Annals of Nuclear Energy 111 (2018) 487–498

Contents lists available at ScienceDirect

Annals of Nuclear Energy

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

Thermal stratication and mixing in a suppression pool induced by

direct steam injection
Hua Li a, Walter Villanueva a,⇑, Markku Puustinen b, Jani Laine b, Pavel Kudinov a
a
Division of Nuclear Power Safety, Royal Institute of Technology (KTH), Roslagstullsbacken 21, SE-10691 Stockholm, Sweden
b
Nuclear Engineering, LUT School of Energy Systems, Lappeenranta University of Technology (LUT), FIN-53851 Lappeenranta, Finland

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

Article history: An experimental and numerical investigation of thermal stratication and mixing in a suppression pool is
Received 25 February 2017 presented. Steam injected into a drywell ows through a blowdown pipe and then down to the pressure
Received in revised form 6 September 2017 suppression pool where direct contact condensation occurs. The steam venting and condensation is a
Accepted 10 September 2017
source of heat and momentum. A complex interplay between the two leads either to thermal stratica-
Available online 23 September 2017
tion or mixing of the pool. The experiments are conducted in a scaled down PPOOLEX facility at
Lappeenranta University of Technology (LUT). The corresponding numerical simulations are performed
Keywords:
using GOTHIC with the Effective Heat Source (EHS) and Effective Momentum Source (EMS) models.
Pressure suppression pool
Direct steam injection
The EHS/EMS models, that have been previously proposed, predict the development of thermal strati-
Thermal stratication and mixing cation and mixing during a steam injection into a large pool of water. The experiments exhibit the devel-
opment of thermal stratication in the pool at relatively low mass ow rates and then pool mixing when
the mass ow rates are increased but later thermal stratication can re-develop even at the same rela-
tively high mass ow rates, which is due to the increasing pool temperature that shifts the condensation
to a different regime. The numerical simulations quantitatively capture this complex transient pool
behavior and are in excellent agreement with the transient averaged pool temperature and water level
in the pool.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction and mixing phenomena is necessary for safety analysis of pressure

Heat exchange systems that involve direct steam injection into
a pressure suppression pool can be found in different industrial
applications. In a Boiling Water Reactor (BWR), an important com-
ponent of its passive safety system is the pressure suppression pool
(PSP) that serves primarily as a heat sink in case of a loss of coolant
accident (LOCA) and station blackout (SBO) such as during the
Fukushima Daiichi accident. The suppression pool is also a source
of water for the reactor’s cooling systems and for possible severe
accident mitigation measures. The steam injection is a source of
heat and momentum in the pool that directly affects its thermal
behavior and as a result the pool is either thermally mixed or strat-
ied. The corresponding top surface temperature of the pool deter-
mines partial pressure of steam and thus pressure in the
containment. The thermal state and water inventory of the pool
has a signicant inuence on the possibility of containment failure
(Gamble et al., 2000). Hence a reliable prediction of stratication
⇑ Corresponding author.
E-mail address: walterv@kth.se (W. Villanueva).
suppression pool operations.
There have been numerous experimental studies on stratica-
tion and mixing in a pool (see Kataoka et al., 1991; Fox, 1992;
Smith et al., 1992; Norman et al., 2006; Song et al., 2014; Cheng
et al., 2006; Laine and Puustinen, 2006; Laine et al., 2013, 2014
and references therein). Most experimental tests with steam injec-
tion have been carried out with small diameter pipes and not all
experimental data is readily available for model development
and code validation. On the other hand, computational studies on
stratication and mixing phenomena have been performed widely
using lumped parameter codes and 1 D codes (Peterson, 1994;
Zhao, 2003; Zhao and Peterson, 2007), since the use of ne resolu-
tion CFD methods (e.g., RANS, LES, DNS) in modeling 3D high Ray-
leigh number natural convection ows in a large pool, and most
importantly, direct contact condensation on the steam-water
interface is not practical.
Previously, we have proposed two effective models, namely,
Effective Heat Source (EHS) and Effective Momentum Source
(EMS) models (Li et al., 2013, 2014a). These validated models (Li
et al., 2014b) can be utilized to predict thermal stratication or
mixing during steam injection into a large pool of water.

