Applied Thermal Engineering 65 (2014) 1e13

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Study on two-phase flow instabilities in internally-ribbed tubes

by using frequency domain method
Yifan Zhang, Huixiong Li*, Liangxing Li, Xianliang Lei, Tai Wang
State Key Laboratory of Multiphase Flow in Power Engineering, Xi’an Jiaotong University, Xi’an 710049, China

h i g h l i g h t s

A new model is made to study two-phase flow instabilities in internally-ribbed tubes.

Two significant improvements are made to the conventional models.

Two-phase flows in internally-ribbed tubes are more stable than that in smooth tubes.

The 6-head ribs internally-ribbed tube is more stable than the 4-head one.

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

Article history: A new mathematical model was proposed in this paper based on the frequency domain theory to
Received 2 December 2013 investigate two-phase flow instabilities in internally-ribbed tubes. Two significant improvements were
Accepted 27 December 2013 made to conventional models, one of which was that the heat storage of the tube wall metal, the internal
Available online 7 January 2014
wall heat flux and the external wall heat flux were considered as dynamic parameters, and another one
was that a new feedback relationship was established. The new model was verified by comparing the
Keywords:
prediction results to both the corresponding experimental data in literature and the results obtained
Two-phase flow instability
with SIMULINK. Then the two-phase flow instabilities in internally-ribbed tubes were studied system-
Frequency domain theory
Internally-ribbed tube
atically. It was shown that the stability of two-phase flows increased with pressure, mass flux or inlet
Critical heat flux resistance coefficient. It was also found that there existed a worst point of the stability (WPS) of the two-
SIMULINK phase flows when the inlet fluid subcooling increased. Two-phase flows in the 6-head ribs internally-
ribbed tube (f38.1  7.5 mm) was more stable than that in the 4-head ribs internally-ribbed tube
(f28  5.41 mm). The present model and results could provide guidance for the design and operation of
boilers.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction there are non-periodic migrations of flow rate of the two-phase

In recent decades, the two-phase flow instability attracts lots of
scholars’ attentions, because it may cause oscillations of flowing
and thermal parameters of the two-phase flow systems, such as the
boiler water wall in thermal power plants, the steam generators in
nuclear power plants and the heat exchangers in chemical engi-
neering, etc. Undesirable oscillations may result in the boiling crisis,
and even induce the thermal fatigue of the pipes and the disrepair
of the equipment. Therefore, much effort had been made to study
the two-phase flow instabilities [1e5].
Based on the results of early research, the two-phase flow
instability is usually divided into two categories [6], i.e., the static
instability and the dynamic instability. The static instability means
* Corresponding author. Tel.: þ86 29 82665870; fax: þ86 29 82669033.
E-mail address: huixiong@mail.xjtu.edu.cn (H. Li).
flows in tubes, and the dynamic instability refers as the self-
sustained oscillations occurred in tubes. Both of the above
mentioned two-phase flow instabilities may occur in the smooth
tubes and other enhanced heat transfer tubes, such as the
internally-ribbed tubes.
Smooth tubes have been widely used in two-phase flow insta-
bility studies due to the fact that smooth tubes are the most
commonly used pipes in various heat exchangers. The density wave
instabilities in smooth tubes of a once-through boiling flow system
was analyzed by Takitani et al. [7e9], based on experimental
studies and theoretical analyses, and the stability boundaries of the
joule-heated and sodium-heated systems were proposed. Pustylnik
et al. [10] investigated the instabilities of two-phase flows in par-
allel smooth channels, and found that the stability of system was
affected by the number of channels and the inlet mass flux of
the flow. Kakaç & Cao [11] got the stability boundary by using the

1359-4311/$ e see front matter Ó 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.applthermaleng.2013.12.069
2 Y. Zhang et al. / Applied Thermal Engineering 65 (2014) 1e13

Nomenclature Dp pressure difference

q heat flux
A area, m2 t temperature
C intermediate variable z height
Cw metal specific heat capacity, J/(kg K)
D diameter, m Superscripts
g acceleration due to gravity, m/s2 * static value
G mass flux, kg/(m2 s) 0 intermediate variable in the derivation of transfer
h enthalpy of fluid, J/kg functions
K coefficient of local friction
l length of the tube, m Subscripts
L width of rib, m sr superheating region
N number of ribs of an internally-ribbed tubes tr two-phase flow region
P pressure, Pa pr preheating region
Q heat power per unit length, kW/m in inlet of the tube
s Laplace transfer factor out outlet of the tube
t time, s pt interface between preheating region and two-phase
T temperature, K flow region
V volume of water filled in the internally-ribbed tube, m3 ts interface between two-phase flow region and
x dryness fraction superheating region
Z height, m f water
ExternalInput the external input perturbation of the feedback g steam
system w metal wall
Input the input perturbation of the feedback system wout metal wall at the outlet point
Output the output perturbation of the feedback system hsm heat storage of wall metal
G(s) the forward transfer function ex external of the tube
H(s) the feedback transfer function wpp water physical property coefficient
K(s) the pre-compensator of external input perturbation cir the circumferential direction
avg average
Greek letters
a convective heat transfer coefficient, W/(m2 K) Prefixes
l coefficient of frictional resistance d perturbation value of the dynamic parameter
p pi digits (3.1415926) D difference value
r density, kg/m3
Abbreviation
Non-dimensional numbers CHF critical heat flux
g acceleration due to gravity M-CHF minimum critical heat flux
h enthalpy of fluid WPS the worst point of the stability

drift-flux model in both vertical and horizontal tubes. Later, a
multivariable frequency domain mathematical model was estab-
lished by Hou et al. [12], and was applied in the study on the two-
phase flow instabilities in once-through steam generators, and it
was shown that the methods could conveniently predict the two-
phase flow instabilities in parallel smooth channels. Su et al. [13]
studied flow instabilities in parallel smooth channels by using the
time domain analysis method, and influences of various parameters
on system’s instabilities were also revealed. Similar results also can
be found in some other works [14e18].
Besides the smooth tubes, there are also a few studies on the
two-phase flow instabilities in enhanced heat transfer tubes.
Mentes et al. [19] studied the two-phase flow instabilities in six
vertical enhanced heat transfer tubes, and found that the two-
phase flows in these enhanced heat transfer tubes were more
stable than that in the smooth tubes. Widmann et al. [20] studied
the two-phase flow instabilities in horizontal enhanced heat
transfer tubes, and obtained the similar results. Later, Karsli et al.
[21] also carried out experiments to investigate the two-phase flow
instabilities in five different enhanced heat transfer tubes. All these
studies showed that the difference in geometrical structures of
tubes might cause great differences in the instability behavior of
the two-phase flow.
The internally-ribbed tube, as one of the most widely used
enhanced heat transfer tubes, plays an important role in the design
and flow stability control of the boiler water-cooled wall system
and other heat exchanger systems. Hence, many experimental re-
searches have been conducted to study the instability phenomena
in internally-ribbed tubes. Gao et al. [22,23] proposed four semi-
empirical equations to predict the critical points of two-phase
flow instabilities in the smooth tube (f25  2.5 mm) and in the
internally-ribbed tube (f38.1  7.5 mm), based on the experimental
studies. Moreover, experiments were conducted on high pressure
two-phase test loops by Huang et al. [24], and revealed the in-
fluences of heat flux, asymmetric heat flux, pressure, mass flux,
inlet sub-cooling on density wave oscillation in vertical parallel
internally-ribbed tubes (two internally-ribbed tubes with the same
structures, f31.8  6 mm).
It is worthy to note that most of the previous studies on two-
phase flow instabilities in the internally-ribbed tubes are carried
out by experiments. On the other hand, two-phase flow instability
characteristics in different internally-ribbed tubes are greatly
different from each other due to the different two-phase flows and
heat transfer behaviors caused by different internally-ribbed
structures. It is very necessary to build proper theoretical models
to facilitate the study of the two-phase flow instability
 Y. Zhang et al. / Applied Thermal Engineering 65 (2014) 1e13 3

