International Journal of Heat and Mass Transfer 118 (2018) 327–339

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

journal homepage: www.elsevier.com/locate/ijhmt

Experimental investigation of variability in bubble departure

characteristics between nucleation sites in subcooled boiling flow
Zhiee Jhia Ooi, Vineet Kumar, Joseph L. Bottini, Caleb S. Brooks ⇑
Department of Nuclear, Plasma, and Radiological Engineering, University of Illinois, Urbana, IL 61801, USA

a r t i c l e i n f o a b s t r a c t

Article history: Experiments have been conducted in a vertical square channel for investigation of wall nucleation char-
Received 9 June 2017 acteristics in subcooled boiling flow. Bubble departures from multiple sites are measured simultaneously
Received in revised form 6 September 2017 for nine conditions. Existing bubble departure diameter models are benchmarked and are shown to be
Accepted 28 October 2017
satisfactory in predicting condition-average bubble departure diameter. However, significant variations
Available online 7 November 2017
are observed in the bubble departure frequency across different nucleation sites of a given condition lar-
gely due to intermittent periods of inactivity. A benchmark of the existing bubble departure frequency
Keywords:
models shows that the models are generally applicable to an ‘active’ frequency but cannot account for
Subcooled boiling
Departure diameter
the impact of these periods of inactivity. This finding highlights an important issue in the current mod-
Departure frequency eling and understanding of the gas-phase boundary condition. This periodic inactivity, if not incorporated
Wall nucleation into the wall nucleation modeling, will result in a large overprediction of bubbles generated at the wall. A
Active nucleation site density physical justification for this inactivity is discussed based on the modeling of the active nucleation site
density.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction ing flows present additional complexity from exposure of the

The process of wall nucleation in subcooled boiling flow has
generated significant interest among the scientific community
given its importance in a wide range of industries. A substantial
amount of effort has been invested to characterize wall nucleation
in terms of active nucleation site density, bubble departure diam-
eter, and bubble departure frequency. Active nucleation site den-
sity represents the number of bubble-producing sites; bubble
departure diameter describes the size of the average bubble as it
departs from a nucleation site, and bubble departure frequency
defines the average rate of bubbles leaving a nucleation site. Accu-
rate prediction of these parameters of wall nucleation is founda-
tional to understanding boiling heat transfer and transport as
emphasized by the wall heat flux partitioning model [1], the bub-
ble number transport model [2], the Interfacial Area Transport
Equation (IATE) [3,4], and the Multiple Size Group (MUSIG) Popu-
lation Balance Model [5,6]. However, a good understanding of wall
nucleation is still lacking in forced convective flows due to inherent
fluctuations in driving mechanisms [7]. Although wall nucleation
in pool systems can provide a foundational understanding of the
bubble incipience, growth, and departure, forced convective boil-
⇑ Corresponding author at: Department of Nuclear, Plasma, and Radiological
Engineering, University of Illinois, 104 South Wright Street, Urbana, IL 61801, USA.
E-mail address: csbrooks@illinois.edu (C.S. Brooks).
nucleation site and departing bubble to steep temperature gradi-
ents [8] and mechanical and thermal fluctuations [9]. Furthermore,
the imaging technology required for detailed measurement of bub-
ble nucleation characteristics in flow boiling systems is relatively
new, and more comprehensive experimental data is needed to
improve understanding and modeling efforts.
Measurements of bubble size and growth in flow boiling date
back to the 1950s. Gunther [10] carried out a photographic study
to investigate the impact of forced convection on the heat transfer
mechanism of a boiling surface. The size and population of bubbles
were measured in this study. The photographs of the boiling sur-
face were obtained using a high-speed film camera that operated
mechanically. A roll of film was accelerated to reach the desired
frame rate of 20,000 frames per second (FPS), and an electro-
optical shutter was used to expose the film in the camera. Since
the shutter could only be exposed for a short time, roughly 0.05 s
of film was recorded for each condition. A photographic study
was performed by Griffith et al. [11] to study void volumes in sub-
cooled boiling. For each test condition, ten photographs of boiling
phenomena of a heated surface were taken manually by an opera-
tor using sheet films as the negatives. The films were then devel-
oped into photographs using chemical solutions. According to
Griffith et al. [11], one of the challenges of this technique was
ensuring uniformity between films during development.

https://doi.org/10.1016/j.ijheatmasstransfer.2017.10.116
0017-9310/Ó 2017 Elsevier Ltd. All rights reserved.
328 Z.J. Ooi et al. / International Journal of Heat and Mass Transfer 118 (2018) 327–339

Nomenclature

Bo boiling number [–] tG growth time [s]

C non-dimensional constant [–] tW wait time [s]
CD drag coefficient [–] v liquid velocity [m/s]
CDd departure diameter constant [–]
Cfd departure frequency constant [–] Greek symbols
Db instantaneous bubble diameter [m] a thermal diffusivity [m2/s]
Dd departure diameter [m] q density [kg/m3]
D+ dimensionless diameter [–] r surface tension [N/m]
Dp dry patch diameter [m] /s static contact angle [rad]
fa active departure frequency [Hz]
fd classical departure frequency [Hz]
Superscripts
g gravitational constant [m/s2] ⁄ non-dimensional
G mass flux [kg/m2 s] N number mean
hc heat transfer coefficient [W/m2 K] S surface mean
hfg latent heat of vaporization [J/kg]
V volume mean
Ja Jakob number [–]
La Laplace length [m]
m number of active nucleation sites [–] Subscripts
Nn active nucleation site density [1/m2] a active
NT non-dimensional temperature [–] b bulk
c condition
n number of departures [–]
P pressure [Pa] cal calculated
Pr Prandtl number [–] exp experimental
q00w wall heat flux [W/m2] f liquid
g gas
rs characteristic radius of active cavity [m]
T temperature [°C] ONB onset of nucleate boiling
DTw wall superheat [°C] s site
sat saturated
t time [s]
Dt mean time difference between departures [s] sub subcooled
t 0a accumulated active time [s]

