Progress in Nuclear Energy 67 (2013) 104e113

Contents lists available at SciVerse ScienceDirect

Progress in Nuclear Energy

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

Study theoretically on two-phase circulation flow characteristics

under ERVC condition in advanced PWR
Guozhi Zhao, Xinrong Cao*, Xingwei Shi
National Defense Key Subject Laboratory of Nuclear Safety and Simulation Technology, Harbin Engineering University, 145 Nantong Street,
Harbin 15001, China

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

Article history: For the problem of two-phase natural circulation flow in gap clearance between reactor vessel lower
Received 11 January 2013 head and insulator in the condition of severe accident, one-dimensional steady-state natural flow
Received in revised form analysis code was written by utilizing FORTRAN. Based on the code, the effects of different correlations
11 April 2013
for friction coefficient and the number of nodes of heating section on mass flow rate of two-phase
Accepted 13 April 2013
natural circulation flow were studied. And the results are compared with that of Chinese REPEC
experiment and simulation using RELAP5 program so as to verify the rationality and correctness of the
Keywords:
code. Based on the experiment data, simulation results and the model, friction coefficient and the void
ERVC
Two-phase natural flow
fraction condition under ERVC correlation are obtained by fitting. The results calculated by the model
One-dimensional steady-state using fitting friction coefficient correlation agree well with ULPU V test data. Furthermore, the effect of
Mass flow rate power, pressure, inlet area, gap diameter, flooding level and inlet water subcooling on mass flow rate and
Void fraction void fraction of two-phase natural circulation were studied utilizing this code.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction performed before, such as ULPU-V test for obtaining CHF of AP600
(Theofanous et al., 2003), HERMES-HALF test for observing and
As one of the important measures for in-vessel retention, quantifying the flow in the gap and relevant simulation (Ha et al.,
external reactor vessel cooling (ERVC) under hypothetical severe 2012), SULTAN test for investigating the vessel cooling ability in
accident of pressurized water reactor can remove heat from corium natural convection (Rouge, 1997), CYBL test for investigating the
which relocates into lower head though two-phase natural circu- effect of flooded cavity on IVR (Chu et al., 1997), REPEC test for
lation flow in gap clearance between reactor pressure vessel (RPV) obtaining results of full-scale and full-height ERVC test facility (Li
and reactor vessel insulation (Energy Research Inc., 2004). How- et al., 2012), a simulation of natural circulation in AP1000 ERVC
ever, in conventional light water reactor the insulation cannot system utilizing RELAP5 code (Tao et al., 2011), etc.
provide a flow path which surrounds the RPV because preventing The study in theory focuses primarily on two-phase natural
the coolant heat from losing to surrounding environment is the circulation characteristics in order to provide designers and oper-
primary function of the insulation. ators with optimal configuration parameters of the ERVC system
The characteristic of the two-phase natural circulation in the and the effect of these parameters on natural circulation flow.
annular gap is an important factor of the evaluation of IVR though However, the studies from previous literatures on two-phase cir-
ERVC since it is considered that high flow rate in the gap corresponds culation flow in ERVC system mostly use the average value of
with large critical heat flux (CHF) of the lower head wall (Park, 2009). important hydraulic parameters like two-phase mixture density
Furthermore, the instability of the two-phase circulation can cause and void fraction in path outside of RPV (Park, 2009). Due to the
earlier boiling crisis even if the heat flux is lower than the CHF of great change of two-phase fluid density along the heating porting
corresponding position of lower head wall. For understanding the average value between inlet and outlet may not describe exactly
effect of structural and thermal-hydraulic parameters on the char- the flow state under ERVC condition. Comparing with previous
acteristic of two-phase natural circulation under ERVC condition, a studies, this paper utilizes difference method to deal with the two-
number of experimental and simulative researches have been phase density in heating porting which some important parame-
ters correlate with, like lifting pressure drop and friction coefficient.
And three various empirical formulae for calculating friction coef-
ficient which is important to predict the flow velocity are used for
* Corresponding author. Tel.: þ86 (0)45182569655.
E-mail address: [email protected] (X. Cao). comparison between each other because there are no appropriate

0149-1970/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.pnucene.2013.04.007
 G. Zhao et al. / Progress in Nuclear Energy 67 (2013) 104e113 105

formulae available for flow path in ERVC system configuration from qc  hsub rc uc ac
previous literatures. Furthermore, the solution based on the model cc ¼ (4)
hlg rc uc ac
is compared with that of REPEC experiment and RELAP5 simulation
and the effects of configuration parameters such as inlet area, gap Two-phase density is given by:
width and thermal-hydraulic parameters such as heating power,
rls
containment pressure, inlet subcooling, flooding level are para- rTP;c ¼ (5)
Dr
metrically investigated for advanced reactor system AP1000. 1 þ cc
rgs
2. Analytical study
Substituting Eq. (4) into Eq. (5) yields:

2.1. Process of circulation flow under ERVC uc ac rls rgs hlg  qc Dr

rc ¼
 (6)
Before the corium relocates into lower head (Energy Research uc ac rgs hlg  Drhsub
Inc., 2004), water from water tank has already flowed by gravity
Substituting Eq. (6) into Eq. (4) yields:
into gap clearance between sump shield wall and insulation. Then
water enters the annular gap between PRV and insulation through
qc rgs  uc ac rls rgs hsub
inlets located below the lower head. From there, it flows upwards cc ¼ (7)
along the spherical portion of RPV where boiling and steam pro- uc ac rls rgs hlg  qc Dr
duction occurs due to removing the heat from corium. The escaping
steam/water mixture flows into the gap between the cylindrical 2.2.2. Loop momentum equation
portion of RPV and insulation, and finally passes through the out- The momentum equation is rewritten in pressure difference
lets on the top of RPV cylindrical section. The steam is condensed in form, that is, the two-phase pressure drop along the flow path of
the containment free space and flows back into water tank. At the insulation inside and outside can be represented as the sum of
same time, the escaping water flows back to drain tank room via an lifting pressure drop induced by the height difference, friction drop
opening between the vertical channel and drain tank room, and induced by the two-phase flow, shape resistance drop induced by
enters the cavity again through a passively actuated damper. So the geometric change of flow path and acceleration pressure drop
long time natural circulation flow under passive ERVC is estab- induced by the flow velocity difference between liquid and stream
lished, and the process mentioned above is shown in Fig. 1. as given by Eq. (8).
     
2.2. Analytical model dP dP dP dP
 ¼ rTP g þ  þ  þ  (8)
dz dz fr dz fo dz a
2.2.1. Mass and energy conservation equation
For steady-state, one-dimensional flow, the mass and energy To determine the two-phase circulation flow velocity, the mo-
conservation equation of homogenous system can be written as mentum equation is integrated over the entire circulation flow loop
(Reyes and Hochreiter, 1998): inside and outside of the insulation. The entire loop pressure drop
must be zero as given in Eq. (9)
rin uin ain ¼ rc uc ac ¼ G (1) I   I I   I  
dP dP dP
 dz ¼ rTP gdz þ  dz þ  dz
rc uc ac ðhTP  hl Þ ¼ qc (2) dz dz fr dz fo
I   (9)
dP
Considering hTP  hl ¼ cc hlg þ hsub , from its definition as shown þ  dz ¼ 0
dz a
in Eq. (3) and Eq. (2) the mass vapor quality can be given by Eq. (4).
The lifting pressure drop term in Eq. (9) can be rearranged as
h  hls shown in Eq. (10), accounting for two-phase flow in heating portion
cc ¼ TP (3)
hgs  hls and riser, single-phase flow outside of the insulation. The densities
of vaporeliquid mixture in riser and single-phase water outside of
the insulation can be regarded as constants as the effect of height
difference can be neglected.

I ZLh ZL Z0
rTP gdz ¼ rTP gdz þ rout gdz þ rl gdz
0 Lh LþLh

Z h
LþL

¼  ðrl  rout  rTP Þgdz (10)

0

The two-phase mixture density in heating portion in Eq. (10) in

previous analyses on natural circulation flow under ERVC condition
is represented as the average value of the entire heating portion
(Park, 2009). Generally, the larger the circulation flow rate of the
ERVC system, the stronger the coolability of the corium and natural
circulation flow rate hardly lies on the power shape but on the total
power (Theofanous et al., 2003). So, though such as in ULPU test or
REPEC test, non-uniform power shape is utilized in heated sections
Fig. 1. Schematic of ERVC system. for measuring critical heat flux at outside surface of lower head or
106 G. Zhao et al. / Progress in Nuclear Energy 67 (2013) 104e113