https://doi.org/10.1016/j.anucene.2017.09.014
0306-4549/Ó 2017 Elsevier Ltd. All rights reserved.
488 H. Li et al. / Annals of Nuclear Energy 111 (2018) 487–498
The EHS model provides thermal effect of steam injection in the
form of a distributed heat source with the purpose to conserve
mass and thermal energy of the injected steam. In Fig. 1 a sche-
matic diagram of the EHS model is shown. It is assumed that only
hot saturated water ows out of the blowdown pipe, i.e. all steam
is condensed inside the blowdown pipe. Such approach correctly
preserves the mass balance in the system even if some fraction
of injected steam is condensed outside the pipe outlet.
The EMS model (Fig. 1) provides time averaged momentum
source induced by steam injection. This momentum creates large
scale circulation in the pool which can lead to erosion of thermally
stratied layer and mixing of the pool. Different regimes of steam
condensation in the pool (Li et al., 2014c) result in different
dynamics of the free surface oscillations. It was proposed (Li
et al., 2013, 2014a) to use ‘‘synthetic jet” model (Smith and
Swift, 2003) in order to predict effective momentum generated
by the oscillations of steam-water interface. Specically, for a sin-
gle harmonic oscillation, the velocity scale based on the momen-
tum ux (Smith and Swift, 2001; Krishnan and Mohseni, 2009) is
given as
p
U0 ¼ 2fL ð1Þ
where f is the frequency of oscillation and L is the amplitude of
oscillation. In (Villanueva et al., 2015), we have proposed a scaling
approach to determine the amplitude and frequency of oscillations
given the Froude number (which relates the inertial forces to grav-
itational forces). The momentum rate is then given as,
2
M ¼ pqU 20 d =4 ð2Þ
where q is the liquid density and d is the diameter of blowdown
pipe.
In the current paper, the EHS/EMS models are implemented and
validated against PPOOLEX-MIX experiments (Laine and Puustinen,
2006; Laine et al., 2013). Six tests from the PPOOLEX-MIX series are
Fig. 1. Schematic of proposed Effective Heat Source (EHS) and Effective Momentum
Source (EMS) models.
used in this study. We will demonstrate the connection of the
steam-water interface oscillations to the condensation regimes
and consequently the pool’s thermal behavior. At low steam mass
uxes, most of the steam condenses inside the blowdown pipe. The
steam-water interface stays close to the pipe’s exit and the oscilla-
tions are negligible. Given the EMS model, the effective momen-
tum due to the oscillations is then zero. However, condensed hot
water is also injected into the pool and is a source of momentum
but its magnitude is relatively small. Hence, in this regime, the
steam injection into the pool will likely result in thermal stratica-
tion of the pool. At relatively high steam mass uxes, the steam-
water interface can oscillate that can generate enough momentum
to mix the pool provided that the frequency and amplitude of oscil-
lations are large and frequent enough. However, it will be shown
here that even if the steam mass uxes are kept relatively high,
the steam-water interface oscillations can die down and the effec-
tive momentum decreases signicantly thereby resulting in re-
stratication of the pool. In connection to the condensation regime
map, the pool’s bulk temperature increases and the steam conden-
sation regime goes to ‘transition’ which is exhibited by a reduction
in frequency and amplitude of oscillations.
In the succeeding section, some details of the PPOOLEX facility
are presented. Next, details of the implementation of EHS/EMS in
Ò
GOTHIC 8.0 are provided. Then it is followed by discussions of
the validation of EHS/EMS against the PPOOLEX experimental tests.
First, validation of EHS/EMS against PPOOLEX tests that involve the
development of thermal stratication and then mixing is pre-
sented. Second, validation of the EHS/EMS models against PPOO-
LEX tests that also include re-stratication is discussed. Finally,
conclusions are provided.
2. PPOOLEX-MIX experiments
A series of experiments on steam condensation, thermal strati-
cation, and mixing in a large water pool have been performed at
Lappeenranta University of Technology (Finland) with POOLEX
(POOL EXperiment) and later modied PPOOLEX (Pressurized POO-
LEX) facility (Laine and Puustinen, 2006; Laine et al., 2013). In
Laine and Puustinen (2006) and Laine et al. (2013, 2014) details
on the scaling of the facility in comparison to Olkiluoto 1 and 2
have been provided. The POOLEX/PPOOLEX series are among the
few experiments on water pool mixing/stratication at such large
scales, and the availability of data was very instrumental for the
validation of the EHS/EMS models (Li et al., 2014a,b).
The PPOOLEX facility is a closed cylindrical stainless steel tank
with an outer diameter of 2.4 m (see Fig. 2a). It has both a drywell
(3.2 m height) and a wetwell (4.2 m height) and is considered
to be realistically closer to a containment of BWRs than POOLEX.
The bottom is close to hemispherical and the drywell wall was
insulated while the wetwell was not insulated during the tests.
First, steam is injected through a horizontal inlet plenum, then into
the drywell, and nally it discharges into the wetwell through a
vertical blowdown pipe which is installed close to the central axis
of the tank, with a length of about 3.1 m. A vacuum valve is
installed between the drywell and the wetwell in order to balance
the pressure between the compartments once the steam discharge
is stopped. A single train of 16 TCs was installed in the wetwell at
different elevations to measure the temperature distribution in the
pool.
There are 12 experimental tests performed in the PPOOLEX-MIX
series (Laine et al., 2013, 2014). The rst subseries, MIX-01 to 06,
was done with a 0.209 m diameter of the blowdown pipe while
the second, MIX-07 to 12, was done with a smaller 0.109 m diam-
eter of the blowdown pipe. The second subseries covers a wider
region in the chugging regime map than the rst subseries. A total
 H. Li et al. / Annals of Nuclear Energy 111 (2018) 487–498 489