Nearly two decades later, Treshchev [12] studied the formation
of vapor on a heated surface during boiling. The influence of exper-
imental conditions such as pressure, flow rate, and heat flux on
bubble diameters and active nucleation site density was the focus
of the study where the author reported that the bubble diameter
decreased with the system pressure. Abdelmessih et al. [13] used
the same method to investigate the effect of fluid velocity on the
growth and collapse of bubbles on a heated surface where the size
of ten bubbles from each nucleation site was measured. The study
reported that bubbles nucleating from the same active nucleation
site under the same condition had varying maximum sizes and
lifespans. Abdelmessih et al. [13] attributed the variation to the
microscopic eddies in the liquid. Akiyama and Tachibana [14] used
a photographic technique to investigate the growth, collapse, and
motion of bubbles on a heated surface in subcooled boiling under
atmospheric pressure. Photographs of bubble formation and col-
lapse were taken using a high-speed camera at 14,000 FPS. Similar
to the results by Abdelmessih et al. [13], Akiyama and Tachibana
[14] observed bubbles with varying sizes and lifetimes even
though the bubbles nucleated from the same site under the same
condition.
With improving imaging capability came more flow boiling
nucleation data and attempts to present the statistical distribution
of bubble departure sizes. Unal [15] studied subcooled boiling flow
to investigate the size and growth of bubbles on the heated surface.
Bubble sizes were measured by recording bubble formation at a
frame rate of 5000 FPS and enlarging the developed films. Unal
[15] noted that the maximum bubble diameter followed an
approximate normal distribution from samples of 65–450 bubbles
and reported an average for seven conditions. Klausner et al. [16]
carried out flow boiling experiments with Refrigerant-113 to
measure bubble departure diameter focusing on the size distribu-
tions. Klausner et al. [16] observed that the departure diameter
resembled a normal distribution. Furthermore, it was observed
that even though the mean departure diameter varied with the
flow condition, the standard deviation of departure diameter from
the mean was similar to the mean value for different cases which
demonstrated the random nature of the bubble departure process.
Klausner et al. [16] suggested that the randomness exhibited by
the experimental data most likely originated from both the turbu-
lent fluctuations inherent in two-phase flow and the variations of
wall superheat on the heated surface, also later cited by
Martinez-Cuenca et al. [7] and Dhir [17]. The authors then
extended their analysis of the bubble size distribution in Klausner
et al. [18] by modeling the distribution based on an assumed nor-
mal distribution for the wall superheat and the liquid velocity. The
good agreement between the predicted distribution and existing
data suggested that the normal distribution of bubble departure
diameter was caused by the variations of liquid velocity and wall
superheat in a boiling system. Thorncroft et al. [19] investigated
bubble detachment in upward and downward forced convection
subcooled boiling flow with refrigerant FC-87, reported bubble
distributions based on approximately fifty measured bubble depar-
tures, and observed a distribution resembling a Gaussian although
slightly skewed in some cases. Thorncroft et al. [19] also measured
the nucleation site wait time, which is the time from a bubble
departure until the next bubble is initiated, and suggested evi-
dence of a correlation between the bubble departure size and the
subsequent wait time. Zou and Jones [20] investigated the effects
of heater surface material on subcooled flow boiling heat transfer
of R134a using high-speed photography. Using a copper surface
and a stainless steel surface, it was observed that the bubble
 Z.J. Ooi et al. / International Journal of Heat and Mass Transfer 118 (2018) 327–339 329

departure diameters and frequencies of both surfaces did not exhi-
bit a significant difference. Similar to the results of Treshchev [12],
Zou and Jones [20] observed that bubble departure diameters
decreased as system pressure increased. Brooks et al. [21] pre-
sented data based on the measurement of one hundred consecu-
tive departures for ninety-two conditions (as tabulated in [9,21]),
providing a parametric study of the effect of mass flux, pressure,
heat flux, and bulk liquid subcooling on bubble departure diameter
and frequency. Martinez-Cuenca et al. [7] and Brooks [22] analyzed
the bubble departure diameter distribution in terms of its impact
on calculating bubble surface area and volume, showing that for
the dataset of Brooks et al. [21] a normal distribution assumption
leads to significant averaging covariance. By normalizing the
departure diameter based on the mean, Martinez-Cuenca et al.
[7] showed that a generalized log-normal distribution could cap-
ture the bubble departure distribution for the forty low-pressure
conditions reported by Brooks et al. [21].
Martinez-Cuenca et al. [7] also highlighted the complexity in the
distribution of time between departures and suggested that this dis-
tribution is impacted by three phenomena: coalescence with bub-
bles generated upstream, evaporation of bubbles passing the
nucleation site suppressing bubble formation by reducing the wall
superheat, and a low frequency dormancy of the nucleation site.
Basu et al. [23,24] partitioned the time between departures into wait
time and growth time for calculating the departure frequency. One
active nucleation site per condition was studied where sites with
at least three cycles of nucleation and departure were the only ones
considered. The bubble growth time and wait time were averaged
from 1220 FPS recordings of these cycles, and the departure fre-
quency was determined to be the inverse of the sum of the average
wait and growth times. It is inferred that the effects highlighted by
Martinez-Cuenca [7] were not considered here considering the
three-cycle criterion for this study. Podowski [25] also studied the
bubble wait time, growth time, and frequency in order to benchmark
an analytical model of boiling heat transfer. No details of the exper-
iment were provided, but the frequency was shown to be sensitive to
liquid subcooling, heat flux, and cavity radius.
Many other experimental works can be found with bubble
departure diameter and departure frequency based on measure-
ments taken from a single nucleation site per condition. Bibeau
and Salcudean [26] measured bubble formation with a high-
speed camera at 5000 or 6000 FPS, depending on the flow condi-
tions. In order to eliminate bubble size variation due to the differ-
ence in cavity sizes, the authors only measured bubbles produced
from the same active nucleation site for all conditions. Hong et al.
[27] measured bubble departure size in forced convective sub-
cooled boiling flow under static and heaving conditions photo-
graphically using a high-speed camera at 2000 FPS where only
bubbles departed from the most active site were measured. Depar-
ture diameters measured under each static condition were aver-
aged while the transient values of bubble diameter were
presented for the heaving conditions to reflect the influence of
heaving on bubble diameters. Lin and Chen [28] investigated bub-
ble behaviors photographically under subcooled boiling flow using
R134a with frame rates ranging from 1000 to 8000 FPS. The bubble
departure diameters were reported to exhibit a normal distribution
as presented by the previous works [15,16,18,19,29].
On the other hand, Del Valle and Kenning [29] photographed
bubble formation in water at one atmosphere on a heated surface
at 10,000 FPS to study the heat transfer coefficients and the distri-
butions of bubble size, frequency, and active nucleation sites. The
bubble formation from thirty randomly selected sites were mea-
sured for each condition. The exact number of bubbles measured
per site was not specified. Based on the authors’ statement that
only forty consecutive frames were analyzed for each condition
and that bubbles appeared every three to five frames, it is esti-
mated that fewer than fifteen bubbles were measured for each site.
The study determined that the maximum bubble diameters were
normally distributed while the total bubble periods did not exhibit
a recognizable distribution. Yang et al. [30] carried out flow boiling
experiments under atmospheric pressure for narrow flow channels
where departure diameter and departure frequency from multiple
nucleation sites were studied for every condition. Bubble departure
diameter was measured by tracking the bubble growth from
twenty active nucleation sites in the viewing area using a high-
speed camera at a frame rate of 2000 FPS. Ten departures were
measured from each site and the authors concluded a normal dis-
tribution of departure sizes for three conditions. Furthermore,
Yang et al. [30] propose that the departure frequency of the
observed sites can be correlated as proportional to the wall
superheat to the third power. A summary of the experimental
investigations reviewed in this work is provided in Table 1.

