horizontal cylinders. In developing the correlations,
which include the effect of radiation, the functional form of the correlations was
obtained by solving the conservation equations for a two-layered configuration
of subcooled film boiling. Sakurai et al also developed a database that covered
a large range of system variables such as heater diameter, wall superheat, liquid
subcooling, and pressure. The correlations agreed well with the data, including
that obtained for cryogenic liquids.
Sakurai & Shiotsu (1992) extended their correlations to include a vertical
surface and a sphere. By measuring the size of the bubble departing a heater
under saturated film boiling conditions and bubble release frequency, Sakurai &
Shiotsu also developed an expression for minimum heat flux on horizontal cylin-
ders. In addition, they showed that Lienhard & Wong’s (1964) semi-mechanistic
correlation tends to overpredict the effect of system pressure, and that the col-
lapse of vapor film is influenced by the heater surface temperature rather than
the heat flux. For saturated film boiling, the minimum film boiling tempera-
ture increased with pressure and asymptotically approached the homogeneous
nucleation temperature. In correlating the minimum film boiling temperature,
Sakurai & Shiotsu (1992) accounted for the reduction in the heater surface tem-
perature that occurs upon spontaneous contact of the liquid with the surface.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

390 DHIR

Mechanistic Studies
Studies have also focused on understanding the structure of the vapor-liquid
interface in film boiling. Bui & Dhir (1985a) investigated saturated film boiling
on a vertical surface. They noted that both long and short waves exist on
the interface. The long waves evolve into large bulges, and vapor from the
intervening thin-film region feeds the large bulges. As a result, the flow path
length is shortened, and higher average heat transfer rates occur than those
predicted for a continuous flow path. They also noted that a significant variation
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

in local heat transfer coefficient occurs with time as large bulges and thin-film
regions sweep over the heated surface. In a study of subcooled film boiling
on a vertical surface, Vijaykumar & Dhir (1992a) noted that, as in saturated
film boiling, two types of waves (termed ripples and large waves) exist on the
by University of Nottingham on 08/04/10. For personal use only.

interface.
Figure 7 shows photographs of the frontal view of film boiling on a vertical
surface. The waves tend to form at a short distance downstream of the leading
edge. The amplitude of the long waves controls the interface velocity. Based
on the amplitude-to-wavelength ratio, the interface behavior in subcooled film
boiling was divided into two regimes. In the low subcooling regime, three-
dimensional waves exist on the interface. The flows that result from the de-
veloping liquid boundary layer and from the rapid acceleration of the interface
merge downstream of the leading-edge vapor layer. The fluid in the merged
region is entrained by the moving interface.
Figure 8 shows profiles of axial velocity in the liquid boundary layer when
water had a subcooling of 6.7 K. At the top of the first wave peak downstream
of the leading edge, the velocity profile in the liquid shows a steep gradient.
In the valley behind the first peak, flow expands and velocities decrease in the
boundary layer. A local minimum and a maximum in velocity profile can be
noted in the boundary layer. Flow contracted and expanded at the succeeding
wave peaks and valleys, respectively. At higher subcoolings, wave structure
degenerates into two-dimensional waves. The liquid boundary layer thins, but
the phenomenon of expansion and contraction of the flow in the valleys and at
the peaks, respectively, persists.
Using a holographic technique, Vijaykumar & Dhir (1992b) also obtained
temperature profiles and heat fluxes in the liquid layer adjacent to the vapor
film. Figure 9 shows the interface shape and the heat flux into the liquid at
the interface for water subcooling of 1.6 K. A relatively large heat flux exists
at the frontal region of the leading-edge layer (location a). The rate of heat
transfer decreases to a minimum value (location f) and thereafter remains fairly
constant in the valley region. At the frontal region of the first peak, the heat
flux increases gradually, reaching a maximum at the top of the peak before
decreasing again.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 391


Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org
by University of Nottingham on 08/04/10. For personal use only.

Figure 7 Photographs of frontal view of film boiling on a vertical surface.

