You are on page 1of 18

Numerical Heat Transfer, Part A: Applications

An International Journal of Computation and Methodology

ISSN: 1040-7782 (Print) 1521-0634 (Online) Journal homepage: https://www.tandfonline.com/loi/unht20

Modeling of steam–water direct contact


condensation using volume of fluid approach

Priyankan Datta, Aranyak Chakravarty, Koushik Ghosh, Achintya


Mukhopadhyay & Swarnendu Sen

To cite this article: Priyankan Datta, Aranyak Chakravarty, Koushik Ghosh, Achintya
Mukhopadhyay & Swarnendu Sen (2018) Modeling of steam–water direct contact condensation
using volume of fluid approach, Numerical Heat Transfer, Part A: Applications, 73:1, 17-33, DOI:
10.1080/10407782.2017.1420308

To link to this article: https://doi.org/10.1080/10407782.2017.1420308

Published online: 16 Jan 2018.

Submit your article to this journal

Article views: 638

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=unht20
NUMERICAL HEAT TRANSFER, PART A
2018, VOL. 73, NO. 1, 17–33
https://doi.org/10.1080/10407782.2017.1420308

none defined

Modeling of steam–water direct contact condensation using


volume of fluid approach
Priyankan Dattaa, Aranyak Chakravartya,b, Koushik Ghosha, Achintya Mukhopadhyaya, and
Swarnendu Sena
a
Department of Mechanical Engineering, Jadavpur University, Kolkata, India; bSchool of Nuclear Studies and
Applications, Jadavpur University, Kolkata, India

ABSTRACT ARTICLE HISTORY


Direct contact condensation (DCC) of steam in subcooled water has Received 23 September 2017
paramount importance in many heat transfer devices in different industrial Accepted 12 December 2017
areas like nuclear, thermal, chemical plants. This work aims at the exploration
of underlying physics of steam–water DCC in two dimensions using ANSYS
FLUENT 14.5. The volume of fluid method is utilized for performing direct
simulation of the phenomenon at the phase interface. In this work, thrust is
given on the modeling of the interphase heat transfer using interfacial jump
approach instead of proposed empirical correlations which have different
applicability limits. User-defined functions in the FLUENT software are used for
evaluating the interfacial mass transfer rate, thermal gradient across the phase
interface, and interface curvature. This study also emphasizes on the
prediction of transient temperature field and interface characteristics under
different parametric conditions (e.g., variation of water injection velocity and
water temperature). Observation reveals that the present condensation model
is capable of capturing the transient temperature history as well as the flow
regime transition (stratified to slug flow) induced by the interfacial instability.

1. Introduction and literature survey


Direct contact condensation (DCC) between steam and subcooled water is a very important thermo-
hydraulic phenomenon in various industrial sectors such as nuclear reactor safety systems, direct
contact feed water heater, mixing-type heat exchanger in chemical industry, etc. DCC phenomenon
has gained attention among the researchers since it enables intense heat and mass transfer through
the two-phase interface and typical fast transients [1–4] which could have serious implications on
structural integrity and safety, especially in nuclear plants.
Past studies reveal that DCC is a complex process which is strongly dependent on the thermal
energy transport across the two-phase interface. On the other hand, this transport of thermal energy
increases with an increase in the interfacial area density which in turn is affected by the turbulence
intensity in the liquid phase. The complex interplay among these thermal and mechanical factors
ultimately drives the condensation rate during steam–water DCC [5]. Literature survey depicts that
due to the diversity in flow pattern, no universal correlation exists for assessing interphase heat trans-
fer in the context of DCC. Thus, thrust is given on the realistic modeling of different condensation
models in horizontal stratified flow situation for correct understanding of DCC process using either
CFD code or dedicated thermohydraulic code such as RELAP, CATHARE, etc.
The group of Štrubelj and others [4, 6, 7] performed numerical simulations to replicate the experi-
ments performed on the integral test facility PMK-2 at Hungarian Atomic Energy Research Institute

CONTACT Koushik Ghosh kghoshjdvu@gmail.com Department of Mechanical Engineering, Jadavpur University, Kolkata
700032, India.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/unht.
© 2018 Taylor & Francis
18 P. DATTA ET AL.

Nomenclature
aint Interfacial area density (m 1) Greek Letters
Cp specific heat capacity at constant α vapor volume fraction
pressure (J kg 1 K 1) ρ density (kg m 3)
g acceleration due to gravity μ dynamic viscosity (kg m 1 s 1)
(m s 2) λ thermal conductivity (W m 1 K 1)
h enthalpy (J kg 1) κ turbulence kinetic energy (m2 s 2)
m
_ mass transfer rate E turbulence dissipation rate (m2 s 3)
(kg m 3 s 1)
^
n unit normal vector Subscripts
P pressure (Pa) eff effective
Pr Prandtl number g vapor phase
q00 heat flux (W m 2) in injection
Re Reynolds number int interface
Sm energy transfer rate k phase index
(J m 3 s 1) l liquid phase
T temperature (K) sat saturation
V velocity (m s 1) t turbulent
x, y Cartesian co-ordinates w wall