Table 1
Summary of experimental data for wall nucleation departure characteristics in flow boiling.

Study Fluid Geometry Frame rate [FPS] No. of conditions [–] No. of sites per No. departure bubbles
condition [–] measured per site [–]
Gunther [10] Distilled Water Rectangular 20,000 38 1 NAa
Treschev [12] Distilled Water Rectangular NA 57 1 NAa
Abdelmessih et al. [13] Distilled Water Circular 7000 34 1b 10
Akiyama & Tachibana [14] Deionized Water Annular 14,000 33 1 NA
Unal [15] Distilled Water Annular 5000 7 NA NA
Klausner et al. [16] Refrigerant-113 Square NA 35 NA >3c
Thorncroft et al. [19] Refrigerant FC-87 Square 2000 20 NA 50d
Zou & Jones [20] Refrigerant-134 Square 2000–4000 48 1 NA
Brooks et al. [9,21] Deionized Water Annular 10,000 92 1 100
Basu et al. [23,24] Distilled Water Rectangular & Rod Bundle 1220 16 1 >3c
Bibeau & Salcudean [26] Distilled Water Annular 5000–6000 24 1 5–10
Hong et al. [27] Deionized Water Rectangular 2000 47 1 NA
Lin & Chen [28] Refrigerant-134 Rectangular 1000–8000 1 10 NAe
Del Valle & Kenning [29] Distilled Water Rectangular 10,000 4 30 <15f
Yang et al. [30] Deionized Water Rectangular 2000 32 20 10

NA: Not Available.

a
Average bubble departure size not measured.
b
Prefabricated artificial site.
c
Considered nucleation sites with at least three nucleations.
d
Studied vertical pool, upflow and downflow conditions.
e
For one condition on smooth surface, 240 total bubbles measured across 10 nucleation sites.
f
Number estimated based on reported departure frequency and frames captured.
330 Z.J. Ooi et al. / International Journal of Heat and Mass Transfer 118 (2018) 327–339
Fig. 1. Layout of test facility.
Many previous studies focus on one nucleation site per condi-
tion assuming it is sufficient to provide an accurate average of
departure characteristics across nucleation sites. Even though a
tremendous amount of effort has been invested by many research-
ers to model the process of wall nucleation, the generally poor
accuracy of bubble departure diameter and frequency models sug-
gests the existence of an inherent variation among bubble depar-
ture characteristics that were not captured by these models [31].
This work studies wall nucleation in subcooled boiling flows under
various conditions, considering multiple nucleation sites in each
Fig. 2. Layout of test section (left) and schematic of the wall heater with thermocouple locations (right).
condition to characterize the variation of bubble departure diame-
ters and frequencies between sites. The primary focus of this work
is the characterization of bubble departure diameter and fre-
quency, and the goal is to build on the current understanding,
specifically through investigating the variability in departure
characteristics between nucleation sites. Due to the consideration
of multiple nucleation sites in a condition and the existence of
intermittency in site activity, the conventional method to calculate
bubble departure frequency must be investigated further.
Therefore, the objective of this work is to analyze the variation in
departure characteristics across multiple nucleation sites simulta-
neously to determine a condition-average property and to evaluate
the predictive capability of current models in representing the
condition-average departure diameter and frequency.
2. Experimental approach
A closed-loop test facility is used to carry out subcooled boiling
flow experiments. The schematic layout of the facility is shown in
Fig. 1. A gear pump drives the circulation of distilled water in the
facility, and the flow rate is measured using a turbine flow meter
which has an accuracy of ±1% of the reading. The facility is pressur-
ized with compressed nitrogen through a pressurizing tank up to
600 kPa where the pressurizing tank is separated from the main
system by a long stainless-steel tube to prevent the entry of non-
condensable gases into the main test loop. Liquid subcooling is
controlled by a pre-heater located upstream of the test section.
T-type thermocouples with an accuracy of ±1 °C measure water
temperature at the inlet and the outlet of the test section while
system pressure is measured using a pressure transducer with an
accuracy of ±0.025% of the 80 kPa span.
The main component of the facility is the stainless-steel vertical
test section. It has a square flow channel with dimensions 1.27 cm
 1.27 cm  99.80 cm. The schematic of the test section is shown
in Fig. 2. A copper heater block with seven cartridge heaters is
inserted into the test section with the bottom surface approxi-
mately thirty-two hydraulic diameters downstream of the inlet.
The heater surface has dimensions of 12.70 mm  107.95 mm with
a 1.0 mm thick stainless-steel piece at the surface. K-type
 Z.J. Ooi et al. / International Journal of Heat and Mass Transfer 118 (2018) 327–339 331

Table 2
Summary of test conditions.

Cond. q00w [kW/m2] P [kPa] vf [m/s] G [kg/m2 s] DTsub [°C] No. of measured sites [–] No. of departures measured [–]

1 231.0 143.3 0.43 404.2 12.7 11 1012

2 235.5 147.6 0.28 261.7 12.1 10 821
3 216.1 143.3 0.28 262.8 23.9 9 792
4 257.8 340.8 0.43 406.6 13.7 6 474
5 267.6 344.1 0.27 260.1 14.5 5 380
6 260.4 323.0 0.28 265.1 23.4 5 301
7 294.5 440.9 0.39 372.6 13.9 2 200
8 285.8 445.4 0.39 373.1 24.3 5 318
9 295.3 419.2 0.27 259.9 22.8 7 662

