You are on page 1of 10

Applied Thermal Engineering 132 (2018) 595–604

Contents lists available at ScienceDirect

Applied Thermal Engineering


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

Research Paper

Boiling incipience of subcooled water flowing in a narrow tube using


wavelet analysis
M. Shibahara a,⇑, K. Fukuda a, Q.S. Liu a, K. Hata a, S. Masuzaki b
a
Graduate School of Maritime Sciences, Kobe University, 5-1-1Fukaeminamimachi, Higashinada, Kobe, Hyogo 658-0022, Japan
b
National Institute for Fusion Science, 322-6, Oroshi-cho, Toki, Gifu 509-5292, Japan

h i g h l i g h t s

 Incipient boiling phenomena of subcooled water in a narrow tube were observed.


 The boiling signal was analyzed by the wavelet decomposition method.
 The semi-empirical correlation of the boiling incipience was obtained.

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

Article history: Various incipient boiling phenomena for subcooled water flowing in a uniformly heated narrow tube
Received 25 August 2017 were observed experimentally. The boiling signal was analyzed using the wavelet decomposition
Revised 15 November 2017 method. The boiling incipience of subcooled water in the narrow tube was recorded by a sound level
Accepted 28 December 2017
meter at various flow velocities. A platinum tube was used as the experimental tube with an inner diam-
Available online 28 December 2017
eter of 1.0 mm. The length of the experimental tube was 23.2 mm. The tube was heated by the Joule effect
using a direct current. The inlet temperature and flow velocities ranged 285–346 K and 2.5–14 m/s,
Keywords:
respectively. The surface superheat ascended with an increase of the heat flux until the incipient boiling
Incipient boiling
Subcooled boiling
point was reached. The initial temperature overshoot did not appear as the outlet pressure increased.
Upward flow Since the existing correlations underestimated the incipient heat flux, a semi-empirical correlation of
Narrow channel the boiling incipience was obtained based on the experimental data. The predicted value of the new
Wavelet correlation is in agreement with the experimental data within ±30%.
Ó 2018 Elsevier Ltd. All rights reserved.

1. Introduction prediction of the boiling incipience in a cooling tube is an issue


to be considered in the thermal design of the PFC.
Understanding of the boiling incipience in tubes is important Since their convective heat transfer coefficients are higher than
for the safety assessment of industrial applications, since sub- those of gases, the convective heat transfer of various liquids was
cooled liquid is used in various fields, such as plasma facing com- investigated experimentally and empirical correlations have been
ponents (PFC) in fusion reactors, rocket engines, and hydrogen suggested [4–6]. For mini- and micro-channels, many studies of
storage that utilize narrow channels [1]. The liquid channel cooling heat transfer characteristics have been conducted over the past
is also used in insulated gate bipolar transistor (IGBT) modules that decade [7]. Even though micro channels are known to enhance
are used in hybrid vehicles, electric vehicles, and electric propul- the heat transfer coefficient, the boiling incipience has not been
sion ships [2]. With the increase in the IGBT power density, clearly investigated so far. For conventional tubes up to an inner
dynamic cooling technology is required and the detection of the diameter of 2.0 mm, the boiling incipience can be predicted by
boiling incipience is more important for the safe operation of these the Bergles and Rohsenow [8] or Sato and Matsumura [9] correla-
power electronics. Furthermore, the knowledge of boiling incipi- tions. In contrast, Ghiaasiaan and Chedester [10] mentioned that
ence is necessary to design the divertor of the PFC. Since the edge the predictions calculated by the correlations in [8] or [9] were
localized mode incurs high heat flux at the divertor [3], the lower than the results of Ghiaasiaan and Chedester’s [10] and Ina-
saka et al.’s [11] experiments. Since the accuracy of these correla-
tions is not satisfactory for micro scale phenomena, Ghiaasiaan and
⇑ Corresponding author. Chedester suggested the use of a new correlation modified by
E-mail address: sibahara@maritime.kobe-u.ac.jp (M. Shibahara). Davis and Anderson’s model [12] using a ratio of thermocapillary

https://doi.org/10.1016/j.applthermaleng.2017.12.110
1359-4311/Ó 2018 Elsevier Ltd. All rights reserved.
596 M. Shibahara et al. / Applied Thermal Engineering 132 (2018) 595–604