(KFKI). They used “two-fluid model”-based NEPTUNE-CFD code for the prediction of temperature
field at the salient locations within the test section. The authors considered the “large interface model”
proposed by Coste et al. [8] and the “Hughes-Duffey model” [9] based on “surface renewal theory
(SRT)” for modeling interphase momentum transfer and phase change, respectively, during steam–
water DCC. They claimed that NEPTUNE-CFD code, based on the large interface model along with
the surface renewal model, can capture the stratified to slug flow regime transition adequately. Ceuca
et al. [10, 11] performed experimental as well as numerical investigation at Lithuanian Energy Insti-
tute on DCC phenomenon during steam–water cocurrent flow situation. Their study focused on the
verification and validation of the “ATHLET” system code (which involves both hybrid SRT-based
heat transfer model and interfacial area concentration transport equation) results with the experi-
mental data obtained at PMK-2 test facility. The authors found that two-phase flow turbulence
(Reynolds number and Froude number), condensation potential (Jakob number), and density ratio
(Atwood number) between the two phases primarily govern the DCC-driven rapid condensation
events. Hohne et al. [12] performed the numerical simulations using two-fluid model in ANSYS
CFX and compared the results with the steam–water DCC experiments performed at Lithuanian
Energy Institute. They used three different models, namely, “Egorov DCC model [13],” “Hughes-
Duffey model,” and “Coste DCC model [14]” for modeling the interface heat transfer during DCC
scenario and claimed all three condensation models showed good accuracy with the experimental
result. Patel et al. [15] performed numerical simulations using two CFD software—OpenFOAM
and NEPTUNE-CFD, for assessing the interfacial condensation rate during low Reynolds number
DCC of steam in subcooled water. The authors used SRT-based “Hughes-Duffey model” as well as
direct numerical simulation-based “surface divergence model [16]” to compare their simulation
results with the experimental outcomes of the “POOLEX” test facility. They claimed that “surface
divergence model” can predict the condensation phenomena more accurately than the “Hughes-Duffey”
model. Szijártó et al. [17, 18] implemented three different condensation models, namely, “numerical
iteration technique,” “heat flux balance model,” and “phase filed model” in the FLUENT
software and assessed their results with the experimental cases of “LAOKOON” test facility during
steam–water DCC. They used volume of fluid (VOF) method and observed that prediction of the
temperature field with the phase field theory gives the best agreement with the experimental data among
the three models.
The objective of the recent work of Li et al. [19] is slightly different. They performed 3D numerical
simulations using VOF method in FLUENT 6.3 for investigating the superheated steam and
NUMERICAL HEAT TRANSFER, PART A 19

subcooled water direct contact-induced “chugging” phenomenon in a tee junction and compared the
results with their own experiment. The authors used LES turbulence model along with the interfacial
heat transfer coefficient correlation and claimed a good agreement between simulation and experi-
mental data. Patel et al. [20] also performed simulations with NEPTUNE-CFD code as well as with
OpenFOAM to replicate the experiment performed at the “PPOOLEX” facility of Lappeenranta
University of Technology in the context of chugging DCC mode. They used “Hughes-Duffey model”
based on small eddies, “Hughes-Duffey model” based on large eddies, and “Coste model” to assess the
interfacial condensation rate. The authors also studied the influence of “Rayleigh-Taylor Instability”-
based interfacial area density model (as proposed by Pellegrini et al. [21]) and claimed a better agree-
ment with the experimental outcomes.
It is evident from the above survey that most of the past studies focused on the modeling of steam–
water DCC phenomenon using two-fluid approach. This two-fluid modeling approach is based on the
space and time averaging of the governing equations [22, 23] and hence, introduce an additional
“modeling uncertainty.” In addition, previous studies involve the use of different proposed empirical
correlations for evaluating the interfacial condensation rate. However, these proposed correlations are
not generic and have different applicability limits. Table 1 summarizes the different empirical
condensation correlations that are used in the previous researches in the context of DCC.
Therefore, the present study focuses on the exploration of the underlying physics of the steam–
water DCC phenomenon in a two-dimensional horizontal pipe geometry using VOF method which
explicitly allows the direct simulation of the phenomena at the two-phase interface. Furthermore, in
this work, instead of using proposed empirical correlation, basic conservation principles are used to
assess the interface condensation rate across the two-phase interface (as proposed in the work of
Welch and Wilson [24] and Sun et al. [25, 26]). It is to be noted that VOF model used in the present
work accurately estimates the temperature gradient across the phase interface and thus enables a high
precision of heat and mass transfer calculation. The effect of variation of water injection rate and inlet
water temperature on the interface morphology as well as on the temperature field within the test
section is investigated using ANSYS FLUENT 14.5. User-defined functions (UDF) in the FLUENT
software such as “DEFINE_MASS_TRANSFER” macro is used for obtaining the energy and mass
transfer across the interface. To the best of authors’ knowledge, though a few recent works
[25–28] have been performed on vapor bubble condensation during subcooled boiling flow situation
based on the interfacial jump approach, such detailed systematic numerical investigation in the
context of steam–water DCC phenomenon using this fundamental approach along with pure
interface tracking method has not been addressed yet.

2. Problem definition
2.1. Description of the problem
The schematic of the physical model along with the initial and boundary conditions which are
considered in the present work is shown in Figure 1.
In this study, steam–water DCC phenomenon is investigated in a two-dimensional horizontal pipe
having length of 2500 mm and internal diameter of 66 mm (L/D ratio ¼ 38). Initially it is assumed
that the test section is filled with dry saturated steam. Simulation begins as subcooled water starts

Table 1. Different condensation models used in the earlier works in the context of DCC.
Model name Condensation Nusselt number
Hughes and Duffey model [9] 1=2
p2 Pr
p l Ret
Egorov DCC model [13] 1=3 1=2
2 þ 0:6Prl Ret
Coste DCC model [14] 1=2 7=8
2:7Prl Ret
Surface divergence model [16] CPrn f ½Ret �Rem
t
DCC, direct contact condensation.
20 P. DATTA ET AL.