thermocouples are embedded in the heater block to measure tem-
perature at various locations. Four thermocouples are located at a
depth of 1.1 mm from the stainless-steel surface and 6.35 mm
apart from each other, while the other four are placed diagonally
at either 12.7 mm or 19.05 mm depth from the first four thermo-
couples. Power to the cartridge heaters is controlled using three
3-kW autotransformers. Heat flux is determined based on the tem-
perature profile obtained by averaging the temperature measured
at different depths of the heater and is assumed to be uniform
throughout the heater surface due to the high thermal conductivity
of the copper heater. A clear quartz viewport is installed on the
opposite side of the heater surface to observe bubble nucleation
and departure.
A high-speed camera records the process of bubble nucleation
and departure from the stainless-steel surface at a frame rate of
10,000 FPS. The camera is mounted on a traverse with two-
degrees of freedom and the lens is pointed orthogonally to the hea-
ter surface. An adjustable 150-W light source with two flexible
fiber optic conduits is used to illuminate the heater surface. Two
resolutions, 384  512 pixels for low-pressure conditions and
480  640 pixels for elevated-pressure conditions, are chosen to
account for the pressure effect on active nucleation site density
for the conditions considered. The resolution is higher for the
elevated-pressure conditions as fewer active nucleation sites are
present for the conditions considered compared to the other condi-
tions. As a result, for the same file storage capacity, the recording
time is 2.91 s for the low-pressure conditions and 1.87 s for the
elevated-pressure conditions. A calibration was conducted for the
camera and lens settings indicating a 3.5 mm/pixel resolution. In
order to observe the process of wall nucleation more extensively,
a maximum of one hundred departing bubbles were manually
measured for each nucleation site. The departure diameter of a
bubble is determined by manually measuring the horizontal dis-
tance between two opposite points on the edge of the bubble at
the frame of departure. However, due to long periods of intermit-
tent inactivity and low bubble departure rate, some active nucle-
ation sites produced fewer than one hundred bubble departures
throughout the recording. Nevertheless, the departures from these
nucleation sites are also considered in the dataset.
Nine conditions spanning system pressure, subcooling, and flow
rate are used. The test matrix consists of three sets of pressures,
referred to as low (approximately 150 kPa), medium (approxi-
mately 340 kPa), and elevated (approximately 450 kPa) pressure
conditions. Two sets of flow rates with corresponding ranges of
259.9–265.1 kg/m2-s and 372.6–406.6 kg/m2-s were selected.
Lastly, two sets of subcooling temperatures with corresponding
ranges of 12.1–15.4 °C and 22.8–25.6 °C were selected. A detailed
summary of the dataset is shown in Table 2.
This work is focused on highlighting the variability between
sites observed simultaneously and therefore conditions with at
least two active nucleation sites within the viewing area are
included. The variation of the number of measured sites is due to
the differences in active nucleation site density across the condi-
tions considered. Although only two active nucleation sites were
observed for Condition 7, the data is included in this work as it
reflects the variability between nucleation sites under conditions
with low active nucleation site density.
3. Results and discussion
3.1. Bubble departure diameter
The probability density function (PDF) of bubble departure
diameters are constructed to describe the variation of bubble
departure diameters in a condition. Fig. 3 shows the distributions
of normalized bubble departure diameter, Dd . The PDF is defined
as the number of bubbles of a certain range of diameters normal-
ized by the total number of bubbles of a site. The diameters are
normalized by the site mean bubble diameter, Dd,s, which can then
be averaged to obtain the condition mean diameter, Dd,c. The defi-
nitions of these quantities are given by,
2 , !3
1 4Xm Xn
1X m
Dd;c ¼ Dd;i n 5 ¼ ðDd;s Þj ; ð1Þ
m j¼1 i¼1 m j¼1
j
where Dd,i is the i-th bubble departure diameter, n is the number of
departures observed, and m is the number of nucleation sites. The
distributions of the site normalized departure diameters, defined as,
Dd;i
Dd;i ¼ ; ð2Þ
Dd;s
are shown in Fig. 3 where the black circles represent the mean val-
ues of the PDF and are connected with a smooth line to show the
overall trend of all sites for a given condition.
The normalized bubble departure diameter distributions
resemble normal distributions at medium (Conditions 4–6) and
elevated-pressure (Conditions 7–10), but at low-pressure (Condi-
tions 1–3) they resemble a positive-skewed distribution such as
a log-normal or gamma distribution with the peak shifted to the
left.
The distribution of bubble departure diameter was previously
investigated in several studies using various refrigerants where
the bubble departure diameters were reported to resemble a nor-
mal distribution at low-pressure [16,19,28]. As the properties of
refrigerants and water are different, the effects of pressure on the
distribution of bubble departure diameter cannot be compared
directly. Nevertheless, the underlying physics behind the distribu-
tion of bubble departure diameters could still be applicable. Klaus-
ner et al. [18] suggested that the normal distributions of the wall
superheat and the liquid velocity at the bubble center of mass were
the main contributors to the normal distribution of the bubble
departure diameter. The hypothesis was further strengthened by
the success of the mechanistic model built by Klausner et al.
[18], based on these assumptions, in predicting the normal distri-
bution of the bubble departure diameter. On the other hand, the
332 Z.J. Ooi et al. / International Journal of Heat and Mass Transfer 118 (2018) 327–339

Fig. 3. Probability density function of normalized bubble departure diameters of all active nucleation sites.

positive-skewed distribution exhibited by bubble departure diam-
eters in low-pressure cases observed in this study might be due to
the fact that the comparatively larger departing bubbles at low
pressures are affected more significantly by bulk liquid subcooling
as they extend further into the turbulent core [7,9,21]. Previous
studies [7,9,21] that investigated bubble departure diameters rela-
tive to the wall length scale from basic single-phase turbulence
theory found that the difference in the bubble departure diameter
distribution between low- and elevated-pressures could be due to
the extent which the bubble grows through the boundary layer.
Similar observations were made by Brooks et al. [21] where it
was found that at low-pressure the effect of local bulk subcooling
on bubble departure diameter was more significant than at
elevated-pressure.
In addition to the number mean diameters, the variations of
bubble departure diameter distributions are investigated in terms
 Z.J. Ooi et al. / International Journal of Heat and Mass Transfer 118 (2018) 327–339 333

Fig. 4. Comparison of (a) surface mean departure diameter and (b) volume mean departure diameter with number mean departure diameter.

of the surface and volume mean diameters which conserve the Table 3
higher order uses of the bubble departure diameter. The definitions Available bubble departure diameter models.
of the number mean departure diameter, DNd (used by Dd,s and Dd,c), Authors Models
surface area mean diameter, DSd , and volume mean departure diam-  
Unal [15] q00 pD2b D2 pD2 3
1  Dp2 ¼ hc DT sub 2 b þ p6 qg hfg dtb
dD
eter, DVd are, 4
b
 0:9 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
, vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
, ffi vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
, ffi
Kocamustafaogullari and Ishii
Dd ¼ 1:584  103 Dqq r
u m u m [8] gðqf qg Þ
X uX 2 u X 3
g
m
Dd ¼ t Dd;i m; Dd ¼ t Dd;i m:
Dd r
DNd ¼ Dd;i m; S V 3
Prodanovic et al. [32] Dþ
d  qf a2 ¼ 236:75Jaw NT q
0:58 0:88 1:77
Bo0:14
f

i¼1 i¼1 i¼1 Brooks and Hibiki [9] Dd  DLad ¼ C Dd Ja0:49
T q 0:78 Bo0:44 Pr1:72
sat

