Chemical Engineering Science 181 (2018) 326–340

Contents lists available at ScienceDirect

Chemical Engineering Science

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

A numerical investigation on heat transfer dynamics of a periodically

forced boiling channel in low amplitudes
Swapan Paruya ⇑, Jyoti Bhati, Tanusree Ghoshal, Bitan Mankar
Department of Chemical Engineering, NIT Durgapur, 713209, India

h i g h l i g h t s

Mechanism of wall heat transfer dynamics in a vertical boiling channel under periodic forcing.

Role of chaotic bubble dynamics and bubble mixing in wall heat transfer.

Effect of forcing amplitude and frequency on boiling heat flux.

Effect external flux on wall heat flux.

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

Article history: In this article, the authors report wall-heat transfer characteristics of a periodically forced vertical boiling
Received 10 November 2017 channel with the help of numerical simulation based on four-equation drift flux model. The periodic forc-
Received in revised form 26 January 2018 ing on the boiling channel was initiated by the sinusoidal perturbation of the inlet velocity. The effect of
Accepted 20 February 2018
the frequency x and the amplitude ã of the forcing on the heat transfer enhancement were studied to
formulate a mechanism for the enhancement of wall heat flux q00w . q00w increases with ã in the range of
x less than a critical value xcrit. The comparison of the numerical solution and the asymptotic solution
Keywords:
of a reduced-order model through harmonic balance method (HBM) shows that both the solutions for
Heat transfer
Subcooled boiling
the amplitude of exit mixture enthalpy decrease with x consistently and asymptotically, when x is
Boiling channel greater than the characteristic frequency xc of the channel. The enhancement is also facilitated by chaotic
Periodic forcing instability (mixing) at higher ã and higher external heat flux.
Drift flux model Ó 2018 Elsevier Ltd. All rights reserved.

1. Introduction economic return. Jet impingements and liquid droplet spray are
being successfully used in metal industries for high heat removal.
Studies on boiling heat flux from the heated surface under flow The effect of flow oscillations on the wall heat flux in single-
oscillations or unstable conditions have been an interesting subject phase and two-phase flow, studied by many authors (West and
of research in many applications which involve transfer of a large Tayor, 1952; Baird et al., 1966; Cheng and Zhao, 1998; Bouhadji
amount of heat. In particular, the applications include fast heat and Djilali, 2003; Chen et al., 2010; Okawa et al., 2010; Wang
removal from nuclear fuel rods (pins) in maintaining thermohy- et al., 2013; Chen et al., 2016; Paruya et al., 2016) causes to facili-
draulic safety of nuclear reactors (Tong, 1967; Bergles, 1999; tate the heat transfer. West and Taylor reported 60–70% increase in
Kakac, 1999; Sinha and Kakodkar, 2006; Saha et al., 2013) and cool- single-phase heat transfer rate in pulsating flow. Cheng and Zhao
ing of extremely hot bodies in metal industries (Chattopadhyay concluded that pulsating flow or periodic flow could cause a signif-
et al., 2002; Warrier and Dhir, 2013). Subcooled nucleate boiling icant increase of single-phase heat transfer. Bouhadji and Djilali
is practiced in the water-cooled nuclear reactors for safe and fast performed a two-dimensional simulation of the unsteady sepa-
heat removal despite the highly complex mechanism of the sub- rated flow over a bluff rectangular plate subjected to an oscillatory
cooled boiling. High fuel-surface heat flux can be facilitated by sev- bulk flow over a broad range of forcing amplitudes and frequencies
eral heat transfer mechanisms (Tong, 1967). Improving the heat up to the 60th harmonic of the natural vortex shedding frequency.
transfer in these applications adds an appreciable amount of They showed that the periodic forcing had a significant impact on
the heat transfer and an appreciable enhancement of local Nusselt
number. Chen et al. (2010) experimentally investigated the effect
⇑ Corresponding author. of the forced periodic flow rate of refrigerant using a triangular
E-mail address: [email protected] (S. Paruya). wave function on flow boiling heat transfer characteristics and

https://doi.org/10.1016/j.ces.2018.02.037
0009-2509/Ó 2018 Elsevier Ltd. All rights reserved.
 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340 327

Nomenclature

a interfacial heat transfer area, m2/m3 x vapour quality

a2F fraction of wall area occupied by bubbles z axial distance, m
an surface density of active nucleation sites, m2
~
a amplitude of velocity perturbation, m/s Greek letters
Co void distribution parameter a void fraction
cf correction factor D difference
dbw bubble departure diameter, m C boiling rate, kg/s-m3
db mean bubble diameter, m Ch rate of vaporization due to heating, kg/s-m3
D diameter of channel, m Cf rate of vaporization due to flashing, kg/s-m3
fbd bubble departure frequency, s1 x cyclic frequency of forcing signal, Hz
g acceleration due to gravity, m/s2 u contact angle, degree
G mass flux, kg/m2-s q density, kg/m3
Gc condensation rate, kg/s-m3 w heat transfer coefficient, W/m2-K
Gc space-averaged condensation rate, kg/s-m3 r surface tension, N/m
H height of boiling channel, m s averaging time, s
h enthalpy, J/kg 0 inhibition factor
j volumetric flux of mixture or superficial velocity, m3/
m2-s
Subscripts
k thermal conductivity, W/m-K b bubble
L length, m c single-phase convection/condensation/characteristic
p local pressure, Pa crit critical
Pr Prandtl number
e exit/equilibrium
q00w wall heat flux, W/m2 eV evaporative
q00w space-averaged wall heat flux, W/m2 g gas/vapour
q00 uniform external heat flux to the wall, W/m2 gl vapour to liquid
00
qc single-phase heat flux, W/m2
00 htf heat flux
qe evaporative heat flux, W/m2 i inlet
q00q quenching heat flux, W/m2 lg liquid to vapour
q000 volumetric heating rate, W/m3
ld OSV point
R space-averaged and time-averaged ratio m mixture
Re Reynolds number q quenching
t time, s sat saturation
T temperature, K
 s steady state
T space-averaged temperature, K sub subcooling
u phase velocity, m/s w wall
W mass flow rate, kg/s

the associated bubble characteristics of refrigerant R-134a in a hor- The above survey indicates that a large number of studies on
izontal narrow annular duct. They observed aggressive oscillations single-phase heat transfer subjected to oscillating flow have been
in the bubble characteristics at higher amplitude of forcing. The carried out. However, to the authors’ knowledge, the numerical
forced oscillation caused to dominate the size of the departing studies and the experimental studies on promoting wall heat
bubble and active nucleation site density over the bubble depar- flux in boiling flow by the periodic flow perturbation have not
ture frequency. Wang et al. conducted an experimental investiga- matured; the evidence of numerical study in this context, in par-
tion to see the effect of periodic forcing in the form of a ticular, was not found in literature. The mechanism for the
triangular wave on subcooled flow boiling heat transfer and the enhancement of heat flux by periodic forcing still remains
associated bubble characteristics of refrigerant R-134a in a hori- unclear due to complex bubble dynamics. In this work, we
zontal narrow annular duct. They noted that the heated surface examine the influence of the periodic flow perturbation with
temperature, bubble departure diameter, bubble frequency and low forcing amplitudes on the dynamics of boiling heat transfer
active nucleation site density also oscillated strongly with large from the heated wall of a vertical boiling channel. The high forc-
amplitudes. Okawa et al. (2010) showed that the periodic oscilla- ing amplitudes would make the channel unstable or would seri-
tion facilitated the axial liquid transport to the film region, which ously affect the critical heat flux (CHF). This research proposal
subsequently enhanced CHF and dryout. Chen et al. (2016) exper- originated from the study of Paruya et al. (2016) on frequency
imentally examined the effect of periodic flow on bubble charac- entrainment during density wave oscillations in the boiling
teristics to show the increase of active nuclear site density, channel, in which the authors claimed the possibility of augmen-
bubble diameter and bubble frequency with forcing amplitude of tation of boiling heat transfer at resonance frequencies. We have
heat flux. Paruya et al. (2016) studied the effect of the periodic numerically studied the effect of the sinusoidal variation of the
forcing on oscillating boiling flow in low forcing frequencies using inlet velocity on heat transfer enhancement during subcooled
numerical simulation and harmonic balance. They showed aggres- boiling based on a numerical simulation. The parametric varia-
sive oscillations with high amplitude leading to frequency entrain- tions of the frequency and the amplitude of the forcing on the
ment at a lower frequency depending on inlet subcooling and heat flux augmentation allow us to propose the mechanism of
external heat flux. The high-amplitude oscillations facilitate mix- enhancement. The effect of external heat flux also contributes
ing and heat transfer in boiling. to reinforcing the mechanism.
328 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340

