Heat TransferJapanese Research, 27 (2), 1998

Macrolayer Formation and Mechanisms of Nucleate Boiling,

Critical Heat Flux, and Transition Boiling

Hiroto Sakashita and Toshiaki Kumada

Division of Quantum Energy Engineering, Hokkaido University, Japan

The drying process of a macrolayer on a 15 mm diameter boiling surface was

observed with high speed video in the region of nucleate and of transition boiling close
to the critical heat flux (CHF). It was found that the macrolayer rests beneath a large
vapor mass. It partially dries in nucleate boiling and completely dries in transition
boiling at the detachment of the vapor mass. The macrolayer thickness at CHF and in
transition boiling was determined on the basis of the energy balance relation proposed
by Katto and Yokoya. The macrolayer thickness at low heat flux was obtained by
decreasing CHF with downward-facing heating surfaces and agreed well with the
correlation proposed previously by the present authors. The macrolayer thickness in
transition boiling with a vertical surface also agrees fairly well with the correlation,
when the heat flux at macrolayer formation, given on the nucleate boiling curve, is
extrapolated to surface superheat of transition boiling and when the surface tempera-
ture at macrolayer formation is equal to a time-averaged value. 1998 Scripta
Technica. Heat Trans Jpn Res, 27(2): 155168, 1998

Key words: Nucleate boiling, transition boiling, macrolayer

Introduction

It is known that in fully developed nucleate pool boiling, a thin liquid-rich layer, a so-called
macrolayer, exists beneath a large vapor mass and on the heater surface. Based on the possible role
of this macrolayer, Katto and Yokoya [1] proposed a critical heat flux (CHF) model, in which CHF
occurs when the macrolayer dries just before the departure of a large vapor mass. This model is
expressed by the following energy balance:

qCHF = 1Hfg1f. (1)

Katto and Yokoya [1] applied this model to a transition boiling region to explain the mechanism for
heat transfer in transition boiling. If there is no inflow of bulk liquid onto the heating surface before
detachment of the large vapor mass and the heat flux after surface dryout is not considered, then a
heat balance relation similar to Eq. (1) is obtained:
_
qtr = 1Hfg1f, (2)
CCC0096-0802/98/020155-14
1998 Scripta Technica
155
 _
where qtr is the time-averaged heat flux in the transition boiling region. For a heating surface which
is so small that the vapor mass completely covers the surface 1, appearing in Eqs. (1) and (2), can be
regarded as an average thickness of the liquid that rests on the heating surface on the formation of the
vapor mass.

Other CHF and transition boiling models have been proposed by Dhir and Liew [2] and by
Unal and others [3]. Dhir and Liew postulated stationary vapor stems on a heater surface and attempted
to predict the boiling curves from nucleate to transition boiling by considering the change in diameter
and contact angle of the stem base. Unal and coworkers proposed a model where CHF occurs when
some point on a heating surface reaches a temperature at which the liquid can no longer maintain
contact. Although the macrolayer dryout model of Katto and Yokoya explains CHF of pool boiling
with various heater configurations, the validity of the model based on direct observations of the boiling
surface has not been well established.

In this paper, the drying process of the macrolayer formed on a vertical heater surface was
observed by using high speed video to verify the macrolayer dryout model at CHF and in the transition
boiling region. The macrolayer thickness with low heat fluxes for water and ethanol under atmospheric
pressure was determined based on the energy balance Eq. (1) by changing the orientation of the heating
surface from a vertical to a downward direction. The macrolayer thickness was compared with data
previously reported and with correlations in the nucleate boiling region. The macrolayer thickness
for water and ethanol in transition boiling was also determined based on the energy balance Eq. (2)
under the condition that no liquid inflow onto the heating surface was permitted during the hovering
of the large vapor mass.