Only when the distribution of the bubble departure diameter is
normal are these three mean diameters equal [7]. Hence, using the
mean diameter to describe the associated area or volume when the
departure diameter distribution is not normal could be problem-
atic as the inherent bias in these quantities will lead to an overpre-
diction, for example, in the wall nucleation source term and the
wall nucleation volume source term of the IATE [7,9,21]. One
approach to overcome this issue is by relating the traditional num-
ber mean diameter, DNd , used to model the departure diameter, to
the respective equivalent diameters through the distribution fac-
tors obtained from the slopes of the plots of DSd or DVd against DNd ,
as shown in Fig. 4. In the work by Martinez-Cuenca et al. [7], the
distribution factors, CS and CV, are defined as
DSd ¼ C S DNd ; DVd ¼ C V DNd
and their values were determined to be 1.06 and 1.12, respectively.
CS and CV of this work are determined to be 1.056 and 1.109, respec-
tively. Additionally, it is observed from Fig. 4 that the surface mean
departure diameter shows a marginal pressure effect and is slightly
above the fitted line of CS = 1.056 at low-pressure conditions, as
shown by the black circles. Similar observations can be made on
the volume mean departure diameter with a higher slope at low-
pressure conditions and the reverse at elevated-pressure condi-
tions, represented by the green triangles. Fig. 4 also shows that
the distribution factors are overpredicted in the elevated-pressure
conditions but underpredicted in the low-pressure conditions. Nev-
ertheless, more data is needed to better understand the impact of
pressure on the distribution factors.
The dataset of bubble departure diameter collected in this
experimental study is used to evaluate several bubble departure
diameter models in the literature. The evaluated models are listed
in Table 3 and the results are plotted in Fig. 5 where the experi-
ð3Þ
mental and calculated mean departure diameters of all active
nucleation sites observed in this study are compared. These models
are chosen due to their wide usage in the literature and relative
success in past benchmarks across a wide range of data. The major-
ity of the bubble departure models benchmarked in this work
show relatively good results. However, the model of Koca-
mustafaogullari and Ishii [8] shows substantially large errors by
overpredicting the bubble departure diameter as it was modeled
by modifying a pool boiling correlation to account for the effect
of pressure. The model by Unal et al. [15] marginally captures
the effects of pressure on bubble departure diameters where most
of the departure diameters in medium and elevated-pressures are
predicted to within 50% accuracy but those in low-pressure are
slightly overpredicted. Similar results are obtained from the model
ð4Þ
by Prodanovic et al. [32] at low-pressure as bubble departure
diameters are overpredicted but at medium and elevated-
pressure the predicted diameters show reasonably good agreement
with the experimental values as most of them are predicted within
50% accuracy. On the other hand, the model developed by Brooks
and Hibiki [9] gives relatively good predictions of the bubble
departure diameter across all conditions, particularly at low-
pressures, as a majority of the departure diameters are predicted
within 50% of the experimental values. However, the model mar-
ginally underpredicts the results at medium and elevated-
pressure conditions. Furthermore, Fig. 5 also highlights the spread
of departure diameters from different active nucleation sites
within a condition which has not been addressed by past studies.
The maximum and the minimum site mean departure diameters
of a condition can vary significantly as shown by the low-
pressure cases (Conditions 1–3). Clearly, choosing only one nucle-
ation site can lead to drastically different conclusions of modeling
capability. The spread in departure diameter also highlights the
334 Z.J. Ooi et al. / International Journal of Heat and Mass Transfer 118 (2018) 327–339

Unal (1976) Kocamustafaogullari

& Ishii (1983)

Prodanovic et al. (2002) Brooks & Hibiki (2015)

Fig. 5. Evaluation of bubble departure diameter models with experimental data.

difficulty of model development based on one nucleation site per
condition.
3.2. Bubble departure frequency
The traditional method of determining the departure frequency
of a condition by investigating only the departures from one active
nucleation site per condition assumes a largely uniform time dif-
ference between each departure and disregards the variation of
bubble departure frequencies from different sites within the condi-
tion. The validity of this method is questioned as multiple active
nucleation sites studied in this work are observed to undergo inter-
mittent periods of inactivity of varying lengths, and the departure
frequencies across different sites within a condition show varia-
tions. The intermittencies experienced by active nucleation sites
are highlighted by the plots of departure number against time in
Fig. 6. Some active nucleation sites are observed to produce bub-
bles at a uniform rate, such as Site 5 of Condition 1, while most
show pauses of different lengths, such as Site 10 of the same con-
dition. Despite these interruptions in production, a closer investi-
gation shows that the rate of bubble departure by the sites are
similar when they are actively producing bubbles. This observation
indicates that during the ‘active’ period, bubbles are produced by
the active nucleation sites within a condition at a rate characteris-
tic of the condition. However, due to the intermittent inactivity,
the departure frequency of each site will be significantly different
if it is given by the total number of departures over time. Therefore,
the issue lies with the definition of ‘active’ in the active nucleation
site density. The steady-state experiments result in active sites that
have periods of dormancy, in that the continuous ebullition cycle is
interrupted. The variability in the ‘active’ frequency between
nucleation sites can be studied by separating the periods of dor-
mancy from the periods of activity.
Classically, the mean departure frequency is calculated from the
number of departures observed at a nucleation site over the time
between the first departure, t1, and last departure, tn,
n
fd ¼ ð5Þ
tn  t1
where n is the number of departures. This method assumes that the
time elapsed between departures is relatively constant throughout
the observation such that the frequency becomes independent of
the number of bubbles observed. Based on this assumption, the
mean bubble departure frequency of an active nucleation site, fd,s,
is defined as,
!1
1 X n1
1
f d;s ¼ ðt iþ1  ti Þ ¼ ; ð6Þ
n  1 i¼1 Dtn;s
where n is the total number of departures observed for a particular
site and Dt n;s is the mean time difference between departures of a
site. The method is adopted to determine the average departure
 Z.J. Ooi et al. / International Journal of Heat and Mass Transfer 118 (2018) 327–339 335