From information such as that shown in Figure 9, Vijaykumar & Dhir (1992b)
concluded that liquid-side heat transfer over each wavelength shows a cyclic
behavior with the highest heat flux occurring at the peaks and the minimum
occurring in the wave valleys. Invariably, the highest heat flux exceeds the heat
flux by conduction through the vapor bulge. High local liquid-side heat flux
at the peaks suggests the possibility of local condensation. Evaporation in the
valleys and condensation at the peaks results in little increase in the substrate
film thickness in the vapor flow direction. The average liquid-side heat transfer
is enhanced both by the cyclic behavior and by the increased interfacial area.
Thus, Vijaykumar & Dhir were able to explain that the underprediction of
subcooled film boiling heat transfer by the two-layer models that assume a
plane interface is largely the result of the underprediction in the liquid-side heat
flux.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

392 DHIR
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org
by University of Nottingham on 08/04/10. For personal use only.

Figure 8 Velocity profiles of the component of the velocity (1T = 207 K, 1Tsub = 6.7 K) along
the vertical plate.

Complete Numerical Simulation


Son & Dhir (1996) carried out a complete numerical simulation of the evolu-
tion of the liquid-vapor interface during saturated film boiling on a horizontal
plate. They invoked the assumption of axi-symmetry and considered the cir-
cular regions around the nodes and anti-nodes of the Taylor wave. Each of the
circular regions was assumed to have an area equal to half the square of the two-
dimensional “most dangerous” Taylor wavelength. Figure 10a shows at 1-atm
pressure the calculated shapes of the evolving interface for a β (≡ c pv 1T / h f g )
value of 0.09, which corresponds to a wall superheat of 100 K for water.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 393


Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org
by University of Nottingham on 08/04/10. For personal use only.

Figure 9 Liquid-side heat transfer as a function of distance along the vertical plate for 1T = 160
K, 1Tsub = 1.6 K.

Figure 10b shows the dependence of Nusselt number on dimensionless time


and dimensionless radial position for β = 0.09. Most of the heat is transferred
in the thin-film region and in the bubble region in the vicinity of the point of
the minimum film thickness. Little heat transfer takes place under the bubble
core. The magnitude of the highest heat transfer coefficient increases with time,
and the location at which the film is the thinnest moves radially inward as the
interface evolves into a bubble.
These observations run counter to Berenson’s (1961) assumption of a uniform
heat transfer rate in the thin-film region connecting neighboring bubbles. The
magnitude of the highest heat transfer coefficient increases as the wall superheat
or β decreases. Since the film is the thinnest where the heat transfer rate is the
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

394 DHIR
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org
by University of Nottingham on 08/04/10. For personal use only.

Figure 10 Prediction of film boiling on a horizontal surface. (a) Evolution of the interface.
(b) Heat transfer coefficient as a function of position and time. (c) Nusselt number based on
area-averaged heat transfer coefficient.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 395


Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org
by University of Nottingham on 08/04/10. For personal use only.

Figure 10 (Continued )

highest, the magnitude of the minimum thickness decreases as wall superheat


decreases. At a certain superheat, the film may become so thin that it ruptures.
Any perturbations near the interface can accelerate the rupture process. Film
rupture can lead to local liquid-solid contacts and can, in turn, cause the stable
film boiling to cease. However, the exact nature of these contacts and the
spreading behavior can be determined only if a conjugate problem involving
conduction in the solid is solved simultaneously. This was not done in Son &
Dhir’s study (1996).
Figure 10c shows the Nusselt numbers based on the heat transfer coefficient
averaged over the cell area. With increase in wall superheat, vapor film thickness
in the thin-film region increases, and as a result, the heat transfer coefficient
decreases. The average Nusselt number also shows some dependence on time,
but it is much less than that seen in the local heat transfer coefficient.
The Nusselt numbers based on the area- and time-averaged heat transfer co-
efficients obtained by integrating the area under the curves (see Figure 10c) are
about 30 to 35% lower than those obtained from Berenson’s (1961) correlation.
Also, the numerically calculated time- and area-averaged heat transfer coeffi-
cients are closer to the lower bound of Hosler & Westwater’s (1962) data. One
possible reason for the underprediction of the heat transfer rate from the numeri-
cal simulation is the use of axi-symmetric analysis instead of three-dimensional
analysis, which is more appropriate for the low-pressure film boiling on a hor-
izontal surface.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

396 DHIR

Nevertheless, the analysis provides a better physical description of the film


boiling process on a horizontal surface. The numerical analysis such as that
used by Son & Dhir can be a useful experimental tool for investigating the
effect of various system variables on the film boiling process, including the
thermal properties of the heater material. Numerical simulation of phase change
processes is expected to be used much more in the future.