Nomenclature

A inner surface area of the experimental tube, m2 U expanded uncertainty


aj approximation coefficient at the decomposition level of u flow velocity, m/s
j (–) V volume, m3
b systematic standard uncertainty (–) VI voltage of the standard resistor, V
C coefficient in Eqs. (23), (24) and (31) VR voltage of the experimental tube, V
c constant in Eq. (13) VT unbalanced voltage, V
ch specific heat of the platinum tube, J/kg K v specific volume, m3/kg
cp specific heat at constant pressure, J/kg K w weighting factor
d inner diameter, m X Fourier transform
dj detail coefficient at the decomposition level of j (–) y distance from the wall, m
f boiling signal (–)
fs frequency, Hz Greek symbols
h heat transfer coefficient, (W/m2 K) a coefficient of RT (3.78  103) (–)
hfg latent heat of vaporization, J/kg b coefficient of RT (5.88  107) (–)
I direct current, A e emissivity of platinum (–)
J maximum decomposition level of signal (–) / scaling function (–)
k coverage factor (=2) (–) k thermal conductivity, W/mK
L length, m l viscosity, Ns/m2
MAE mean absolute error,% v number of degrees of freedom (–)
n number of experimental data (–) q density, kg/m3
Nu =hd/k, Nusselt number (–) r surface tension, N/m
P pressure, kPa rsf Stefan-Bolzmann
Rt constant (=5.67  108 W/m2 K4) (–)
Pr =lcp/k, Prandtl number (–) s = 0 Q ðtÞdt=Q ðtÞ; e-folding time, s
Pcr critical pressure, kPa w wavelet (–)
Pre =Pout/Pcr, reduced pressure, kPa f ratio of thermocapillary forces and aerodynamic forces
Q heat transfer rate, W in Eq. (25) (–)
Qi heat transfer rate at the boiling incipience, W x frequency response, dB
Q0 initial exponential heat input, W/m3
Q_ heat generation rate, W/m3
Subscripts
q heat flux, W/m2 exp experimental value
qi incipient heat flux, W/m2 B bubble
Re =qud/l, Reynolds number (–) BR Bergles and Rohsenow
R0 electrical resistance at 0 °C (=1.38  103), X
GC Ghiaasiaan and Chedester
R electrical resistance, X f liquid
R⁄ critical cavity radius g vapor
RB Bubble radius, m h heater
Ra average roughness, lm
i inner
Rs standard resistor, X in inlet
RT electrical resistance of the experimental tube, X MIC microphone
Ry maximum height, lm o outer
Rz ten-spot average roughness, lm
out outlet
r radius, m pred predicted value
rx lead resistance, X s surface
s random standard uncertainty of the mean of N mea-
sat saturation
surements (–) SLM sound level meter
T temperature, K sur surrounding
Ta average temperature, K SM Sato and Matsumura
Ti incipient surface temperature, K
sub subcooling
t time, s w wall
t95 student’s t value at a specified confidence level with v WSE windscreen effect
degrees of freedom (–) WSC windscreen correction
DTsat =Ts  Tsat, surface superheat, K
DTsub,in =Tsat  Tin, inlet liquid subcooling, K

and aerodynamic forces. However, there is a limitation on the To obtain a high accuracy correlation of boiling incipience in
applicability of the model due to experimental restrictions. narrow tubes, this study focuses on the measurement of the boiling
Other investigators studied the heat transfer characteristics, incipience using the wavelet decomposition method (WDM) [16].
including the critical heat flux (CHF) for water, using small plat- Various incipient boiling phenomena for subcooled water flowing
inum tubes and obtained the correlations of forced convection in in a uniformly heated narrow tube were observed and the boiling
the tubes at various experimental conditions [13–15]. However, signal was analyzed using WDM. Based on the experimental data
the boiling incipience was not clarified because the pump noise and their analysis, an empirical correlation of the boiling incipience
was too high at high flow velocities. was obtained.
M. Shibahara et al. / Applied Thermal Engineering 132 (2018) 595–604 597

2. Experimental setup for the boiling incipience resistance RT, can be obtained. The relationship between RT and
the average temperature Ta, was previously calibrated as RT =
The experimental facility of the subcooled flow boiling is shown R0(1 + aTa + bTa2) using the water circulation.
in Fig. 1. The experimental facility consists of three parts: a data The heat flux q, was calculated using the following energy
acquisition system for the temperature of the experimental tube balance:
(ET), the fluid temperature, and the pressure measurements; a heat  
input control system; and a water loop with a pressurizer, a non- Vh _ dT a
q¼ Q  qh ch ð1Þ
seal pump, a preheater, a Coriolis mass flow meter, and a cooler. Ah dt
A platinum tube (d = 1.0 mm, L = 23.2 mm), with a material purity
of 99.99%, was used as the ET. The tube was soldered to the silver- where Vh, Ah, Q_ , qh, ch, and t, denote, respectively, the volume of the
coated copper plates in the test section. The ET was installed ver- ET, the inner surface area of the ET, the heat generation rate, the
tically in the test section. To measure the sound of the boiling density of the ET, the heat capacity of the ET, and time. The heat
incipience, a sound level meter that can analyze the spectrum of generation rate Q_ was defined with a long e-folding time s:
the boiling incipience was used. Q_ ¼ Q 0 expðt=sÞ.
The exponential signal of the heat generation rate was gener- To calculate the surface temperature of the ET, the following
ated in a programmable computer. The signal was supplied to steady heat conduction equations were solved:
the controllable DC power supply through a digital to analog (D/
A) convertor. The heat generation was calculated by Ohm’s law, kh r2 T þ Q_ ¼ 0 ð2Þ
VRI, where VR is the voltage drop of the ET and I is the current cal-

culated using the voltage drop across a standard Manganin electric @T 
kh ¼q ð3Þ
resistor, with resistance Rs. The average temperature of the ET was @r r¼ri
measured by a double bridge circuit in the data acquisition system.
Platinum was used for the ET for high accuracy temperature mea- 
@T 
surements of electrical resistance thermometry. Using the unbal- ¼0 ð4Þ
@r r¼ro
anced voltage VT, of the double bridge circuit, the electrical