Cond. 2 Cond. 3

Cond. 1

Cond. 4 Cond. 5 Cond. 6

Cond. 7 Cond. 8 Cond. 9

Fig. 6. Time sequence of bubble departure for all active nucleation sites.

frequency of a condition, fd,c, where it is defined as the harmonic nucleation site in a consistent manner. The latter is defined as an
mean [33] of fd,s, given by, extended period of inactivity between bubble departure from an
" ! #1 " #1 active nucleation site. Similar phenomenon was observed by Zou
1X m
1 1X m
1 and Jones [34] while investigating the effect of thermal interaction
f d;c ¼ ¼ ðDt n;s Þi ¼ ; ð7Þ on nucleation site distribution in subcooled boiling where active
m i¼1 f d;s
i
m i¼1 Dtn;c
nucleation sites were observed to be randomly activated or deacti-
vated. The authors suggested that activation and deactivation of a
where Dtn;c is the arithmetic mean of Dt n;s and m is the total number
site were caused by the interaction between nearby bubbles [34] or
of sites in a condition.
bubbles sliding from upstream. The partitioning of the two quanti-
The periods of inactivity of the nucleation sites have an over-
ties is done by visually identifying a threshold between the active
whelming effect on the average nucleation time, as made clear
and dormant departure times based on the PDF of the time
by Fig. 6. However, the figure also suggests that there is a charac-
between departures for each site. An example of this method is
teristic frequency while the site is actively producing bubbles. This
shown in Fig. 7 using the PDF of Site 10 of Condition 1, where
is shown by the approximately parallel lines in Conditions 1, 2, 4,
the red line represents the threshold.
and 8, indicating the same departure frequency but separated by
With the dormant departure time eliminated, the results are
departure number due to dormancy. Therefore, the time between
plotted in Fig. 8 as active departure number against accumulated
departures, Dtn, is separated into two groups, namely the active
active time, t 0a . The elimination of the dormant time removes the
departure time, Dta, and the dormant departure time, Dtd. The for-
mer is described as the time elapsed between individual depar- large variation of the mean departure frequency of each site and
tures during a period where bubbles depart from an active results in a more uniform value that is more characteristic of each
336 Z.J. Ooi et al. / International Journal of Heat and Mass Transfer 118 (2018) 327–339

Table 4
Available bubble departure frequency models.

Authors Models
Cole [35] 4gðq q Þ0:5
fd ¼ f g
3qf C D Dd
 2 
Basu et al. f d ¼ tG þt
1 4:1
tw ¼ 139:1DT w
D
tG ¼ 0:0222e0:027Jasub af Jad
W
[23,24] sup

Brooks and  f D2
f d  daf d ¼ C fd Ja0:82 q
1:46 0:93
Pr2:36
w NT sat
Hibiki [9]
Yang et al. [30] f d ¼ 0:032DT w3:08

results in a lower site departure frequency. The departure fre-

Fig. 7. Partitioning of PDF of time difference between departures to separate active
and dormant departure times.
site. This highlights the fact that the classical site departure fre-
quency is strongly influenced by the number of departures mea-
sured. A larger number of departures increases the likelihood of
observing the inherent periods of inactivity which subsequently
quency of a site can be better characterized when the inactivity
is removed as it reflects the uninterrupted rate of bubble depar-
ture, but this is not the true generation rate of bubbles from the
site. Furthermore, it is this active departure frequency that has
been experimentally reported and modeled as past works have
chosen to observe highly active sites and have considered only a
small number of consecutive departures. The mean active

Cond. 1 Cond. 2 Cond. 3

Cond. 4 Cond. 5 Cond. 6

Cond. 7 Cond. 8 Cond. 9

Fig. 8. Time sequence of bubble departure for all active nucleation sites without considering periodic dormancy.
 Z.J. Ooi et al. / International Journal of Heat and Mass Transfer 118 (2018) 327–339 337

Cole (1960) Cole (1960)

Basu et al. (2005) Basu et al. (2005)

Brooks & Brooks &

Hibiki (2015) Hibiki (2015)

Yang et al. (2016) Yang et al. (2016)