FLOW BOILING
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

External Flow Boiling


Two of the most often studied geometries under external flow conditions are
those of a liquid jet impinging on a horizontal surface and of flow normal to
by University of Nottingham on 08/04/10. For personal use only.

a horizontal cylinder. In these two cases, the flow can be along the direction
of gravity, against the direction of gravity, or normal to gravity. Fully devel-
oped nucleate boiling data obtained under forced flow conditions represent an
extension of the pool boiling curve. However, in partial nucleate boiling, the
functional dependence of the boiling heat flux on wall superheat is weaker under
forced flow condition.
Several investigators have developed semi-theoretical correlations for the
maximum heat flux obtained with impinging jets. Monde (1987) extended Hara-
mura & Katto’s (1983) critical liquid-layer model, whereas Sharan & Lienhard
(1985) used the mechanical energy stability criterion initially proposed by Lien-
hard & Eichhorn (1979). According to this criterion, the maximum heat flux
occurs when the rate at which kinetic energy added to the coolant (as a result of
evaporation) exceeds the rate at which energy is consumed in the formation of
new droplets. During jet impingement cooling, substantially higher maximum
heat fluxes are obtained at relatively low jet velocities.
Lay & Dhir (1995b) showed that macro- or micro-modification of the surface
can enhance twofold to threefold the maximum or critical heat flux during jet
impingement cooling. The macro-modification is in the form of radial channels,
which help retain the liquid on the surface. The micro-modifications, in contrast,
provide a high density of active nucleation sites at low wall superheats.
Lienhard (1988) reviewed the area of prediction of maximum heat flux
on cylinders. He also gave correlations applicable to gravity-influenced and
gravity-free data. Jensen & Hsu (1988) showed that for upflowing cross-flow
over tube bundles, nucleate-boiling heat transfer coefficients are influenced by
the location of the tube in the bundle. Because of an accumulation of vapor
along the flow direction, flow regimes change. A liquid film with vapor core
is observed on tubes far away from the inlet. Thus, different types of critical
heat flux mechanisms can prevail in the lower and upper parts of a tube bundle.
Jensen (1988) reviewed cross-flow boiling on horizontal tube bundles.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 397

Internal Flow Boiling


Extensive studies of boiling in tubes have been reported in the literature because
of the need for researchers to understand the cooling limits of nuclear reactor
cores and steam generators. As a result of the addition of heat along the axis
of the tube, the enthalpy of the liquid entering the tube increases as it flows
through the tube. When a subcooled liquid enters the tube, forced convection is
followed by subcooled boiling at the wall, which in turn gives way to saturated
or bulk boiling. After initiation of bulk boiling, the addition of vapor along
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

the tube axis causes flow regimes to change from bubbly flow, to slug flow, to
annular flow, and eventually to entrained flow.
In bubbly flows, discrete vapor bubbles exist in the continuous phase (liquid).
With an increase in the vapor flow rate, small bubbles merge to form long
by University of Nottingham on 08/04/10. For personal use only.