Heat Input control System


Electric
Electromagnetic
Switch Ref. Amp. DC
Signal + Valve
D/A
-

Shut-off Temp.
Comparator I
Ta Q Pressurizer T
A/D ÷ Divider × Multiplier
AC
Diff. Amp.
VT VI VR VI VI RS

VI VT VI VTL
Water loop
Double
Data acquisition system Bridge
R2 circuit R1
VT
R2
r0 : Thermocouple
R3 T P
: Pressure Gauge
Sound Level
R3 Test Section Meter
VR Exp. Tube
(RT)
Ion Exchanger
T P
Bypass Loop

Cooler

Preheater P

Coriolis Mass
Flow Meter Non-Seal Pump
AC

Fig. 1. Schematic of the experimental set up.


598 M. Shibahara et al. / Applied Thermal Engineering 132 (2018) 595–604

Z ro sX
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 n
Ta ¼ 2prTðrÞdr ð5Þ sR ðRk  RLS;k Þ2
p ðr 2o  r 2i Þ ri ¼ k¼1
ð13Þ
R Nc
where ri, ro, and kh are the inner diameter, the outer diameter, and
n and c in Eq. (13) are the number of data points and a constant
the thermal conductivity of the ET, respectively. The calculations of
value for the least-square (LS) method, respectively.
surface temperature and the heat loss from the outer surface of the
Inner surface temperature
ET are shown in Appendix A.1.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
The liquid temperature and pressure at the inlet and outlet of  2  2  2  2
the test section were measured by a K-type sheathed thermocou- dT s @T s dT a @T s dq
¼ þ ð14Þ
ple and a strain gauge transducer, respectively. Ts @T a Ta @q q
During the experiment, the inlet temperatures ranged from 285 Inlet and outlet temperatures
to 349 K, with corresponding subcooling ranging from 76 to 156 K.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
The deionized water was pressurized up to 800 kPa. The flow  2  2  2  2
dT TC @T TC dV TC @T TC dV AMP
velocity ranged from 2.5 to 14 m/s and the inlet pressure ranged ¼ þ ð15Þ
T TC @V TC V TC @V AMP V AMP
from 238 to 855 kPa. To determine the saturated temperature in
the ET, the outlet pressure of the ET was calculated using Hall Inlet and outlet pressures
and Mudawar’s correlation [17]. The experimental conditions are sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2  2  2
listed in Table 1. The surface roughness of the ET was measured dP @P dV P @P dV AMP
by a HANDYSURF-E35-A profilometer. ¼ þ ð16Þ
P @V P VP @V AMP V AMP
The relative uncertainties were estimated using ANSI/ASME PTC
19.1 [18] guidelines as follows: Mass flow rate
Heat generation rate sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2  2  2
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
! !2   dG @G dV G @G dV AMP
u ¼ þ ð17Þ
dQ_ u @ Q_ 2 dV 2 @ Q_
2
G @V G VG @V AMP V AMP
¼ t R
þ
dV I
ð6Þ
Q_ @V R VR @V I VI
The maximum relative errors of these measurement parameters
are summarized in Table 2. Since the sound measurement is
Heat flux
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi affected by the wavelength, the uncertainties of the frequency
u 2 !2  2  2 responses were shown in Appendix A.2.
u
dq t @q dQ _ @q dT a
¼ þ ð7Þ
q @ Q_ Q_ @T a Ta
3. Experimental method
Average temperature
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi The experiment was conducted at various experimental condi-
 2  2  2  2
dT a @T a dV T @T a dRT tions described in Table 1. Deionized (DI) water was used in the
¼ w21 þ w22 ð8Þ
Ta @V T VT @RT RT experiment. The water flowed through an ion exchanger at least
30 min before the experiment. The DI water was deaerated by boil-
where
ing it at around atmospheric pressure in the pressurizer, where it
U RSS; V was subsequently pressurized up to the prescribed pressure. After
w1 ¼ ð9Þ
U RSS; V þ U RSS; R the pressurization in the water loop, the DI water was supplied by
a circulation pump, through the preheater, to the test section.
w2 ¼ 1  w1 ð10Þ When the DI water flowed to the prescribed temperature and pres-
sure, the heat input provided from the DC power supply was
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2 applied to the ET. As the heat flux of the ET increased with the
U RSS; V T bV T sV T
¼ þ t95 ð11Þ increasing electric current, nucleate boiling began. When the heat
VT VT VT
flux reached a critical heat flux, measurements such as the average
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi temperature of the ET, the flow velocity of DI water, the inlet and
 2 s 2
U RSS;R bR R outlet pressures were stopped at each experimental condition.
¼ þ t 95 ð12Þ Fig. 2 shows the schematic of the test section. The boiling incip-
R R R
ience was detected by a microphone (13.2 mm dia.) in the sound
level meter (RION NL-42) during the experiment. The microphone
was placed across the ET. The distance between the microphone
and the ET was approximately 20 mm and the sensitivity of the
sound measurement are shown in Appendix A.3. The height of
Table 1 the microphone was approximately 10 mm from the outlet of the
Experimental conditions. ET since the location of boiling incipience was estimated around
Heater material Platinum
Inner diameter 1.0 mm
Table 2
Outer diameter 1.8 mm
Measurement uncertainties.
Heated length 23.2 mm
Surface roughness Ra 0.31 lm Variables Uncertainties
Ry 1.48 lm
Rz 1.10 lm Heat generation rate ±2.0%
Heat flux ±2.4%
e-folding time 9, 18 s Average temperature ±2.8%
Flow velocity 2.5–14 m/s Inner surface temperature ±2.8%
Inlet temperature 285–349 K Inlet and outlet temperatures ±0.7 K
Inlet pressure 238–855 kPa Inlet and outlet pressures ±2.6 kPa
Outlet pressure 213–827 kPa Mass flow rate ±0.011 kg/min
M. Shibahara et al. / Applied Thermal Engineering 132 (2018) 595–604 599