(a) (b)
Fig. 9. Comparison of (a) classical and (b) active departure frequencies with predicted values.
338 Z.J. Ooi et al. / International Journal of Heat and Mass Transfer 118 (2018) 327–339
departure frequency of an active nucleation site, fa,s can be
obtained by modifying Eq. (6) as,
!1
1 nX a 1
1  h q 6:5
f a;s ¼ ðt a  tai Þ ¼ ;
na  1 i¼1 iþ1 Dt a;s
where Dt a;s is the mean time difference between active bubble
departures of a site, na is the total number of active departures of
a site, and ta is the time of an active bubble departure. Similarly,
by modifying Eq. (7), the mean active departure frequency of a con-
dition, fa,c is defined as,
" ! #1 " #1
1X m
1 1X m
1 els like the IATE will result in an overprediction of the bubble
f a;c ¼ ¼ ðDt a;s Þi ¼ ;
m i¼1 f a;s
i
m i¼1 Dt a;c
where Dta;c is the arithmetic mean of Dt a;s and m is the number of
nucleation sites in a condition.
Several bubble departure frequency models, presented in
Table 4, are benchmarked with the classical departure frequencies
obtained from this study, and the results are plotted in Fig. 9(a).
The models by Cole [35], Basu et al. [23,24], and Brooks and Hibiki
[9] show overprediction of the classical departure frequency while
the model by Yang et al. [30] shows a significant underprediction
for the elevated-pressure conditions. The models by Brooks and
Hibiki [9] and Yang et al. [30] are more sensitive to pressure while
the models by Cole [35] and Basu et al. [23,24] show little influence
by pressure. In addition, a significant spread is present as the min-
imum and maximum site departure frequencies in a condition are
observed to differ by several orders of magnitude. Consequently,
only a portion of the classical departure frequencies are predicted
by the models to within 50% accuracy. In general, the models show
poor prediction on the classical departure frequency as the accura-
cies range from as low as 20% to as high as 2000%, depending on
the presence and length of dormancies of a site.
Overall, the newly defined active departure frequency values
are comparatively less scattered, although even with the dorman-
cies eliminated, the departure frequencies of different active nucle-
ation sites of a condition still have an inherent distribution.
Therefore, caution should be taken in validating and developing
models based on observing only one active nucleation site per con-
dition. This is also highlighted by the predictions of the models
plotted in Fig. 9 where the variability between model predictions
is smaller than the spread of the different sites. The distribution
of the departure characteristics is likely due to the strong
dependence of the wall nucleation characteristics, especially the
departure frequency on the wall superheat, DTw, defined as the dif-
ference between the wall temperature and the saturation temper-
ature at the system pressure. The dependence is highlighted in the
departure frequency, presented in Table 4, where the wall super-
heat is raised to several powers as shown by the model by Basu
et al. [23,24] and Yang et al. [30]. This means that a small fluctua-
tion in the wall superheat may drastically increase or decrease the
bubble departure frequency of an active nucleation site.
This experimental study highlights the existence of intermittent
periods of inactivity (or dormancies) between bubble departures
that are not accounted for by the traditional modeling. The dor-
mancies, if not considered, will result in a large overprediction of
the bubbles nucleated. It is found that this dormancy has gone lar-
gely undocumented by most departure frequency models in the lit-
erature as most of them were created based on experimental data
collected from only one active nucleation site per condition. Typi-
cally, sites with the most consistent production of bubbles were
selected. The current models’ improved accuracy in predicting
the active departure frequency shows that these models predict
the active departure frequency instead of the classical departure
frequency of a site. The application of these models without correc-
Table 5
Available active nucleation site density models.
Authors Models
ð8Þ Mikic and
N n ¼ C 1 r 6:5 fg g
DT 6:5
Rohsenow s 2T sat r w
[37]
Basu et al. N n ¼ 0:34½1  cosð/s ÞDT 2:0
W DT w:ONB < DT w < 15  C
[23,24] N n ¼ 3:4  105 ½1  cosð/s ÞDT 5:3
W DT w P 15  C
Yang et al. [30] N n ¼ 0:28DT 2:66
w
tion in Computational Fluid Dynamics codes and mechanistic mod-
ð9Þ departure frequency as well as the gas-phase contribution of wall
nucleation in two-phase flows. Thus, it is paramount to invest
more effort to study the inherent dormancy of bubble departures
to improve the capability in predicting the bubble generation rate
of an active nucleation site.
This dormancy should be addressed in the modeling of the ‘ac-
tive’ nucleation site density. As shown by several common active
nucleation site density models listed in Table 5, even a small fluc-
tuation in the wall superheat can translate into a significant change
in nucleation site density. By measuring the patterns of wall tem-
perature in pool boiling with thermochromic liquid crystal, Ken-
ning [36] showed that local wall superheat could differ by 20–
150% from the spatial mean value. Also, according to Klausner
et al. [18], the localized cooling at active nucleation sites as well
as the highly turbulent nature of two-phase flows are the main rea-
sons for the spatial and temporal variation of the wall superheat in
subcooled boiling flow. The effect of localized cooling on the activ-
ity of an active nucleation site is explained by Zou and Jones [34]
where the authors argue that, during bubble growth, evaporation
in the thin liquid film between the surface and bubble interface
causes the local wall temperature near an active nucleation site
to drop. Through lateral heat conduction in the solid, this localized
drop in wall temperature subsequently reduces the local wall tem-
perature at an adjacent nucleation site to a level lower than that
required to activate bubble nucleation. In addition, Zou and Jones
[34] also investigated the spatial distribution of active nucleation
sites by comparing the probability densities of nucleation site dis-
tribution and the probability densities of nearest-neighbor nucle-
ation site distance to a spatial Poisson distribution. It was
concluded that the nucleation site distribution on a heater surface
is more uniform than the random Poisson distribution due to the
interactions between nucleation sites [34]. The uniform spatial dis-
tribution of active nucleation sites suggests the existence of the
temporal distribution of active nucleation sites. However, more
experimental data and effort are necessary to gain a deeper under-
standing on this subject. Considering the large variation of the wall
superheat and its significant impact on the bubble departure fre-
quency and the active nucleation site density, the modeling and
prediction of total bubble generation (which is proportional to
the product of the active nucleation site density and the departure
frequency) is currently limited by the nature of the wall tempera-
ture and perhaps the use of only a mean temperature value.
4. Conclusions
Experiments were conducted to study wall nucleation charac-
teristics in subcooled boiling flow in a vertical channel with vary-
ing system pressure, subcooling, and flow rate. Bubble departures
from multiple active nucleation sites in each condition are mea-
sured to determine the variation of bubble departure diameter
and frequency between nucleation sites. From the analysis of the
data set and comparison with existing departure characteristic
models, the following conclusions can be made:
 Z.J. Ooi et al. / International Journal of Heat and Mass Transfer 118 (2018) 327–339