Nomenclature

A constant in the nucleate boiling curve

a thermal diffusivity
B constant in the transition boiling curve
Cp specific heat
C right hand side of Eq. (3) except for term (q / vHfg)5 / 6
f detachment frequency of vapor mass
G0 vapor blowing rate, (q / vHfg)
g gravity
Hfg latent heat of evaporation
m superheat exponent in the nucleate boiling curve
p superheat exponent in the transition boiling curve
q heat flux
qCHF
_ critical heat flux
qtr time averaged heat flux in transition boiling
Tsat superheat

Greek Symbols

1 thickness of macrolayer
angle of inclined surface measured from the downward position

156
 thermal conductivity
n dynamic viscosity
density
surface tension

Subscripts

l liquid
v vapor

Observation of Boiling Surface

Experimental procedure

Experiments were carried out using the apparatus shown in Fig. 5. In the early stage of the
experiments we used a clean copper surface as the boiling surface. The drying process of the
macrolayer formed on the clean surface was visible on a monitor screen connected to a high speed
video camera (Photron, max.2066f/s), but the monitor images were unclear when those were
developed by using a video printer. Thus, a boiling surface with calcium carbonate (CaCO3) thinly
deposited to create a contrast between the dry and wet regions on the surface was used. This surface
was prepared as follows: calcium chloride (CaCl2) and sodium bicarbonate (NaHCO3) were dissolved
in distilled water, and carbon dioxide gas was blown into the solution to restrain crystal nucleation in
the solution. The clean copper surface was placed in the solution and heated to a temperature just
below boiling, and maintained at that temperature for several tens of minutes. This procedure deposits
a uniform, thin CaCO3 layer on the clean copper surface.

Result of observation

Figure 1 shows high speed video pictures from nucleate and transition boiling close to CHF.
The time written on each photo is the elapsed time from the departure of the preceding vapor mass.
The white part in each photo is the region where the heating surface is dry and the dark part is the
region where the heating surface is wet. In nucleate boiling, about one-third of the heating surface is
wet at the departure of the vapor mass. In transition boiling, almost all of the surface dries out before
the vapor mass detaches. Furthermore, the periphery of the surface is completely dry and there is no
liquid inflow while the vapor mass hovers. We also observed that: 1) primary bubbles nucleate on the
surface when the macrolayer is thicker, 2) as the macrolayer becomes thinner, nucleation terminates
and the macrolayer evaporates by heat conduction through it, and 3) no vapor stems were observed.
Further observations with larger magnification will be needed to identify the existence of vapor stems.
These observations appear to confirm that the macrolayer dryout model by Katto and Yokoya is valid
as a physical model for CHF and transition boiling.

Macrolayer Thickness at CHF

Previous studies have measured the macrolayer thickness by using a conductance probe
technique. However, it is quite difficult to measure the macrolayer thickness accurately by this
technique, because a thicker macrolayer includes many primary bubbles and its thickness varies
spatially and temporally. On the other hand, the preceding section showed that Eqs. (1) and (2) are

157
 Fig. 1. Drying of macrolayer formed on a CaCO3 deposited heating surface.

close approximations to the physical CHF and transition boiling model. The present authors [4]
determined the macrolayer thickness based on Eq. (1) by measuring CHF and the detachment
frequency, f, of vapor masses for boiling in the case where a large vapor mass completely covers and
departs regularly from the heating surface. Measurements were made on horizontal and vertical, 20
mm diameter, surfaces for water, ethanol, methanol, and acetone in the pressure range from 0.03 MPa
to 0.5 MPa. Based on these data and assuming that the macrolayer is formed by the coalescence of
coalesced bubbles (secondary bubbles) except at low pressures, the following semi-empirical corre-
lation for the macrolayer thickness was proposed [4]:

Fig. 2. Apparatus for measuring macrolayer thicknesses at low heat fluxes.

158
 Fig. 3. Macrolayer thicknesses for water at 0.1 MPa.

n1l / 311 / 24 q
5 / 6
1 = 0.786 H . (3)
1l / 4 {g(l v)}5 / 24
v fg

The macrolayer thickness results, together with Eq. (3), are shown by solid symbols in Fig. 11.