2. Model descriptions port of heat, mass and momentum was neglected; heat loss to
the ambient air was also neglected. The required four boundary
Four-equation drift flux model (DFM4) (Zuber and Findlay, conditions are shown in Fig. 1. The inlet boundary conditions are
1965; Ishii (1977)) was used to model the two-phase flow dynam- (i) inlet subcooling DTsub (58.2 K); (ii) inlet mass flux velocity (G
ics in a vertical boiling channel (refer Fig. 1) to obtain local mixture = 900 kg/m2-s); (iii) inlet void fraction (a = 0.0). The outlet bound-
enthalpy hm, local pressure p, local mixture volumetric flux or ary condition is the pressure (p2 = 45 bar).
superficial velocity j and local void fraction a from mixture energy
equation, mixture momentum equation, mixture continuity equa- 2.1. Numerical modeling
tion and vapour-phase continuity equation. The liquid temperature
was estimated from hm. The model also takes into account of the Mixture energy balance:
thermal inertia of heated wall by solving an unsteady heat balance
@p @hg @p @hg @p
equation for the heated wall. We used the thermo-physical proper- ðhg  hl ÞC ¼ q000 þ  aqg  qg jg  ql ð1  aÞ
ties of metal wall in the model. In absence of the heated-wall @t @p @t @p @z
dynamics, the model would result in faster dynamics. dhl @p dhl @p
  j1 q l ð1aÞ
DFM4 essentially considers the two-phase flow to be a mixture dp @t dp @z
flow for which the mixture momentum equation along with the
Eq. (1a) can be expanded as:
relative motion between the phases. Thus, DFM4 significantly

 
reduces the difficulties and numerical instabilities associated with @p @hg hlg
hlg C  q000 ¼ 1  aqg  a
two-fluid model, in which the phases are treated separately with @t @p hfg
larger number of constitutive relations at interfaces (Ishii and   
@hm dhl @ a dh
Hibiki (2006)). The constitutive relations are still under investiga- þ ql ð1  aÞ þ ql ð1  aÞ l
@t dhm @t da
tion to obtain the unified ones. A large number of studies related    
@p @hg hlg @hm dhl
to the boiling phenomena in the nuclear reactors have been carried þ qg jg  b þ jl ql
out experimentally and theoretically. Even though, there is a need @z @p hfg @z dhm
 
for an efficient, accurate and stable algorithm to be used for @a dhl
reactor-dynamics simulation for severe malfunctions (Nayak
þ ql ð1bÞ
@z dhm
et al., 2007; Talebi et al., 2012). The low-order models with suffi-
cient accuracy are quite efficient in real-time simulations and con- @hg @hf
trol as well. a ¼ 1  aqg  ð1  aÞqf ð1cÞ
@p @p
In the present formulation, one-dimensional flow is assumed to
occur. Vapour phase was assumed to be saturated (Končar and

@hg @hf
Kljenak, 2002; Krepper et al., 2007); this assumption is supported b ¼  qg jg þ q f jf ð1dÞ
@p @p
by the fact that the possibility of vapour superheating is very little
during subcooled boiling because of bubble condensation. The
hm ¼ ð1  aÞhl þ ahg ð1eÞ
model has also been extended to incorporate the effect of
thermal-nonequilibrium (it is always present in subcooled boiling)
hlg ¼ hg  hl ð1fÞ
using the correlations of Saha and Zuber (1974). Molecular trans-
The symbols h, p, a, q and j are enthalpy, pressure, void fraction,
density and mixture volumetric flux or superficial velocity respec-
tively. The subscript f, g, l and m stand for saturated liquid, satu-
rated vapour, subcooled liquid and two-phase mixture
respectively.
C is boiling rate (kg/m3-s) and is the sum of the vaporization
rate due to heating Ch and the vaporization rate due to flashing Cf .
C ¼ Ch þ Cf ð1gÞ


1 @p @p
Cf ¼ a þb ð1hÞ
hfg @t @z
2 hf hl
3
q000 4 1  hf hld 5
Ch ¼ ðfor subcooled boilingÞ ð1iÞ
hfg 1 þ qf ðhf hl Þ
qg hfg

q000
Ch ¼ ðfor bulk boilingÞ ð1jÞ
hfg
Mixture momentum equation:

@p h i
4f q jv j

j j

  g ql þ aðqg  ql Þ  m m
vm  C g  l
@z Dh 2 a 1a
 
jl
@ 1a



@ jl @ jg @ jg
¼ ql ð1  aÞ þ jl ql þ aqg þ qg jg
@t @z 1  a @t a @z a
ð2aÞ
Fig. 1. Vertical boiling channel in detail. The above equation is expanded as:
 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340 329


 
@j @C @V space-derivatives of jth element of Yk in ith equation; (i, j) of Ck
aðqg  ql Þ j o þ gj þ aC o ðqg  ql Þ þ ql
@t @j @j indicate the coefficients of jth element of Y in ith equation; Dk is


@hm @C o @V gj constant in ith equation. Note that i = 1, 2, 3 and 4 are for Eq.
þ aðqg  ql Þ j þ
@t @hm @hm (1b), (2b), (3b) and (4b) respectively, and j = 1, 2, 3 and 4 are for


 hm, p, j and a respectively. It is required to mention that in the bulk
@p @C o @V gj a jg j
þ aðqg  ql Þ j þ þ  l boiling region, both the phases are at saturation; the energy bal-
@t @p @p hfg a 1  a


  ance equation becomes redundant leading to Ak, Bk, Ck 2 R33, Dk
@a @C o @V gj jg jq 2 3
þ aðqg  ql Þ j þ þ qg C o þ V gj  þ l l p
@t @a @a a 1a


  2R 31
for kth node and Y k ¼ 4 j 5 .
@j ajl ql @C o @V gj jl ql a
þ q g jg  j þ þ ð1  aC o Þ þ qg jg C o k
@z 1a @j @j 1a We applied implicit finite difference scheme to Eq. (5) over the



@hm ajl ql @C o @V gj entire number of nodes in the channel and block factorization
þ q g jg  j þ
@x 1a @hm @hm technique for matrix solver to compute local mixture enthalpy,




@p aj q l @C o @V gj b jg jg local pressure, local mixture velocity and local void fraction. The
þ qg jg  l j þ þ1þ 
@z 1a @p @p hfg a 1  a liquid enthalpy obtained from Eq. (1c) gives the estimate of liquid
"