bubbles separated by liquid filaments. These bubbles occupy almost the entire
cross-section of the tube. As the vapor flow rate increases further, the long
bubbles merge to give rise to annular flow. In annular flow, liquid is confined
to the thin region adjacent to the wall, whereas vapor occupies the core of the
tube. At a still higher vapor flow rate, liquid droplets are entrained in the vapor
phase, which is now continuous.
Bergles & Rohsenow (1964), among others, suggested a correlation for par-
tial nucleate boiling based on interpolation of the data for forced convection and
fully developed nucleate boiling. For fully developed nucleate boiling in forced
flow, the correlations developed for pool boiling are applicable, although cor-
relations specific to flow boiling in tubes have also been reported. According to
these correlations, the exponent m in Equation 1 varies between 2 and 4. These
correlations are valid for both bubbly and slug flows. In annular flow, a liquid
film covers the heated surface. Chen (1963) developed a correlation that ac-
counts for nucleate boiling in the liquid film and forced convection cooling of the
wall by the film. In entrained flow, correlations for single phase forced flow are
applicable. However, these correlations must be corrected for the presence of
droplets in the core flow. The droplets generally enhance the wall heat transfer.
For flow in tubes, a distinction must be made between critical heat flux
under low-flow and high-flow conditions. Under low-flow conditions, the liquid
film can dry out and lead to a rise in wall temperature as the heat removal
rate degrades. Under high-flow conditions, the critical heat flux condition
corresponds to local dryout of the tube surface even though the tube core is
full of liquid. Upon occurrence of the critical heat flux condition, the tube
surface temperature rises rapidly to a very high value, and the tube can fail if
the temperature exceeds the melting temperature of the heater material.
Several correlations for dryout and critical heat flux covering different fluids,
flow rates, tube diameter, local quality, and system pressure have been reported.
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

398 DHIR

“Look-up” tables (see e.g. Groenveld & Lueng 1989) have been developed that
require less computing time and that can be updated easily as new data become
available.
Mechanistic models for critical heat flux in tubes have also been developed.
These models, however, use empirical constants to match the predictions with
the data. Several competing mechanisms for critical heat flux have been pro-
posed. Two of the most commonly proposed mechanisms for critical heat flux
are (a) the inability of the core flow to remove vapor generated at the wall,
and (b) the formation of a persistent dry patch underneath a bubble attached to
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

the wall or resulting from dryout of the thin liquid film between the wall and
a large bubble. Katto (1994) and Celata et al (1994) provided comprehensive
reviews of correlations and models for critical heat flux. Efforts similar to those
by University of Nottingham on 08/04/10. For personal use only.

described earlier for film boiling are currently being made to predict critical
heat flux by carrying out complete numerical simulation of the process (see e.g.
Lahey 1996).

ACKNOWLEDGMENTS
The author appreciates the support from the NASA Microgravity Fluid Physics
Program, with Dr. David Chao as Project Scientist. The author is also grateful
to Ms. Cindy Gilbert for skillfully typing the manuscript.

Visit the Annual Reviews home page at


http://www.AnnualReviews.org.

Literature Cited

Auracher A. 1992. Transition boiling in natural mination of forced convection surface boiling
convection systems. See Dhir & Bergles, pp. heat transfer. J. Heat Transfer 86:365–72
219–36 Bier K, Gorenflow D, Salem M, Tanes Y. 1978.
Bankoff SG. 1958. Entrapment of gas in the Pool boiling heat transfer and size of active
spreading of liquid over a rough surface. nucleation centers for horizontal plates with
AIChE J. 4:24–26 different surface roughness. Proc. Int. Heat
Bar-Cohen A, McNeil A. 1992. Parametric ef- Transfer Conf., 6th, Toronto, 1:151–56
fects on pool boiling critical heat flux in di- Bui TD, Dhir VK. 1985a. Film boiling heat
electric liquids. See Dhir & Bergles, pp. 171– transfer on an isothermal vertical surface. J.
76 Heat Transfer 107:764–71
Barthau G. 1992. Active nucleation site den- Bui TD, Dhir VK. 1985b. Transition boiling
sity and pool boiling heat transfer—an ex- heat transfer on a vertical surface. J. Heat
perimental study. Int. J. Heat Mass Transfer Transfer 107:756–63
33:271–78 Buyevich YA, Webber BW. 1996. Towards
Berenson PJ. 1961. Film boiling heat transfer a new theory of nucleate pool boiling.
form a horizontal surface. J. Heat Transfer Presented at Eur. Thermal-Sciences Conf.,
83:351–58 Rome, Italy
Berenson PJ. 1962. Experiments on pool boil- Carvalho RDM, Bergles AE. 1992. The effects
ing heat transfer. Int. J. Heat Mass Transfer of heater thermal conductance/capacitance
5:985–99 on the pool boiling critical heat flux. See Dhir
Bergles AE, Rohsenow WM. 1964. The deter- & Bergles, pp. 219–24
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 399