As discussed in Ref. 4, gravity, g, appearing in the denominator of Eq. (3) should be excluded,
because coalesced bubbles forming the macrolayer grow fast and are small enough to be able to neglect

Fig. 4. Macrolayer thicknesses for ethanol at 0.1 MPa.

159
 Fig. 5. Apparatus for observation of boiling surfaces and measurement of macrolayer thicknesses
in transition boiling.

the buoyancy effect. However the present authors have not yet found a model for macrolayer formation
for which it is possible to exclude g from the correlation, and this paper treats g as a dimensional
constant with the value 9.8 m/s2.

Macrolayer Thickness at Low Heat Flux

Experimental procedure

Previous studies have measured the relation between the macrolayer thickness and heat flux
in the nucleate boiling region under atmospheric pressure. For comparison, the macrolayer thickness
was determined for water and ethanol using the apparatus shown in Fig. 2. The heating surface is a
20 mm diameter disk with orientation varied from vertical ( = 90) to a downward direction ( = 0).
With this apparatus the detachment frequency of vapor masses decreased from f = 17 (1iter/s) at
= 90 to f = 2.5 (1iter/s) at = 0, and the CHF at = 0 decreased to about 35% of the CHF at
= 90. The detachment frequency of the vapor masses was recorded on a high speed video camera
for water and ethanol and also with a conductance probe for water.

Previous correlations and measured macrolayer thickness data

Haramura and Katto [5] proposed the following theoretical correlation:

0.4
v v q 2
1 = 0.00536v 1 + . (4)
l l Hfg

Iida and Kobayasi [6] were the first to measure the macrolayer thickness for water by the conductance
probe technique. Bhat and coworkers [7] obtained the following empirical relation for water at
atmospheric pressure based on their measurements with a conductance probe:

160
 1 = 1.59 105 q1.527. (5)

Rajvanshi and others [8] measured the macrolayer thicknesses with a conductance probe for various
liquids at atmospheric pressure and provided a modified form of Eq. (4) by Haramura and Katto

0.4 2
v q .
1 = 0.0107v Hfg (6)
l

Recently Ohta and colleagues [9] measured the macrolayer thickness of water at atmospheric pressure
under microgravity. They provided a 55 mm diameter sapphire glass as the heating surface which was
heated from its back. The macrolayer thickness was determined from the electrical resistance between
very thin electrodes (0.1 m) coated on the glass surface.

Experimental results

Figures 3 and 4 show a comparison of the previously reported data and the present data with
the various correlations. Our data for water and ethanol lie close to Eq. (3). The data from the probe
method show thicker macrolayers than our data, and the previous correlations are steeper for G0 than
Eq. (3). On the other hand, the data of Ohta and coworkers show thinner macrolayers than the previous
data and the correlations and agree quite well with Eq. (3), where g in this equation was 9.8 m/s2 as
mentioned in an earlier section. This may be because the method of Ohta makes it possible to measure
the net amount of water in the macrolayer without effects such as turbulence of the macrolayer surface
and bubble nucleation in the macrolayer. From the these results Eq. (3), which applies at the CHF,
may also be considered to apply in the nucleate boiling region.