 # temperature.
@a ajl ql @C o @V gj qg jg jg jl ql
2
þ q g jg  j þ þ C o þ V gj  þ
@z 1a @a @a a a ð1  aÞ2 2.2. Closure relations



4f qm jv m j j j
¼ g ql  g aðqg  ql Þ  v m  Csub g  l
Dh 2 a 1a The wall heat flux, interfacial heat transfer and mass transfer
were calculated using the closure relations.
ð2bÞ
Mixture mass equation: 2.2.1. Wall heat flux in subcooled boiling
The externally-applied heat flux to the tube wall mainly pro-
@½ð1  aÞql þ aqg  @½ql jl þ qg jg 
þ ¼0 ð3aÞ vides the sensible heat to the subcooled liquid to raise its temper-
@t @z ature above the saturation temperature and subsequently the
Expanding Eq. (3a), we obtain: latent heat q00ev to evaporate the liquid. The liquid superheating in
     terms of wall superheating is essential for nucleation and bubble
@p @ ql @ qg @ a @ ql
ð1  aÞ þ a þ ð1  aÞ þ qg  ql formation. The sensible heat is a sum of the heat flux q00c due to tur-
@t @p @p @t @p bulent convection and the quenching heat flux q00q due to transient
 

@C  
@a  @V  heat conduction. The quenching heat transfer occurs due to the
þ qg  ql a j o þ gj þ jC o þ V gj
@z @a @a additional liquid mixing induced by the bubbles emerging from

 
@C 
@j @V the wall. It takes place through the patches of fresh liquid coming
þ ql ð1  aC o Þ þ qg aC o þ a qg  ql j o þ gj to the wall from the bulk to replace the departing bubbles (Tu and
@z @j @j
  
 Yeoh, 2002; Basu et al., 2005; Krepper et al., 2007). The total wall
@p @ ql @ qg @C o @V gj
þ j þ jg þ a qg  ql j þ heat flux during subcooled boiling therefore essentially comprises
@z l @p @p @p @p

 of q00ev ; q00q and q00c :
@hm @ ql   @C @V gj
þ a qg  ql j
o
þ j þ ð3bÞ q00 ¼ q00c þ q00q þ q00ev ð6Þ
@z l @zhm @hm @hm
Vapour-phase mass equation: The quenching heat flux occurs at the fraction a2F of wall area
populated by the bubbles and the convection heat flux occurs at
@½aqg  @½qg jg  the remaining fraction a1F of the wall area (a1F = 1.0 – a2F).
þ ¼C ð4aÞ
@t @z Single-phase convective heat flux is:
Expanding the above equation, we write: q00c ¼ a1F wl ðT w  Tl Þ# ð7Þ
  wl is heat transfer coefficient of subcooled liquid. 0 is inhibition fac-
@ a @p @ qg a
qg þ a  tor (Eq. (8)) such that the wall heat flux at saturation is fully used to
@t @t @p hfg

 generate vapour (Končar and Kljenak, 2002).
@a  @C @V
þ qg jC o þ V gj þ aqg j o þ gj h h
2 h fh l
@z @a @a #¼ f ld
ð8Þ

  hf hl
@p @C o @V gj b @ qg 1þ hf hld
þ aqg j þ  þ jg
@z @p @p hfg @p

 
 hld is the liquid enthalpy at the onset of significant void (OSV) and is
@j @C o @V gj @hm @C o @V gj
þ aqg C o þ j þ þ j þ obtained from the Saha-Zuber correlations. Recommended by Hari
@z @j @j @z @hm @hm and Hassan (2002), we used the correlation of Zeitoun and
¼ Csub ð4bÞ Shoukri (1997) for pumping factor. The pumping factor is defined
in literature as the ratio of q00q to q00ev and is used for the realistic cal-
Each of Eqs. (1b), (2b), (3b) and (4b) can be written as:
culation of active nucleation site density.
@Y k @Y k Quenching heat flux given by Mikic and Rohsenow (1969) is:
Ak þ Bk þ C k Y k ¼ Dk ð5Þ
@t @z q00q ¼ a2F wq ðT w  Tl Þ# ð9aÞ
2 3
hm p 
6 p 7 2
Ak, Bk, Ck 2 R44 and Dk 2 R41 6
for kth node; Y k ¼ 4 7 a2F ¼ min ðaf dbW Þ ; 1 ð9bÞ
j 5 4
a k qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
(i, j) of Ak indicate the coefficients of time-derivatives of jth ele- wq ¼ pffiffiffiffi f bd t w kl ql cpl ð9cÞ
ment of Yk in ith equation; (i, j) of Bk indicate the coefficients of p
330 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340

wq is quenching heat transfer coefficient. af is the bubble influ- 2.3. Harmonic balance method (HBM) for asymptotic solution
ence factor and is taken as 2.0 (Končar and Kljenak, 2002).
dbw is bubble departure diameter and is given by Fritz correla- Due to the lack of related experimental data, we compare our
tion (Fritz, 1935): numerical simulation of oscillating boiling flow due to the periodic
perturbation with asymptotic solution. Since the inlet flow of the
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r channel is periodically perturbed, the desired solution will be peri-
dbw ¼ 0:0208u ð10Þ odic in nature. We therefore apply HBM based on the work of
gðqL  qG Þ
Paruya et al. (2016) to derive the asymptotic solution from the
r is surface tension; g is acceleration due to gravity. The Fritz cor- mixture energy balance equation given by Eq. (18) to find the mix-
relation is simple to use and provides a correct characteristic length. ture enthalpy at the exit of the channel hme in the form of
In the present boiling channel with high inlet subcooling, small hme ¼ a0 þ a1 sin2pxt þ b1 cos2pxt. We assume the first harmonics
bubbles with spherical geometry form due to the dominating effect to be dominant.
of surface tension. Consequently, one may use the correlation in ~ sin 2pxt; x is cyclic forcing
The periodic forcing is ui ¼ uis þ a
highly subcooled conditions. Some uncertainties may arise from frequency in Hz; a ~ is forcing amplitude in m/s; uis is water velocity
the correlation in very high pressures and from neglecting the effect at the inlet of the channel at steady state. The subscript i represents
of local subcooling, for which the other correlations may be useful. the inlet of the channel.
The contact angle u would be moderately large. u for water- We assume the followings to derive a low-order process model
stainless steel is taken as 80° in this study. In the numerical simu- that yields asymptotic solution:
lation, we did not find any significant change in exit void fraction
due to the variation of u from 30° to 80°. (i) Linear profile of enthalpy throughout the channel (Clausse
Bubble departure frequency fbd given by Cole (1960) is: and Lahey, 1991; Paruya et al., 2012).
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (ii) Incompressible and homogenous flow.
4gðql  qg Þ (iii) Thermal equilibrium between the phases at the channel exit.
f bd ¼ ð11Þ
3dbw ql (iv) The pressure change in the channel is negligible.