characteristics in different internally-ribbed tubes. Up to now,

limited theoretical studies of two-phase flow instabilities in
internally-ribbed tubes have been found in published literature,
because of not only the complicated two-phase flow phenomena in
the internally-ribbed tube but also the complex geometric struc-
tures of the kind of tubes. Moreover, conventional models for the
study of the two-phase flow instabilities in tubes in previous
research are mainly established for smooth tubes, and the heat
storage of wall metal is usually neglected in the conventional
models, and the perturbation of external wall heat flux is usually
ignored. In view of the fact that the perturbations of the external
wall heat flux and the wall metal heat storage have a great influence
on the stability of system, the neglect of them may cause unrea-
sonable results in the study of two-phase flow instabilities in tubes,
especially when the heat transfer behaviors and geometry struc-
tures are complex. Therefore, the conventional models are not
appropriate for the study of the two-phase flow instabilities in
internally-ribbed tubes, because different internally-ribbed tubes
have great different geometry structures, and the frictional resis-
tance coefficient and the heat transfer coefficient of it are also
different.
Hence, a new mathematical model was established based on the Fig. 1. Schematic diagram of a tube.

frequency domain theory in the present paper, and was used to

theoretically investigate the two-phase flow instabilities in
ignored. In view of the fact that the perturbations of the external
internally-ribbed tubes.
wall heat flux and the wall metal heat storage have a great influence
In the new mathematical model, two obvious improvements
on the stability of system, the neglect of them may cause unrea-
were made to the conventional model, one of which was that the
sonable results in the study of two-phase flow instabilities in tubes,
heat storage of the tube wall metal, the internal wall heat flux and
especially when the heat transfer behaviors and geometry struc-
the external wall heat flux were considered as dynamic parameters
tures are complex. Therefore, the conventional models are not
which were influenced by the perturbation in system, and another
appropriate for the study of the two-phase flow instabilities in
one was that a new feedback relationship with the new input
internally-ribbed tubes, because different internally-ribbed tubes
perturbation and the new output perturbation was established to
have great different geometry structures, and the frictional resis-
facilitate the study on two-phase flow instabilities in internally-
tance coefficient and the heat transfer coefficient of it are also
ribbed tubes.
different.
In order to verify the feasibility and the reliability of the present
Hence, in the present model, the heat storage of the tube wall
model, two works had been done: firstly, the results of the present
metal, the internal wall heat flux and the external wall heat flux are
model were compared with the experimental results obtained from
considered as dynamic parameters, which are influenced by the
literature [7,23], secondly, the two-phase flow instabilities in tubes
perturbation in the system. The relation of the three dynamic pa-
under different working conditions were simulated by using
rameters is expressed as Eq. (1) according to Fig. 2.
SIMULINK, and the results of the present model were compared
with the simulation results as well. All these verifications proved
qex $pDex ¼ q$pD þ Qhsm (1)
that the present model is reliable and can be employed to predict
the two-phase flow instabilities in tubes.
where, qex is the external wall heat flux, q is the internal wall heat
The present model was then employed to study the two-phase
flux, Dex is the external diameter of a tube, D is the inner diameter of
flow instability behaviors in internally-ribbed tubes, and was also
the smooth tube or the equivalent inner diameter of the internally-
used to study the two-phase flow instability behaviors in a smooth
tube for the comparison. The influences of pressure, mass flux, inlet
resistance coefficient and inlet fluid subcooling on the two-phase
flow instabilities were also investigated with the present model
respectively. Furthermore, the differences in two-phase flow
instability behaviors of different tubes were also analyzed.

2. Mathematical model

Fig. 1 shows the schematic diagram of the tube studied in the

present paper. As shown in Fig. 1, the heating section of a tube is
divided into three regions: preheating region, two-phase flow re-
gion and superheating region. Zpt, Zts are the interfaces which
represent the starting point of boiling and the starting point of
superheating region, respectively. In the present paper, the values
of Zpt and Zts are influenced by the perturbations in the system.
Fig. 2 shows the detailed heat transfer situations at the metal
wall. In conventional models, because of the simple geometry
structures of smooth tubes, the heat storage of wall metal is usually
neglected, and the perturbation of external wall heat flux is usually Fig. 2. Schematic diagram of the heat transfer process.
4 Y. Zhang et al. / Applied Thermal Engineering 65 (2014) 1e13

ribbed tube, Qhsm is the heat storage of wall metal per unit length in 2.1. Fundamental equations
axial direction.
Due to the different structures of a smooth tube and an Dynamic characters of two-phase flows are analyzed by the
internally-ribbed tube, as shown in Fig. 3, the calculation method of equations based on the following assumptions, (1) Subcooling
the heat storage of wall metal (Qhsm) for different tubes is also boiling is neglected; (2) Homogeneous flow model is used to
different. Detailed information of Qhsm for different tubes can be describe flows in the two-phase flow region; (3) The compress-
found in Eqs. (2) and (3), respectively. ibility of liquid in a single phase region can be neglected; (4) Dif-
Heat storage equation of the wall metal of the smooth tube is: ferential pressure from inlet to outlet is constant; (5) Heat flux is
uniform in the axial direction.
The details of the conservation equations are expressed as
"   2 #
vTw vTw Dex 2 D follows.
Qhsm ¼ Cw rw Aw ¼ Cw rw p  (2)
vt vt 2 2
2.1.1. Equations for the preheating region
Heat storage equation of the wall metal of the internally-ribbed
Because of the incompressibility of flows in the one-
tube is:
dimensional single phase region, the conservation equations of
mass, energy and momentum can be expressed as Eqs. (4)e(6)
( "    # respectively. Eq. (7) is the local pressure drop of inlet throttling.
vTw vTw Dex 2 Dmax 2
Qhsm ¼ Cw rw Aw ¼ Cw rw p  Mass conservation equation:
vt vt 2 2
" 2  2 #) vGpr
NLcir Dmax Dmin ¼ 0 (4)
þ $p  vZ
pDmin 2 2
Energy conservation equation:
(3)
vhpr vGpr $hpr Q
where, Cw is the specific heat capacity of the metal wall, rw is the rf þ ¼ (5)
vt vZ A
density of the metal wall, Aw is the area of the cross section of the
metal wall, Tw is the temperature of the metal wall, Dex is the Momentum conservation equation:
external diameter of a tube, D is the inner diameter of a smooth
tube, Dmax is the maximum inner diameter of an internally-ribbed vGpr vG2pr =rf vPpr lpr G2pr
þ ¼   rf $g  $ (6)
tube, Dmin is the minimum inner diameter of an internally-ribbed vt vZ vZ 2D rf
tube, Lcir is the width of rib in the circumferential direction and N
Local pressure drop of the inlet throttling:
is the number of ribs in an internally-ribbed tube.
According to the two heat storage equations, Eqs. (2) and (3), the
Kin G2in
influences of the different heat storage capacities of different tubes DPin ¼ $ (7)
2 rf
are embodied in the present model, which have not been
mentioned in previous models. In these equations, differences in flow areas between smooth
To establish a new mathematical model for the frequency tubes and internally-ribbed tubes are mainly embodied in the
domain analysis of two-phase flow instabilities in the internally- parameter D and A, while difference in flow conditions, especially
ribbed tubes, three steps are needed. The first step is establishing the effect of rotational flow in the internally-ribbed tubes on the
the fundamental equations which can be used to describe the two-phase flow instabilities, are embodied in the coefficient of
physical phenomenon accurately. The second step is getting the frictional resistance lpr. Calculation methods of D, A and lpr are the
transfer functions from the fundamental equations by using Lap- same as that published in literature [26e28], which provide related
lace transform. The last step is obtaining the feedback relations, semi-empirical equations to calculate the coefficients of frictional
according to the classic automatic control theory [25]. The detailed resistance of internally-ribbed tubes. Detailed calculation methods
process of derivation of the model in the present paper is shown as of D, A and lpr are shown in Section 2.1.6.
follows.