Microphone Electrode (-) Microphone Electrode (-) Cu


Cu

10
20

Pt Pt
Electrode Electrode(+)
(+) Cu Cu

Bakelite

Bakelite

(a) Perspective view (b) Side view


Fig. 2. Schematic of the test section.

8 8
10 10
Pt Pt
d = 1.0 mm Tsub, in = 97 K d = 1.0 mm Tsub,in = 118 K
L/d = 23.2 Pout = 229 kPa L/d = 23.2 Pout = 415 kPa
u = 7 m/s u = 7 m/s CHF
CHF
q [W/m2]

q [W/m2]

7 7
10 10

Experimental data Experimental data


Eq.(20) Eq.(20)
Incipient Boiliing point Incipient Boiling Point
6 6
10 10
2 2
1 10 10 1 10 10
Tsat [K] Tsat [K]

(a) (b)
Fig. 3. Typical boiling curves at outlet pressures of (a) 229 kPa and (b) 415 kPa.

11 mm from the thermal entry region of the ET based on the exper- 4. Results and discussion
imental result. Then, the boiling spectrum was analyzed by WDM
using a commercial application from the MATLAB wavelet tool Fig. 3 shows typical boiling curves at outlet pressures of (a) 229
box [19]. The WDM was proposed by Mallat [16]. The advantages kPa and (b) 415 kPa and a 7 m/s flow velocity. The surface super-
are the de-noising of the signal from pump noise and the separa- heat DTsat, ascends with an increase in heat flux q, along the fol-
tion of the boiling sound. The discrete wavelet transform of a lowing convection correlation [14], until the incipient boiling
square integral function can be expressed as point is reached:

wj;k ðtÞ ¼ 2j=2 wð2 j t  kÞ ð18Þ  0:08  0:14


L l
Nu ¼ 0:001Re1:2 Pr0:4 ð20Þ
where, w and j are a wavelet and a decomposition level, respec- d lw
tively. The Daubechies’ order of 4 was used in this study. The boiling As shown in Fig. 3(a), after boiling inception occurs, DTsat
signal f(t), can be decomposed as decreases, then fluctuates slightly due to nucleate boiling, and sub-
sequently increases again with an increase in q. We believe that the
X X
0 X
f ðtÞ ¼ aj;k uj;k ðtÞ þ dj;k wj;k ðtÞ ð19Þ initial temperature overshoot is due to the activation of neighbor-
k j¼J k ing large cavities that are initially flooded [20]. As DTsat increases, it
seems that cavities whose mouth radii are smaller begin to serve as
where, aj, /(t), J and dj, denote, respectively, the approximation bubble nuclei and the number of nucleation sites on the inner
coefficient at the decomposition level of j, the scaling function, surface increases. Then, originally flooded cavities of mouth radii
the maximum decomposition level of signal, the detail coefficient larger than that of the maximum unflooded cavities will be acti-
at the decomposition level of j [19]. vated by the bubbles originating from the neighboring cavities.
600 M. Shibahara et al. / Applied Thermal Engineering 132 (2018) 595–604

Fig. 4. Detection of the boiling incipience point by WDM at a flow velocity of 7 m/s and an outlet pressure of 229 kPa.