investigating the flow instability, the predicted flow rates according Table 1
Design and thermal-hydraulic parameter values in REPEC test, RELAP5 simulation
to uniform power distribution agree well with the measured flow
for AP1000 and ULPU V test.
rates in ULPU experiment (Park, 2009). Furthermore, the void
fraction, obtained from RELAP5 simulation of REPEC test in which Parameter (Unit) REPEC RELAP5 ULPU V test
(base-case) simulation
power distribution is uniform in 0e40 and 40e70 range of heating
for AP1000
section respectively as shown in Fig. 2 and Table 2, shows that void
Environment pressure (atm) 1 1.4 1
fraction is nearly linear distribution in 0e40 and 40e70 range
Riser height (m) 4.1 4.37 4
respectively. And the void fraction is non-uniform in 70e90 range Riser clearance width (m) 0.15 0.15 0.152
because heat flux in eighth zone is not equal to its two adjacent Radius of heating section (m) 2.05 2 2
zones. Hence, it can be assumed that the void fraction is an Maximum gap clearance width (m) 0.2 0.15 0.076e0.152
approximate linear when uniform power distribution is used along Inlet area (m2) 0.01767 0.748 0.018
Inlet water temperature (K) 338 360 283
whole heating section. In this study, differentiation is utilized to
Shape resistance lose coefficient 0.01 0.01 0.01
obtain the density distribution in the heating portion which is of heating section, Kheat
divided into n sections, where void fraction in every section can be Riser shape resistance lose 0 0 0.01
represented as the arithmetic mean value of its inlet and outlet coefficient, Kup
Downcomer shape resistance 0 0 0
value as given in Eq. (11).
lose coefficient, Kdown
Average/maximum heat 790/1235 524/1270 800/1974
flux (kW/m2)
Maximum total power (MW) 0.399 16 0.502

1
ai ¼ ða þ aiþ1 Þ (11)
2 i
And void fraction can be obtained though mass vapor quality as
follows:

1
ai ¼ (12)
1  ci rgs
1þ
ci rls
The average density in ith section can be given by:


rTP;i ¼ rls  ai rls  rgs (13)

Substituting Eqs. (4), (11) and (12) into Eq. (13) the average
density in ith section can be represented by:

0:5Drrls rgs ðci þ ciþ1 Þ þ rls r2gs

rTP;i ¼ (14)
Dr2 ci ciþ1 þ 2Drrgs ðci þ ciþ1 Þ þ r2gs
Substituting Eq. (14) into Eq. (10) yields:
ZLh X
n
Lh
rTP gdz ¼ rTP;i g
i¼1
n
0

gLh Xn 0:5Drrls rgs ðci þ ciþ1 Þ þ rls r2gs

¼ (15)
n i ¼ 1 Dr ci ciþ1 þ 2Drrgs ðci þ ciþ1 Þ þ r2gs
2

Once iterative calculation begins with initial value uout, rout can
be calculated by Eq. (6) as well as G by Eq. (1). Because hinlet, namely
h1, is known, h of inlet and outlet point in every section can be
obtained by Eq. (2) as follows:

Table 2
Copper information of heating section in REPEC experiment.

Heating Corresponding Maximum

zone heating zone (degree) heat flux (kW/m2)

1 0e10 420
2 10e20 420
3 20e30 420
4 30e40 420
5 40e50 880
6 50e60 880
7 60e65 1235
8 65e80 1198
9 80e90 1235
Fig. 2. (a) copper block distribution, (b) outlet void fraction in heating section.
 G. Zhao et al. / Progress in Nuclear Energy 67 (2013) 104e113 107

ðh2  h1 ÞG ¼ qc fSP ¼ 0:184 Re0:2

SP (19a)
«
ðhiþ1  hi ÞG ¼ qc (16) fTP ¼ 0:184 Re0:2
TP (19b)
«
ðhn  hn1 ÞG ¼ qc or by Colebrook formula as follows:

So the mass vapor quality of inlet and outlet point in every    

section can be obtained by Eq. (3). 68 ε 0:25 68 ε 0:25
fSP ¼ 0:11 þ ; fTP ¼ 0:11 þ (20a)
The friction and shape resistance pressure drop in Eq. (9) can be ReSP D ReTP D
expressed as
1
fSP ¼ fTP ¼  2 (20b)
D
I   I   n 
X  2lg þ 1:74
dP dP Lh rTP;i u2i 2ε
 dz þ  ¼ fTP;i þ Kheat;i
dz fr dz fo i¼1
ndH 2
  or by annular channel empirical formula as follows:
L rout uout
2
þ fTP;up þ Kup
dH 2 96
! fSP ¼ (21a)
L þ Lh rin u2in ReSP
þ fSP þ Kdown
dH;down 2
  96
1 1 G fTP ¼ (21b)
þ 2 ReTP
ac ain ac rin
(17) The comparative results using the three formulae respectively
are shown in Section 3.1.
where the first to third terms represent the friction and shape Correlation for Reynolds num is given by its definition as
resistance pressure drop in heating porting, riser and insulation follows:
outside respectively. The forth term represent shape resistance
rl uout dH
pressure drop at insulation inlet. ReSP ¼ (22)
Substituting Eq. (1) and Eq. (6) into Eq. (17) yields: ml