The bubble waiting time tw given by Kurul and Podowski (1991) It is worthwhile to mention that the assumptions of homoge-
is: neous equilibrium flow model (HEFM) and incompressibility of
the mixture restrict to the boiling flows with low vapour-quality
0:8
tw ¼ ð12Þ and high inlet subcooling. These conditions arise in our boiling
f bd
channel, in which the vapour quality is low (5.96% for q00 = 570
2 dbW kW/m2) and high inlet subcooling (DTsub = 58.2 K). We also com-
q00ev ¼ a2F f bd 2 qg hfg an ð13Þ puted hme based on HEFM and DFM4 in low vapour quality (see
3 a
Table 1). The small gap between the enthalpies allows us to apply
The surface density of active nucleation sites an given by
the above assumptions in working out the asymptotic solutions
Lemmert and Chawla (1977) is:
using low-order model.
an ¼ ½185ðT w  T sat Þ1:805 ð14Þ
@ðqm hm Þ @ðqm hm um Þ
þ ¼ q000 ð18Þ
@t @z
2.2.2. Interfacial heat transfer
Interfacial heat transfer, which takes between vapour and liquid The mixture continuity equation given by Eq. (19) is substituted
through the interfaces has been expressed for two cases – (a) the in Eq. (18) to obtain Eq. (20a).
vapour bubbles are in the bulk of subcooled liquid (bubble conden- @ qm @ðqm um Þ
sation) and (b) the vapour bubbles are in the bulk of superheated þ ¼0 ð19Þ
@t @z
liquid (bubble growth). The computation of bulk liquid tempera-
ture enables us to identify these cases. @hm @hm q000
Case (a): bubble condensation: þ um ¼ ð20aÞ
@t @z qm
Q gl ¼ wgl ðT sat  T l Þagl ð15aÞ
While integrating Eq. (20a) in z 2 [0, H], we applied mean value the-
Case (b): bubble growth: orem (see Eq. (20b)) to obtain the periodic solution.
Z
Q lg ¼ wlg ðT l  T sat Þalg ð15bÞ H
fðz; tÞdz ¼ fðtÞH ð20bÞ
0
In Eq. (15), the subscript gl means the heat transfer from vapour
to liquid and lg the heat transfer from liquid to vapour. The inter- f ðtÞ is the mean value of f(t) in z 2 [0, H]. Now,
facial heat transfer coefficients wgl and wlg are determined from the hme ¼ a0 þ a1 sin2pxt þ b1 cos2pxt is substituted in Eq. (20a) to
Ranz-Marshall correlation (Ranz and Marshall, 1952) given below. obtain a0, a1 and b1 in the following form:
kl 1=2 a0 ¼ hmes ð21aÞ
wgl ¼ ð2 þ 0:6Reb Pr1=3 Þ ð16Þ
db
db is the bubble mean diameter usually obtained from its linear ~
2uis ðhmes  hi Þa
a1 ¼  ð21bÞ
relation with DTsub given by Anglart et al. (1997). Reb is bubble Rey- 4u2is þ 4p2 x2 H2
nolds number computed using db and the relative velocity of bub-
ble. Interfacial area agl or alg (m3/m2) is taken as 3.6a/db. 4pxHðhmes  hi Þa
~
The condensation rate (kg/m3-s) is given by: b1 ¼ ð21cÞ
4u2is þ 4p2 x2 H2
Q gl
Cgl ¼ ð17Þ hmes is the steady-state mixture enthalpy at channel exit, which is
hfg
defined below:
 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340 331

Table 1
Comparison of exit mixture enthalpy based on HEFM and DFM4.

Heat flux q00 (kW/m2) Vapour quality, xe % hme (kJ/kg) (from HEFM) hme (kJ/kg) (from DFM4) Gap between the enthalpies (%)
500.0 3.24 1176.244 1152.770 2.03
540.0 4.80 1202.331 1165.073 3.19
570.0 5.96 1221.890 1174.101 4.07
600.0 7.13 1241.460 1182.770 4.96

h i
ð1  aÞql hl þ aqg hg q000 H
hmes ¼ es
¼ hi þ ð21dÞ
½qm es uis qm
The subscripts e and s stand for exit and steady state respec-
tively. Based on the assumption of thermal equilibrium, one can
also compute the asymptotic solution for ae.
The amplitude ahe of the oscillation of hme is expressed as:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ~q000 H
a
2 qm uis
ahe ¼ a21 þ b1 ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð22Þ
4u2is þ 4p2 x2 H2

The following can be inferred from Eq. (22):

Proposition 1. For a higher forcing amplitude ã, higher x>xcrit is Fig. 2a. Model validation with experimental results (0 = 1.0).
required to make ahe asymptotically zero.

Proposition 2. ahe ! 08x ! 1

Proposition 3. At higher external heat flux, the amplitude of hme is

higher. A high x is also required to make ahe asymptotically zero.
We have established the three propositions stated above with
the help of our numerical simulation presented in the next section.

3. Numerical results

3.1. Comparison of numerical simulation and experimental results at

steady state
Fig. 2b. Effect of interfacial heat transfer correlations and correction factors (0 =
For the comparison, we used the geometries and the operating 1.0).
conditions of Bartolomej and his group (Bartolomej and
Chanturiya, 1967; Bartolomej et al., 1982) shown in Fig. 1. The local
void fraction, the local liquid temperature and the local surface Fig. 2c describes the effect of grid numbers on the calculation of
temperature are presented in Fig. 2a. The experimental results of void fraction profile for which 0 was calculated using Eq. (8). The
Bartolomej experiments were adapted from the work of Krepper results remain almost same. It indicates that N = 150 can be used
et al. (2007). The predicted liquid temperature agrees well with for the calculation of boiling flow under periodic forcing. Fig. 2c
the experimental data while the void profile shows an appreciable also shows that the void profile with 0 = 1 (the inhibition factor
overprediction by maximum 40.5%, particularly in the bulk boiling was neglected) is appreciably higher (maximum 40.2% overpredic-
region (z  1.5 m). The reason is that the Chen correlation used for tion at exit) than the profile with 0 computed by Eq. (8) (maximum
computing the boiling heat flux in the bulk boiling region com-
putes high heat transfer coefficient. The simplifications we made
are that the characteristic temperature is equal to the wall temper-
ature and the effect of inhibition factor (0 = 1) is negligible. In the
subcooled boiling region, the agreement is fairly good. The steep
decrease of the wall temperature indicates the onset of nucleate
boiling (ONB). The correction factor (cf) in Ranz-Marshall correla-
tion for interfacial heat transfer coefficient is 0.8. The effect of
interfacial heat transfer correlations and the associated correction
factor is presented in Fig. 2b. The results of void fractions were
computed using Akiyama correlation (Akiyama,1973) and Ranz-
Marshall correlation with cf = 0.45, 0.8 and 0.7. The influences of
the correlations and the correction factors are insignificant.
Akiyama correlation for cf = 0.45 shows a little improvement in
the subcooling boiling zone. Slightly higher void fraction is
obtained in the saturation boiling zone using the Akiyama correla-
tion with cf = 0.8. Fig. 2c. Effect of grid numbers on void fraction profile.
332 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340