Fig. 2. Schematics of the PPOOLEX experimental facility (Laine and Puustinen, 2006; Laine et al., 2013) (not to scale). The initial water pool depth is about 2.1 m.

of 17 TCs were installed inside the blowdown pipe with 20 Hz blowdown pipe during chugging in the PPOOLEX MIX-06 test. The
measurement frequency, and the temperature readings were used corresponding water level positions inside the pipe are shown in
to estimate the level of water inside the pipe especially during the Fig. 3b. Given the TC measurements inside the blowdown pipe,
chugging regime. the effective momentum can be calculated with the following steps.
For the rst subseries MIX-01 to 06 (see Table 1), a clearing
phase is done at relatively high steam ow rate (about 200 g/s) 1. Convert the TC measurements to water level positions (see
at the beginning of each test in order to (i) heat-up the drywell Fig. 3 as an example).
structures to the level of 130 °C so that steam condensation is min- 2. Calculate the velocities u ¼ dz
dt
, where z is the water level
imized in the drywell compartment during the stratication and position.
mixing stages, and (ii) ‘clear’ the spaces/compartments between 3. Calculate the moving time-averaged velocities
the steam source and the exit of the blowdown pipe of non- s
condensable gases. The clearing phase lasted about 500 s in every 1
Z t
experiment and resulted in an increased pool bulk temperature UðtÞ ¼ u2 ðsÞds ð3Þ
Dt tDt
of about 2 °C. The clearing phase is followed by a low steam ow
rate (about 100 g/s) with the aim to develop thermal stratication with an averaging time scale Dt ¼ 100 s.
in the pool. After several thousand seconds, the ow rate is 4. Calculate the effective (jet) velocity U 0 ðtÞ ¼ p2  UðtÞ. This relation
increased up to 300–425 g/s with the aim to mix the pool. In all can be shown simply by taking a single harmonic signal
tests, pool mixing has been observed after 150–500 s. In MIX-05 zðtÞ ¼ L  sinð2pftÞ where f ¼ 1=T and use Eqs. (1) and (3).
to 06, re-stratication has been observed. 5. Calculate the effective momentum rate M eff given by Eq. (2).

3. Implementation of the EHS-EMS models using the GOTHIC
code
The amplitude and frequency of the water-level oscillations in
the pipe can be obtained experimentally, e.g. by temperature mea-
surements on the pipe’s inner surface or by a level meter. Fig. 3a
shows a sample 5 s time window of TC measurements inside the
For the 5 s time window given in Fig. 3b, the time-averaged
velocities are around 0.55 m/s and the momentum rates are
around 11 kgm/s2.
The start of the mixing phases in the simulation of the MIX 03
and MIX-06 tests are t = 1288 s and 2045 s, respectively. Fig. 4
shows the effective momentum rates based on the water level
oscillations in the pipe during the mixing phases in the MIX-03

Table 1
Test conditions in the MIX-01 to MIX-06 experiments.

Experiment Initial water level [m] Initial water temperature [°C] Steam source pressure [MPa] Steam ow rate* [g/s]
MIX-01 2.09 13 0.55–0.6 50–350
MIX-02 2.09 13 0.53–0.6 10–430
MIX-03 2.09 13 0.5–0.6 80–440
MIX-04 2.10 16 0.58–0.65 50–320
MIX-05 2.10 15 0.57–0.63 80–365
MIX-06 2.10 15 0.55–0.64 90–475
*
These values were measured by the ow meter. At low steam ow rates, there was a considerable uncertainty in ow meter readings, thus a correction is implemented based
on the water level in the pool (see Li et al., 2014b for more details and the next section for the actual values).
490 H. Li et al. / Annals of Nuclear Energy 111 (2018) 487–498

Fig. 3. A sample of (a) temperature measurements inside the blowdown pipe and corresponding (b) water level positions, and (c) momentum rates and time-averaged
velocities during the chugging regime in the MIX-06 test. The outlet of the pipe is at 0 m.
 H. Li et al. / Annals of Nuclear Energy 111 (2018) 487–498 491

Fig. 4. Effective momentum rates based on the water level oscillations in the pipe during the chugging (mixing) phase in the (a) MIX-03, and mixing and re-stratication
phases in the (b) MIX-06 tests.