2
rTP uout dH
ReTP ¼ (23)
I   uout aout rls rgs hlg  qc Dr X n mTP
dP 1
 dz ¼

dz fr
2 rgs hlg  Drhsub
2 a 2 r
i ¼ 1 i TP;i where two-phase kinetic viscosity is given by
   
L
 fTP;i h þ Kheat;i þ fTP;up
L
þ Kup
mTP;c ¼ cc mgs þ ð1  cc Þmls (24)
nd d

H H The acceleration pressure drop along the entire circulation flow
uout uout aout rls rgs hlg  qc Dr loop in Eq. (9) must be zero as follows:


2aout rgs hlg  Drhsub I  
dP
!  dz ¼ 0 (25)
L þ Lh dz a
þ fSP þ Kdown
dH;down

2
uout aout rls rgs hlg  qc Dr Gðain  ac Þ 2.2.3. Flow velocity equation

2 þ 2
ac ain rin Substituting Eqs. (15), (18) and (25) into Eq. (9), a cubic implicit
2rin a2in rgs hlg  Drhsub
equation for flow velocity can be obtained as follows:
(18)
u3c þ j1 u2c þ j2 uc  jc ¼ 0 (26)
where friction coefficient is given by McAdams formula as follows
(Levy, 1999): where


qc 2F1 rin Dr  F2 4rin Drrls rgs hlg  2F3 Dr=a2in þ 2ðain  aout Þ4=a3out ain
j1 ¼ (27a)
aout F1 rin rls rgs hlg þ F2 rin 4=aout þ aout F3 rls rgs hlg =a2in

 

qc Dr F1 qc Dr þ F3 qc Dr=rin a2in  2ðain  aout Þ4=a3out ain rin þ 4gðL þ Lh Þ 3rls rgs hlg  rin 4
j2 ¼
 (27b)
aout rls rgs hlg F1 aout rls rgs hlg þ F2 4=aout þ F3 aout rls rgs hlg =rin a2in
108 G. Zhao et al. / Progress in Nuclear Energy 67 (2013) 104e113

3ðL þ Lh Þg4qc Dr
j3 ¼

aout rls rgs hlg F1 a2out rls rgs hlg þ F2 4 þ F3 a2out rls rgs hlg =rin a2in
(27c)

where

4 ¼ rgs hlg  Drhsub (28)

X
n  
1 Lh
F1 ¼ fTP;i þ Kheat;i (29a)
r
i ¼ 1 TP;i
ndH

L
F2 ¼ fTP;up þ Kup (29b)
dH

L þ Lh
F3 ¼ fSP þ Kdown (29c)
dH;down

The cubic implicit Eq. (26) is solved though NewtoneRaphson

Method and iteration accuracy is 106. The configuration and
thermal-hydraulic parameters as input data for the equations up-
wards are referred to AP1000 ERVC system design and literatures
available.

3. Results and analyses

Flow velocity under ERVC condition can be calculated by using

Eq. (9) for various conditions. However, from the derivation
process of Eq. (9) we can find that the empirical formulae are
used for friction coefficient calculation. For the special configu- Fig. 3. Effect of friction coefficient formula on, (a) mass flow rate, (b) void fraction.

ration of insulation flow path, there is no available suggestion

that which friction coefficient formula is proper from previous
literatures. So four cases are studied: the first is the comparison 3.3. Comparison with experiment and simulation data
between each the model calculation using three empirical
The system design and thermal-hydraulic parameters that pre-
formulae mentioned above respectively, that is, Eqs. (19), (20)
sented in Eq. (26) and its auxiliary equations should be determined
and (21); the second is the evaluation of the effect of the num-
according to ERVC system design expected prior to obtain the flow
ber of the nodes in heating section; third is comparison between
velocity solutions for various conditions under ERVC condition.
the model calculation with REPEC experiment results and RELAP5
Hence, two cases are studied to investigate the feasibility of the
simulative results respectively; the fourth is the evaluation of the
homogeneous flow model for ERVC system. One is the comparison
effect of diverse parameters on the two-phase natural circulation
with the REPEC experiment data (Li et al., 2012) and the other is the
flow characteristics under ERVC condition based on AP1000 ERVC
comparison with the simulation results using RELAP5 code (Tao
system design.
et al., 2011; Li et al., 2012).