Figure 1. Schematic of the simulation geometry, initial and boundary conditions.

entering into the steam-filled test section at a specific velocity from the left. All the walls are
considered as adiabatic and nonslip to the fluids. IAPWS-95 formulation-based thermophysical
properties are used for both the phases. In addition, present work also assumes that both steam
and water are incompressible and flow is turbulent in nature.

2.2. Mathematical modeling


2.2.1. Governing equations
In our work, VOF approach in ANSYS FLUENT 14.5 is used for modeling of the steam–water DCC
events. The governing equations for continuity, momentum and energy are presented as follows:
Continuity equation:
For the kth phase, continuity equation can be written as:
qðak qk Þ
_k
þ r � ðak qk VÞ ¼ m ð1Þ
qt
The transport equation for the kth phase volume fraction can be derived from Eq. (1) which will
take the following form
qak _k
m
þ V � rak ¼ ð2Þ
qt qk
In ANSYS FLUENT, Eq. (2) is solved only for the dispersed phase. Therefore, the primary phase
volume fraction is computed subject to the following constraint
X
n
ak ¼ 1 ð3Þ
k¼1

Momentum equation:
qðqVÞ �
þ r � ðqVVÞ ¼ rP þ r � mðrV þ rVT þ qg þ F ð4Þ
qt
In Eq. (4), V represents the resulting velocity field which is shared by the individual phases.
Furthermore, ρ and μ in Eq. (4) are the volume-averaged density and viscosity respectively that
are evaluated as follows
X
n
q¼ ak qk ð5Þ
k¼1

X
n
m¼ ak mk ð6Þ
k¼1

The last term in the RHS of Eq. (4) represents the volumetric equivalent of the surface tension
force which is implemented in the FLUENT solver in a similar fashion (continuum surface force
model) as proposed in the work of Brackbill et al. [29].
NUMERICAL HEAT TRANSFER, PART A 21

Energy equation:
qðqEÞ
þ r � ðVðqE þ PÞÞ ¼ r � ðkeff rT Þ þ Sm ð7Þ
qt
In Eq. (7), E and T represent the mass-averaged energy and temperature, respectively, which are
calculated as
Pn
ak qk Ek
E ¼ Pk¼1n ð8Þ
k¼1 ak qk

where Ek for individual phase is based on the specific heat of that phase and shared temperature.
In the present study, the last term in the RHS of Eq. (7) is the volumetric energy source term
corresponding to the mass transfer across the two-phase interface. The effective thermal conductivity
(λeff) in the energy equation can also be expressed in a similar fashion as volume-averaged density and
viscosity defined in Eqs. (5) and (6)–
X
n
keff ¼ ak kk ð9Þ
k¼1

2.2.1.1. Turbulence modeling. In the present work, “standard j �o” model in ANSYS FLUENT [30]
is used for modeling of the turbulence. The transport equations for obtaining turbulence kinetic
energy (κ) and its dissipation rate (E) have the following form
�� � �
qðqjÞ qðqjui Þ q mt qj
þ ¼ mþ þ Gk þ Gb q�o ð10Þ
qt qxi qxj rk qxj
�� � �
qðq�oÞ qðq�oui Þ q mt q�o �o E2
þ ¼ mþ þ C1E ðGk þ C3E Gb Þ C2E q ð11Þ
qt qxi qxj rE qxj j j
where Gk and Gb in Eqs. (10) and (11) are the turbulence kinetic energy generation term due to mean
velocity gradient and buoyancy, respectively. C1E , C2E , and C3E are the constant terms. In our work,
C1E and C2E are taken as 1.44 and 1.92, respectively. Furthermore, σk (¼1.0) is the turbulent kinetic
energy Prandtl number and rE (¼1.3) is the turbulence dissipation rate Prandtl number,
respectively. The similar values of these model constants can also be observed in the recent work
of Patel et al. [15, 20] in the context of steam–water DCC.

2.2.1.2. Modeling of interphase mass transfer. User-defined functions in ANSYS FLUENT [31] are
utilized for implementing the mass transfer across the two-phase interface. For this purpose,
DEFINE_MASS_TRANSFER macro is used. The detailed steps for calculation of the mass transfer
term are discussed in this section as
Interface identification:
The computational cells which contain the steam–water two-phase interface are identified subject
to the following volume fraction constraint
0 < ak < 1 ð12Þ
In Eq. (12), αk represents the vapor phase volume fraction.
Mass transfer rate:
Total mass transfer rate across the interface, thus identified, at a specific time can be expressed in
the similar way as described in the work of Welch and Wilson [24] as-
jjq00 jj � n
^
_k¼
m aint ð13Þ
hlg
22 P. DATTA ET AL.

Figure 2. Schematic of heat flux condition across the two-phase interface.