2.1.2. Equations for the two-phase flow region

Eqs. (8)e(10) are the conservation equations of an arbitrary
point in the two-phase flow region. Homogeneous flow model is
used in the two-phase flow region, and Lumped parameter method
is adopted in integral processes of conservation equations. The
differences in flow conditions between smooth tubes and
internally-ribbed tubes are solved in a way similar to that for the
preheating region.
Mass conservation equation:

vrtr vGtr
þ ¼ 0 (8)
vt vZ
Energy conservation equation:

vrtr htr vGtr $htr Q

þ ¼ (9)
vt vZ A
Fig. 3. Geometrical structures of a smooth tube and an internally-ribbed tube (cross
section). Momentum conservation equation:
 Y. Zhang et al. / Applied Thermal Engineering 65 (2014) 1e13 5

vGtr vG2tr =rtr vPtr ltr G2tr 2.1.6. Remarks

þ ¼   rtr $g  $ (10) The detailed calculation methods of geometry parameters,
vt vZ vZ 2D rtr
physical properties of two-phase mixture, heat transfer coefficients
and frictional resistance coefficients, which are used in Eqs. (4)e(18),
2.1.3. Equations for the superheating region are shown as follows.
Eqs. (11)e(13) are conservation equations of an arbitrary point Inner equivalent diameter of an internally-ribbed tube [26]:
in the superheating region. Eq. (14) is the local pressure drop of the rffiffiffiffiffiffiffi
outlet throttling, and the outlet throttling is considered as a part of 4V
D ¼ (19)
the superheating region. pl
Mass conservation equation: Flow area of an internally-ribbed tube:
vrsr vGsr
þ ¼ 0 (11) pD2
vt vZ A ¼ (20)
4
Energy conservation equation: Density of two-phase mixture:

vrsr hsr vGsr $hsr Q 1

þ ¼ (12) rtr ¼ (21)
vt vZ A x 1x
rg þ rf
Momentum conservation equation: Enthalpy of two-phase mixture:

htr ¼ hg $x þ hf $ð1  xÞ (22)

vGsr vG2sr =rsr vPsr lsr G2sr
þ ¼   rsr $g  $ (13)
vt vZ vZ 2D rsr Average heat transfer coefficient of an internally-ribbed tube
[26]:
Local pressure drop of the outlet throttling:
GACp ðTout  Tin Þ 1
a ¼   (23)
Kout G2out pDl Twavg  Tavg
DPout ¼ $ (14)
2 rout
Frictional resistance coefficient of two-phase flows in an
internally-ribbed tube [26]:
2.1.4. Equations for the dynamics of heat transfer on the metal wall
rtr 2
According to heat transfer processes shown in Fig. 2, the equa- ltr ¼ $f $l (24)
tions that describe dynamic processes of heat transfer can be
rpr fo pr
written as follows.
Heat storage of metal wall: where, f2fo is the two-phase flow frictional pressure drop factor, lpr
is the frictional resistance coefficient of subcooling water in an
vTw internally-ribbed tube.
Cw rw Aw ¼ Qhsm (15) In this paper, frictional resistance coefficients of single phase in
vt
internally-ribbed tubes are empirical coefficients [26e28].
Convective heat transfer on the internally side of tube wall:

Q 2.2. Equation transformation and transfer functions

ðTw  TÞa ¼
pD
Thermal equilibrium:
Qex ¼ Q þ Qhsm
For Eqs. (15)e(17), convective heat transfer property differences
between smooth tubes and internally-ribbed tubes are mainly
embodied in D and a, volume differences and heat storage property
differences of metal walls are embodied in Aw. Calculation methods
of D and a are the same as that published in literature by Cheng &
Chen [27], Wang et al. [29]. Calculation methods of Aw is shown in
Eqs. (2) and (3).
2.1.5. Equations for the relationship of the total pressure drop and
the pressure drop of every region
According to the fourth assumption, the total pressure drop
between the inlet and outlet is expressed as Eq. (18).
DPall ¼ DPpr þ DPtr þ DPsr
where, DPall is constant. The pressure drop of every region can be
calculated by the momentum equations of fluids in corresponding
regions.
(16)
In previous models for the study of two-phase flow instabilities
in smooth tubes, when fundamental equations are translated into
transfer functions, the pressure, mass flux, enthalpy and tempera-
(17)
ture are considered as dynamic parameters which are influenced by
the perturbation in system, and the heat storage of wall metal,
internal heat flux and external flux are dealt with as constants
[8,12,30]. But the heat transfer behaviors of two-phase flows in
internally-ribbed tubes are more complex than that in smooth
tubes, the previous models are not appropriate for the study of the
two-phase flow instabilities in internally-ribbed tubes.
Hence, to well present the complex heat transfer behaviors of
two-phase flows in internally-ribbed tubes under dynamic working
conditions, the heat storage of the tube wall metal, the internal wall
heat flux and the external wall heat flux are considered as dynamic
parameters in the present model.
To get the transfer functions, dynamic parameters, which are
influenced by the perturbation, need to be written in forms as
follows:
(18)
Q ¼ Q * þ dQ (25)
Eq. (25) shows the heat flux of the system in the dynamic
working condition. Q is the heat power per unit length in axial
direction under dynamic working condition, Q* is the heat power
6 Y. Zhang et al. / Applied Thermal Engineering 65 (2014) 1e13

per unit length in axial direction under the corresponding static Table 2
working condition and dQ is the perturbation of heat power per Detailed equations of intermediate variables.