3.1. Calculation results using different friction coefficient formulae 3.3.1. Comparison with REPEC test
The REPEC experiment facility is a full-scale and full-height slice
Eq. (19) given by McAdams is fit for turbulent in round cross loop with changeable inlet and outlet area, inlet water subcooling
section flow path and Eq. (21) is fit for annular channel. For Eq. (20) and heating power. The scheme of the experiment facility and
given by Colebrook, when Re > 26.98(D/ε)8/7, Eq. (20a) is called and RELAP5 nodalization are shown in Fig. 5(a) and (b) (Li et al., 2012).
when Re > 191.2/f0.5(D/ε) Eq. (20b) is called. Fig. 2 shows the The facility contains an experiment loop, data acquisition system,
calculation results using the three formulae. For subcooling higher heating system and tank condenser. REPEC has an orifice plate and
than 20 K, the results using the three formulae are almost close to butterfly valve to control inlet area. And there is a tank in the
each other. And for that lower than 20 K, the results using McA- bottom of the facility which connects the heating section and
dams and Colebrook empirical formula are close to each other but downcomer for simulating the cavity. The design and thermal-
different from the one using annular channel empirical formula. hydraulic parameter values of REPEC experiment are presented in
Table 1. The copper blocks in the heating section which are sub-
3.2. Evaluation of the number of the nodes in heating section divided into 9 groups to get the non-uniform power distribution
shape are installed along the curved flow path as shown in Fig. 4
The effect of the number of the nodes in heating section and heat flux in each group is presented in Table 2. The mass
(parameter n in Eq. (17)) in various inlet water subcooling condi- flow rate (uc  ac  rTP ) results predicted by this model are
tions is presented in Fig. 3. It shows that the variation of the number compared with the ones calculated by RELAP5 code and REPEC test
of nodes from 5 to 200, the mass flow rate varies just from 2.4% to data in Fig. 5(c). It shows that the minimum and maximum errors
3%. Hence, the results show that the flow rate is not sensitive to the between prediction and experimental data are 2.3% and 12.1%
number of nodes of the heating section in the homogeneous model. respectively. The reason may be that the homogeneous model can
 G. Zhao et al. / Progress in Nuclear Energy 67 (2013) 104e113 109

Riser

Downcomer

Heating section 4
3
2
1

(a)

(b)
Fig. 4. Effect of the number of nodes in the present model: (a) the nodalization scheme
of the present model, (b) The flow rate at various number of nodes condition.

not embody the two-phase velocity difference, i.e., underestimate

the two-phase average velocity because vapor phase velocity is
faster than liquid phase in curved the flow path upward in the low
pressure. Besides, rTP is also underestimated because Eqs. (3) and
(5) base on heat balance without considering the effect of low
pressure on the cross-section quality (the void fraction), namely, in
low pressure, small mass vapor quality corresponds to relative large
void fraction (Levy, 1999). And for prediction and simulation, re-
sults the errors are 2% and 7.7% respectively. The error can be
explained by the difference between the homogeneous flow model
and the actual flow condition which is considered as separate flow
model in RELAP5 code (Nuclear Safety Analysis Division, 2001).

3.3.2. Comparison with simulation results for AP1000 ERVC system

The AP1000 ERVC system is simulated with RELAP5 code from
previous literature and the nodalization scheme are shown in
Fig. 6(a) (Tao et al., 2011). In the RELAP5 simulation of the natural
circulation flow at ERVC condition, the sump free space is included
and the non-uniform power distribution is used and the area of
inlet and annular gap clearance is actual cross-section area values
of ERVC system in AP1000 as presented in Table 2. In the analytical
model from Eq. (26), the total power is equal to RELAP5 simulation
power but uniform power distribution is used and the configura-
tion parameters are equal to RELAP5 simulation data. The param-
eter n in subsidiary Eq. (29a) is set to 5 which corresponds with the
node number of volume 120 in RELAP5 model (Tao et al., 2011) as
shown in Fig. 4. The mass flow rate results predicted by this mode
are compared with the ones calculated by RELAP5 code as shown in Fig. 5. REPEC test: (a) facility, (b) RELAP5 nodalization scheme, (c) comparison with
REPEC test data.
Fig. 6(b). It shows that the minimum and maximum errors between
prediction and simulation results are 2.1% and 7.5% respectively.
110 G. Zhao et al. / Progress in Nuclear Energy 67 (2013) 104e113

Fig. 7. Comparison of volumetric flow rate with ULPU V test data.