8
10
8 10
30 %
Pt 30 % Pt
d = 1.0 mm d = 1.0 mm
L/d = 23.2 L/d = 23.2
u 2.5-14 m/s -30 % u 2.5-14 m/s -30 %
7 7
[W/m ]

10
[W/m ]

10
2
2

i,exp
i,exp

q
q

6 6 Pout
10 Pout 10
825 kPa 825 kPa
412 kPa 621 kPa
621 kPa 412 kPa
232 kPa 232 kPa

5 5
10 10
5 6 7 8 5 6 7 8
10 10 10 10 10 10 10 10
2 2
qi,BG [W/m ] qi,SM [W/m ]

(a) (b)
Fig. 5. Comparison between the experimental data and proposed ONB correlations of (a) Bergles and Rohsenow [8] and (b) Sato and Matsumura [9].

DTsat would decrease with an increase in the number of activated corresponds to the onset of d2 at various experimental conditions,
cavities. Boiling from originally unflooded cavities would be super- such as the liquid subcooling and the flow velocity. These heat flux
seded by the boiling from the activated cavities, which require and surface temperatures at the incipient boiling point are defined
smaller DTsat to maintain boiling. After the initial temperature as qi and Ti.
overshoot, nucleate boiling developed, accompanied with the acti- Fig. 5 shows comparisons between the measured qi,exp and the
vation of neighboring initially flooded large cavities, and q reached proposed onset nucleate boiling (ONB) correlations of (a) Bergles
a maximum value. Namely, the maximum heat flux reached the and Rohsenow [8] in Eq. (21) and (b) Sato and Matsumura [9] in
so-called well-known critical heat flux (CHF). Eq. (22), defined as follows:
On the contrary, in Fig. 3(b), the initial temperature overshoot
0:0234
does not appear as the system pressure increases. Since the activa- qi ¼ 5:3P1:156 ½1:8ðT i  T sat Þ2:41=P ð21Þ
tion of neighboring initially flooded large cavities is dependent on
the experimental condition, (i.e., system pressure, subcooling, flow
velocity, and surface condition) the prediction of the boiling incip- kf hfg ðT i  T sat Þ2
qi ¼ ð22Þ
ience is complicated. 8rT sat ðv g  v f Þ
Fig. 4 shows a typical time variation of a heat flux q, a surface
temperature Ts, and a detail coefficient d2 of wavelet transform at where P, Ti, Tsat, kf, hfg, r, and v are the system pressure in kPa, the
a decomposition level of 2. As shown in Fig. 4, Ts gradually surface temperature at the boiling initiation, the saturated temper-
increases with an increase in q, which changes at around 85 s ature, the thermal conductivity, the latent heat of vaporization, the
due to the boiling incipience, corresponding to d2. Since the incip- surface tension, and the specific volume, respectively. As shown in
ient boiling point is often defined as the discrepancy point of the Fig. 5, these correlations underestimated the measured qi in the
convective correlation, we confirmed that the discrepancy point experiment.
M. Shibahara et al. / Applied Thermal Engineering 132 (2018) 595–604 601

Kennedy et al. [21] and Ghiaasiaan and Chedester [10] also 10


3

compared their experimental data and the correlations proposed Pt


by (a) Bergles and Rohsenow [8] and (b) Sato and Matsumura = 76-156 K
d = 1.0 mm sub,in
[9]. Since the predicted values of these correlations show a similar
2
trend to Fig. 5, Ghiaasiaan and Chedester [10] modified the Sato 10 L/d = 23.2
and Matsumura correlation using the ratio of the thermocapillary Pout
and aerodynamic forces, as follows:
825 kPa
kf hfg ðT i  T sat Þ2 10 621 kPa
qi ¼ ð23Þ 412 kPa
C8rT sat ðv g  v f Þ
232 kPa

C
C ¼ 22f0:765 ð24Þ 1

where f is the ratio of the thermocapillary forces and aerodynamic


forces, given as
rf  rw 10
-1 Ghiaasiaan and
f¼ ð25Þ Chedester [10]
qf u2 R
Eq. (24)
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2rT sat ðv g  v f Þkf
-2