unit length in axial direction. Other dynamic parameters can be C11 ¼ rf ðhin  hf Þz*pt =2 C12 ¼ Q * Zpt =A
expressed in the same forms. C13 ¼ ðhf  hin Þ$G*pt C14 ¼ Q * Zpt =A
In order to facilitate the analysis, fifteen non-dimensional C15 ¼ G*pt Zpt
* * G*2 =ðDr Þ þ K G*2 =r
C16 ¼ lpr Zpt pt f in pt f
parameters (Shown in Table 1) are defined, which represent the C17 ¼ pr Zpt G*2
* * C18 ¼ DPpr
*
l pt =ð2Drf Þ þ rf gZpt
dynamic parameters in system respectively. zpt represents the start *  Z * Þ=2
C21 ¼ ðrf  rg ÞðZts * =2
C22 ¼ ðrf  rg ÞZpt
pt
boiling point perturbation, zts represents the perturbation of the C23 ¼ G*ts C24 ¼ G*pt
start point of superheating region, gin represents the perturbation of *  Z * Þ=2
C25 ¼ ðrf hf  rg hg ÞðZts pt C26 ¼ Q * ðZts
*  Z * Þ=A
pt
inlet mass flux, gpt represents the perturbation of mass flux at the * =2
C27 ¼ ðrf hf  rg hg ÞZpt C28 ¼ Q * Zpt
* =A
start boiling point, gts represents the perturbation of mass flux at the
C29 ¼ G*ts hg C210 ¼ G*pt hf
start point of superheating region, gout represents the perturbation
C211 ¼ Q * ðZts
*  Z * Þ=A
pt C212 ¼ DPtr
*

of outlet mass flux, Dppr represents the perturbation of pressure C213 ¼ ðrf þ rg ÞgðZts*  Z * Þ=2 þ l ðG*2 =r þ G*2 * *
pt tr ts g pt =rf ÞðZts  Zpt Þ=ð4DÞ
drop of subcooling region, Dptr represents the perturbation of * =2  l ðG*2 =r þ G*2 =r ÞZ * =ð4DÞ
C214 ¼ ðrf þ rg ÞgZpt tr ts g pt f pt
pressure drop of two-phase flow region, Dpsr represents the *  Z * ÞG*2 =ð2Dr Þ
C215 ¼ ltr ðZts pt ts g
*  Z * ÞG*2 =ð2Dr Þ
C216 ¼ ltr ðZts pt pt f
perturbation of pressure drop of superheating region, q represents C31 ¼ G*out h*out C32 ¼ G*out h*out
the perturbation of internal wall heat flux, qex represents the C33 ¼ G*ts hg C34 ¼ Q * ðZout  Zts
* Þ=A

perturbation of external wall heat flux, qhsm represents the pertur- C35 ¼ Q * ðZts
*  Z * Þ=A
pt C36 ¼ DPsr
*

bation of heat storage of wall metal, hout represents the perturbation C37 ¼ fðrg þ rout Þg=2 þ lsr G*2 * *
out =½Dðrg þ rout ÞgðZts  Zpt Þ
* ÞG*2 =½Dðr þ r
C38 ¼ 2lsr ðZout  Zts *2
of the enthalpy of fluid at outlet, tout represents the perturbation of out g out Þ  Kout Gout =rout
0
C41 0 C0
¼ C32 C17 0
C42 0 C0
¼ C32 C17
the temperature of fluid at outlet, twout represents the perturbation 0
36
0 C0 0
37
0 C0 þ C0 C0
C43 ¼ C32 C17 C53 ¼ C48
of the temperature of tube wall at outlet. These non-dimensional 0
38
0 C0 þ C0 C0 0
51 41 52
0 C0 þ C0 C0
C54 ¼ C49 51 42 52 C55 ¼ C50 51 43 52
parameters equal to the ratios of value differences (value differ- 0 * * Þ
C56 ¼ Cw rw Aw Twout =ð2Qex
ences between dynamic values and static values) and static values.
In fact, these non-dimensional parameters are also equal to the
ratios of small perturbation values and static values. This can be
described by Eq. (26), and the other non-dimensional parameters
C21 $s$zts þ C22 $s$zpt þ C23 gts þ C24 gpt ¼ 0 (30)
are similar.

Gin  G*in dGin ðC25 $s þ C26 Þzts þ ðC27 $s þ C28 Þzpt þ C29 gts þ C210 gpt þ C211 q ¼ 0
gin ¼ ¼ (26)
G*in G*in (31)
C212 Dptr þ C213 zts þ C214 zpt þ C215 gts þ C216 gpt ¼ 0 (32)
To get the transfer functions, three steps are carried out one by
one. Firstly, the fundamental differential equations are integrated
gout ¼ gts (33)
in their own regions. Secondly, the small perturbations are taken
into the system, and the fundamental equations are translated into
perturbation equations, and higher order infinitesimals of small C31 gout þ C32 hout þ C33 gts þ C34 q þ C35 zts ¼ 0 (34)
perturbations are neglected. Thirdly, instead of dimensional pa-
rameters, non-dimensional parameters are used in these equations, C36 Dpsr þ C37 zts þ C38 gout ¼ 0 (35)
and then Laplace transform is carried out.
Eqs. (27)e(39) show the detailed transfer functions derived
* * *
from the fundamental equations. In these equations, all the non- Dppr $DPpr þ Dptr $DPtr þ Dpsr $DPsr ¼ 0 (36)
dimensional parameters are functions of s (s is the Laplace trans-
fer factor) instead of time, because of Laplace transform. 0
C56 $s$twout ¼ qhsm (37)

gin ¼ gpt (27) 0 0

C51 $tout þ C52 $q ¼ twout (38)

ðC11 $s þ C12 Þzpt þ C13 gpt þ C14 q ¼ 0 (28) qex ¼ q þ qhsm (39)

C16 gpt þ C17 zpt þ C18 Dppr ¼ 0 (29) where, C11 w C18 are the intermediate variables for preheating re-
gion, C21 w C216 are the intermediate variables for two-phase flow
region, C31 w C38 are the intermediate variables for superheating
region, and C10 wC56
0 are the intermediate variables for other equa-

tions. The calculation equations of the key intermediate variables

Table 1
Non-dimensional parameters in the present paper.
are shown in Table 2.
*
Zpt Zpt Zts Zts* Gin G*in
zpt ¼ *
zts ¼ * *
Zpt
gin ¼ G*in
Zpt Zin Zts

gpt ¼
Gpt G*pt
gts ¼
Gts G*ts
gout ¼
Gout G*out 2.3. Feedback relations
G*pt G*ts G*out
DP DP *
Dptr ¼ DPtrDP *DPtr Dpsr ¼ DPsrDP*DPsr
* *
Dppr ¼ prDP* pr sr To form the feedback relations, transfer functions must be
pr tr

Q Q * *
Qex Qex qhsm ¼ Qhsm written in the classic forms like Eqs. (40) and (41).
q ¼ Q* qex ¼ * *
Qex
Qex
hout h*out *
Tout Tout *
Twout Twout
hout ¼ tout ¼ twout ¼
h*out *
Tout *
Twout
Output ¼ GðsÞ$Input (40)
 Y. Zhang et al. / Applied Thermal Engineering 65 (2014) 1e13 7