The distribution of mean bubble departure diameter in a condi-
tion resembles a positive-skewed distribution in low-pressure
conditions and a more Gaussian distribution in medium and
elevated-pressure conditions. Therefore, the assumption that
the number mean diameter is equal to both the surface mean
diameter and the volume mean diameter is not valid for low-
pressure conditions.
Comparison with existing departure diameter models shows
that models formulated specifically for subcooled boiling flow
can predict the bubble departure diameter relatively accurately.
However, the spread of the bubble departure diameters con-
firms the inherent variation from one active nucleation site to
another within the same condition, which highlights the need
to measure many nucleation sites to obtain a condition average
value.
The classical bubble departure frequency in a condition varies
significantly from one site to another due to periods of dor-
mancy, thus causing existing models to significantly overpre-
dict the departure frequency. The elimination of dormancy
from the experimental bubble departure frequency results in
a new quantity known as the active departure frequency which
describes the rate of bubble departure from a site during the
‘active’ period. The removal of dormancy reduces the distribu-
tion of departure frequency of a given condition and, more
importantly, provides a measurement that is independent of
the number of departures observed.
More experimental studies are required to understand the dor-
mancy experienced by bubble departure. New modeling of the
active nucleation site density or bubble departure frequency is
necessary to include the impact of bubble dormancy on bubble
generation rate from a heated surface. Employing current mod-
eling based on the mean wall superheat without considering
these periods of inactivity will result in an overprediction of
the number of bubbles generated.
Conflict of interest
The authors declare that there is no conflict of interest.
References
[1] N. Kurul, M.Z. Podowski, Multidimensional effects in forced convection
subcooled boiling, in: Proceedings of the 9th International Heat Transfer
Conference, Hemisphere Publishing Corporation, Jerusalem, Israel, 1990, vol. 2,
pp. 21–26.
[2] J. Riznic, M. Ishii, Bubble number density and vapor generation in flashing flow,
Int. J. Heat Mass Transf. 32 (1989) 1821–1833.
[3] M. Ishii, T. Hibiki, Thermo-Fluid Dynamics of Two-Phase Flow, Springer, 2011.
[4] M. Ishii, S. Kim, J. Kelly, Development of interfacial area transport equation,
Nucl. Eng. Tech 37 (2005) 525–536.
[5] G. Yeoh, J. Tu, Two-fluid and population balance models for subcooled boiling
flow, Appl. Math. Model. 30 (2006) 1370–1391.
[6] G. Yeoh, S. Cheung, J. Tu, Mo Ho, Fundamental consideration of wall heat
partition of vertical subcooled boiling flows, Int. J. Heat Mass Transf. 51 (2008)
3840–3853.
[7] R. Martínez-Cuenca, C.S. Brooks, J.E. Juliá, T. Hibiki, M. Ishii, Stochastic nature
of wall nucleation and its impact on the time average boundary condition, J.
Heat Transf. 137 (2015) 021504.
[8] G. Kocamustafaogullari, M. Ishii, Interfacial area and nucleation site density in
boiling systems, Int. J. Heat Mass Transf. 26 (1983) 1377–1387.
339
[9] C.S. Brooks, T. Hibiki, Wall nucleation modeling in subcooled boiling flow, Int. J.
Heat Mass Transf. 86 (2015) 183–196.
[10] F.C. Gunther, Photographic study of surface-boiling heat transfer to water with
forced convection, Trans. ASME (1951) 115–123.
[11] P. Griffith, J.A. Clark, W.M. Rohsenow, Technical report no. 12: Void volumes in
subcooled boiling systems, The Office of Naval Research, D.S.R. Project Number
7-7673, 1958.
[12] G.G. Treschev, The number of vapor-formation centers in surface boiling,
Convect. Heat Transf. Two-Phase One-Phase Flows (1969) 97–105.
[13] A.H. Abdelmessih, F.C. Hooper, S. Nangia, Flow effects on bubble growth and
collapse in surface boiling, Int. J. Heat & Mass Transf. 15 (1971) 115–125.
[14] M. Akiyama, F. Tachibana, Motion of vapor bubbles in subcooled heated
channel, Bullet. JSME 17 (1974) 241–247.
[15] H.C. Unal, Maximum bubble diameter, maximum bubble growth time and
bubble growth rate during subcooled nucleate flow boiling of water up to 17.7
MN/m2, Int. J. Heat Mass Transf. 19 (1976) 643–649.
[16] J.F. Klausner, R. Mei, D.M. Bernhard, L.Z. Zeng, Vapor bubble departure in
forced-convection boiling, Int. J. Heat Mass Transf. 36 (1993) 651–662.
[17] V.K. Dhir, Nucleate and transition boiling heat transfer under pool and external
flow conditions, Int. J. Heat Fluid Flow 12 (1991) 290–314.
[18] J.F. Klausner, R. Mei, L.Z. Zeng, Predicting stochastic features of vapor bubble
detachment in flow boiling, Int. J. Heat Mass Transf. 40 (1997) 3547–3552.
[19] G.E. Thorncroft, J.F. Klausner, R. Mei, An experimental investigation of bubble
growth and detachment in vertical upflow and downflow boiling, Int. J. Heat
Mass Transf. 41 (1998) 3857–3871.
[20] L. Zou, B.G. Jones, Heating surface material’s effect in subcooled flow boiling
heat transfer of R134a, Int. J. Heat Mass Transf. 58 (2013) 168–174.
[21] C.S. Brooks, N. Silin, T. Hibiki, M. Ishii, Experimental investigation of wall
nucleation characteristics in flow boiling, J. Heat Transf. 137 (2015) 051501-1–
051501-9.
[22] C.S. Brooks, Wall nucleation and the two-fluid model in subcooled boiling flow,
Ph.D. Thesis, Nuclear Engineering, Purdue University, West Lafayette, Indiana,
2014.
[23] N. Basu, Modeling and experiments for wall heat flux partitioning during
subcooled flow boiling of water at low pressures, Ph.D. Thesis, Mechanical
Engineering, University of California, Los Angeles, CA, 2003.
[24] N. Basu, G.R. Warrier, V.K. Dhir, Wall heat flux partitioning during subcooled
flow boiling: Part 1 - model development, J. Heat Transf. 127 (2005) 131–140.
[25] R.M. Podowski, D.A. Drew, R.T. Lahey, M.Z. Podowski, A mechanistic model of
the ebullition cycle in forced convection subcooled boiling, Eighth Int. Topical
Meet. Nucl. React. Therm.-Hydraul. 3 (1997) 1535–1542.
[26] E.L. Bubeau, M. Salcudean, A study of bubble ebullition in forced-convective
subcooled nucleate boiling at low pressure, Int. J. Heat Mass Transf. 37 (1994)
2245–2259.
[27] G. Hong, X. Yan, Y.H. Yang, T.Z. Xie, J.J. Xu, Bubble departure size in forced
convective subcooled boiling flow under static and heaving conditions, Nucl.
Eng. Des. 247 (2012) 202–211.
[28] M. Lin, P. Chen, Photographic study of bubble behavior in subcooled flow
boiling using R-134a at low-pressure low pressure range, Ann. Nucl. Energy 49
(2012) 23–32.
[29] V.H.M. Del Valle, D.B.R. Kenning, Subcooled flow boiling at high heat flux, Int. J.
Heat Mass Transf. 28 (1985) 1907–1920.
[30] L.X. Yang, A. Guo, D. Liu, Experimental investigation of subcooled vertical
upward flow boiling in a narrow rectangular channel, Exp. Heat Transf. 29
(2016) 221–243.
[31] C.S. Brooks, T. Hibiki, Modeling and validation of interfacial area transport
equation in subcooled boiling flow, J. Nucl. Sci. Technol. 53 (2016) 1192–1204.
[32] V. Prodanovic, D. Fraser, M. Salcudean, Bubble behavior in subcooled flow
boiling of water at low pressures and low flow rates, Int. J. Multiphase Flow 28
(2002) 1–19.
[33] J.F. Kenny, E.S. Keeping, Mathematics of Statistics, Pt. 1, 3rd ed, Van Nostrand,
1962.
[34] L. Zou, B.G. Jones, Thermal interaction effect on nucleation site distribution in
subcooled boiling, Int. J. Heat Mass Transf. 55 (2012) 2822–2828.
[35] R. Cole, A photographic study of pool boiling in the region of the critical heat
flux, AIChE J. 6 (1960) 533–538.
[36] D.B.R. Kenning, Wall temperature patterns in nucleate boiling, Int. J. Heat Mass
Transf. 35 (1992) 73–86.
[37] B.B. Mikic, W.M. Rohsenow, A new correlation of pool-boiling data including
effect of heating surface characteristics, J. Heat Transf. 91 (1969) 245–246.