10
R ¼ ð26Þ 10
-4
10
-3
10
-2 -1
10 1 10
qi hfg

where R⁄ is the critical cavity radius that satisfies the surrounding


Fig. 7. Relationship between C and f.
superheated liquid layer. A linear temperature profile was used
with boundary conditions obtained using the Clausius-Clapeyron
relation, as follows:
the dashed line shows their correlation calculated from Eq. (24),
Tjy¼yB ¼ T B ð27Þ while the filled symbols indicate the authors’ experimental data.
As the outlet pressure of the tube increases, C decreases with a
 decrease in f. It was found that C is dependent on f and Pout, as
@T  @T B
¼ ð28Þ
@y y¼yB @RB
shown in Fig. 7.
Since the value of C depends on Pout at the same f, the relation-
The restriction for the correlation is C = 1. ship between an exponent of f and the reduced pressure Pre, is plot-
Fig. 6 shows a comparison between the Ghiaasiaan and Chede- ted in Fig. 8. The exponent, n, increases with an increase in Pre, thus,
ster correlation in Eq. (23) and the experimental data. C in Eq. (24) the following correlation can be obtained by the LS method.
was calculated using Eqs. (25) and (26) with the qi measured in the
experiment. Although the predicted values of the correlation are n ¼ 8:3P0:5
re ð29Þ
improved compared with those of Fig. 4, the accuracy of the corre-
Fig. 9 shows the relationship between the values of lnðCÞ=8:3P 0:5
re
lation is insufficient because C was lower than 1.0. Hence, their
correlation is not applicable to this experiment. and f. In this figure, the values of lnðCÞ=8:3P0:5re increase with
Fig. 7 shows the relationship between C and f in narrow tubes. increasing f. Consequently, the empirical correlation was obtained
The open symbols depict Ghiaasiaan and Chedester’s data [10] and as follows:

8 10
10
30 % Pt
Pt d = 1.0 mm Pout 232-825 kPa
d = 1.0 mm
L/d = 23.2
L/d = 23.2
-30 % Eq. (29)
u 2.5-14 m/s
7
10
[W/m ]
2

1
n
i,exp
q

6
10 Pout
825 kPa
621 kPa
412 kPa
232 kPa

5
10
5 6 7 8
10 10 10 10
0.1
2 0.01 0.1
qi,GC [W/m ]
Pre
Fig. 6. Comparison of the Ghiaasiaan and Chedester correlation [10] and the
current experimental data. Fig. 8. Relationship between an exponent of f and Pre.
602 M. Shibahara et al. / Applied Thermal Engineering 132 (2018) 595–604

10 8
10
Pt Pt 30 %
d = 1.0 mm sub,in
= 76-156 K d = 1.0 mm
L/d = 23.2 L/d = 23.2
5 -30 %
= 9, 18 s u 2.5-14 m/s
7
10
Pout
ln(C)/8.3Pre0.5

[W/m ]
2
825 kPa
621 kPa
0
412 kPa

i,pred
232 kPa

q
Pout 10
6

825 kPa
-5
621 kPa
412 kPa
232 kPa
Eq. (30) 5
10
-10 10
5
10
6
10
7
10
8
-10 -8 -6 -4 -2 0
2
ln qi,exp [W/m ]

Fig. 9. Relationship between ln(C)/8.3Pre0.5 and f. Fig. 10. Comparison of the empirical correlation and the experimental data.

lnðCÞ ¼ 8:3P0:5
re ð2:89 þ lnðf
0:87
ÞÞ ð30Þ Table 3
MAE of the correlations.

This leads to Correlation Ref. MAE (%)


pffiffiffiffiffi Bergles and Rohsenow [8] 52.3
8:3 P re
C ¼ ð18f0:87 Þ ð31Þ Sato and Matsumura [9] 58.6
Ghiaasiaan and Chedester [10] 48.0
Eq. (31) is substituted into Eq. (23), then Ti can be obtained Present study 26.5
since qi can be calculated by the experimental range.
A comparison between this correlation and the experimental
data is plotted in Fig. 10. The predicted value of the correlation is To evaluate the correlations of references [8–10] and present
in agreement with the experimental data within ±30%. The excess the study quantitatively, the following mean absolute error
errors of the correlation are due to the initial temperature over- (MAE) formula was used and the MAE of each correlation is listed
shoots, as referenced in Fig. 3. The accuracy of the correlation could in Table 3. The correlation obtained by Eqs. (23) and (31) is consid-
be improved to take into account the influence of initial tempera- ered to be suitable to express the experimental data in this
ture overshoots. research.

Reference sound Reference sound


o
180 180
o

o o
0 210 150 0 210
o
150
o

-5 -5
o
120 o
dB

120
dB

-10 -10

-15 -15

-20 90
o -20 o
90

o o o o
300 60 300 60

1 kHz 1 kHz
2 kHz 2 kHz
o o o o
330 30 4 kHz 330 30 4 kHz
o
8 kHz o
8 kHz
0 12.5 kHz 0 12.5 kHz

(a) Horizontal direction (b) Vertical direction


Fig. 11. Directional characteristics of the sound measurement on the reference sound.
M. Shibahara et al. / Applied Thermal Engineering 132 (2018) 595–604 603

Fig. 12. FFT profile for the sound waves at a flow velocity of 7 m/s.