Fig. 5. (a) GOTHIC model schematic of MIX-06 and (b) corresponding 2D mesh with 48  75 cells for the wetwell.

and MIX-06 tests. For the MIX-03 test (Fig. 4a), the non-constant
momentum rates are between 16.5–25 kgm/s2 which are based
on effective jet velocities between 0.71–0.85 m/s. For the MIX-06
test (Fig. 4b), the maximum momentum rate in the mixing phase
is around 24 kgm/s2 while the momentum rate during the
re-stratication stage is nearly zero.
The GOTHIC 8.0 model schematic is shown in Fig. 5a while the
4875 mesh of the wetwell is shown in Fig. 5b. Mesh indepen-
dence is established and is reported elsewhere (Li et al., 2014a,c;
Gallego-Marcos et al., 2017). Since GOTHIC supports only a Carte-
sian coordinate system, the porosities of volume and surface area
of all cells are adjusted to represent a cylindrical symmetry. The
drywell is not modelled. However, an internal thermal conductor
is used to provide heat transfer through the bottom wall of the dry-
well and is used in boundary conditions at the top of the wetwell
gas space. The oor surface temperature of the drywell is measured
during the experiment and this is used as a boundary condition at
the top of the wetwell. Two external thermal conductors are used
to simulate the heat loss from the wetwell to the lab. In GOTHIC, a
pump is a component that can be used to drive the ow between
two cells by providing the corresponding ow rate and ow path
area. The effective momentum source (EMS) is implemented with
a pump. In GOTHIC, the use of pump is more convenient and prac-
tical than using a boundary condition. The EHS-EMS calculation
only includes part of the thermal stratication and mixing phases
and excludes the clearing phase. It should be noted that direct
steam injection using GOTHIC cannot predict accurately the ther-
mal stratication and mixing in the pool (Gallego-Marcos et al.,
2017). A typical calculation time is about 3 h on 4 processors of
an i7 3.4 GHz processors.
492 H. Li et al. / Annals of Nuclear Energy 111 (2018) 487–498

Fig. 6. Steam condensation regime map by Lahey and Moody (1993).

4. Thermal stratication phase to mixing phase
The MIX-01 to MIX-06 tests cover three regimes in the conden-
sation regime map (Li et al., 2014c) as shown in Fig. 6: condensa-
tion within the blowdown pipe (regime 1), chugging (regime 2),
and transition (regime 5). The chugging regime is characterized
by periodic overexpansion of steam in the blowdown pipe followed
by implosion of the void that causes a cyclic suction of water inside
the pipe. Chugging results in the oscillatory motion of the steam-
water interface (see Fig. 2b) with downward direction during the
injection phase and upward direction during the suction phase.
The chugging oscillations are relatively of lower frequency and
higher amplitude compared to the condensation oscillations. All
tests start from regime 1 followed by regime 2 and then go to
regime 5. In this work we present analysis and validation of the
EHS-EMS models against the MIX-02 to MIX-06 tests. Validation
of the models against other tests is a subject for the future work.
Validation of the EHS-EMS models against the MIX-01 test can be
found in Li et al. (2014).
During the tests, the steam mass ow rates are measured in the
steam line and not in the inlet of the blowdown pipe. Thus, the mea-
sured steam ow rate is not necessarily the same as the steam ow
rate from the drywell to the wetwell through the blowdown pipe.
This is especially true during the clearing phase as the drywell is ini-
tially cold (about 28 °C) and lled with non-condensable gases.
During the clearing phase, steam pushes all the non-condensable
gases from the drywell to the wetwell rst while steam condenses
on the drywell walls and heats up the drywell compartment. For
example, at 200 g/s steam mass ow rate, a transient time of
500 s (which is the time period that is set for the clearing phase)
is more than enough to push all the gas to the wetwell. After the
clearing phase, it is expected that the steam ow rate in the steam
line is almost the same as the steam ow rate to the wetwell and
this is found to be true in GOTHIC lumped simulations taking into
account the insulated drywell.
A GOTHIC lumped calculation has also conrmed that at low
steam mass ow rates, about 100 g/s and below (typically during
the stratication phase), there is considerable uncertainty in the
ow meter readings in the PPOOLEX facility. This can be corrobo-
rated also with estimation of the water level in the pool. Using
the measured steam mass ow rates during the stratication
phase, the predicted average temperature as well as the water level
in the pool increases much faster in the lumped simulation than in
the experiment. To resolve this issue, the steam mass ow rates are
estimated based on the collapsed water level in the pool (which
takes into account the variable water level in the blowdown pipe
and dependence of the densities on the increasing temperature
in the pool), see Fig. 7. The clearing phase has resulted in complete
mixing in the pool and has also generated a strong circulation ow
that took time to stagnate, about 600 s after the steam ow rate

Fig. 7. Measured steam mass ow rates with (i) ow meter and (ii) based on the water level in the pool in (a) MIX-02 and (b) MIX-03 tests. The timelines are adjusted to
reect the initial time t = 0 s in the simulations.
 H. Li et al. / Annals of Nuclear Energy 111 (2018) 487–498 493

Fig. 8. MIX-03 test (a) averaged pool temperature and (b) water level: Comparison between experiment and EHS-EMS simulation.