Input ¼ KðsÞ$ExternalInput  HðsÞ$Output (41)

where, G(s) is the forward transfer function, H(s) is the feedback

transfer function, K(s) is the pre-compensator of external input
perturbation.
Theoretically, choosing different parameters as external input
perturbation, input perturbation and output perturbation does not
influence the stability of the system, but the chosen must fit the
research system and the assumptions in the model, or else the
Fig. 4. Feedback relation of the system.
prediction of two-phase flow instabilities by using the model may
be great unreasonable.
In the conventional model, the perturbation of pressure drop is Table 3
usually defined as the input perturbation, and the perturbation of Working conditions of the two-phase flows in the smooth tube [7].
mass flux is defined as the output perturbation, but this method
Parameter Range
may not reflect the complex heat transfer behaviors in internally
ribbed tubes well. So in the present model, the external input Pressure (MPa) 2e4.3
Heat flux (kW/m2) 260e748
perturbation, input perturbation and output perturbation are
Mass flux (kg/m2 s) 150e350
determined in specialty, i.e., qex is defined as the external input Inlet fluid subcooling ( C) 20e225
perturbation, qhsm is defined as the input perturbation and twout is Inlet resistance coefficient 300e1200
defined as the output perturbation. According to the later verifi-
cations of the present model, it is proved that the new de-
terminations of the external input perturbation, input perturbation
and output perturbation in the present model are competent for FðsÞ ¼ 1 þ GðsÞHðsÞ ¼ 0 (48)
the prediction of the two-phase flow instabilities in internally-
The open-loop transfer function can be expressed:
ribbed tubes.
To change the transfer functions into the classic forms as Eqs. 0 $s2 0 $s þ C 0
þ C42
1 C41 43
(34) and (35), some simple transformations are carried out as GðsÞHðsÞ ¼ 0 $ 0 0 $s þ C 0
(49)
C56 $s C53 $s2 þ C54 55
follows.
From Eq. (37): Based on Nyquist stability criterion, the necessary and sufficient
conditions of the system’s stability is [12]:
1
twout ¼ 0 $s qhsm (42)
C56 Nd ¼ P0 (50)
Solve equations from Eqs. (27) to (36): where, Nd is the number of anticlockwise encirclements of the
0 $s2 þ C 0 $s þ C 0
point (1, j0) by Nyquist locus, P0 is the total number of the poles
C41 42 43 (poles of the open-loop transfer function) in the right-half plane.
q ¼ 0 $s2 þ C 0 $s þ C 0
twout (43)
C53 54 55 According to this necessary and sufficient conditions of the
system’s stability, the stability of two-phase flows in tubes can be
According to Eqs. (39) and (43):
predict easily, if the necessary transfer functions of the system are
0 $s2 þ C 0 $s þ C 0 determined.
C41 42 43
qhsm ¼ qex  0 $s2 þ C 0 $s þ C 0
twout (44) To facilitate the analysis of two-phase flow instabilities in tubes,
C53 54 55 a computer code was also developed by using both FORTRAN and
Eqs. (42) and (44) are transfer functions which are just in the MATLAB.
classic forms. Then the forward transfer function, the feedback
transfer function and the pre-compensator function of external 4. Verifications of the present model
input perturbation can be expressed as follows:
In order to verify the mathematical model and the code made in
1 this study, the results of the present model are firstly compared
GðsÞ ¼ 0 $s (45)
C56 with the experiment results which are got from literature [7,23].
Then, a simulation system is established by using SIMULINK, and
0 $s2 þ C 0 $s þ C 0
C41 the results of the present model are further compared with that
42 43
HðsÞ ¼ 0 $s2 þ C 0 $s þ C 0
(46) obtained by the simulation.
C53 54 55

4.1. Verifications by experimental results

KðsÞ ¼ 1
Hence, the new feedback relation can be exposed in Fig. 4, ac-
cording to classic automatic control theory [25].
3. Stability analysis method and Nyquist stability criterion
From the feedback relation, which is shown in Fig. 4, the char-
acteristic equation of system is:
(47)
Some typical experimental data are collected from literature for
the verifications. Takitani & Takemura [7] experimentally studied
two-phase flow instabilities in the smooth tube, the length, outer
diameter and inner diameter of which are 5.250 m, 0.015 m and
0.0125 m respectively. Gao & Luo [23] carried out experiments to
study the two-phase flow instabilities in a 6-head internally-ribbed
tube (f38.1  7.5 mm), the length of which is 15 m.
The two-phase flow instability phenomena in the smooth tube
are calculated and analyzed by using the present model, with the
8 Y. Zhang et al. / Applied Thermal Engineering 65 (2014) 1e13

same working conditions with Takitani & Takemura [7], shown in

Table 3. For the internally-ribbed tube, the same working condi-
tions with Gao & Luo [23] are employed, shown in Table 4.

4.1.1. Two-phase flow instabilities in a smooth tube

Fig. 5 is the comparison of the result of the present model to that
of the experiment of Takitani & Takemura [7]. Curves in Fig. 5 are
the stability boundaries of system pressure varying with inlet
resistance coefficient. The stability boundary represents the critical
working conditions, and it divides the working condition region
into two parts, i.e., the stable region and the unstable region. The
trend and value of the stability boundary can represent the stability
behaviors of the two-phase flows in tubes. In Fig. 5, the smooth
curve represents the boundary from the results of the present
model, and the other one represents the boundary from the ex-
periments of Takitani & Takemura [7]. Generally, these two results
curves have the similar trends, and the physical conclusions are the
same, that is, the system may become more stable with the rising of
pressure and inlet resistance coefficient. Although the results of the Fig. 5. Comparison of the result of the present model to that of the experiment
present model are approximately 15% higher than that of Takitani’s, (Takitani and Takemura, 1978).
such a deviation of the calculation results from the experimental
data doesn’t affect the verification conclusion, because parts of data
of Takitani & Takemura’s on static parameters are snapped from the the present model and the simulation system respectively. Table 5
diagrams in the corresponding literature, and possibly, limited er- gives detailed parameters of three working conditions.
rors may exist in the snapped data. Therefore, the present model
and the related code are generally quite reliable for the study of the
4.2.1. Case 1
two-phase flow instabilities in smooth tubes.
Fig. 8 shows the Nyquist analysis of Case 1. It can be seen from
Fig. 8 that the Nyquist curve surrounds the point (1, j0) twice
4.1.2. Two-phase flow instability in an internally-ribbed tube
along the anticlockwise direction. This means Nd ¼ 2 (Key param-
Fig. 6 is the comparison of the result of the present model to that
eter in Nyquist stability criterion, the detailed meaning is shown in
of the experiment of Gao et al. [23]. Curves in Fig. 6 are the stability
Eq. (50)). According to the present model, the detailed values of
boundaries. The results of the present model are approximately
these poles equal to 0, 0.40651 þ 2.31647i, 0.40651  2.31647i
equal to the experimental results that are published by Gao et al.
respectively. Two of these poles are in the right side of imaginary
[23]. The two curves also have the similar trends, despite there is a
axis, that means P0 ¼ 2. According to Nyquist stability criterion,
little difference of values between them. This difference may be
Nd ¼ P0 ¼ 2 means Case 1 agrees with the necessary and sufficient
caused by the deviations of semi-empirical equations [26e29],
condition of stability, so the closed-loop system is stable in Case 1.
which are used for this f38.1  7.5 mm internally-ribbed tube.
Fig. 9 shows the response curves of main parameters of Case 1
Overall, the conclusions from the experimental results and the
that are got from the simulation system. A small step disturbance of
results of the present model agree with each other well. Therefore,
the external heat power is added to the system in the beginning of
it is powerful proved that the present model and code can predict
the simulation. Then, the response characteristics of main param-
the two-phase flow instabilities in the internally-ribbed tubes.
eters (Such as Gout, Tout and Twout) are revealed compendiously.
From Fig. 9, the fluctuations of Gout, Tout and Twout become smaller
4.2. Verifications by simulation results

A simulation system for the two-phase flow instability study is

established by using SIMULINK, since SIMULINK is usually used for
dynamic simulation, and it can simulate the transfer processes of
dynamic signals in a system compendiously.
In this simulation system, qex is the external input perturbation,
q is the input perturbation and twout is the output perturbation. The
detailed schematic diagram of this simulation system is shown as
Fig. 7.
Three representative working conditions (stable condition,
critical condition and unstable condition) are analyzed, based on

Table 4
Working conditions of the two-phase flows in the internally-ribbed
tube [23].