In Eq. (13), hlg is the latent heat in Eq. (13), hlg is the latent heat of vaporization while jjq00 jj
represents the heat flux jump across the interface. A schematic of the heat flux condition across
the two-phase interphase is shown in Figure 2.
The detailed calculation of the heat flux jump term ðq00 Þ is performed in the following manner

kq00 k ¼ q001 þ q00g ð14Þ

In this work, it is assumed that the vapor phase will remain at the saturation state which signifies
q00g ¼ 0 and thus, Eq. (14) yields

jjq00 jj ¼ q001 ¼ kl rT ð15Þ


where ∇T represents the temperature gradient across the interface. A user-defined macro “C_T_G” in
FLUENT software is used for evaluating the temperature gradient across the interface.
The interfacial area density (aint) in Eq. (13) is evaluated in terms of the magnitude of the vapor
phase volume fraction gradient as

aint ¼ jrak j ð16Þ


Furthermore, n ^ in Eq. (13) represents the unit normal vector that signifies the orientation of the
interface in the two-phase cells. This unit normal vector is evaluated in the similar fashion as defined
in the work of Welch and Wilson [24] as
rak

n ð17Þ
jrak j
In Eq. (17), αk is considered as the vapor phase volume fraction. The user-defined scalar macro
“C_UDSI_G” in FLUENT is used to calculate the volume fraction gradient for obtaining both the
interfacial area density and unit normal vector for the present study. Table 2 gives a compact overview
about different macros in FLUENT software which are used in the present work.
Energy transfer rate:
After evaluating the interfacial mass transfer rate, volumetric energy source term in Eq. (7) is
obtained as follows:

_ k hlg
Sm ¼ m ð18Þ

Table 2. Different macros used in the present work.


Macros Purpose
DEFINE_MASS_TRANSFER Calculation of mass transfer rate
C_T_G Temperature gradient calculation
C_UDSI_G Volume fraction gradient calculation
NUMERICAL HEAT TRANSFER, PART A 23

Table 3. Discretization schemes used in the present work.


Variables Spatial discretization scheme
Pressure PRESTO (pressure staggering option)
Volume fraction QUICK (quadratic upwind interpolation for convective kinematics)
Momentum, energy, κ and �o Second-order upwind

3. Numerical procedure
In our work, numerical simulation of Eqs. (1), (2), (4), (7), (10), and (11) is obtained by a control
volume approach using computational fluid dynamics software ANSYS FLUENT 14.5. Pressure–
velocity coupling is achieved using the well-known “SIMPLE” [32] algorithm. Table 3 summarizes
about different discretization schemes that are used in this work.
For the current analysis, solution domain is divided into 2,01,284 number of computational cells
which is corresponding to 2,04,240 nodes. The convergence criterion for continuity, momentum,
energy, turbulence kinetic energy, and turbulence dissipation rate are set to 10 5.

4. Results and discussion


4.1. Model validations
To assess the capability and reliability of the present phase change model, three different validation
case studies are performed prior to the presentation of the detailed simulation results of the
steam–water DCC phenomenon.

4.1.1. Stefan problem


The one-dimensional Stefan problem, defined by Welch and Wilson [24] as well as Son and Dhir [33],
is chosen as a benchmark problem for assessing the accuracy of the phase change model used in the
present study. Figure 3 is a schematic of the problem considered.
Both the phases, liquid and vapor—are assumed to be incompressible and initially in equilibrium
with each other. A higher wall temperature, however, induces a thermal gradient within the vapor
layer and resulting in mass transfer at the vapor–liquid interface. As a consequence of this interfacial
mass transfer, vapor layer grows with time which is evident from a movement of the two-phase
interface away from the solid wall as time progresses.
The analytical solution of the instantaneous interface position (δ) is given by Alexiades and
Solomon [34] as
sffiffiffiffiffiffiffiffiffiffiffi
kg t
dðtÞ ¼ 2e ð19Þ
qg Cpg
where e is the solution of the following transcendental equation
� Cpg ðTw Tsat Þ
e exp e2 erf ðeÞ ¼ pffiffiffi ð20Þ
hlg p

Figure 3. Definition of the Stefan problem.


24 P. DATTA ET AL.

Figure 4. Comparison of the instantaneous interface position between simulation and analytical solution for Stefan problem.

The thermophysical properties chosen for this case study are ρg ¼ 0.001 kg m 3, ρl ¼ 1 kg m 3,
λg ¼ 0.005 W m 1 K 1, Cpg ¼ 200 J kg 1 K 1, hlg ¼ 104 J kg 1, and (Tw Tsat) ¼10 K.
For the simulation purpose, computational domain is divided into 200 cells with element size of
1 mm. The initial vapor film thickness is taken as 1 mm. Figure 4 compares the simulated interface
location with the analytical solution (given by Eq. 19) proposed by Alexiades and Solomon [34]. The
analytical solution here is computed for the abovementioned thermophysical properties. Observation
shows that the simulation result obtained with the present phase change model (interface jump
approach) is in good agreement with the analytical solution of the problem.

4.1.2. Sucking interface problem


The sucking interface problem defined by Welch and Wilson [24] is also investigated in our
validation study for assessing the accuracy of the present phase change model. It is assumed that a
stagnant vapor layer at saturation condition initially coexists with a metastable liquid (superheated
liquid) layer. This leads to the development of a thin moving thermal boundary layer along with
the vapor–liquid interface. Capturing of a very thin boundary layer around the vapor–liquid interface
makes the sucking problem more computationally challenging. Figure 5 schematically represents the
initial and boundary conditions of the problem.
To simulate the above-defined problem, the computational domain is divided into 200 cells with
element size of 0.05 mm. The thermophysical properties for both vapor and liquid phases are taken
corresponding to 1.013 bar pressure. Initial vapor film thickness is considered 0.5 mm. The vapor–
liquid interface temperature is maintained at 373.15 K (saturation temperature corresponding to
1.013 bar), whereas the far stream liquid temperature is maintained at 378.15 K (5° superheat). A time
step size of 5 � 10 5 s is selected for the present simulation.

Figure 5. Definition of Sucking interface problem.


NUMERICAL HEAT TRANSFER, PART A 25

Figure 6. Instantaneous interface position comparison between present simulation result and the result obtained by Welch and
Wilson [24] for sucking interface problem.