Fig. 9. Comparison of pool temperature between (a) MIX-02 measured data and (b) EHS-EMS simulation.

has been decreased. It should be noted that the initial time t = 0 s
correspond to the actual start of the thermal stratication phase.
The estimated ow rates during the stratication phase and the
measured steam ow rates during the mixing phase are then used
as input for the EHS-EMS simulation (see Li et al., 2014a,b for more
details).
In Fig. 8a, the averaged liquid temperature in the pool predicted
by the EHS-EMS simulation shows an excellent agreement with the
MIX-03 data. At time t = 0 s, the averaged liquid temperature is
about 16 °C. At the end of the development of stratication, it
increases to 18.5 °C in the test, and 19 °C in the simulation. It
reaches 34.5 °C at the end of the mixing in the test and the simu-
lation follows the same trend and rate of increase. An excellent
agreement in the water level in the pool between the test and sim-
ulation is shown in Fig. 8b. The water level increases from 2.112 m
to only 2.12 m for stratication phase. And during the mixing

Fig. 10. Comparison of pool temperature between (a) MIX-03 measured data and (b) EHS-EMS simulation.
494 H. Li et al. / Annals of Nuclear Energy 111 (2018) 487–498
phase, the water level increases to 2.175 m. This also conrms that
the heat losses through the wetwell walls are modeled properly.
The comparison in pool temperature between the MIX-02 mea-
sured data and EHS-EMS simulation is shown in Fig. 9. In general,
the predicted pool temperature agrees very well with the mea-
sured data. The development of thermal stratication in different
layers is well captured in the simulation except for the small region
in the vicinity of the pipe outlet (at 1.027 m) which may be attrib-
uted to the assumption of uniform heat ux distribution in the
simulation. There is a strong circulation ow in the pool during
the chugging with high steam ow rate, and it leads to complete
mixing. In the simulation, the time scale for mixing is about
100 s while in the MIX-02 experiment it is about 150 s. The exper-
imental data shows oscillating temperature at the free surface of
the pool, which is not reproduced in the simulations.
Fig. 10 shows the good agreement in pool temperature between
the MIX-03 measured data and EHS-EMS simulation. The uniform
heat ux distribution assumed in the simulation may cause the dif-
ference on the temperature at the top layer (at 1.969 m) of the wet-
well pool. At the end of the stratication phase, the predicted
temperature at the top layer (at 1.969 m) is about 25.5 °C while
the measured temperature is about 27 °C. For mixing phase with
high steam ow rate, the time scale for mixing is about 80 s in
the simulation while in the MIX-03 experiment it is about 130 s.
Part of the reason is the slight (1.5 degree) under-prediction of
temperature difference between the top (at 1.969 m) and bottom
layers (at 0.34 m) at the end of the stratication phase.
In Fig. 11, good agreement in pool temperature between the
MIX-04 measured data and the EHS-EMS simulation is shown.
Again the development of thermal stratication is well captured
in the simulation as well as the erosion of the layers when the
steam mass ow rate has been increased. These observations con-
rm that EMS provides an accurate estimation of the momentum
rates for each regime of steam injection and GOTHIC can represent
the various ow patterns.
5. Re-stratication in the pool
We have shown in the previous section that an increase in
steam mass ow rate resulted in thermal mixing. However, in
MIX 05 and MIX-06 tests, there is a redevelopment of thermal
stratication at the same high steam mass ow rate due to a
change in condensation regime, which occurs at an increasing pool
bulk temperature.
Fig. 12 shows the steam mass ow rates in MIX-05 and MIX-06
tests that are measured with the ow meter and based on the col-
lapsed water level in the pool. As stated earlier, the discrepancy in
Fig. 11. Comparison of pool temperature between (a) MIX-04 measured data and (b) EHS-EMS simulation.
ow rates during the stratication phase is due to the uncertainty
in the ow meter measurements when used around 100 g/s and
below. It is important to note that the ow rates after the rst
stratication phase remains high until the end of the transient
while the pool thermal behavior changes from mixing to
re-stratication.
Fig. 13 shows the agreement in the averaged liquid temperature
and water level in the pool predicted by the EHS-EMS simulation
and measured in the MIX-06 test. At time t = 0 s, the averaged liq-
uid temperature is about 18 °C and it increases to 26 °C during the
stratication phase matching both experiment and simulation. The
increase in temperature is more pronounced at the end of the tran-
sient as it reaches 73 °C while the predicted temperature is 75 °C,
since large steam mass ow rate is used for injection. The predicted
increase in water level in the pool (see Fig. 13b) shows an excellent
match before 2800 s. The initial water level is at 2.12 m and during
the stratication phase it increases to 2.15 m. During the mixing
phase the water level shows an abrupt increase to 2.37 m both in
the experiment and simulation.
Fig. 14 shows the comparison of pool temperature between the
MIX-05 measured data and the corresponding EHS-EMS simula-
tion. The development of the rst thermal stratication in different
layers is well captured in the simulation except for the small region
in the vicinity of the pipe outlet. A ne grid resolution around the
pipe exit can be used to improve the accuracy. In the simulation,
the time scale for mixing is about 250 s while in the MIX-05 exper-
iment it is about 400 s. The discrepancy can be attributed to the
fact that the temperature measured at the top layer is not stable
during this transition and such instability is not predicted by the
simulation. The continuous mixing and escalation of temperature
in the pool is also well captured except that the re-stratication
starts earlier in the simulation than in the experiment. This early
re-stratication leads to different top and bottom layer tempera-
tures, although the general behavior of the re-stratication is cap-
tured in the simulation. The main possible reason is that with
increasing pool bulk temperature the condensation regime
changes from chugging to transition regime (see Fig. 6). More
steam owing outside of the pipe can be expected as it takes a
longer distance to condense in the pool. The additional momentum
generated due to the buoyancy effects and possible entrainment
should be considered in this regime of condensation for more accu-
rate analysis.
For the MIX-06 test, the comparison of the experimental and
predicted pool temperature is shown in Fig. 15. The development
of the rst thermal stratication in different layers is well captured
in simulations. The predicted time scale for mixing is about 200 s
while in the MIX-06 test it is about 250 s. The redevelopment of
 H. Li et al. / Annals of Nuclear Energy 111 (2018) 487–498 495