Macrolayer Thickness in Transition Boiling

Experimental procedure

Figure 5 shows the experimental apparatus. Measurements were made in the steady-state mode
for nucleate boiling and the transient cooling mode for transition boiling. To cool the heating surface
gradually in the transition boiling experiments, one part of the copper cylinder was machined to form
a cone and the narrow end was used as the heating surface. Three 0.5 mm diameter thermocouples
were embedded 1.5 mm, 10 mm, and 20 mm from the heating surface with their tips soldered at the
bottoms of the holes. For transition boiling on upward-facing heating surfaces, Katto and Yokoya
[10] with a 10 mm diameter heater, and Nishikawa and others [11] with an 8 mm diameter heater,
demonstrated that there was inflow of bulk liquid to the edge of the heating surface while the vapor
mass hovered on the heating_ surface. If liquid flows onto the surface, the macrolayer thickness
determined from Eq. (2), qtr, and f, apparently becomes thicker. To avoid this, we adopted a vertical
surface since gravity does not promote the movement of vapor masses in the direction normal to the
surface. Furthermore, a glass plate was located in front of the heating surface, so that the vapor mass
was pressed against the glass plate and spread to the bakelite plate surrounding the heating surface,
preventing inflow of liquid to the heating surface.

161
 The departure frequency of vapor masses and the motion of vapor and liquid around the
heating surface were measured by two conductance probes, located 2 mm from the center of the
heating surface and on the bakelite plate 0.5 mm from the edge of the heating surface. High speed
video was also used to measure the departure frequency of vapor masses and to observe the boiling
performance.

Boiling curve

Figures 6 and 7 show examples of boiling curves obtained from the present experiment. The
data plot almost linearly on the log-log graph for both nucleate and transition boiling regions and are
well approximated by the following relations:

in the nucleate boiling region: q = AT m

sat,
(7)

_
and in the transition boiling region: qtr = BT psat. (8)

In the range of superheat of the present experiments, the film boiling heat flux predicted by Bromleys
correlation is 0.01 to 0.02 MW/m2 for both water and ethanol, and the film boiling curves are not
shown in Figs. 6 or 7.

Inflow of bulk liquid beneath the vapor mass

To determine the macrolayer thickness based on Eq. (2), it is necessary to make sure that there
is no liquid inflow onto the heating surface while the vapor mass hovers. Figure 8 shows voltage traces
from the conductance probe signals measured during transition boiling. Figure 8(a) is a trace of the
probe on the bakelite plate 0.5 mm from the edge of the heating surface, and Fig. 8(b) is from the
probe 2 mm from the heating surface. The higher voltage shows that the tip of the probe contacts
vapor and the lower voltage shows contact with water. The trace in Fig. 8(b) is periodic, corresponding

Fig. 6. Boiling curves for water.

162
 Fig. 7. Boiling curves for ethanol.

to the growth and departure of vapor masses. The trace in Fig. 8(a) is similar to that in Fig. 8(b), and
it shows that the tip of the probe on the bakelite plate contacts vapor for most of the period during
which the vapor mass covers the heating surface. The periphery of the heating surface can, therefore,
be assumed to be dry until the departure of the vapor mass. This is consistent with the results from
high speed video images shown in Fig. 1, and it may be concluded that there is no liquid inflow to
the heating surface when the vapor mass is hovering on the heating surface in our experiments. The
conductance probe could not be used for ethanol, but the high speed video showed that the periphery
of the heating surface was completely dry when the vapor mass covered the heating surface.

Most previous experiments on transition boiling were conducted using upward-facing flat
surfaces. The measurements were repeated for an upward-facing surface, created by changing the
orientation of the heating surface toward the upward direction and also removing the glass plate. The
voltage traces from the probes are shown in Figs. 8(c) and 8(d). These figures show that the edge of
the upward-facing heating surface is in contact with water during most of the time for which the vapor
mass hovers.

Macrolayer thickness

Equation (3) was derived based on a model in which a macrolayer is formed by the coalescence
of coalesced bubbles. In transition boiling, the surface superheat and hence the number of primary
bubbles are larger than is the case for nucleate boiling and at the CHF. Therefore, similar to nucleate
boiling and at the CHF point, the macrolayer in the transition boiling is formed by the coalescence of
coalesced bubbles, and the thickness of the macrolayer need not be dependent on the surface superheat
but does depend on the vapor blowing rate G0. However, differing from nucleate boiling and at the
CHF point, the transition boiling heat flux and the superheat calculated using temperatures recorded
from the thermocouples are not the instantaneous values at macrolayer formation but are time
averaged values.