Parameter Range

Pressure (MPa) 3e10

Heat flux (kW/m2) 0e190
Mass flux (kg/m2 s) 300e600
Inlet fluid subcooling ( C) 30e110
Fig. 6. Comparison of the result of the present model to that of the experiment (Gao
Inlet resistance coefficient 0
et al., 2005b).
 Y. Zhang et al. / Applied Thermal Engineering 65 (2014) 1e13 9

Fig. 7. Simulation system of two-phase flow instabilities in tubes.

and smaller with time until disappearing. It means that the closed-
Table 5 loop system of Case 1 is stable.
Detail parameters of three working conditions.
It can be concluded from Figs. 8 and 9 that the system is stable in
Parameter Case1 Case 2 Case 3 Case 1. The present model can predict the stable condition well.
Total length of the channel (m) 5.250 5.250 5.250
Inner diameter (mm) 12.5 12.5 12.5
Outer diameter (mm) 15 15 15 4.2.2. Case 2
Pressure (MPa) 3.8 3.8 3.8 Fig. 10 shows the Nyquist analysis of Case 2. The curve passes
Heat power (kW) 76.24 79.83 80.23 through the point (1, j0) twice along the anticlockwise direction,
Mass flux (kg/m2 s) 240 240 240
Inlet fluid subcooling ( C) 152 152 152
i.e., Nd ¼ 2. According to the present model, the poles of the open-
Inlet resistance coefficient 600 600 600 loop transfer function equal to 0, 0.59153 þ 2.29130i,
Working condition Stable Critical Unstable 0.59153  2.29130i, i.e., P0 ¼ 2. Despite Nd ¼ P0, Case 2 is the critical
condition, because Nd is the times that the Nyquist curve passes
through the point (1, j0).
Fig. 11 shows the response curves of Case 2. After importing a
small perturbation to the external heat power, Tout and Twout rise
slowly in the form of fluctuation. Seconds later, self-sustaining

Fig. 8. Result of Nyquist analysis of Case 1 by using the present model. Fig. 9. Result of response curves of a step disturbance in Case 1 by using SIMULINK.
10 Y. Zhang et al. / Applied Thermal Engineering 65 (2014) 1e13
Fig. 10. Result of Nyquist analysis of Case 2 by using the present model.
fluctuations occur, with the same frequency and amplitude. From
Fig. 11, the oscillation period of this system equals to 2.89 s, which
means a classical density wave oscillation.
It can be concluded from Figs. 10 and 11 that the system in Case
2 is at the critical state and the critical heat power equals to
79.83 kW. The present model can predict the critical condition well.
4.2.3. Case 3
Fig. 12 shows the Nyquist analysis of Case 3. The Nyquist curve
does not surround point (1, j0), and Nd ¼ 0. According to the
present model, the poles of the open-loop transfer function are 0,
0.61067 þ 2.28856i, 0.61067  2.28856i, and P0 ¼ 2. Nd sP0 means
the closed-loop system is unstable.
Fig. 13 shows the response curves of Case 3. Oscillation ampli-
tudes of Tout, Twout and Gout become bigger and bigger with time,
especially after the 100th second. Obviously, this system is
unstable.
It can be concluded from Figs.12 and 13 that the system is unstable
in Case 3. The present model can predict the unstable condition well.
Generally all results of the present model agree well with that
obtained by SIMULINK. It indicates indirectly that the present
model can be used to predict the two-phase flow instabilities in
tubes accurately.
Fig. 11. Result of response curves of a step disturbance in Case 2 by using SIMULINK.
Fig. 12. Result of Nyquist analysis of Case 3 by using the present model.
5. Effect of flow parameters on the two-phase flow
instabilities in internally-ribbed tubes
After the verifications, two-phase flow instabilities in internally-
ribbed tubes were studied systematically by using the present
model. As contrasts, study of two-phase flow instabilities in smooth
tubes was made to verify the difference between two-phase flow
instabilities in internally-ribbed tubes and that in smooth tubes.
In the present paper, three representative tubes are studied,
which are widely used in manufacture of boiler water-cooled walls.
Table 6 gives the detailed parameters of tubes, and Table 7 shows
the ranges of working variables.
5.1. Effects of inlet resistance coefficient
According to the definition of CHF, CHF can be considered as the
demarcation point of stable region and unstable region [24], and
CHF is widely used to represent the two-phase flow instability
behaviors in tubes [22e24,31,32]. The system is unstable when the
heat flux is greater than CHF, and it is stable when the heat flux is
lower than CHF. A system with high CHF means a large stable re-
gion, i.e., this system can be more stable than other systems. In this
paper, the effect of inlet resistance coefficient on two-phase flow
instabilities is studied by using the CHF parameter.
Fig. 13. Result of response curves of a step disturbance in Case 3 by using SIMULINK.
 Y. Zhang et al. / Applied Thermal Engineering 65 (2014) 1e13 11

Table 6 that of Tube 1 (the smooth tube, f28  6 mm), while CHF of Tube 3
Detailed parameters of the three tubes. (the 6-head ribs internally-ribbed tube, f38.1  7.5 mm) is 50%
Parameter Tube 1 Tube 2 Tube 3 greater than that of Tube 1. Two-phase flows in the internally-
Type Smooth Internally- Internally-
ribbed tubes show greater stability than that in the smooth one.
tube ribbed tube ribbed tube Further analyses of the stability differences of the two-phase flows
Outside diameter (mm) 28 28 38.1 in these tubes are showed in Section 5.5 in specialty.
Inside diameter (mm) 16 17.1 23.23
Maximum wall thickness 6 5.41 7.5
(mm) 5.2. Effects of inlet fluid subcooling
Number of ribs / 4 6
Rib width (mm) / 9 5.59
Rib height (mm) / 0.81 0.99
Inlet fluid subcooling is one of the factors which may affect the
Pitch (mm) / 21 21.06 flow instabilities, and how does it work is a meaningful question.
Lead (mm) / 92.53 126.37 Keeping mass flux, inlet resistance coefficient and pressure as
Lead angle ( ) / 61.5 60 constants, values of inlet fluid subcooling from 10  C to 300  C are
selected for this analysis. According to the present model, CHF can
be determined and detailed CHF curves are shown in Fig. 15.
Table 7 Every CHF curve shown in Fig. 15 has an extreme point, which is
Ranges of working variables. the worst point of the stability (WPS) of two-phase flows. In the left
side of the extreme point, CHF become lower with the inlet fluid
Parameter Range
subcooling, but in the right side, CHF become greater with the inlet
Pressure (MPa) 2e15
fluid subcooling. The WPS corresponds to the Minimum Critical
Heat flux (kW/m2) 0e800
Mass flux (kg/m2 s) 200e1500
Heat Flux (M-CHF) of two-phase flows in tubes, while it represents
Inlet fluid subcooling ( C) 10e300 that the two-phase flows at this inlet fluid subcooling show the
Inlet resistance coefficient 0e600 worst stability. According to Fig. 15, M-CHFs of the three tubes equal
to 53.4  C, 51.7  C and 72.6  C respectively.
Physically, M-CHF is the point where the proportion of two-
Keeping mass flux, inlet fluid subcooling and pressure as con- phase flow region occupies the most major status in the system,
stants, inlet resistance coefficient from 0 to 600 are selected for this and the influences of two-phase region are more greater than other
analysis. Values of CHFs under different working conditions in parts in this condition. That means two-phase flow instability may
every tube are calculated by using the present model, and then the occur more easily at M-CHF. When inlet fluid subcooling is lower
CHF curves are formed, shown in Fig. 14. than M-CHF, the lower the inlet fluid subcooling is, the greater the
The CHF curves, shown in Fig. 14, increase with the inlet resis- proportion of preheating region may become, and the more stable
tance coefficient for all three tubes. This means the inlet resistance the system may be, because the non-compressible water in pre-
coefficient plays a positive role in the system’s stability. The heating region plays a stable role in the system. When inlet fluid
consequence agrees with the physical mechanism well. Because the subcooling is greater than M-CHF, the higher the inlet fluid sub-
rising of inlet resistance coefficient may cause the inlet pressure cooling is, the greater the proportion of preheating region may
drop increasing, then the increasing of inlet pressure drop may become, and the more stable the system may be, because the
reduce the proportion of the two-phase flow pressure drop rela- superheating region is also non-compressible according to the
tively. That means the influences of two-phase flow on the stability hypothesis in this paper. Hence, the results based on the present
of system decrease, and it may make the system more stable finally. model in Fig. 15 reveal the physical phenomena compendiously and
So the two-phase flows in the system may become more stable exactly.
with the increasing of inlet resistance coefficient, based on the Generally, in the design and operating process of a boiler sys-
calculations and analyses here. tem, M-CHF is a notable point to be avoided, and the value of M-
In the situations showed in Fig. 14, CHF of Tube 2 (the 4-head CHF can be determined by using the present model.
ribs internally-ribbed tube, f28  5.41 mm) is 10% greater than