Conflict of interest
Table 4
Total expanded uncertainty of the frequency response [23].
On behalf of all authors, the corresponding author states that
Nominal frequency (Hz) Total expanded uncertainty (dB)
there is no conflict of interest.
63 0.3
80 0.3
100 0.3
125 0.3 Appendix A
160 0.3
200 0.2 A.1. Calculations of the surface temperature and the heat loss from the
250 0.2
outer surface of the ET
315 0.2
400 0.2
500 0.2 The temperature distribution of the ET can be obtained by solv-
630 0.2 ing the conduction equation in steady state (see Eq. (2)) as follows:
800 0.2
1000 0.2
1250 0.3 Q_ r 2 Q_ r2o
TðrÞ ¼  þ lnr þ T a
1600 0.3 4k 2k
2000 0.3
    
qr i 1 1
2500 0.3  2
4r 2
o r 2
o lnr o   r 2
i lnr i   ðr 4
o  r 4
i Þ
3150 0.3 4ðr2o  r 2i Þ k 2 2
4000 0.4
ðA1Þ
5000 0.4
6300 0.4 The inner and outer surface temperatures of the ET are calcu-
8000 0.4
lated by the following equations:

    
! qri 1 1
1 X jqi;pred  qi;exp j T s ¼ Tðr i Þ ¼ T a    
2 2 2
2
4r o r o ln r o r i ln r i
MAE ¼  100 ð32Þ 4ðr 2o  r 2i Þ k 2 2
N qi;exp qr i
ðr4o  r 4i Þ  ðr2  2r 2o ln ri Þ ðA2Þ
2ðr 2o  r 2i Þk i

5. Concluding remarks
    
qr i 1 1
T o ¼ Tðro Þ ¼ T a  2
 4r 2o r2o lnr o   r 2i lnr i 
Various incipient boiling phenomena for subcooled water flow- 4ðr2o  r 2i Þ k 2 2
ing in a uniformly heated narrow tube were observed experimen- qri r2o
tally. The boiling signal was analyzed using wavelet decomposition ðr4o  r 4i Þ  ð1  2lnr o Þ ðA3Þ
2ðr 2o  r2i Þk
method. The results showed that the initial temperature overshoot
did not appear as the outlet pressure increased. It was found that The radiation loss at the boiling incipience was calculated by
the incipient heat flux is dependent on pressure and non- the Stefan-Bolzmann law as follows:
dimensional parameter expressed by a ratio of thermocapillary
forces and aerodynamic forces (f). Since existing correlations Q r ¼ rsf e½T 4o  T 4sur pdo L ðA4Þ
underestimated the incipient heat flux, a semi-empirical correla-
tion of the boiling incipience was obtained based on the experi- where rsf, e, To, Tsur and do denote, respectively, the Stefan-
mental data. The predicted value of the new correlation is in Bolzmann constant, the emissivity of platinum (=0.0536 at 505.6
agreement with the experimental data within ±30%. K [22]), the outer surface temperature, the surrounding tempera-
ture (=283.65 K) and the outer diameter (=0.0018 m). The total heat
transfer rate at the boiling incipience can be estimated by the fol-
Acknowledgements
lowing equation:

This work was partly supported by the JSPS KAKENHI Grant


Q total ¼ Q i þ Q r ðA5Þ
Numbers JP16K18322 and JP15K05828. This work was partly per-
formed with the support and under the auspices of NIFS Collabora- Since the ratio of the Qr and Qtotal was 0.002%, the heat loss was
tion Research program (NIFS17KEMF100). negligible in the experiment.
604 M. Shibahara et al. / Applied Thermal Engineering 132 (2018) 595–604