32.3% overprediction at exit), particularly in the bulk boiling zone. two pressure sensors P1P2 are WIKA make (Germany) with an
The inclusion of inhibition factor is therefore essential to compute uncertainty less than 0.25%, the response time of 1.0 ms and the
more accurate void fraction profiles. The computed profile is also range of 0–10 bar. A capacitive differential pressure transmitter
found to be consistent with the experimental results and the (DPT) measures the pressure drop with an uncertainty less than
numerical results of Krepper et al. (2007) using CFX-5.0 presented 0.25% and response time of 1 ms. The electromagnetic flow meter
in Fig. 2c. The results of Krepper et al. are relatively close to the (EFM) measures the flow rate of water with an uncertainty less
experimental results, because CFX-5.0 does an elegant treatment than 0.5%.
of interfacial transports using two-fluid model. The subcooled boiling experiments were carried out at q00 =
39.93 kW/m2, Ti = 85 °C, p2 = 1.4 bar, mean G = 119.3 kg/m2-s and
3.2. Experimental validation for the oscillations xe = 0.0075%. All measuring devices have uncertainties within
±0.5%. The variations of the SDs of G, p1 and T2 with time were
In this section, we evaluate the ability of the present DFM4 to to found to remain within ±8.5% (maximum). It indicates that the
predict oscillations of boiling flow in the heated section by com- time series is almost stationary and the oscillations are sustained.
paring them with the experimental results. The test section (boil- It has been found that the uncertainty of the temperature con-
ing channel) in Fig. 3 includes the process tube and its troller is in between ±2.8% to ±3.7%. The heat loss is about 3.5%.
instrumentation in the heater section, which is a part of test natu- The flow oscillations are at boiling incipience in which the
ral circulation boiling loop (Karmakar et al., 2016). The process vapour quality is very low (xe = 0.0075%). The measured inlet
tube made of SS 316 is 1 m long. Inside diameter (ID) and outside flow oscillation in Fig. 4 was introduced in the model to compute
diameter (OD) of the tube are 20.96 mm and 26.71 mm respec- inlet pressure P1 and outlet temperature T2. T2 was computed
tively. The two pressure sensors P1 and P2 are at the inlet and exit using DFM4 and HEFM, and P1 using DFM4. The computed T2
of the tube respectively. Two resistance temperature detectors and p1 are close to the measured values with maximum errors of
(RTDs), namely T1 and T2 are at the inlet and the exit of the tube 0.75% and 5.12% respectively.
while another RTD measures the tube-wall temperature. The video The T2-predictions by DFM4 and HEFM are almost same. This is
and the impedance probe is placed at the exit of the tube to study due to very low vapour quality. Overall, DFM4 predicts the oscillat-
the bubble phenomena. The differential pressure transmitter (DPT) ing behavior of the boiling channel, which is reasonably close to
with a range of 0–10 bar measures pressure drop across the dis- the experiments with flow oscillation amplitudes of 45–85%. This
tance of 0.6 m of the tube. The electromagnetic flow meter (EFM) experimental validation of DFM4 and HEFM in oscillating flow con-
with 20 mm polytetrafluoroethylene lining and the range of 0– ditions also indicates that the correlations for heat transfer and
1.8 m3/hr measures the flow rate at the inlet of channel. For uni- fluid flow used in the models are applicable within a limit of oscil-
form and safe heating, six segmental heater-coils (Kanthal super: lation amplitudes (45–85%). The correlations therefore work well
MoSi2) are wound around an aluminium tube co-axially placed in the present work. Further studies are required to estimate the
outside the tube. The aluminium tube helps us avoid the overheat- permissible limit of forcing amplitudes, within which those corre-
ing of the process tube. The heat transfer from the surface of the lations perform well.
aluminium tube to the surface of the process tube takes place by We present another case study for comparison with RELAP5
the process of radiation and convection. (1999) in steady-state condition and unsteady-state condition.
The RTDs (T1 and T2 in Fig. 3) are Pt-100 (sensitivity of 0.384 X/ The a-profile obtained by the present model were also compared
°C and uncertainty less than 0.5%) with a range of 0–400 °C. The with RELAP5-computations in Fig. 5 for inlet temperature of 248
°C, G = 998 kg/m2-s, p = 68.0 bar and q00 = 440 kW/m2. The void
fraction profile at steady state (Fig. 5(a)) computed by the present
DFM4, RELAP5 and numerical results of Talebi et al. (2012) based
on DFM for a vertical boiling channel are in close agreement with
experiments. The results obtained from DFM4 and RELAP5 for a
transient of power ramp-up are consistent and closely agree
(Fig. 5(b)). Although the RELAP5-results and the DFM4-results ini-
tially show a small difference between, they closely agree
eventually.

3.3. The effect of periodic forcing on wall heat flux

The temporal and spatial variations of the quantities such as

wall temperature, liquid temperature, exit enthalpy, condensation
rate, void fraction, quenching heat flux, single-phase heat flux,
evaporative heat flux and overall wall-heat flux in subcooled boil-
ing, as resulted from the model computation, were averaged over
time and space after the periodic forcing was initiated. The effect
of the periodic forcing on the mentioned quantities can be quanti-
fied by the ratio R of their averages defined above to their space-
averaged values at steady state. For example, RTl is expressed as:
R Lsub
1
Lsub 0
ðT l ðz; tÞÞdz
RTl ðtÞ ¼ ð23aÞ
T ls
Z s
1
RTl ¼ RTl ðtÞdt ð23bÞ
s 0

Fig. 3. Schematic of the test section in the loop. The space-averaged steady-state values are:
 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340 333

Fig. 4. Comparison of measured T2 and p1 with DFM4-computation.

Z Lsub
1 We averaged the quantities over a duration of the time period of
T ls ¼ T ls ðzÞdz ð23cÞ
Lsub 0
forcing function, i.e., s = x1. Time-averaging starts from t = 40 s
when sustained oscillations occur. R > 1.0 indicates the increase
Z Lsub of the corresponding quantities compared to their steady-state val-
1
T ws ¼ T ws ðzÞdz ð23dÞ ues. Lsub is the length of subcooled boiling region.
Lsub 0
Now, we explain the time series of R(t) for Tw, Tl, q00ev , Gc and a in
Z Lsub the subcooled boiling zone, and q00w obtained as a result of introduc-
1
Gcs ¼ Gcs ðzÞdz ð23eÞ ing a periodic perturbation of the velocity
Lsub 0
~ ¼ 0:020 ms and x ¼ 0:31 HzÞ (at the inlet of the boiling channel.
a
Z Lsub These results are shown in Figs. 6. As the inlet velocity is decreased
1
q00ev s ¼ q00ev ðzÞdz ð23fÞ (Fig. 6a), both the liquid temperature and the surface temperature
Lsub 0
in the region increase with a delay of 0.62 s for a given heat flux. As
Z Lsub the wall heats up, the superheat increases resulting in the increase
1
q00qs ¼ q00q ðzÞdz ð23gÞ of evaporative heat flux and total wall heat flux, void fraction and
Lsub 0
decrease of condensation rate (Fig. 6b). We note a sharp rise of
Eqs. (23c)–(23g) represent the space-averages in subcooled boiling evaporative heat flux and total heat flux (Fig. 6b), which causes
zone. The space-average of total wall heat flux over entire length of to cool down the heated surface in the next instant. However,
the channel is: when the inlet velocity is increased, more cold-water flow causes
Z H the cooling of the heated surface and liquid (Fig. 6a); the bubble
1
q00w ¼ q00w ðzÞdz ð23hÞ condensation increases and void fraction decreases (Fig. 6b). The
H 0 effect of decreasing velocity in one half of a cycle is more signifi-
RTw for wall temperature Tw, RTl for liquid temperature Tl, Rgc for cant over the effect of increasing velocity in the next half of the
bubble condensation rate Gc, Rqe for evaporative heat flux q00ev and cycle. The overall rise of wall heat flux is due to overall increase
Rhtf for wall heat flux q00w can be defined in the similar manner. Rhe of evaporative heat flux q00ev .
for exit enthalpy and Rae for exit void fraction are defined below: To further confirm the reason of heat flux enhancement, we
present the effect of x on R for q00ev , departure bubble diameter
Z s
1 he ðtÞ db, bubble frequency fb and active nucleation density An in Fig. 7
Rhe ¼ dt ð23iÞ (a) q00w and R for Tw and Tl in Fig. 7(b). The variations clearly show
s 0 hes
that the variation of q00ev and q00w closely follow the variation of An
Z s (Fig. 7(a)) and An follows the variation of Tw in low x. Although
1 ae ðtÞ the variations of fb and db with x are relatively small, bubble diam-
Rae ¼ dt ð23jÞ
s 0 aes eter decreases (R < 1.0) and bubble frequency increases (R > 1.0)
334 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340