3.3.3.2. Void fraction correlation. The flow rate can be obtained

through test or the present model easily even if there is a maximum
about 12.1% error which is acceptable due to the difference between
the homogeneous flow model and actual flow condition. And there
are many factors affecting the outlet void fraction in low pressure.
Hence, based on the experiment data, simulation results and this
prediction model, the outlet void fraction at 1 atm pressure can be
obtained by fitting in the form of the void fraction correlation for
annular channel (Levy, 1999):

" #1=2
ð1  ce Þ3
a ¼ 1 (31)
1 þ K1 ce

rls 1=2
K1 ¼ s u expðK2 Q Þ (32)
rgs

where s and K2 can be determined in the following Fig. 8.

The parameter u in Eq. (32) is homogeneous flow velocity but it
can be substituted by liquid phase apparent velocity with the un-
Fig. 6. Comparison with RELAP5 results for AP1000 ERVC system, (a) the nodalization changed type correlation but the different s value. And the void
scheme, (b) the flow rate for various inlet area. fraction data are hardly available for comparison from previous
literatures. Hence, it needs more experiments like three-
3.3.3. Data fitting dimensional test or better established model like separate flow
3.3.3.1. Friction coefficient correlation. The comparison results show model.
that the homogeneous flow model can not totally reflect the actual
flow under ERVC condition. And if it is considered that the friction
coefficient calculation is contributed mostly to the error in the pre-
sent homogeneous flow model, based on the experiment data and
this prediction model, the following friction coefficient correlation
for gap flow path under ERVC condition can be obtained by fitting:

fTP ¼ 0:071 Re0:25

TP (30)

where Reynolds num and two-phase average kinetic viscosity are

still calculated by Eqs. (23) and (24) respectively.
The thermal-hydraulic and configuration parameters of the
ULPU slice test facility which is similar with REPEC test facility are
presented in Table 1. And the volumetric flow rate (uc  ac) results
using Eq. (30) which are compared with the ULPU V test data (the
flow rates are measured in volumetric flow rate form) (Dinh et al.,
2003) are presented in Fig. 7. The results agree well with ULPU V
test data and the maximum error is 4.6%. Fig. 8. The relation between the inlet subcooling and values of s and K2.
 G. Zhao et al. / Progress in Nuclear Energy 67 (2013) 104e113 111

3.4. Effect of thermal-hydraulic and configuration parameters heating power but on pressure, inlet area, gap width and flooding
level which will be discussed in next sections. However, heating
The effects of various thermal-hydraulic and configuration pa- power only determines the subcooling corresponding to mass flow
rameters on the two-phase natural circulation flow rate under rate peak, that is, subcooling corresponding to mass flow rate peak
ERVC condition are discussed. The parameters are obtained on the increases as heating power increases.
bases from AP1000 correlative literature (Energy Research, Inc., Fig. 9(a) also shows that in the condition of the heating power
2004; Tao et al., 2011). The values of these base-case parameters higher than 10 MW, for the same subcooling lower than the value
are summarized in Table 1. corresponding to mass flow rate peak, the mass flow rate decreases
as heating power increases; for the same subcooling higher than
3.4.1. Effect of heating power the value corresponding to mass flow rate peak, the mass flow rate
Fig. 9 shows the effect of heating power. The void fraction is increases as heating power increases. This is mainly because two-
small in high subcooling condition, so the water in gap clearance phase friction pressure drop increases as power increases for sub-
takes the corium decay heat through subcooled boiling as shown in cooling higher than the value corresponding to the mass flow rate
Fig. 9(b). It is obvious that there is a mass flow rate peak in every peak and decreases as power increases for subcooling lower than
curve as shown in Fig. 9(a). For subcooling higher than the one to the value.
which mass flow rate peak is corresponding, natural circulation
mass flow rate decreases as subcooling increases. And for sub- 3.4.2. Effect of pressure
cooling lower than that, mass flow rate increases rapidly as sub- The pressures are set to be 1, 1.4, 2, 3 and 3.94 atm in accordance
cooling increases. The reason is that subcooled boiling is dominant with AP1000 containment designed pressure. For the heating po-
in the gap when inlet subcooling is high and two-phase friction wer of 30 MW the effect of pressure on the mass flow rate and void
pressure drop is small. The driven head mainly overcomes shape fraction is shown in Fig. 10. Fig. 10(a) shows that pressure affects
resistance because of the density difference between riser and flow rate and void fraction only when subcooling is low and rarely
downcomer. So mass flow rate decrease as subcooling increases affects flow rate when subcooling is higher than 10 K. And for the
due to the decrease of driven head. However, when subcooling is same low subcooling, the mass flow rate increases as pressure in-
lower or near saturated value, mass flow rate increase as subcooling creases. Fig. 10(b) shows that void fraction increases as pressure
increases as two-phase friction pressure drop decreases and driven increases for subcooling lower than 10 K. This may be explained by:
head increases. The intense boiling occurs in the gap and void the pressure has effect on flow rate through influencing the hy-
fraction is large as shown in Fig. 9(b). The mass flow rate increases draulic parameters like density and viscosity. For the high sub-
due to the increasing entrained water in the vapor as subcooling cooling, the driven force is weak because of density difference
increases. For the heating power of 50, 30, 18, 10 and 1 MW, sub- between the insulation inside and outside which is not large due to
cooling corresponding to mass flow rate peak is 10, 6, 3, 2 and 1 K subcooled boiling in the gap clearance and friction pressure drop
respectively. increases as subcooling increases; for the low subcooling, void
Fig. 9(a) also shows that when heating power is higher than fraction is large as shown in Fig. 10(b), and then driven force is large
10 MW, the peak value of mass flow rate does not depend on