163
 Fig. 8. Voltage traces of probes for vertical and upward-facing surfaces.

Surface superheat at macrolayer formation. There have been many investigations into
fluctuations in surface temperatures in the transition boiling regions, but it is quite difficult to obtain
relations between instantaneous temperatures at macrolayer formation and time-averaged tempera-
tures. The discussion in this paper is based on a numerical analysis of surface temperature fluctuations.
The analysis modeled a 20 mm thick heated copper plate with its lower side kept at a constant
temperature. At the upper side, it was assumed that the surface temperature drops to the liquidsolid
contact temperature just as the liquid contacts the dry surface, and that the liquid macrolayer forms
instantaneously at the contact temperature. It was also assumed that Eq. (3) applies and that the heat
flux in this equation could be obtained from the nucleate boiling curve extrapolated to the transition
boiling region. Furthermore, in the analysis the macrolayer was considered to evaporate by the greater
of the following two heat fluxes: the heat flux given on the nucleate boiling curve extrapolated to the

164
instantaneous superheat in the transition boiling region, or the heat flux due to heat conduction through
the macrolayer. The heating surface after drying was considered to be adiabatic. The calculation was
started from a uniform temperature distribution and terminated when the change in surface tempera-
ture with time became periodic. Figure 9 shows the results calculated from the nucleate boiling curve
in Fig. 6 and a departure frequency of the vapor mass, f = 20. The maximum amplitude of the
temperature fluctuations is large and increases with increasing surface superheat, but the difference
between the time-averaged temperature and the temperature at macrolayer formation reached a
maximum of 2 K near the CHF and decreased with increasing surface superheat. For ethanol, the
maximum amplitude and the difference were about one-third of the corresponding values for water.
From these results we conclude that the time averaged surface superheat can be used as the
instantaneous temperature at macrolayer formation.

Relation between nucleate boiling and transition boiling. Figure 10 shows the relation
between, m, the Tsat exponent for nucleate boiling curves, and, p, the Tsat exponent for transition
boiling curves measured in the present experiments. Both exponents are proportional, and this implies
that the transition boiling curve is strongly dependent on the nucleate boiling curve. The occurrence
of nucleate boiling when liquid contacts the dry surface was verified by high speed video in our
experiment. At subcooled boiling and forced convection boiling, the nucleate boiling curve extends
beyond the CHF point for saturated pool boiling. The present study, therefore, extrapolated the
measured nucleate boiling curve to the surface superheat of transition boiling to obtain the heat flux
at macrolayer formation, as shown by the solid lines in Figs. 6 and 7.

If the macrolayer thickness in transition boiling is expressed by Eq. (3), the surface superheat
at macrolayer formation is equal to the time-averaged surface superheat, and also the heat flux at

Fig. 9. Calculated results of surface temperature fluctuations for water.

165
macrolayer formation is obtained by extrapolating the nucleate boiling curve, then the relation
between the heat flux and surface superheat in transition boiling is

1 / 24
_ n1811
qtr = l5v / 6H11
fg
/6
0.786 5 5 6
fA5 / 6T5m
sat
/6
. (9)
g (l v) l

Since the departure frequency of vapor masses in the present experiments was constant, independent
of heat flux, the exponents of nucleate boiling and transition boiling are related by

p = (5/6)m. (10)

This relation is shown as a solid line in Fig. 10, and agrees well with the data.

Arrangement of macrolayer thickness. According to the above discussion, the macrolayer

thickness was compared with Eq. (3) by equating the surface superheat at macrolayer formation to
the measured surface superheat and by determining the vapor blowing rate from the extrapolated
nucleate boiling curve. The results are shown in Fig. 11, and the solid line in the figure is from Eq.
(3). Figure 11 also shows the data previously determined by the present authors at CHF [4] and the
data of Shoji and coworkers [12] measured using a conductance probe in transition boiling. The data
in transition boiling scatter somewhat, but agree fairly well with Eq. (3). Comparison of the data with
Eq. (9) is shown as the broken lines in Figs. 6 and 7.