A.2. The uncertainty of the frequency response for the sound [6] N. Sider, E, Heat transfer and pressure drop of liquids in tubes, Ind Eng Chem.
28 (1936) 1429–1435.
measurement
[7] G.L. Morini, Single-phase convective heat transfer in microchannels: a review
of experimental results, Int. J. Therm. Sci. 43 (2004) 631–651, https://doi.org/
Table 4 shows the uncertainty of the frequency response [23] 10.1016/j.ijthermalsci.2004.01.003.
for the microphone (MIC) and the sound level meter (SLM) based [8] W.M. Bergles, A.E. Rohsenow, The determination of forced-convection
surfaceboiling heat transfer, J. Heat Trans., Trans. ASME, Ser. C. 86 (1964).
on JIS C 1509-1 [24]. The total expanded uncertainty can be [9] H. Sato, T. Matsumura, On the conditions of incipient subcooled-boiling with
calculated as follows: forced convection, Bull. JSME. 7 (1963) 392–398.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi [10] S.M. Ghiaasiaan, R.C. Chedester, Boiling incipience in microchannels, Int. J.
Heat Mass Transf. 45 (2002) 4599–4606, https://doi.org/10.1016/S0017-9310
U x ¼ k x2MIC þ x2SLM þ x2WSE þ x2WSC ðA6Þ (02)00167-9.
[11] F. Inasaka, H. Nariai, T. Shimura, Pressure drops in subcooled flow boiling in
where x is an uncertainty of frequency response and k is used as a narrow tubes, Heat Transf. Jpn. Res. 18 (1989) 70–82.
[12] E.J. Davis, G.H. Anderson, The incipience of nucleate boiling in forced
covering factor to have a level of confidence of approximately 95%.
convection flow, AIChE J. 12 (1966) 774–780, https://doi.org/10.1002/
aic.690120426.
[13] M. Shibahara, K. Fukuda, Q.S. Liu, K. Hata, Correlation of high critical heat flux
A.3. The sensitivity of the sound measurement
during flow boiling for water in a small tube at various subcooled conditions,
Int. Commun. Heat Mass Transf. 82 (2017) 74–80, https://doi.org/10.1016/j.
Fig. 11 shows the directional characteristics of the sound icheatmasstransfer.2017.02.012.
measurement on the reference sound [23]. The sensitivity for [14] M. Shibahara, K. Fukuda, Q.S. Liu, K. Hata, Steady and transient forced
convection heat transfer for water flowing in small tubes with exponentially
sound waves is affected by the horizontal and vertical directions increasing heat inputs, Heat Mass Transf. 53 (2017) 787–797, https://doi.org/
as the frequency increases. However, for the boiling incipience 10.1007/s00231-016-1860-z.
sounds, since the spectrum peak of the sound waves analyzed by [15] M. Shibahara, K. Fukuda, Q.S. Liu, K. Hata, Effects of outlet subcoolings and heat
generation rates on transient critical heat flux for subcooled flow boling of
Fast Fourier Transform (FFT) was within 1 kHz as shown in water in a vertical tube, Heat Mass Transf. (2017), https://doi.org/10.1007/
Fig. 12, the influence of the measurement location for the micro- s00231-017-2036-1.
phone is negligible in the experiment. [16] S.G. Mallat, A theory for multiresolution signal decomposition: the wavelet
representation, IEEE Trans. Pattern Anal. Mach. Intell. II (11) (1989) 674–693.
[17] D.D. Hall, I. Mudawar, Critical heat flux (CHF) for water flow in tubes—II, Int. J.
References Heat Mass Transf. 43 (2000) 2605–2640, https://doi.org/10.1016/S0017-9310
(99)00192-1.
[18] ANSI/ASME PTC19.1-1985, Measurement Uncertainty, Supplement on
[1] I. Mudawar, Two-phase microchannel heat sinks: theory, applications, and
Instruments and Apparatus, Part 1, 1987.
limitations, J. Electron. Packag. 133 (2011) 41002, https://doi.org/10.1115/
[19] M. Misiti, Y. Misiti, G. Oppenheim, J.-M. Poggi, Wavelet Toolbox TM User’s
1.4005300.
Guide, MathWorks, 2017.
[2] B.J. Baliga, B.J. Baliga, Chapter 9 – IGBT Applications: Transportation, in: IGBT
[20] A. Sakurai, M. Shiotsu, Transient pool boiling heat transfer, Part 2, boiling heat
Device, 2015: pp. 223–275. 10.1016/B978-1-4557-3143-5.00009-2.
transfer and burnout, ASME J. Heat Transf. 99 (1977) 554–560.
[3] A.R. Raffray, R. Nygren, D.G. Whyte, S. Abdel-Khalik, R. Doerner, F. Escourbiac,
[21] J.E. Kennedy, G.M. Roach, M.F. Dowling, S.M. Ghiaasiaan, The onset of flow
T. Evans, R.J. Goldston, D.T. Hoelzer, S. Konishi, P. Lorenzetto, M. Merola, R. Neu,
instability in uniformly heated horizontal microchannels, Trans. ASME. 122
P. Norajitra, R.A. Pitts, M. Rieth, M. Roedig, T. Rognlien, S. Suzuki, M.S. Tillack, C.
(2016) 118–125.
Wong, High heat flux components—readiness to proceed from near term
[22] Y.S. Touloukian, D.P. DeWitt, Thermophysical Properties of Matter, Thermal
fusion systems to power plants, Fusion Eng. Des. 85 (2010) 93–108, https://doi.
Radiative Properties, Metallic Elements and Alloys, IFI/Plenum, New York-
org/10.1016/j.fusengdes.2009.08.002.
Washington, 1970.
[4] F.W. Dittus, L.M.K. Boelter, Heat transfer in automobile radiators of the tubular
[23] RION CO. LTD., Technical Notes Sound Level Meter NL-42/NL-52, No.55755 17-
type, Int. Commun. Heat Mass Transf. 12 (1985) 3–22, https://doi.org/10.1016/
10.
0735-1933(85)90003-X.
[24] Japanese industrial standards committee, JIS C 1509-1. <http://www.jisc.go.jp/
[5] V. Gnielinski, New equations for heat and mass transfer in turbulent pipe and
app/jis/general/GnrJISSearch.html>, 2017 (accessed 8 December 2017).
channel flow, Int. Chem. Eng. 16 (1976) 359–368.

You might also like