Celata GP, Cumo M, Mariani A. 1994. Assess- cleate pool boiling on a horizontal surface.
ment of correlations and models for the pre- ASME J. Heat Transfer 87:17–29
diction of CHF in water subcooled flow boil- Gaertner RF, Westwater JW. 1960. Population
ing. Int. J. Heat Mass Transfer 33:237–55 of active sites in nucleate boiling heat trans-
Chen JC. 1963. A correlation for boiling heat fer. Chem. Engr. Prog. Symp. Ser. 56:39–48
transfer to saturated fluids in convective flow. Golobic I, Bergles AE. 1992. Effects of thermal
ASME Pap. 63-HT-34 properties and thickness of horizontal verti-
Chen JC, Hsu KK. 1995. Heat transfer dur- cally oriented ribbon heaters on the pool boil-
ing liquid contact on superheated surfaces. ing critical heat flux. See Dhir & Bergles, pp.
J. Heat Transfer 117:693–97 213–18
Cole R, Rohsenow W. 1969. Correlations of Gorenflow D, Knabe V, Bieling V. 1986. Bubble
bubble departure diameters for boiling of sat- density on surfaces with nucleate boiling—its
urated liquids. Chem. Eng. Prog. 65:211– influence on heat transfer and burnout heat
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

13 flux at elevated saturation processes. Proc.


Cooper MG. 1984a. Heat flow rates in saturated Int. Heat Transfer Conf., 8th, San Francisco
nucleate pool boiling—a wide-ranging exam- 4:1995–2000
ination using reduced properties. Adv. Heat Groenveld DC, Leung LKH. 1989. Tabular ap-
Transfer 16:155–239 proach for predicting critical heat flux and
by University of Nottingham on 08/04/10. For personal use only.

Cooper MG. 1984b. Saturation nucleate post dryout heat transfer. Proc. Int. Topical
boiling—a simple correlation. IChemE Meet. Nuclear Reactor Thermal-Hydraulics,
Symp. Ser. 86:786–93 4th, Karlsruhe. 1:109–114
Cooper MG, Judd AM, Pike RA. 1978. Shape Hahne E, Diesselhorst T. 1978. Hydrodynamic
and departure of single bubbles growing at and surface effects on the peak heat flux in
a wall. Proc. Int. Heat Transfer Conf., 6th, pool boiling. Proc. Int. Heat Transfer Conf.,
Toronto, 1:115–20 6th, Toronto, 1:209–19
Cooper MG, Lloyd AJP. 1969. The microlayer Haramura Y. 1991. Steady state pool transi-
in nucleate pool boiling. Int. J. Heat Mass tion boiling heat with the condensing steam.
Transfer 12:895–913 Proc. ASME/JSME Thermal Eng. Joint Conf.,
Cornwell K, Brown RD. 1978. Boiling surface Reno, NV. 2:59–64
topography. Proc. Int. Heat Transfer Conf., Haramura Y, Katto YA. 1973. A new hydrody-
6th, Toronto, 1:157–161 namic model of critical heat flux applicable
Costello CP, Frea WJ. 1965. A salient non- widely to both pool and forced convection
hydrodynamic effect in pool boiling burnout boiling on submerged bodies in saturated liq-
of small semi-cylinder heaters. Chem. Eng. uids. Int. J. Heat Mass Transfer 26:389–99
Prog. Symp. Ser. 61:258–68 Hosler ER, Westwater JW. 1962. Film boiling
Dhir VK. 1991. Nucleate and transition boil- on a horizontal plate. ARS J. 32:553–58
ing heat transfer under pool and external flow Houchin WR, Lienhard JH. 1966. Boiling
conditions. Int. J. Heat Fluid Flow 12(4): burnout in low thermal capacity heaters.
290–314 ASME Pap. 66-WA/HT-40
Dhir VK, Bergles AE, eds. 1992. Proc. Eng. Hsu YY. 1962. On the size range of active nu-
Found. Conf. Pool Extern. Flow Boil. ASME cleation sites on a heating surface. J. Heat
Pub. Transfer 84:207–16
Dhir VK, Liaw SP. 1989. Framework for a uni- Hsu YY, Graham RW. 1976. Transport Process-
fied model for nucleate and transition pool ing in Boiling and Two Phase Systems. Wash-
boiling. J. Heat Transfer 111:739–45 ington, DC: Hemisphere
Elkassabgi Y, Lienhard JH. 1988. The peak pool Jensen MK. 1988. Boiling on the shellside
boiling heat flux from horizontal cylinders in of horizontal tube boilers. In Two Phase
subcooled liquids. J. Heat Transfer 110:479– Heat Exchangers, ed. S Kakac, pp. 707–46.
86 Boston: Kluwer
Forest TW. 1982. The stability of gaseous nu- Jensen MK, Hsu JT. 1988. A parametric study
clei at liquid-solid interfaces. J. Appl. Phys. of boiling heat transfer in a horizontal tube
53:6191–201 bundle. J. Heat Transfer 110:976–81
Forster DE, Greif R. 1959. Heat transfer to a Judd RL, Chopra A. 1993. Interaction of the
boiling liquid—mechanism and correlation. nucleation process occurring at adjacent nu-
J. Heat Transfer 81:43–53 cleation sites. J. Heat Transfer 115:955–62
Fritz W. 1935. Maximum volume of vapor bub- Judd RL, Hwang KS. 1976. A comprehen-
bles. Physik Zeitschr. 36:379–84 sive model for nucleate boiling heat trans-
Fujita Y. 1992. The state of the art—nucleate fer including microlayer evaporation. J. Heat
boiling mechanism. See Dhir & Bergles, pp. Transfer 98:623–29
83–98 Katto Y. 1994. Critical heat flux. Int. J. Multi-
Gaertner RF. 1965. Photographic study of nu- phase Flow 20:53–90
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