Fig. 12. Measured steam mass ow rates with (i) ow meter and (ii) based on the water level in the pool in the (a) MIX-05 and (b) MIX-06 tests. The timelines are adjusted to
reect the initial time t = 0 s in the simulations.

Fig. 13. MIX-05 test (a) averaged pool temperature and (b) water level: Comparison between experiment and EHS-EMS simulation.

Fig. 14. Comparison of pool temperature between (a) MIX-05 measured data and (b) EHS-EMS simulation. The level of the pipe outlet is at 1.045 m.

thermal stratication starts earlier in the simulation than in the
experiment. Similar to MIX-05 test, the re-stratication leads to
different top and bottom layer temperatures but the general
behavior of the re-stratication is captured in the simulation.
A more detailed comparison of the temperature behavior in the
pool is shown in Fig. 16. In Fig. 16a, the temperature proles of the
middle (about 0.48 m from the bottom) and bottom layers both in
the EHS-EMS simulation and MIX-06 experiment are shown. In
GOTHIC 8.0, we model the bottom of the vessel as at to simplify
the heat transfer at the bottom with one thermal conductor. In this
GOTHIC version, although it is possible to represent the nearly
hemispherical shape of the bottom by blocking cells, it requires,
however, to use one thermal conductor for each blocked cell and
there are 48 blocked cells in the given mesh. For the comparison,
the corresponding locations of the bottom and middle layers are
then adjusted accordingly. In the experiment, the pool bottom is
at 0 m and the location of the rst TC is at 0.07 m. In the
simulation, the pool bottom is at 0.27 m and the corresponding
location of the rst TC for comparison is at 0.34 m. As can be seen
in Fig. 16c, the temperatures of the bottom layers both in the
496 H. Li et al. / Annals of Nuclear Energy 111 (2018) 487–498

Fig. 15. Comparison of pool temperature between (a) MIX-06 measured data and (b) EHS-EMS simulation. The level of the pipe outlet is at 1.045 m.

Fig. 16. MIX-06 test: Comparison of pool temperature (a) snapshot of vertical temperature prole at time t = 500 s, (b) snapshot of vertical temperature prole at time
t = 2000 s, (c) snapshot of vertical temperature prole at time t = 2500 s, (d) snapshot of vertical temperature prole at time t = 3500 s.