Interface position is tracked in the present study and it is compared with the result obtained by
Welch and Wilson [24]. Comparison of the results in Figure 6 shows that prediction of the present
phase change model is fairly close with the published result.

4.1.3. Validation of steam–water DCC phenomenon with experimental work


For assessing the capability of the interface jump approach during steam–water direct contact
scenario, we performed a validation study by comparing our simulation results with the experimental
result of Štrubelj and Tiselj [7]. The problem geometry, initial and boundary conditions for this
validation case are shown in Figure 7.
In this validation study, it is assumed that test section is initially filled with saturated steam at
14.5 bar pressure. Phase front starts to move as the cold water (22°C) is injected into the steam-filled
test section at a constant flow rate of 1.01 kg s 1.
We focused on the comparative assessment of the local temperature variation by our simulation with
the experimental data given in the work of Štrubelj and Tiselj [7]. Figure 8 shows the comparison between
the results at about 993 mm away from the injection point. Observation reveals that the prediction of
temperature variation at the specified location with the interface jump approach is in well agreement with
the published experimental data. Results show that as the water front reaches at the specific location, a
sudden decrease in temperature is observed. It can be observed that the onset of water flooding (around
5 s) predicted by our simulation is in good agreement with the published experimental result and
thus establishes the reliability of the present condensation model in the context of steam–water DCC.

4.2. Grid independence study


The grid independence study for the present problem (defined in Section 2.1) is performed by
visualizing temperature and liquid volume fraction variation at 800 mm from the beginning of the

Figure 7. Schematic of the problem geometry along with the initial and boundary conditions for the validation case.
26 P. DATTA ET AL.

Figure 8. Comparison of the temperature variation, predicted by the present simulation using interface jump approach with the
experimental work of Štrubelj and Tiselj [7] at 993 mm away from the water inlet.

test section. Five different element sizes (Δx ¼ 4, 2, 1, 0.5, and 0.25 mm) are considered for this study.
In addition, for each case, a fixed time step (Δt ¼ 5 μs) is used.
It is observed from Figure 9 that as the element size is reduced from 0.5 to 0.25 mm, the change in
the temperature as well as liquid volume fraction distribution at the specified location is almost
insignificant. Therefore, Δx ¼ 0.5 mm (corresponds to 2,01,284 number of computational cells) is
selected and used for further numerical simulations of the present work.

4.3. Study of base case and parametric variations


We segregated our detailed study in two parts—analysis of a base case followed by different
parametric variations. In this work, we aimed to study the effect of water subcooling and its injection
velocity magnitude on the temperature field as well as on the interface behavior (by visualizing liquid
volume fraction distribution) in the context of steam–water DCC. Table 4 represents a compact
overview of different parametric cases which are studied in this work.

4.3.1. Base case


The observations corresponding to the base case are shown in Figure 10. In this case, test section is
filled with dry saturated steam at 6 bar pressure (corresponding Tsat ¼ 158.826°C). Transience

Figure 9. Grid independence study in terms of (a) temperature variation and (b) liquid volume fraction history at 800 mm from
the beginning of the test section for Psat ¼ 6 bar, Vl,in ¼ 5 m s 1, andTl,in ¼ 20°C.
NUMERICAL HEAT TRANSFER, PART A 27

Table 4. A compact overview of the study matrix.


Case study Tl(°C) Tg(°C) Psat(bar) Vl,in(m s 1)
Base case 20 158.826 6 3
Variation of water injection velocity 20 158.826 6 1
5
Variation of inlet water temperature 40 158.826 6 3
80
100

initiates as the subcooled water at 20°C starts entering within the steam-filled section. During
simulation period, water injection velocity is kept at 3 m s 1. Figure 10a shows the transient tempera-
ture variation at 800 and 1400 mm from the beginning of the steam-filled test pipe. Observation
reveals that about 0.3 s subcooled water front reaches at 800 mm which results in sudden drop of
the temperature from its saturation value and thus local thermal equilibrium between the two phases
is attained. It can be observed that the variation of the temperature field at the distant section (at
1400 mm) also follows the similar distribution only with a time lag.
Figure 10b shows the liquid volume fraction history at the specified locations (at 800 and
1400 mm). This variation can be explained with the visualization of the instantaneous water front
position within the test section. Figure 11 shows water front propagation history at six different time
instants (t ¼ 0.2, 0.4, 0.6, 0.8, 1.2, and 1.4 s). Water front starts to flood the specified location
(800 mm) around 0.3 s and consequently liquid volume fraction in that location also starts to increase.
Observation shows that initially as the subcooled water enters into the test section, both the phases
(water and vapor) coexist in a stratified flow regime (Figure 11a) and condensation occurs continu-
ously at the phase boundary. As a result of this phase change process, steam from the surrounding
region accelerates over the liquid phase into the area of condensation to replace the condensed steam
and thus results in the formation of wavy interface due to “Kelvin-Helmholtz instability”. As time
progresses, it can be observed that surface wave grows and the wave crest touches the pipe wall. This
in turn leads to the stratified to slug flow regime transition and resulting in bridging within the test
section. It can be seen that as the water slug bridges the pipe, vapor gets entrapped within the
surrounding liquid and thus a vapor pocket is formed at about 0.4 s (Figure 11b). The formation
of vapor pocket can be attributed to the decrease in liquid volume fraction (Figure 10b). The
oscillations in the liquid volume fraction distribution at 800 mm in the subsequent times are seen
due to the accelerating water front propagation throughout the section. Observation also inferred that
liquid volume fraction variation at 1400 mm follows the similar distribution with a time shift.