Fig. 14. The effects of inlet resistance coefficient on CHF. Fig. 15. The effects of inlet fluid subcooling on CHF.
12 Y. Zhang et al. / Applied Thermal Engineering 65 (2014) 1e13
5.3. Effects of pressure
As is well-known, the physical properties of water are changing
with the pressure. The lower the pressure is, the more obvious the
two-phase flow characters may be. So the working pressure may also
affect the system stability. Keeping mass flux, inlet resistance coeffi-
cient and inlet fluid subcooling as constants, values of pressure from
3 MPa to 15 MPa are used for this analysis. The CHF curves are got
based on the present model. Detailed CHF curves are shown in Fig. 16.
As shown in Fig. 16, no matter the smooth tube or the internally-
ribbed tubes, the CHF curves increase with the pressure. As the
pressure rise from 3 MPa to 15 MPa, CHFs of three tubes increase by
25.63%, 24.09% and 33.24% respectively. It means that the stabilities
of the two-phase flows in tubes intensify obviously with the
pressure increasing. With the pressure increasing, differences of
density between gas and liquid decrease, the characters of two-
phase flow become faint. So the higher the pressure is, the more
stable the two-phase flows in tubes may be.
5.4. Effects of mass flux
Keeping inlet resistance coefficient, inlet fluid subcooling and
pressure as constants, values of mass flux from 200 kg/m2 s to
1200 kg/m2 s are used for this analysis. For Tube 3, there are no
results with 1100 kg/m2 s and 1200 kg/m2 s, because the CHFs are
so high that the temperature of metal wall is out of range in these
conditions. According to this model, the CHF curves are made,
detailed in Fig. 17.
Fig. 17 shows that the CHF curves rise with the mass flux
conspicuously. It means that, the mass flux is one of the most
important factors which determine the stability of two-phase flows
in a system.
Theoretically, the increasing of mass flux will cause the increasing
of Zpt, which equals to the preheating region length, when other
parameters remain unchanged. As is well-known, the preheating
region only contains subcooling water, which is non-compressible.
The more the non-compressible liquid exists, the more stable the
flows in the system are. So it is reasonable that the rising of the mass
flux may make the flows in the system more stable.
5.5. Differences between internally-ribbed tubes and smooth tubes
From Figs. 14 to 17, the results show that the geometry param-
eters play an important role in the system stability. According to
Fig. 16. The effects of pressure on CHF.
Fig. 17. The effects of mass flux on CHF.
CHF curves, Tube 2 (the 4-head ribs internally-ribbed tube,
f28  5.41 mm) and Tube 3 (the 6-head ribs internally-ribbed tube,
f38.1  7.5 mm), two internally-ribbed tubes, both show better
stability than Tube 1 (the smooth tube, f28  6 mm). Under the
same working condition, CHF of Tube 2 is 5%e10% greater than that
of Tube 1 approximately, and CHF of Tube 3 is 40%e60% greater
than that of Tube 1 approximately, and CHF of the Tube 3 was 30%e
50% greater than that of Tube 2. Flows in Tube 3, show the best
stability in the three tubes. Internally-ribbed tubes, with rotational
flow influences and fantastic internal wall structures, have greater
frictional resistances than smooth tubes, so this character may
cause the two-phase flows in internally-ribbed tubes show more
stable. Other factors, such as differences of heat transfer property
and differences of local resistance, also affect the two-phase flow
instabilities in tubes respectively.
Generally, two-phase flows in internally-ribbed tubes express the
greater stability than that in smooth tubes. Two-phase flows in
different internally-ribbed tubes show great differences of stability,
because the different geometrical structures of the internally-ribbed
tubes. Under the same working conditions, two-phase flows in the 6-
head ribs internally-ribbed tube (f38.1  7.5 mm) are more stable
than that in the 4-head ribs internally-ribbed tube (f28  5.41 mm).
6. Conclusions
A new mathematical model is established in the present paper
for the study of two-phase flow instabilities in tubes by using fre-
quency domain method. In the new mathematical model, two
obvious improvements are made to the conventional models, one
of which is that the heat storage of the tube wall metal, the internal
wall heat flux and the external wall heat flux are considered as
dynamic parameters which are influenced by the perturbation in
system, that is, the different flow and heat transfer behaviors in
internally-ribbed tubes are taken into account especially; another
one is that a new feedback relationship is established for the study
of two-phase flow instabilities in internally-ribbed tubes. Then the
present model is verified against the experimental data in
literature and the simulation results by SIMULINK. By using the
present model, two-phase flow instabilities in both internally-
ribbed tubes and smooth tubes are analyzed, and the following
conclusions can be reached:
1) According to the comparison among the results got from the
new mathematical model, simulation system and other
 Y. Zhang et al. / Applied Thermal Engineering 65 (2014) 1e13
scholars’ experiments, the new mathematical model established
by using the frequency domain theory in the present paper is
proved to be accurate, and can be used to study two-phase flow
instabilities in both smooth tubes and internally-ribbed tubes.
2) Two-phase flows in tubes tend to be more stable, if the pressure,
mass flux or inlet resistance coefficient in the tube increase.
Among the three factors, mass flux is the most influential
parameter to raise the stability of two-phase flows in tubes.
3) Inlet fluid subcooling of the flow in a tube affects the two-phase
flow instabilities greatly. According to the present study, there
exists a Minimum Critical Heat Flux (M-CHF), and the M-CHF
represents the point where the two-phase flows in a tube show
the worst stability. In the design and operating process of a
boiler system, the M-CHF is a notable point to be avoided, and
the value of M-CHF can be determined by using the present
model.
4) Because of geometrical structure differences of tubes, the