Fig. 5. Comparison among DFM4, RELAP5 and Talebi et al. (2012) (a) steady state and (b) dynamic state.

due to the oscillating flow. We therefore claim that the periodic et al. (2010) and Wang et al. (2013) experimentally observed a con-
forcing of inlet velocity in low x triggers overall increase of heated siderable increase of active nucleation density in the section of a
surface temperature (wall superheat) that heavily influences the cycle where the mass flux decreases. The periodic forcing of inlet
active nucleation density. Consequently, the evaporative heat flux flow causes out-of-phase variation of pressure drop across the boil-
increases. All variations reduce to unity in high x indicating that ing channel that intensifies axial and radial mixing; the mixing also
the enhancement of heat flux does not occur in high x. Chen contributes to enhancement of wall heat flux (Chen et al., 2010).

Fig. 6a. Dynamical behavior of the boiling channel due to periodic perturbation of inlet velocity at ã = 0.020 m/s and x = 0.31 Hz.
 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340 335

Fig. 6b. Dynamical behavior of the boiling channel due to periodic perturbation of inlet velocity at ã = 0.020 m/s and x = 0.31 Hz.

Fig. 7. Variation of Rqe, RDb, Rfb, RAn, Rhtf, RTw and RTl with frequency at ã = 0.020 m/s.

We have also noted that increasing the number of nodes (from velocity was varied from 0.010 to 0.025 m/s (1.0–2.5% fluctua-
N = 150 to 350) did not really affect the accuracy of computations tion) and the frequency from 0.01 Hz to 1.5 Hz.
during periodic forcing. The average relative deviation at ã = 0.020 Fig. 8 presents the variation of Rhtf (for average q00w ) with x at
m/s is for is 0.0322% for the increase from N = 150 to N = 250 and different amplitudes. As mentioned in the previous section, the
0.032% for N = 150 to N = 350. enhancement of heat flux occurs in low x. Increasing x over a
critical value xcrit does not improve heat flux; Rhtf approaches
3.4. Effect of forcing amplitude and frequency to unity as x > xcrit (Proposition 1). As noted in the figure, xcrit
increases with the forcing amplitude (xcrit = 1.15 Hz, 0.69 Hz,
In this section, we examine the effect of forcing amplitude and 0.41 Hz and 0.25 Hz for the forcing amplitudes of 0.025 m/s,
the forcing frequency on the heat flux. The amplitude ã of inlet 0.020 m/s, 0.015 m/s and 0.010 m/s respectively). Number of
336 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340

Fig. 8. Variation of Rhtf with frequency at different amplitudes.

Fig. 9. Variation of Rqe with frequency at different amplitudes.

Fig. 10. Variation of RTw with frequency at different amplitudes.

 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340 337

Table 2
Enhancement of wall heat flux at different forcing amplitudes.

Sl. # Amplitude (m/s) x (Hz) Enhancement of q00w Time average over

a period of 1/x
Max 1 Max 2 Max3
1 0.010 0.13a 3.43% 0.16%
2 0.015 0.23a 5.0% 0.28%
3 0.020 0.53a 3.0% 3.30% 0.67%
4 0.025 0.31a 3.20% 3.10% 3.30% 0.56%
5 0.020 0.31b 0.40%
6 0.015 0.31b 0.22%
a
Frequency for dominant peak.
b
The peaks at 0.31 Hz.

peaks also increases as we increase the forcing amplitude. The these peaks of q00w correspond to the peaks of evaporative heat flux
qev presented in Fig. 9. xcrit remains same. The amplitude of qev
00 00
number of fundamental peaks is one, three, four and five for
00
the amplitude of 0.025 m/s, 0.020 m/s, 0.015 m/s and 0.010 m/s increases with forcing amplitude. The amplitude of average qw fol-
00
respectively. The dominant peak at 0.53 Hz for the amplitude lows the variations of qev and Tw. Fluctuation of Tw plays a pivotal
of 0.020 m/s spearheads over those for the amplitude of 0.025 role in the enhancement of q00w . The higher number of peaks at
m/s. We noted that it was not due to high An; we do not also higher forcing amplitudes (Figs. 8–10) indicate that high-
00
see the dominant peak of qev at 0.53 Hz in Fig. 9 whereas we amplitude forcing induces quasi-periodic or chaotic oscillations
see the dominant peak of Tw at 0.53 Hz in Fig. 10. This aggressive (chaotic mixing) leading to fast decrease of thermal boundary
rise of Tw is due to significant bubble condensation. It clearly layer thickness or superheated liquid-film thickness (confirmed
reveals that a large amount of quenching heat flux and single- by increase of bubble frequency in Fig. 7). And the decreased bub-
phase convective heat flux contributes to the steep rise of wall ble diameter (Fig. 7) also enhances bubble mixing. Consequently,
heat flux at 0.53 Hz and amplitude of 0.020 m/s. However, all the wall heat flux increases. We noted that the chaotic oscillation

Fig. 11. Comparison of time series of numerical simulation and asymptotic solution (a) a = 0.015 m/s and x = 0.31 Hz (b) a = 0.020 m/s and x = 0.31 Hz(c) a = 0.025 m/s and
x = 0.35 Hz.
338 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340

Fig. 12. Asymptotic decrease of amplitude of exit enthalpy oscillation at (a) a = 0.010 m/s, (b) a = 0.015 m/s and (c) a = 0.020 m/s (xc = 0.165 Hz).

occurred much at low forcing frequency. Our observation fairly (ae) and the amplitude of the mixture enthalpy at channel exit
agrees with the results of Garg et al. (2015), who numerically (hme). Fig. 11 presents three cases of the time-varying Ra at differ-
showed that the chaotic mixing occurred favorably at low Rey- ent forcing amplitudes (0.015 m/s, 0.020 m/s and 0.025 m/s) and
nolds number and low slip velocity. different frequencies (0.31 Hz and 0.35 Hz) obtained by numerical
Table 2 summarizes the enhancement of q00w at different ampli- simulation and asymptotic simulation. The results of numerical
tudes of 0.01–0.020 m/s. The average and maximum enhancement simulation and asymptotic simulation are highly consistent. We
of the dominant peak increase with forcing amplitude except for note that the average error and the maximum error are 0.57%
except for the 0.020 m/s (see sl. #1–4 of Table 2). For the peaks cor- and 1.5% respectively for a = 0.015 m/s and x = 0.31 Hz (Fig. 11
responding to the frequency of 0.31 Hz in Fig. 8, the average (a)), 0.73% and 1.8% respectively for a = 0.020 m/s and x = 0.31
enhancement of q00w increases with amplitudes (see sl# 4–6 of Hz (Fig. 11(b)) and 2.22% and 4.96% respectively for a = 0.025 m/s
Table 2). Increased number of irregular and relatively high peaks and x = 0.35 Hz (Fig. 11(c)). The error becomes little higher with
of RTw, Rqe and Rhtf at higher amplitudes indicate that high- increasing amplitude. It is due to the fact that the asymptotic solu-
amplitude forcing induces chaotic mixing leading to fast decrease tions were subjected to three major assumptions – incompressible
of thermal boundary layer thickness or superheated liquid-film phases, linear axial profile of mixture enthalpy and attainment of
thickness that results in high evaporative heat flux and quenching thermal equilibrium condition during subcooled boiling. The level
heat flux. All these high peaks of RTw, Rqe and Rhtf occur at much of error indicates that those assumptions seem to be reasonably
lower frequencies. The results are very consistent with the results well in working range of the forcing amplitude. In high-
of Paruya et al. (2016). amplitude forcing, linear enthalpy profile is a major source of
uncertainty, although the assumption of linear enthalpy profile
3.5. Comparison of numerical solution and asymptotic solution and thermal equilibrium were successfully used by Clausse and
Lahey (1991) and Paruya et al. (2012). Thermal equilibrium condi-
In this section, we compare the results of numerical simulation tion assumed for obtaining asymptotic solution is more likely to be
and the asymptotic solution (based on HBM) to benchmark our valid in high-amplitude forcing, which allows the subcooled boil-
numerical solution of the boiling channel subjected to periodic ing to reach thermal equilibrium due to a large decrease of inlet
forcing. Figs. 11, 12 present the comparisons for void fraction velocity.
 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340 339