Figure 10. Variation of (a) temperature and (b) liquid volume fraction at two different locations within the test section for
Psat ¼ 6 bar, Vl,in ¼ 3 m s 1, and Tl,in ¼ 20°C.
28 P. DATTA ET AL.

Figure 11. Water front (blue colored) propagation history within the test section (x ¼ 0 indicates the beginning of the steam filled
test section) for Vl,in ¼ 3 m s 1 and Tl,in ¼ 20°C.

4.3.2. Variation of water injection velocity


The effect of variation of water injection velocity on the temperature field as well as on the interface
behavior during steam–water DCC is investigated in this section. For this case study, two different
injection velocity magnitude (Vl,in ¼ 1 and 5 m s 1) is considered while water temperature (Tl,in) is
maintained at 20°C.
Observation shows that for the lowest injection velocity case (Vl,in ¼ 1 m s 1), as the water front
starts to flood the test section, a stratified flow regime is established between the two phases (steam
at the top on water as is seen from Figure 13a, c). In this case, since the momentum of the injected
water is small, it can be seen that two-phase interface never touches the pipe wall. Thus, flow regime
remains stratified over the rest of the simulation period and a sluggish water flooding is observed at
the specified locations (at 800 mm as well as at 1400 mm section of Figure 13c). This results in the
decrease in local temperature slowly from its saturation value and hence, thermal equilibrium is
attained at about 2.5 s (Figure 12a) at 800 mm section.
The variation of local temperature as well as liquid volume fraction corresponding to the higher
injection velocity case (Vl,in ¼ 5 m s 1) is shown in Figures 12b, d. It can be observed that as the
injection velocity is increased to 5 m s 1, bridging occurs adjacent to the injection point. Due to
the continuous injection of the water, phase boundary is pushed as a column head (Figure 13b)
and thus stratified to slug flow regime transition has not occurred as we observed in the base case
for Vl,in ¼ 3 m s 1. Result reveals that for Vl,in ¼ 5 m s 1, water front floods the 800 mm section
within 0.2 s and consequently a sharp temperature drop is observed (Figure 12b). Transient
temperature history and the liquid volume fraction variation at 1400 mm section also show similar
behavior as 800 mm with a time lag. It can be seen that (Figure 13d), within 1.2 s, almost entire steam
in the test section gets condensed into water.
Interfacial condensation rate for Vl,in ¼ 1 and 5 m s 1 at 800 and 1400 mm location within the
test section is shown in Figure 14. Observation reveals that for both the cases, mass transfer takes
place only when the water front starts to flood the specific location and ceases as the section is
completely flooded with water. As a result, condensation mass transfer rate becomes substantial
only over a short period of time. Furthermore, it can be observed that at a specific location,
condensation rate is reduced as water injection velocity increases. This can be explained form
NUMERICAL HEAT TRANSFER, PART A 29

Figure 12. Variation of temperature and liquid volume fraction at 800 and 1400 mm within the test section for Vl,in ¼ 1 m s 1
(a, c) and Vl,in ¼ 5 m s 1 (b, d) at Psat ¼ 6 bar, and Tl,in ¼ 20°C.

Figure 13. Water front location (blue colored) within the test section at t ¼ 0.45 and 1.2 s for Vl,in ¼ 1 m s m s 1 (a, c) and for
Vl,in ¼ 5 m s 1 (b, d) with Tl,in ¼ 20°C.
30 P. DATTA ET AL.

Figure 14. Condensation rate at (a) 800 mm and (b) 1400 mm within the test section for Vl,in ¼ 1 m s 1 and 5 m s 1 with
Tl,in ¼ 20°C.

the fact that as the injection velocity increases, the residence time of the two-phase interface at a
particular section reduces that results in the decrease in mass transfer rate across the phasic
interface.

Figure 15. Transient history of temperature (a, b) and liquid volume fraction (c, d) at 400 and 800 mm for Tl,in ¼ 40, 80, and 100°C
with Vl,in ¼ 3 m s 1.
NUMERICAL HEAT TRANSFER, PART A 31

Figure 16. Interfacial condensation rate at (a) 400 mm and (b) 800 mm for Tl,in ¼ 40, 80, and 100°C with Vl,in ¼ 3 m s 1.

4.3.3. Variation of inlet water temperature


In this case study, we varied the inlet water temperature within a range between 40–100°C and
analyzed its effect on the transient temperature as well as on the interface behavior at 400 and
800 mm within the test section. For each case, injection velocity is kept at 3 m s 1.
Results (Figure 15a, b) reveal that water starts to flood the specific locations within the test section
almost at the same instant for different inlet water temperatures. It is, thus, evident the fact that the
onset of water flooding at a specific location primarily depends on the injection velocity magnitude.
However, a minor deviation in the temperature profile is observed due to the variation of local
condensation rate at the two-phase interface. This is also corroborated by the transient history of
the liquid volume fraction at two locations as shown in Figure 15c, d. Interfacial condensation rate
for three different inlet water temperatures at 400 and 800 mm sections is shown in Figure 16. It
can be observed that for the lowest inlet water temperature case (Tl,in ¼ 40°C), as the degree of water
subcooling (Tsat Tl) is highest, condensation rate at the phase interface is found to be maximum at
the specified locations.