Fig. 9. Effect of heating power on mass flow rate and void fraction. Fig. 10. Effect of pressure on mass flow rate and void fraction.
112 G. Zhao et al. / Progress in Nuclear Energy 67 (2013) 104e113

due to two-phase density difference. The pressure has more effect

on two-phase fluid density in high pressure condition than on
signal-phase fluid density in low pressure condition.
Furthermore, the pressure has no effect on subcooling corre-
sponding to the mass flow rate peak.

3.4.3. Effect of inlet area

For the condition of heating power of 30 MW, pressure of 1 atm,
and annular channel cross-section area of 1.452 m2, the effect of
inlet area on two-phase circulation flow characteristics is shown in
Fig. 11. For subcooling higher than 20 K, inlet area has almost no
effect on the mass flow rate and void fraction; for subcooling lower
than 20 K, the mass flow rate increase and void fraction decreases
as inlet area increases. As the inlet area increases, subcooling cor-
responding to the mass flow rate peak moves toward saturated
value. This is because smaller inlet area reduces inlet shape resis-
tance in Eq. (17) and two-phase friction coefficient in Eq. (19).

3.4.4. Effect of gap width

For the heating power of 30 MW, pressure of 1 atm and inlet
area of 0.56 m2, the effect of gap width on two-phase circulation
characteristics is shown in Fig. 12. For subcooling higher than the
value corresponding to the mass flow rate peak, the mass flow rate
decreases as gap width increases; for the subcooling lower than the
value, flow rate decreases as gap width decreases. For the gap width
larger than 0.2 m, gap width has little effect on two-phase natural
circulation flow as shown in Fig. 12(a) and (b).

3.4.5. Effect of flooding level Fig. 12. Effect of gap width on mass flow rate and void fraction.
For the heating power of 30 MW, pressure of 1 atm, inlet area of
0.56 m2 and gap width of 0.1 m, the effect of flooding level (only the increases as flooding level increases since higher flooding level
value L in Fig. 1 is changed) on two-phase circulation characteristics gives rise to increasing static water head. Furthermore, flooding
is shown in Fig. 13. For subcooling higher than 10 K, flooding level level has little effect on the subcooling corresponding to the mass
has no effect on the mass flow rate and void fraction; for subcooling flow rate peak and the peak value itself.
lower than 10 K, the mass flow rate decreases and void fraction

Fig. 11. Effect of inlet area on mass flow rate and void fraction. Fig. 13. Effect of flooding level on mass flow rate and void fraction.
 G. Zhao et al. / Progress in Nuclear Energy 67 (2013) 104e113 113