Fig. 12 compares the amplitudes of ahe at different forcing rium seem to be reasonable in working out the asymptotic solu-
amplitudes obtained by numerical simulation and asymptotic tions. With increasing amplitude, the overall deviation increases
solution (Eq. (22)). Both the amplitudes consistently and asymptot- a bit. In high frequencies, the velocity changes so fast that its
ically decrease to zero as x ? / (Proposition 2). impact on the bubble dynamics during boiling becomes difficult
Now, we rewrite Eq. (22) as: to be reflected; the change in bubble phenomena due to forcing
takes its characteristic time to occur. A high-frequency forcing does
~q000
not allow the time. Consequently, the amplitude of enthalpy
a
2pq u decreases asymptotically fast.
ahe ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
m is
ð24Þ
u2is
p2 H 2
þ x2

Now, we define a characteristic frequency xc:

sffiffiffiffiffiffiffiffiffiffiffiffi 3.6. Effect of external heat flux
u2is
xc ¼ ð25Þ We numerically studied the effect of heat flux to establish the
p2 H 2
Proposition 3. Fig. 13 presents the variations of Rhe and Rhtf with
xc = 0.165 Hz for steady inlet velocity of 1.0378 m/s and height of frequency at ã = 0.01 m/s and q00 = 570, 590 and 600 kW/m2. With
2.0 m, as chosen in the present investigation. We derive two cases increasing external heat flux, the wall heat flux and the number
from Eq. (24) depending on x  xc or x xc: and the height of peaks (of Rhe and Rhtf) increase. It is interesting
Case 1: x  xc (high-frequency asymptotic solution): to note that Rhtf reduces to 1.0 with increasing x in lower heat
fluxes (q00 = 570 and 590 kW/m2) and the peaks of Rhtf continue
f 00
2aq to exist even at higher x at q00 = 600 kW/m2. It is due to the fact
ahe ¼ ðsubstituting q000 by 4q00 =Dh Þ ð26Þ
pxDh qm uis that the boiling channel undergoes thermally-induced flow insta-
bilities at higher heat fluxes above a critical value. This has been
Case 2: x xc (low-frequency asymptotic solution):
confirmed by the existence of out-of-phase oscillation of inlet
f 00
2aq velocity and pressure drop. Paruya et al. (2016) also demonstrated
ahe ¼ ð27Þ that the Eigen values of Jacobian were very sensitive to inlet sub-
pxc Dh qm uis
cooling (inlet temperature) and phase change number (heat flux).
As we see in Fig. 12, the high-frequency asymptotic solution For a given degree of inlet subcooling, the Eigen values shifted from
(Eq. (26)) is very close to the numerical simulation compared to negative real part to zero to positive real part with increasing heat
the asymptotic solution obtained from Eq. (22). Interestingly, the flux. Based on the criterion of amplitude ratio, we also calculated
full asymptotic solution for low-frequencies, obtained by Eq. (22), the threshold q00 of 589.16 W/m2 above which the flow instability
is closer to the numerical solution than the low-frequency solution occurs and the corresponding frequency is 0.033 Hz. Depending
obtained by Eq. (27) in x xc. Eq. (22) for x xc and high- on the level of inlet subcooling, the increasing heat flux allowed
frequency asymptotic solution obtained by Eq. (26) for x  xc, a number of bifurcation phenomena leading to chaos to occur.
therefore, successfully describe our numerical simulation. Thus, Thus, the chaotic flow instability participates in enhancing the wall
the assumptions of phase incompressibility and thermal equilib- heat flux.

Fig. 13. Effect of heat flux on periodically-forced oscillations at ã = 0.01 m/s.