400 DHIR

Kenning DBR. 1977. Pool boiling. In Two by the quenching method. Proc. Int. Heat
Phase Flow and Heat Transfer, ed. D. Butter- Transfer Conf., 7th, Munich, 4:155–60
worth, GF Hewitt, pp. 128–52. Oxford: Ox- Malenkov IG. 1971. Detachment frequency as a
ford Univ. Press function of size of vapor bubbles (translated).
Kenning DBR. 1989. Wall temperatures in nu- Inzh. Fiz. Zhur. 20:99
cleate boiling. Proc. Eurotherm. Semin. Adv. Maracy M, Winterton RHS. 1988. Hysteresis
Pool Boiling Heat Transfer, 8th, Paderborn, and contact angle effects in transition pool
Germany, pp. 1–9 boiling of water. Int. J. Heat Mass Transfer
Klimenko VV. 1981. Film boiling on a horizon- 31:1443–49
tal plate—new correlation. Int. J. Heat Mass Merte H. 1988. Nucleate pool boiling: high
Transfer 24:69–79 gravity to reduced gravity; liquid metals
Kocamustafaogullari G, Ishii M. 1983. Interfa- to cryogenics. Trans. Symp. Space Nuclear
cial area and nucleation site density in boiling Power Systems, 5th, Albuquerque, NM, pp.
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

systems. Int. J. Heat Mass Transfer 26:1377– 437–42


87 Merte H. 1994. Pool and flow boiling in vari-
Kutateladze SS. 1948. On the transition to able and microgravity. Presented at Micro-
film boiling under natural convection. Kotlo- grav. Fluid Phys. Conf., 2nd, Pap. 33, Cleve-
turbostoenie 3:10 land, OH
by University of Nottingham on 08/04/10. For personal use only.