simulation and experiment are higher than the middle layers. This
is caused by the downward ow of heated water from the blow-
down pipe outlet (and can be clearly seen in Fig. 17c). The temper-
ature at the bottom layer though is 2–3 degrees higher in the
simulation than in the experiment. In Fig. 16b, a snapshot of the
vertical temperature prole near the end of the thermal stratica-
tion phase (at time t = 2000 s) is shown. Both the experiment and
simulation shows that the layer below the pipe outlet remains cold
(about 16 °C, same as initial) while in the upper layer thermal
stratication develops.
Fig. 17 shows snapshots of the predicted temperature and
velocity proles in the pool at different times t = 500 s (early strat-
ication phase), t = 2000 s (end of stratication phase), t = 2500 s
(mixing phase), and t = 3500 s (re-stratication phase). At
t = 2000 s, a thermally stratied layer develops at the top while
the lower layer remains cold. The magnitude of the maximum
velocity at this time is just 0.09 m/s. During the mixing phase at
t = 2500 s, the ow circulation changes to counter-clockwise
manner due to the dominant effect of the jet from the pipe outlet.
The jet directed downwards transports hot water to the bottom
 H. Li et al. / Annals of Nuclear Energy 111 (2018) 487–498
Fig. 17. MIX-06 test: EHS-EMS simulation snapshots of temperature proles in the pool with superimposed velocity elds at different times (a) t = 500 s, during early stage of
stratication phase, (b) t = 2000 s, during end of stratication, (c) t = 2500 s, during mixing phase, and (d) t = 3500 s, during re-stratication phase.
layer which is also observed in the experiment. The magnitude of
the maximum velocity at this time is about 0.56 m/s. During the
re-stratication phase at t = 3500 s, the ow circulation changes
to clockwise manner again due to weak momentum downward
and high buoyancy force produced by heat uxes from the pipe.
In MIX-05 and 06 tests, the threshold when re-stratication
occurs seems to be about 12 kgm/s2. This is consistent with the
MIX-01 and 04 tests where the momentum rate is a bit higher than
the threshold and the pool is completely mixed. It should be noted
that such threshold is specic to the facility, test conditions and
respective momentum injected in the pool. In the estimation of
the frequency and amplitude of oscillations, we rely on the TC sig-
nal inside the pipe as mentioned above. The TC signal at the start of
the mixing stage can be divided into three phases: the rst phase is
nearly harmonic, then followed by a saw-type signal with lower
frequency and amplitude of oscillation and nally almost no oscil-
lation. In the current model, our estimation of the frequency and
amplitude assumes a harmonic signal and is applied on all phases.
There is an underestimation of the momentum rate in the phase
with the saw-type signal, hence the re-stratication occurred ear-
lier. Further improvement is necessary. Nevertheless, the model
still correctly represents the physics. The early stratication is cap-
tured where condensation happens inside the pipe and the effec-
tive momentum due to the oscillation of the steam-water
interface is negligible. The mixing phase is also captured where
the steam-water interface starts to oscillate at high frequency
and amplitude generating an effective momentum high enough
to mix the pool. As the oscillation dies down while the steam mass
ow rate remains constant, the effective momentum decreases
which results in a re-stratication of the pool.
6. Conclusions
We have presented an experimental and numerical investiga-
tion of thermal stratication and mixing in a suppression pool
induced by steam injection through a blowdown pipe. The experi-
ments are conducted in the PPOOLEX facility. The effect of steam
condensation regimes on transition between thermal stratication
and mixing are observed in the tests. Specically, the experiments
exhibit the development of thermal stratication in the pool at
497
relatively low mass ow rates. Pool mixing is observed when
steam mass ow rates are increased. However, thermal stratica-
tion can re-develop later on, even at the same relatively high mass
ow rates, which is due to the increasing pool temperature that
shifts the condensation regime from chugging to transition regime.
The previously proposed Effective Heat Source (EHS) and Effec-
tive Momentum Source (EMS) models are implemented in GOTHIC
and validated against the experiments. We found an excellent
agreement between the experiment and simulation in terms of
averaged pool temperature and water level. More importantly,
the numerical simulations quantitatively capture the stratication
and mixing phenomena in the pool, including re-development of
stratication. Furthermore, we also observed different ow pat-
terns from the simulations which can result in either development
of thermal stratication or mixing.
Further improvement of the EHS/EMS models in order to reduce
uncertainties in predictions is a subject of the future work. Speci-
cally, the EHS/EMS models should be extended for other compo-
nents in the suppression pool, which can inuence the thermal
behavior of the pool, i.e. spargers (see, for example Gallego-Marcos