4. Conclusions Dinh, T.N., Tu, J.P., Salmassi, T., Theofanous, T.G., 2003. Limits of Coolability in the
AP1000-related ULPU-2400 Configuration V Facility. Center for Risk Studies and
Safety University of California, Santa Barbara, USA, pp. 1e50.
This study constructs the cubic equations, based on one- Energy Research, Inc., 2004. Analysis of In-vessel Retention and Ex-vessel Fuel
dimensional homogeneous flow model for solving the flow veloc- Coolant Interaction for AP1000. Washington, USA, pp. 1e73.
ity using numerical method. McAdams, Colebrook and annular Ha, K.S., Cheung, F.B., Park, R.J., Kim, S.B., 2012. Evaluations of two-phase natural
circulation flow induced in the reactor vessel annular gap under ERVC condi-
channel empirical formulae evaluated for calculating the velocity of tions. Nuclear Engineering and Design 253, 114e124.
flow in gap clearance. Results are close using three empirical Levy, S., 1999. Two-phase Flow in Complex Systems. Wiley-Interscience.
formulae respectively for high subcooling and the former two Li, F., Li, Y.C., Cheng, X., 2012. Simulation and analysis on REPEC heating experi-
mental mechanism. Atomic Energy Science and Technology 46 (7), 815e820.
formulae for low subcooling. The number of nodes of heating sec- Nuclear Safety Analysis Division, 2001. RELAP5/MOD3.3 Code Manual Volume I:
tion is discussed and the result shows that the flow rate is not Code Structure, System Models, and Solution, Methods. NUREG/CR-5535.
sensitive to the number of nodes in homogeneous model. And the Park, J.W., 2009. Analytical evaluation of two-phase natural circulation flow char-
acteristics under external reactor vessel cooling. Annals of Nuclear Energy 36,
comparison between the mass flow rates for AP1000 ERVC system 1668e1675.
calculated from this work with REPEC test and RELAP5 results Reyes Jr., J.N., Hochreiter, L., 1998. Scaling analysis for the OSU AP600 test facility
shows that the model can reflect the two-phase natural circulation (APEX). Nuclear Engineering and Design 186, 53e109.
Rouge, S., 1997. SULTAN test facility for large-scale vessel cool ability in natural
flow in ERVC system. The friction coefficient and the void fraction convection at low pressure. Nuclear Engineering and Design 169 (1), 185e195.
correlation are obtained by fitting and the results calculated by the Tao, J., Tong, L.L., Cao, X.W., 2011. A study on natural circulation flow under reactor
model using fitting friction coefficient correlation agree well with cavity flooding condition in advanced PWR. Journal of Power and Energy Sys-
tems 5, 429e439.
ULPU V test data.
Theofanous, T.G., Salmassi, T., Dinh, T.N., 2003. Limits of Coolability in the AP1000-
Furthermore, based on the analyses above, some important related ULPU-2400 Configuration V Facility. California, USA, pp. 1e55.
conclusions can be drawn as follows:
Glossary
1. Inlet subcooling greatly affects two-phase natural circulation
flow under ERVC condition. And there is a mass flow rate peak a: cross-section area (m2)
when the subcooling, which depends on heating power, inlet dH: hydraulic diameter (m)
f: friction coefficient
area and gap width but not on pressure and flooding level, is
g: gravitational acceleration (m/s2)
close to the saturated value. For subcooling higher than this h: specific enthalpy (J/kg)
value, the mass flow rate decreases as subcooling increases and hlg: latent heat (J/kg)
K: shape resistance coefficient
contrarily the mass flow rate increase as subcooling increases.
L: height
2. In the condition of heating power of 10 MW, for subcooling P: pressure
lower than the value corresponding to the mass flow rate peak, q: Heat flux (W/m3)
the mass flow rate decrease as heating power increases and R: radius (m)
Re: Reynolds number
contrarily increases as heating power increases. For heating u: linear velocity (m/s)
power higher than 10 MW, heating power rarely affect the mass c: equilibrium vapor quality
flow rate peak. However, higher heating power gives rise to
Greek symbols
increasing subcooling corresponding to the mass flow rate peak. a: void fraction
3. For high subcooling, pressure and inlet area rarely affect the ε: absolute roughness (m)
two-phase natural circulation flow. For low subcooling or r: density (kg/m3)
Dr: density difference (Dr ¼ rls  rgs)(kg/m3)
nearly saturated condition, the mass flow rate increases and
m: kinetic viscosity (kg/m s)
void fraction decreases as pressure or inlet area increases.
4. For high subcooling, the mass flow rate increases as gap width Subscripts
decreases and contrarily decreases as gap width decreases. For a: acceleration drop
c: corium
gap width larger than 0.2 m, it has little effect on two-phase down: downcomer
natural circulation flow. fo: shape resistance
5. For subcooling lower than 10 K, the mass flow rate decreases fr: friction resistance
g: vapor
and void fraction increases as flooding level increases. And for gs: saturated vapor
subcooling higher than 10 K, flooding level rarely affects two- in: inlet
phase circulation flow. i: index of section
l: liquid
ls: saturated liquid
References out: outlet of heating portion
SP: single-phase water
Chu, T.Y., Bentz, J.H., Slezak, S.E., 1997. Ex-vessel boiling experiments: laboratory and sub: subcooled
reactor-scale testing of the flooded cavity concept for in-vessel core retention: TP: two-phase
part II. Nuclear Engineering and Design 169, 89e99. up: riser