Lahey RT Jr. 1996. A CFD analysis of multidi- Mikic BB, Rohsenow WM. 1969. A new corre-
mensional two phase flow and heat transfer lation of pool boiling data including the ef-
phenomena. In Process Enhanced and Mul- fect of heating surface characteristics. J. Heat
tiphase Heat Transfer—A Festschrift for AE Transfer 9:245–50
Bergles, pp. 431–42. New York: Begell Mikic BB, Rohsenow WM, Griffith P. 1970. On
Lay JH, Dhir VK. 1994. A nearly theoretical bubble growth rates. Int. J. Heat Mass Trans-
model for fully developed nucleate boiling fer 13:647–66
of saturated liquids. Proc. Int. Heat Transfer Mizukami K. 1977. Entrapment of vapor in re-
Conf., 10th, Brighton, England. 5:105–10 entrant cavities. Lett. Heat Mass Transfer.
Lay JH, Dhir VK. 1995a. Shape of a vapor stem 2:279–84
during nucleate boiling of saturated liquids. Monde M. 1987. Critical heat flux in satu-
J. Heat Transfer 117:394–401 rated forced convection boiling on a heated
Lay JH, Dhir VK. 1995b. Nucleate boiling heat disk with an impinging jet. J. Heat Transfer
flux enhancement on macro/micro structured 109:991–96
surfaces cooled by an impinging jet. J. En- Moore FD, Mesler RB. 1961. The measure-
hanced Heat Transfer 2:177–88 ment of rapid surface temperature fluctua-
Lee RC, Nydahl JE. 1989. Numerical calcula- tions during nucleate boiling of water. AIChE
tion of bubble growth in nucleate boiling from J. 7:620–24
inception through departure. J. Heat Transfer Nishikawa K, Fujita Y, Ohta H. 1974. Effect of
111:474–79 surface configuration on nucleate boiling heat
Liaw SP, Dhir VK. 1986. Effect of surface transfer. Int. J. Heat Mass Transfer 27:1559–
wettability on transition boiling heat trans- 71
fer from a vertical surface. Proc. Int. Heat Nishio S. 1985. Stability of pre-existing vapor
Transfer Conf., 8th,. San Francisco, 4:2031– nucleus in uniform temperature field. Trans.
36 JSME (Ser. B) 54-503:1802–7
Liaw SP, Dhir VK. 1989. Void fraction measure- Pasamehmetoglu KO, Nelson RA, Gunnersonn
ments during saturated pool boiling of water F. 1987. A theoretical prediction of critical
on partially wetted vertical surface. J. Heat heat flux in saturated pool boiling during
Transfer 111:731–38 power transients. ASME HTD 77:57–64
Lienhard JH. 1988. Burnout on cylinders. J. Peyayopanakul W, Westwater JW. 1978. Evalu-
Heat Transfer 110:1271–86 ation of the unsteady state quenching method
Lienhard JH, Dhir VK. 1973. Extended hydro- for determining boiling curves. Int. J. Heat
dynamic theory of the peak and minimum Mass Transfer 21:1437–45
pool boiling heat fluxes. NASA CR–2270 Plesset MS, Prosperetti A. 1977. Flow of vapor
Lienhard JH, Eichhorn R. 1979. On predicting in liquid enclosure. J. Fluid Mech. 78(3):433–
boiling burnout for heaters cooled by liquid 44
jets. Int. J. Heat Mass Transfer 22:774–76 Plesset MS, Zwick SA. 1954. Growth of vapor
Lienhard JH, Wong PTY. 1964. The dominant bubbles in superheated liquids. J. Appl. Phys.
unstable wavelength and minimum heat flux 25:493–500
during film boiling on a horizontal cylinder. Rajab I, Winterton RHS. 1990. The two transi-
J. Heat Transfer 86:220–26 tion boiling curves and solid-liquid contact on
Lin DYT, Westwater JW. 1982. Effect of metal a horizontal surface. Int. J. Heat Fluid Flow
thermal properties on boiling curves obtained 11:144–53
P1: ARS/ary P2: ARK/plb QC: MBL/bsa T1: MBL
November 13, 1997 13:18 Annual Reviews AR049-13

BOILING HEAT TRANSFER 401

Ramilison JM, Lienhard JH. 1987. Transition stability with application to film boiling and
boiling heat transfer and the film transition wetting. Proc. Japan-U.S. Semin. Two Phase
regime. J. Heat Transfer 109:746–52 Flow Dynamics, pp. 301–9
Rohsenow WM. 1952. A method of correlat- Stephan K, Abdelsalem M. 1980. Heat trans-
ing heat transfer data for surface boiling of fer correlations for natural convection boil-
liquids. Trans. ASME 74:969–76 ing. Int. J. Heat Mass Transfer 23:78–87
Rohsenow WM. 1971. Boiling. Annu. Rev. Sultan M, Judd RL. 1983. Interaction of the nu-
Fluid Mech. 3:211–36 cleation phenomena at adjacent sites in nu-
Rohsenow WM, Griffith P. 1956. Correlation of cleate boiling. J. Heat Transfer 105:3–9
maximum heat flux data for boiling of sat- Tachibana F, Akiyama M, Kawamura H. 1967.
urated liquids. Chem. Eng. Prog. Symp. Ser. Non-hydrodynamic aspects of pool boiling
52:47–49 burnout. J. Nucl. Sci. Tech. 4:121–30
Sakurai A, Shiotsu M. 1977a. Transient pool Vijaykumar R, Dhir VK. 1992a. An experimen-
Annu. Rev. Fluid Mech. 1998.30:365-401. Downloaded from arjournals.annualreviews.org