et al., 2016), injection nozzles, sprays, and strainers.
Acknowledgments
The authors thank Ignacio Gallego-Marcos for fruitful discus-
sions and the referees for their valuable comments and sugges-
tions. This work was performed within the RM3 project funded
by NORTHNET, Sweden, and the ENPOOL project funded by the
Nordic Nuclear Safety Research (NKS). GOTHIC is developed and
maintained by the Numerical Applications Division of Zachry
Nuclear Engineering under EPRI sponsorship. The authors would
like to acknowledge NAI for providing access to the program for
educational and research purposes.
References
Cheng, L., Woo, K.S., Ishii, M., Lim, J., Han, J., 2006. Suppression pool mixing and
condensation tests in PUMA facility. In: Proc. of the International Conference on
Nuclear Engineering, ICONE.
Fox, R.J., 1992.Temperature distribution in pools with shallow buoyant jets. In: Proc.
of the 5th International Topical Meeting on Nuclear Reactor Thermal Hydraulics
(NURETH-5), September 21–24, Salt Lake City, Utah. 1227–1234.
498 H. Li et al. / Annals of Nuclear Energy 111 (2018) 487–498
Gallego-Marcos, I., Villanueva, W., Kudinov, P., 2016. Scaling of thermal
stratication and mixing in a large water pool induced by steam injection
through spargers. In: Proc. of the 11th International Topical Meeting on Nuclear
Thermal-Hydraulics, Operation and Safety (NUTHOS-11), Gyeongju, Korea,
October 9–13, 2016.
Gallego-Marcos, I., Villanueva, W., Kudinov, P., 2017. Modelling of Pool Stratication
and Mixing Induced by Steam Injection through Blowdown Pipes. Ann. Nucl.
Energy, under revision.
Gamble, R.E., Nguyen, T.T., Peterson, P.F., 2000. Pressure suppression pool mixing in
passive advanced BWR plants. Nucl. Eng. Des. 204, 321–336.
Kataoka, Y., Fukui, T., Hatamiya, S., Experimental study on convection heat transfer
along a vertical at plate between different temperature pools. In: ANS National
Heat Transfer Conference. Minneapolis, 28–31 July, 1991.
Krishnan, G., Mohseni, K., 2009. Axisymmetric synthetic jets: an experimental and
theoretical examination. AIAA J. 47, 2273–2283.
Lahey, R.T., Moody, F.J., 1993. . The Thermal Hydraulics of a Boiling Water Reactor,
second ed. American Nuclear Society, Illinois. 582 p..
Laine, J., Puustinen, M., 2006. Thermal stratication experiments with the
condensation pool test rig. NKS-117.
Laine, J., Puustinen, M., Räsänen, A., 2013. PPOOLEX experiments on dynamics of
free water surface in the blowdown pipe. Research Report EXCOP 2/2012,
NKS-281.
Laine, J., Puustinen, M., Räsänen, A., 2014. PPOOLEX mixing experiments. Research
Report EXCOP 1/2013, NKS-309.
Li, H., Villanueva, W., Kudinov, P., 2013. Validation of Effective Momentum and Heat
Flux Models for Stratication and Mixing in a Water Pool. Nordic Nuclear Safety
Research, Research Report NKS-284.
Li, H., Villanueva, W., Kudinov, P., 2014a. Approach and development of effective
models for simulation of thermal stratication and mixing induced by steam
injection into a large pool of water. J. Sci. Technol. Nucl. Installations (STNI).
Article ID 108782.
Li, H., Villanueva, W., Puustinen, M., Laine, J., Kudinov, P., 2014b. Validation of
effective models for simulation of thermal stratication and mixing induced by
steam injection into a large pool of water. J. Sci. Technol. Nucl. Installations
(STNI). Article ID 752597.
Li, H., Villanueva, W., Kudinov, P., 2014. Effective models for simulation of thermal
stratication and mixing induced by steam injection into a large pool of water.
Nordic Nuclear Safety Research, Research Report NKS-316.
Norman, T.L., Park, H.S., Revankar, S.T., Ishii, M., Kelly, J.M., 2006. Thermal
stratication and mixing in an open water pool by submerged mixtures of
steam and air. In: ASME International Mechanical Engineering Congress and
Exposition, IMECE2006 – Nuclear Engineering.
Peterson, P.F., 1994. Scaling and analysis of mixing in large stratied volumes. Int. J.
Heat Mass Transfer 37, 97–106.
Smith, B.L., Swift, G.S., 2001. Synthetic Jets at Larger Reynolds Number and
comparison with Continuous Jets. In: 31st AIAA Fluid Dynamics Conference and
Exhibit, AIAA Paper 2001-3030, Anaheim, CA.
Smith, B.L., Swift, G.S., 2003. A comparison between synthetic jets and continuous
jets. Exp. Fluids 34, 467–472.
Smith, B.L., Dury, T.V., Huggenberger, M., Nöthiger, N., 1992. Analysis of single-
phase mixing experiments in open pools. In: Cheung, F.B., Peterson, P.F. (Eds.),
Thermal Hydraulics of Advanced and Special Purpose Reactors, ASME HTD, vol.
209. ASME, New York, pp. 91–100.
Song, D., Erkan, N., Jo, B., Okamoto, K., 2014. Dimensional analysis of thermal
stratication in a suppression pool. Int. J. Multiph. Flow 66, 92–100.
Villanueva, W., Li, H., Puustinen, M., Kudinov, P., 2015. Generalization of
experimental data on amplitude and frequency of oscillations induced by
steam injection into a subcooled pool. Nucl. Eng. Des. 295, 155–161.
Zhao, H., 2003. Computation of Mixing in Large Stably Stratied Enclosures (Ph.D.
Dissertation). University of California, Berkeley.
Zhao, H., Peterson, P.F., 2007. One-dimensional analysis of thermal stratication in
AHTR and SFR coolant pools. In: Proc. of the 12th International Topical Meeting
on Nuclear Reactor Thermal Hydraulics (NURETH-12).