340 S. Paruya et al. / Chemical Engineering Science 181 (2018) 326–340

4. Conclusions Chen, C.A., Lin, T.F., Yan, W.M., 2016. Bubble characteristics in time periodic
saturated flow boiling of R-134a in a narrow annular pipe due to heat flux
oscillation. Int. J. Heat Mass Transfer 102, 1150–1158.
In our work, the effect of periodic perturbation of inlet velocity Cheng, P., Zhao, T.S., 1998. Heat transfer in oscillatory flows. Annu. Rev. Heat
on subcooled boiling heat flux in a vertical upflow channel was Transfer 9, 359–420.
Clausse, A., Lahey Jr., R.T., 1991. The analysis of periodic and strange attractors
numerically studied using four-equation drift-flux model. The
during density wave oscillations in boiling flow. Chaos, Solitons Fractals 1, 167–
numerical model has been benchmarked with the experiments 178.
and the asymptotic solutions. Our numerical simulations are very Cole, R., 1960. A photographic study of pool boiling in the region of the critical heat
flux. AIChE J. 6, 533–538.
consistent with experiments and asymptotic solutions. All the
Fritz, W., 1935. Maximum volume of vapour bubbles. Physikalsche Zeitschrift 36,
assumptions for obtaining the asymptotic solutions have been jus- 379–384.
tified. We summarize the major findings below: Garg, P., Picardo, J.R., Pushpavanam, S., 2015. Chaotic mixing in a planar, curved
channel using periodic slip. Phys. Fluids 27 (3), 032004.
Hari, S., Hassan, Y.A., 2002. Improvement of the subcooled boiling model for low-
1. A small enhancement of subcooled boiling heat flux occurs in pressure conditions in thermal-hydraulic codes. Nucl. Eng. Des. 216, 139–152.
low forcing frequency (x < xcrit) as a result of periodic pertur- Ishii, M., 1977. One-dimensional drift-flux model and constitutive equations for
bation of inlet velocity. The enhancement is mainly due to the relative motion between phases in various two-phase flow regimes, ANL-77-47,
USA.
overall rise of heated surface temperature (wall superheat) by Ishii, M., Hibiki, T., 2006. Thermo-fluid Dynamics of Two-phase Flow. Springer.
the perturbation, which essentially increases the active nucle- Kakaç, S., 1999. Introduction to heat transfer enhancement, heat transfer
ation density and eventually the evaporative heat flux. The per- enhancement of heat exchangers. In: Kakac, S., Bergles, A.E., Mayinger, F.,
Yüncü, H., (Eds.), pp. 1–11.
turbation also enhances the bubble mixing. Karmakar, A., Dey, R., Paruya, S., 2016. Effect of local phenomena on subcooled
2. Effect of increasing forcing amplitude causes the rise of wall boiling oscillations in natural circulation boiling loop. Nucl. Eng. Des. 310, 205–
heat fluxes in x < xcrit. Chaotic mixing induced by relatively 216.
Končar, B., Kljenak, I., 2002. Subcooled boiling model to simulate upward vertical
high amplitude of forcing causes high evaporative heat flux
flow boiling at low pressures. In: Proceedings of the Internal Conference,
and quenching heat flux. We have provided an appropriate Nuclear Energy for New Europe, Kranjska Gora.
explanation as well. Krepper, E., Končar, B., Egorov, Y., 2007. CFD modelling of subcooled boiling –
concept, validation and application to fuel assembly design. Nucl. Eng. Des. 237,
3. Enhancement of wall heat flux is facilitated by increasing exter-
713–716.
nal heat flux to an extent above which thermally-induced chao- Kurul, N., Podowski, M.Z., 1991. On the modeling of multidimensional effects in
tic flow instabilities plays a dominating role in enhancing wall boiling channels. In: Proceedings of the 27th National Heat Transfer Conference,
heat flux. Minneapolis, USA, pp. 21–26.
Lemmert, M., Chawla, J.M., 1977. Influence of flow velocity on surface boiling heat
4. The assumption of linear enthalpy profile causes a substantial transfer coefficient. In: Hahne, E., Grigull, U. (Eds.), Heat Transfer in Boiling.
discrepancy. Academic Press, New York, NY, pp. 231–247.
Mikic, B.B., Rohsenow, W.M., 1969. A new correlation of pool boiling data including
the effect of heating surface characteristics. J. Heat Transfer 91, 245–250.
In future, we shall examine in detail the assumption of linear Nayak, A.K., Dubey, P., Chavan, D.N., Vijayan, P.K., 2007. Study on the stability
enthalpy profile and the role of chaotic mixing or instability in wall behaviour of two-phase natural circulation systems using a four-equation drift
heat flux. Determining the fractal dimension of the chaotic mixing flux model. Nucl. Eng. Des. 237, 386–398.
Okawa, T., Goto, T., Yamagoe, Y., 2010. Liquid film behaviour in annular two-phase
would be another interesting issue. Dryout condition, which may flow under flow oscillation conditions. Int. J. Heat Mass Transfer 53, 962–971.
arise due to low flow, also needs to be addressed. We shall also Paruya, S., Maiti, S., Karmakar, A., Gupta, P., Sarkar, J.P., 2012. Lumped
examine other correlations for heat transfer and fluid flow. parameterization of boiling channel – bifurcations during density wave
oscillations. Chem. Eng. Sci. 74, 310–326.
Paruya, S., Goswami, N., Pushpavanam, S., Pillai, D.S., Bidyarani, O., 2016.
Acknowledgement Periodically forced density wave oscillations in flow boiling at low forcing
frequencies: non-linear dynamics and control issues. Chem. Eng. Sci. 140,
123–133.
The authors gratefully acknowledge the financial support of Ranz, W.E., Marshall, W.R., 1952. Evapouration from drops, Parts I & II. Chem. Eng.
National Institute Technology Durgapur, India and Department of Prog. 48, 141–173.
RELAP5, 1999. RELAP5/MOD3 Code Manual, NUREG/CR-5535. Science tech, Inc.,
Science and Technology, New Delhi under SERB scheme. Idaho Falls.
Saha, P., Zuber, N., 1974. Point of net vapour generation and vapour void fraction in
subcooled boiling. In: Proceedings of Fifth International Heat Transfer
References Conference 4, pp. 175–179.
Saha, P., Aksan, N., Andersen, J., Yan, J., Simoneau, J.P., Leung, L., Bertrand, F., Aoto, K.,
Akiyama, M., 1973. Bubble collapse in subcooled boiling. Bull. JSME 16, 570–575. Kamide, H., 2013. Issues and future direction of thermal-hydraulics research
Anglart, H., Nylund, O., Kurul, N., Podowski, M., 1997. CFD prediction of flow and and development in nuclear power reactors. Nucl. Eng. Des. 264, 3–13.
phase distribution in fuel assemblies with spacers. Nucl. Eng. Des. 177, 228– Sinha, R.K., Kakodkar, A., 2006. Design and development of the AHWR—the Indian
2150. thorium fuelled innovative nuclear reactor. Nucl. Eng. Des. 236, 683–700.
Baird, M.H., Duncan, G.J., Smith, J.I., Taylor, J., 1966. Heat transfer in pulsed Talebia, S., Kazeminejad, H., Davilu, H., 2012. A numerical technique for analysis of
turbulent flow. Chem. Eng. Sci. 21, 197–199. transient two-phase flow in a vertical tube using the drift flux model. Nucl. Eng.
Bartolomej, G.G., Chanturiya, V.M., 1967. Experimental study of true void fraction Des. 242, 316–322.
when boiling subcooled water in vertical tubes. Therm. Eng. 14, 123–128. Tong, L.S., 1967. Heat transfer in water-cooled nuclear reactors. Nucl. Eng. Des. 6 (4),
Bartolomei, G.G., Brantov, V.G., Molochnikov, Y.S., Kharitonov, Y.V., Solodkii, V.A., 301–324.
Batashova, G.N., Mikhailov, V.N., 1982. An experimental investigation of true Tu, J.Y., Yeoh, G.H., 2002. On numerical modelling of low-pressure subcooled boiling
volumetric vapour content with subcooled boiling in tubes. Therm. Eng. 29 (3), flows. Int. J. Heat Mass Transfer 45, 1197–1209.
132–135. Wang, S.L., Chen, C.A., Lin, T.F., 2013. Oscillatory subcooled flow boiling heat
Basu, N., Warrier, G.R., Dhir, V.K., 2005. Wall heat flux partitioning during subcooled transfer of R-134a and associated bubble characteristics in a narrow annular
flow boiling: Part 1 – Model development. J. Heat Transfer 127, 131–140. duct due to flow rate oscillation. Int. J. Heat Mass Transfer 63, 255–267.
Bergles, A.E., 1999. Enhanced heat transfer: endless frontier, or mature and routine? Warrier, G.R., Dhir, V.K., 2013. Comparison of heat removal using miniature
J. Enhanced Heat Transfer 6, 79–88. channels, jets, and sprays. J. Enhanced Heat Transfer 20, 17–32.
Bouhadji, A., Djilali, N., 2003. Forcing of unsteady separated flow and convective West, F.B., Taylor, A.T., 1952. The effect of pulsation on heat transfer in turbulent
heat transfer via bulk upstream oscillations. Int. J. Heat Fluid Flow 24, 77–90. flow of water inside tubes. Chem. Eng. Prog. 48, 39–43.
Chattopadhyay, H., Biswas, G., Mitra, N.K., 2002. Heat transfer from a moving Zeitoun, O., Shoukri, M., 1997. Axial void fraction profile in low pressure subcooled
surface due to impinging slot jets. J. Heat Transfer 124, 433–440. flow boiling. Int. J. Heat Mass Transfer 40, 869–879.
Chen, C.A., Chang, W.R., Lin, T.F., 2010. Time periodic flow boiling heat transfer of R- Zuber, N., Findlay, J.A., 1965. Average volumetric concentration in two-phase flow
134a and associated bubble characteristics in a narrow annular duct due to flow systems. J. Heat Transfer 87, 453–468.
rate oscillation. Int. J. Heat Mass Transfer 53, 3593–3606.