boiling heat transfer. Part 1: incipience tal study of subcooled film boiling on a verti-
boiling superheat. J. Heat Transfer 99:547– cal surface—hydrodynamic aspects. J. Heat
53 Transfer 114:161–68
Sakurai A, Shiotsu M. 1977b. Transient pool Vijaykumar R, Dhir VK. 1992b. An experimen-
boiling heat transfer. Part 2: boiling heat tal study of subcooled film boiling on a verti-
by University of Nottingham on 08/04/10. For personal use only.

transfer and burnout. J. Heat Transfer cal surface—thermal aspects. J. Heat Trans-
99:554–60 fer 114:169–78
Sakurai A, Shiotsu M, Hata K. 1990a. A general Wang CH. 1992. Experimental and analytical
correlation for pool film boiling heat transfer study of the effects of wettability on nucle-
from a horizontal cylinder to subcooled liq- ation site density during pool boiling. PhD
uid. Part 1: a theoretical pool film boiling thesis. Univ. Calif., Los Angeles
heat transfer model including radiation con- Wang CH, Dhir VK. 1993a. On the gas entrap-
tribution and its analytical solution. J. Heat ment and nucleation site density during pool
Transfer 112:430–40 boiling of saturated water. J. Heat Transfer
Sakurai A, Shiotsu M, Hata K. 1990b. A general 115:670–79
correlation for pool film boiling heat trans- Wang CH, Dhir VK. 1993b. Effect of surface
fer from a horizontal cylinder to subcooled wettability on active nucleation site density
liquid. Part 2: experimental data for various during pool boiling of saturated water. J. Heat
liquids and its correlation. J. Heat Transfer Transfer 115:659–69
112:441–50 Ward CA, Forest TW. 1976. On the relation be-
Sakurai A, Shiotsu M. 1992. Pool film boiling tween platelet adhesion and the roughness of
heat transfer and minimum film boiling tem- a synthetic biomaterial. Ann. Biomed. Eng.
perature. See Dhir & Bergles, pp. 277–301 4:184–207
Serizawa A. 1983. Theoretical prediction of Witte LC, Lienhard JH. 1982. On the existence
maximum heat flux in power transients. Int. of two transition boiling curves. Int. J. Heat
J. Heat Mass Transfer 26:921–32 Mass Transfer 25:771–79
Sharan A, Lienhard JH. 1979. On predict- Zell M, Straub J, Vogel B. 1989. Pool boiling
ing burnout in the jet configuration. J. Heat under microgravity. Proc. Eurotherm. Semin.
Transfer 107:398–401 Adv. Pool Boiling Heat Transfer, 8th, Pader-
Shoji M. 1992. Experimental verification of born, Germany, pp. 70–74
macrolayer evaporation model. See Dhir & Zuber N. 1959. Hydrodynamic aspects of boil-
Bergles 1992, pp. 237–42 ing heat transfer. PhD thesis. Univ. Calif., Los
Snyder NR, Edwards DK. 1956. Summary of Angeles (also USAEC Rep. AECU-4439)
conference on bubble dynamics and boiling Zuber N, Tribus M, Westwater JW. 1961. The
heat transfer. Memo 20-137, pp. 14–15. Jet hydrodynamic crisis in pool boiling of satu-
Propulsion Lab., Pasadena, CA rated and subcooled liquids. Proc. Int. Heat
Son G, Dhir VK. 1996. Nonlinear Taylor in- Transfer Conf., 2nd, Denver, Pap. 27

Вам также может понравиться