Macrolayer thickness for an upward-facing surface

Figure 12 shows the macrolayer thickness for upward-facing heating surfaces, obtained by
turning the heating surface in Fig. 5 in an upward direction and removing the glass plate. The
macrolayer is apparently thicker than Eq. (3) suggests. This result and the results in Fig. 8 indicate

Fig. 10. Comparison of exponents of Tsat of nucleate and transition boiling curves.

166
 Fig. 11. Macrolayer thickness at CHF and in transition boiling.

the main reason why the transition boiling curves previously reported for upward-facing surfaces are
very different in different investigations: the inflow of bulk liquid beneath the vapor mass when it is
hovering on the surface. This complicates the prediction of transition boiling curves because the
inflow changes due to the orientation and conditions at the heating surface, the configuration around
the heating surface, and so on. Further study is required to resolve this problem.

Conclusions

(1) A liquid macrolayer exists on the heating surface from nucleate boiling to transition boiling
close to the CHF, and this macrolayer completely dries out before the detachment of large vapor

Fig. 12. Macrolayer thickness of upward-facing surfaces in transition boiling.

167
masses in transition boiling. The macrolayer dryout model proposed by Katto and Yokoya is
approximately valid as a physical model for CHF and transition boiling.

(2) The macrolayer thickness at low heat flux, obtained by changing the orientation of the
heating surface, was represented well by Eq. (3) proposed by the present authors. Previous data for
macrolayer thicknesses measured by the conductance probe technique showed them to be thicker and
more strongly dependent on heat flux when compared with the present data.

(3) The macrolayer thickness during transition boiling was determined under the condition
that there is no inflow of bulk liquid onto the heating surface. The macrolayer thickness for transition
boiling is represented fairly well by Eq. (3), when the heat flux at macrolayer formation is taken from
the nucleate boiling curve extrapolated to the transition boiling region and the surface superheat at
macrolayer formation is equal to time-averaged surface superheat.

(4) The inflow of bulk liquid onto the heating surface is the primary reason why the transition
boiling curves on upward-facing surfaces are different in different investigations.

Literature Cited

1. Katto, Y., and S. Yokoya. 1968. Int. J. Heat Mass Transfer, 11, 993.
2. Dhir, V.K., and S.P. Liaw. 1989. J. Heat Transfer, 111, 739.
3. Unal, C., V. Daw, and R.A. Nelson. 1992. J. Heat Transfer, 114, 972.
4. Kumada, T., and H. Sakashita. 1995. Int. J. Heat Mass Transfer, 38, 979.
5. Haramura, Y., and Y. Katto. 1983. Int. J. Heat Mass Transfer, 26, 389.
6. Iida, Y., and K. Kobayasi. 1970. Proc. Int. Heat Transfer Conf., 5 (B1.3).
7. Bhat, A.M., R. Prakash, and J.S. Saini. 1986. Int. J. Heat Mass Transfer, 26, 735.
8. Rajvanshi, A.K., J.S. Saini, and R. Prakash. 1992. Int. J. Heat Mass Transfer, 35, 343.
9. Ohta, H., et al. 1995. Proc. 32nd National Heat Transfer Symp. of Japan, 2, 547. (in Japanese)
10. Katto, Y., and S. Yokoya. 1971. Trans. JSME, 37 (295), 535. (in Japanese)
11. Nishikawa, K., T. Fujii, and H. Honda. 1971. Trans. JSME, 37 (297), 1018. (in Japanese)
12. Shoji, M. 1992. Proc. 29th National Heat Transfer Symp. of Japan, 301. (in Japanese)

"F F F"

Originally published in Proceedings of the 3rd KSME-JSME Thermal Engineering Conference, 1,

421426, Oct. 2023, 1996, Kyongju, Korea

168