instability behaviors of two-phase flows are greatly different in
different tubes, according to the present study. Two-phase flows
in both the 4-head ribs internally-ribbed tube (f28  5.41 mm)
and the 6-head ribs internally-ribbed tube (f38.1  7.5 mm)
show better stability than that in the smooth one (f28  6 mm).
Under the same working condition, the CHF of the 6-head ribs
internally-ribbed tube (f38.1  7.5 mm) is 30%e50% greater
than that of the 4-head ribs internally-ribbed tube
(f28  5.41 mm).
Acknowledgements
This work is supported by National Basic Research Program of
China (973 Program) (grant no. 2009CB219805) and National Nat-
ural Science Foundation of China (grant no. 50876090). The authors
thank the staff at the Division of Two Phase Flow and Heat Transfer
in High Pressure and High Temperature for their constructive dis-
cussions and suggestions.
References
[1] N. Brauner, D. Moalem Maron, Stability analysis of stratified liquideliquid
flow, Int. J. Multiphase Flow 18 (1992) 103e121.
[2] K. Fukuda, T. Kobori, Classification of two-phase flow instability by density
wave oscillation model, J. Nucl. Sci. Technol. 16 (1979) 95e108.
[3] R. Khodabandeh, R. Furberg, Instability, heat transfer and flow regime in a
two-phase flow thermosyphon loop at different diameter evaporator channel,
Appl. Therm. Eng. 30 (2010) 1107e1114.
[4] L. Lin, A. Faghri, An analysis of two-phase flow stability in a thermosyphon
with tube separator, Appl. Therm. Eng. 18 (1998) 441e455.
[5] M. Ozawa, K. Akagawa, T. Sakaguchi, Flow instabilities in parallel-channel flow
systems of gaseliquid two-phase mixtures, Int. J. Multiphase Flow 15 (1989)
639e657.
[6] J. Boure, A. Bergles, L.S. Tong, Review of two-phase flow instability, Nucl. Eng.
Des. 25 (1973) 165e192.
[7] K. Takitani, T. Takemura, Density wave instability in once-through boiling flow
system, (I) experiment, J. Nucl. Sci. Technol. 15 (1978) 355e364.
13
[8] K. Takitani, Density wave instability in once-through boiling flow system, (II)
lumped parameter model with moving boundaries, J. Nucl. Sci. Technol. 15
(1978) 389e399.
[9] K. Takitani, K. Sakano, Density wave instability in once-through boiling flow
system, (III) distributed parameter model, J. Nucl. Sci. Technol. 16 (1979) 16e29.
[10] L. Pustylnik, D. Barnea, Y. Taitel, Prediction of two-phase flow distribution in
parallel pipes using stability analysis, AIChE J. 52 (2006) 3345e3552.
[11] S. Kakaç, L. Cao, Analysis of convective two-phase flow instabilities in vertical
and horizontal in-tube boiling systems, Int. J. Heat Mass Transfer 52 (2009)
3984e3993.
[12] S. Hou, Y. Tai, F. Zhao, X. Wei, Instability of OTSG in movable NPP by using
multivariable frequency domain method, Nucl. Sci. Tech. 21 (2010).
[13] Y. Su, J. Feng, H. Zhao, W. Tian, G. Su, S. Qiu, Theoretical study on the flow
instability of supercritical water in the parallel channels, Prog. Nucl. Energy 68
(2013) 169e176.
[14] D. France, R. Carlson, R. Roy, Measurement and analysis of dynamic in-
stabilities in fluid-heated two-phase flow, Int. J. Heat Mass Transfer 29 (1986)
1919e1929.
[15] R. Lahey, Advances in the analytical modeling of linear and nonlinear density-
wave instability modes, Nucl. Eng. Des. 95 (1986) 5e34.
[16] K. Fukuda, Y. Rao, S. Hasegawa, K. Nakagawa, K. Kage, Study on two-phase
flow instability in parallel multi-channels, Heat Transfer e Japanese Res. 23
(1996).
[17] O. Yokomizo, Extended equivalent linearization method for density wave
instability analysis, J. Nucl. Sci. Technol. 27 (1990) 230e249.
[18] O. Yokomizo, M. Sakurai, Y. Yoshimoto, K. Kitayama, T. Enomoto, N. Fukuda,
K. Chuman, Analysis of BWR core nuclear thermal hydraulic oscillation with
three dimensional transient program, Nucl. Eng. Des. 99 (1987) 41e51.
[19] D.-I.A. Mentes, D.-I.O. Yildirim, I.H. Gürgenci, S. Kakaç, T. Veziro, Effect of heat
transfer augmentation on two-phase flow instabilities in a vertical boiling
channel, Wärme-und Stoffübertragung 17 (1983) 161e169.
[20] F. Widmann, O. Comakli, C.O. Gavrilescu, Y. Ding, S. Kakac, The effect of
augmented surfaces on two-phase flow instabilities in a horizontal system,
J. Enhanced Heat Transfer 2 (1995).
[21] S. Karsli, M. Yilmaz, O. Comakli, The effect of internal surface modification on
flow instabilities in forced convection boiling in a horizontal tube, Int. J. Heat
Fluid Flow 23 (2002) 776e791.
[22] F. Gao, T.-K. Chen, Y.-S. Luo, F. Yin, W.-M. Liu, Experimental research on
density wave oscillation of steamewater two-phase flow in parallel inclined
internally ribbed pipes [J], Nucl. Power Eng. 1 (2005) 002.
[23] F. Gao, Y.-S. Luo, T.-K. Chen, D.-H. Lu, H.-X. Li, Comparative research on
steamewater two-phase flow instabilities between internally ribbed tubes
and smooth tubes, Nucl. Power Eng. 26 (2005) 335.
[24] F. Huang, Y.-S. Luo, T.-K. Chen, H.-J. Wang, C.-F. Li, Z.-A. Deng, Experiment on
density wave oscillation instability of two-phase flow in low mass velocity
parallel vertical rifled tubes, Nucl. Power Eng. 1 (2010) 011.
[25] J. Zhang, Control of Nuclear Reactor, Xi’an Jiaotong University Press, Xi’an,
China, 2002.
[26] L. Cheng, T. Chen, Flow boiling heat transfer in a vertical spirally internally
ribbed tube, Heat Mass Transfer 37 (2001) 229e236.
[27] L. Cheng, T. Chen, Study of single phase flow heat transfer and friction pres-
sure drop in a spiral internally ribbed tube, Chem. Eng. Technol. 29 (2006)
588e595.
[28] L. Cheng, T. Chen, Study of vapor liquid two-phase frictional pressure drop in a
vertical heated spirally internally ribbed tube, Chem. Eng. Sci. 62 (2007) 783e792.
[29] J. Wang, H. Li, B. Guo, S. Yu, Y. Zhang, T. Chen, Investigation of forced con-
vection heat transfer of supercritical pressure water in a vertically upward
internally ribbed tube, Nucl. Eng. Des. 239 (2009) 1956e1964.
[30] T.T. Yi, S. Koshizuka, Y. Oka, A linear stability analysis of supercritical water re-
actors, (I) thermal-hydraulic stability, J. Nucl. Sci. Technol. 41 (2004) 1166e1175.
[31] W. Qu, I. Mudawar, Measurement and correlation of critical heat flux in two-
phase micro-channel heat sinks, Int. J. Heat Mass Transfer 47 (2004) 2045e
2059.
[32] K. Chang, C. Pan, Two-phase flow instability for boiling in a microchannel heat
sink, Int. J. Heat Mass Transfer 50 (2007) 2078e2088.