5. Conclusion
We investigated DCC of steam in subcooled water in a two-dimensional horizontal pipe geometry
using ANSYS FLUENT 14.5. An overarching objective of this study is to understand the underlying
physics of the DCC phenomenon using VOF method which explicitly allows the direct simulation of
the phenomenon at the phase interface. Additionally, thrust is given on the assessment of interfacial
heat and mass transfer from the fundamental approach (which is based on the energy jump across
the steam–water interface) instead of proposed empirical correlations. Different UDF (e.g.,
“DEFINE_MASS_TRANSFER” macro) in the FLUENT software are used for evaluating the
interfacial heat and mass transfer rate. Observation reveals that present condensation model along
with VOF method can capture the temperature variation as well as the interface characteristics
adequately during steam–water DCC.
A parametric study is also performed in this study for investigating the effect of variation of water
injection velocity (Vl,in ¼ 1–5 m s 1) and water temperature (Tl,in ¼ 20–100°C) on the interface beha-
vior as well as on the transient temperature field at a specific location within the test section. Obser-
vation shows that for the lowest injection velocity case (i.e., Vl,in ¼ 1 m s 1), flow regime between the
two phases remains stratified. As the injection velocity magnitude is increased to 3 m s 1, interface
instability grows which causes stratified to slug flow regime transition and resulting in formation
of vapor pocket within the surrounding water slug. However, as the velocity magnitude is further
increased (Vl,in ¼ 5 m s 1), water front propagates through the test section as a water column head
32 P. DATTA ET AL.

and stratified to slug flow regime transition is not found. Results show that interfacial condensation
rate becomes substantial over a short period of time only when the subcooled water front starts to
flood the specific location. In addition, we found that as the water injection velocity increases, con-
densation rate at a specific location is reduced.
Our investigation reveals that, as the inlet water temperature decreases, condensation rate at the
phase boundary is also increased. It can be inferred that this change in the condensation rate at
different temperature causes only a minor deviation in the temperature profile. Thus, water front
propagation and flooding time at a specific location within the test section are primarily dependent
on the injection velocity magnitude.

Acknowledgments
The authors would also like to thank Dr. Pallab Sinha Mahapatra, Assistant Professor, Department of Mechanical
Engineering, IIT Madras, India for the valuable suggestions to improve the quality of this work.

Funding
The first author is grateful to Department of Science and Technology (DST), Government of India for providing
fellowship under the INSPIRE program.

References
[1] P. Datta, A. Chakravarty, K. Ghosh, A. Mukhopadhyay, S. Sen, A. Dutta, and P. Goyal, “A numerical analysis on
the effect of inlet parameters for condensation induced water hammer,” Nucl. Eng. Des., vol. 304, pp. 50–62, 2016.
DOI: 10.1016/j.nucengdes.2016.04.035.
[2] I. F. Barna, A. R. Imre, G. Baranyai, and G. Ézsöl, “Experimental and theoretical study of steam condensation
induced water hammer phenomena,” Nucl. Eng. Des., vol. 240, pp. 146–150, 2010. DOI: 10.1016/j.
nucengdes.2009.09.027.
[3] C. Urban, and M. Schlüter, “Investigations on the stochastic nature of condensation induced water hammer,” Int.
J. Multiphas. Flow, vol. 67, pp. 1–9, 2014. DOI: 10.1016/j.ijmultiphaseflow.2014.08.001.
[4] L. Štrubelj, G. Ézsöl, and I. Tiselj, “Direct contact condensaton induced transition from stratified to slug flow,”
Nucl. Eng. Des., vol. 240, pp. 266–274, 2010. DOI: 10.1016/j.nucengdes.2008.12.004.
[5] Y.-S. Kim, J.-W. Park, and C.-H. Song, “Investigation of the steam-water direct contact condensation heat transfer
coefficients using interfacial transport models,” Int. Commun. Heat Mass, vol. 31, pp. 397–408, 2004. DOI:
10.1016/j.icheatmasstransfer.2004.02.010.
[6] L. Štrubelj, and I. Tiselj, “Heat and mass transfer in the stratified flow with ECCS injection,” Proceedings of the
International Conference Nuclear Energy for New Europe, Portorož, Slovenia, pp. 1–9, 2007.
[7] L. Štrubelj, and I. Tiselj, “Condensation of the steam in the horizontal steam line during cold water flooding,”
Proceedings of the International Conference Nuclear Energy for New Europe, Portorož, Slovenia, 2006.
[8] P. Coste, J. Pouvreau, C. Morel, J. Laviéville, M. Boucker, and A. Martin, “Modelling turbulence and friction
around a large interface in a three-dimension two-velocity Eulerian code,” The 12th International Topical Meeting
on Nuclear Reactor Thermal Hydraulics (NURETH-12), Pennsylvania, USA, 2007.
[9] E. D. Hughes, and R. B. Duffey, “Direct contact condensation and momentum transfer in turbulent separated
flows,” Int. J. Multiphas. Flow, vol. 17, pp. 599–619, 1991. DOI: 10.1016/0301-9322(91)90027-z.
[10] S.-C. Ceuca, and D. Laurinavicius, “Experimental and numerical investigations on the direct contact condensation
phenomenon in horizontal flow channels and its implications in nuclear safety,” Kerntechnik, vol. 81, pp. 504–511,
2016. DOI: 10.3139/124.110729.
[11] S.-C. Ceuca, Computational simulations of direct contact condensation as the driving force for water hammer, Ph.
D thesis, Technische Universitat München, 2015.
[12] T. Höhne, S. Gasiunas, and M. Šeporaitis, “Numerical modelling of a direct contact condensation experiment
using the AIAD framework,” Int. J. Heat Mass Tran., vol. 111, pp. 211–222, 2017. DOI: 10.1016/j.
ijheatmasstransfer.2017.03.104.
[13] Y. Egorov, M. Boucker, A. Martin, S. Pigny, M. Scheuerer, and S. Willemsen, Validation of CFD codes with
PTS-relevant test cases, 2004.
[14] P. Coste, J. Pouvreau, J. Laviéville, and M. Boucker, “A two-phase CFD approach to the PTS problem evaluated on
COSI experiment,” Proceedings of the 16th International Conference on Nuclear Engineering, Florida, USA, 2008.
NUMERICAL HEAT TRANSFER, PART A 33

[15] G. Patel, V. Tanskanen, and R. K. Rajamäki, “Numerical modelling of low-Reynolds number direct contact
condensation in a suppression pool test facility,” Ann. Nucl. Energy, vol. 71, pp. 376–387, 2014. DOI: 10.1016/
j.anucene.2014.04.009.
[16] D. Lakehal, and M. Labois, “A new modelling strategy for phase-change heat transfer in turbulent interfacial
two-phase flow,” Int. J. Multiphas. Flow, vol. 37, pp. 627–639, 2011. DOI: 10.1016/j.ijmultiphaseflow.2011.03.004.
[17] R. Szijártó, Condensation of steam in horizontal pipes-model development and validation, Ph.D thesis,
ETH-Zürich, 2015.
[18] R. Szijártó, A. Badillo, B. Ničeno, and H.-M. Prasser, “Condensation models for the water-steam interface and
the volume of fluid method,” Int. J. Multiphas. Flow, vol. 93, pp. 63–70, 2017. DOI: 10.1016/j.
ijmultiphaseflow.2017.04.002.
[19] S. Q. Li, P. Wang, and T. Lu, “CFD based approach for modeling steam-water direct contact condensation in
subcooled water flow in a tee junction,” Prog. Nucl. Energy, vol. 85, pp. 729–746, 2015. DOI: 10.1016/j.
pnucene.2015.09.007.
[20] G. Patel, V. Tanskanen, E. D. Hujala, and J. Hyvärinen, “Direct contact condensation modeling in pressure
suppression pool system,” Nucl. Eng. Des., vol. 321, pp. 328–342, 2017. DOI: 10.1016/j.nucengdes.2016.08.026.
[21] M. Pellegrini, M. Naitoh, C. Josey, and E. Baglietto, “Modeling of Rayleigh-Taylor instability for steam direct
contact condensation,” The 16th International Topical Meeting on Nuclear Reactor Thermal Hydraulics
(NURETH-16), Chicago, USA, 2015.
[22] G. Černe, S. Petelin, and I. Tiselj, “Simulation of the instability in the stratified two-fluid system,” International
Conference Nuclear Energy in Central Europe, Portorož, Slovenia, pp. 201–208, 1999.
[23] G. Yadigaroglu, “Computational fluid dynamics for nuclear applications: from CFD to multi-scale CMFD,” Nucl.
Eng. Des., vol. 235, pp. 153–164, 2005. DOI: 10.1016/j.nucengdes.2004.08.044.
[24] S. W. J. Welch, and J. Wilson, “A volume of fluid based method for fluid flows with phase change,” J. Comput.
Phys., vol. 160, pp. 662–682, 2000. DOI: 10.1006/jcph.2000.6481.
[25] D. L. Sun, J. L. Xu, and Q. Chen, “Modeling of the evaporation and condensation phase-change problems with
fluent,” Numer. Heat Transfer, B, vol. 66, pp. 326–342, 2014. DOI: 10.1080/10407790.2014.915681.
[26] D. L. Sun, J. L. Xu, and L. Wang, “Development of a vapor–liquid phase change model for volume-of-fluid
method in FLUENT,” Int. Commun. Heat Mass, vol. 39, pp. 1101–1106, 2012. DOI: 10.1016/j.
icheatmasstransfer.2012.07.020.
[27] P. Datta, A. Chakravarty, K. Ghosh, A. Mukhopadhyay, and S. Sen, “Modelling aspects of vapour bubble
condensation in subcooled liquid using the VOF approach,” Numer. Heat Transfer, A, vol. 72(3), pp. 236–254,
2017. DOI: 10.1080/10407782.2017.1372673.
[28] H. Ganapathy, A. Shooshtari, K. Choo, S. Dessiatoun, M. Alshehhi, and M. Ohadi, “Volume of fluid-based
numerical modeling of condensation heat transfer and fluid flow characteristics in microchannels,” Int. J. Heat
Mass Tranfer, vol. 65, pp. 62–72, 2013. DOI: 10.1016/j.ijheatmasstransfer.2013.05.044.
[29] J. U. Brackbill, D. B. Kothe, and C. Zemach, “A continum method for modeling surface tension,” J. Comput. Phys.,
vol. 100, pp. 335–354, 1992. DOI: 10.1016/0021-9991(92)90240-y.
[30] ANSYS FLUENT User’s guide 14.5, 2012.
[31] ANSYS FLUENT UDF manual 14.5, 2012.
[32] S. V. Patankar, Numerical heat transfer and fluid flow. New York, NY: Hemisphere, 1980.
[33] G. Son, and V. K. Dhir, “Numerical simulation of film boiling near critical pressures with a level set method,”
J. Heat Transfer-Trans. ASME, vol. 120, pp. 183–192, 1998. DOI: 10.1115/1.2830042.
[34] V. Alexiades, and A. D. Solomon, Mathematical modeling of melting and freezing processes. Washington, DC:
Hemisphere, 1993.

You might also like