ICONE
Conference Paper
American Society of Mechanical Engineers (ASME) owns all copyright and publication
subject to this requirement: “Verbatim reproduction of this paper by anyone will be
permitted by ASME provided appropriate credit is given to the author(s) and ASME.”
1
ICONE10-22206
HYDROGEN MIXING ANALYSES FOR A VVER CONTAINMENT
Pal Kostka, Zsolt Techy
VEIKI Institute for Electric Power Research
H-1016 Gellerthegy u. 17
Budapest, Hungary
Phone: +36 – 1 – 457 8245
Fax: +36 – 1 – 457 8253
e-mail: techy@helka.iif.hu
James J. Sienicki
Argonne National Laboratory
9700 South Cass Avenue
Argonne, Illinois 60439, USA
Phone: +1 – 630 – 252 4848
Fax: +1 – 630 – 252 6080
e-mail:sienicki@anl.gov
ABSTRACT
Hydrogen combustion may represent a threat to containment integrity in a VVER-440/213 plant owing to the combination of high
pressure and high temperature. A study has been carried out using the GASFLOW 2.1 three-dimensional CFD code to evaluate the
hydrogen distribution in the containment during a beyond design basis accident.
The VVER-440/213 containment input model consists of two 3D blocks connected via one-dimensional (1D) ducts. One 3D block
contains the reactor building and the accident localization tower with the suppression pools. Another 3D block models the air traps. 1D
ducts represent the check valves connecting the accident localization tower with the air traps. The VVER pressure suppression system,
called "bubbler condenser," was modeled as a distributed heat sink with water thermodynamic properties. This model accounts for the
energy balance. However, it is not currently possible to model dynamic phenomena associated with the water pools (e.g., vent clearing,
level change).
The GASFLOW 2.1 calculation gave detailed results for the spatial distribution of thermal-hydraulic parameters and gas
concentrations. The range and trend of the parameters are reasonable and valuable. There are particularly interesting circulation patterns
around the steam generators, in the bubbler tower and other primary system compartments. In case of the bubbler tower, concentration and
temperature contour plots show an inhomogeneous distribution along the height and width, changing during the accident. Hydrogen
concentrations also vary within primary system compartments displaying lower as well as higher (up to 13 - 20% and higher) values in
some nodes. Prediction of such concentration distributions was not previously possible with lumped parameter codes.
GASFLOW 2.1 calculations were compared with CONTAIN 1.2 (lumped parameter code) results. Apart from the qualitatively similar
trends, there are, for the time being, quantitative differences between the results concerning, for example, pressure histories, or the total
amount of steam available in the containment. The results confirm the importance of detailed modeling of the containment, as well as of the
bubbler condenser and sump water pools.
The study showed that modeling of hydrogen distribution in the VVER-440/213 containment was possible using the GASFLOW 2.1
code with reasonable results and remarkable physical insights.
KEYWORDS: containment, VVER, accident, hydrogen, CFD, simulation
2
INTRODUCTION
The AGNES program for "Reassessment of the Safety of the Paks NPP ” was carried out in Hungary during 1991-1994. Among the
broad range of issues addressed, AGNES focused on issues pertaining to the robustness of the containment, including containment loads
and containment performance. The AGNES program addressed the issue of the magnitude of containment loads arising from both design-
basis accidents (DBA) and beyond-design-basis-accidents (BDBA). It was identified that hydrogen combustion may be a dominant threat
to containment integrity in a VVER-440/213 owing to the combination of high pressure and high temperature. To better assess the extent of
the threat to containment integrity, a need was identified to perform analyses of hydrogen mixing and combustion for Paks using state-of-
the-art codes. A multi-dimensional mixing code is needed to assess hydrogen distributions under various accident conditions.
Analyses have already been performed at VEIKI in Hungary addressing steam/water discharge rates and hydrogen
production/discharge rates for a number of accident sequences with the MAAP code and the MELCOR code. The results of these analyses
have been used as input to the GASFLOW 2.1 mixing code to evaluate combustible gas distributions and flammability conditions in the
VVER-440/213 containment.
1 THE PAKS VVER-440/213 CONTAINMENT
The functional requirements for VVER-440/213 containment to restrict the release of radionuclides are based on internationally
accepted principles. However, the architecture including the rectangular building with the localization tower is substantially different from
Western-type containments.
1.1 Architecture
The primary circuit and its components are placed in the hermetically sealed main building. The steam generator room is a
rectangular, reinforced concrete structure. The outer walls are 1.5 m thick; the inner walls are generally 0.8 m thick. Diagonal reinforced
concrete walls are located in the corners. These walls reduce the ceiling span and fulfill the function of missile shielding. Inside the upper
portion of the steam generator room there is an annular shaped pump room, located around the reactor shaft. The electric motors of the six
primary coolant pumps and the electric drives of the primary system isolation valves are situated in this room. A plan view of the
containment at 10.5 m level is shown in Fig. 1.
The reactor vessel is installed in the cylindrical reactor cavity shaft. The shaft wall is 2.5 m thick with multiple layers of steel
reinforcement. The steam generator room and the connected neighboring compartments are connected to the localization system by a
horizontal channel.
1.2 Accident Localization System
The accident localization system consists of the bubbler condenser and the air traps. The function of this system is to decrease the
maximum peak pressure and to ensure a near atmospheric containment pressure after a pipe break. The localization tower contains 1500 m
3
of water, which is distributed among 12 levels of bubbler condenser trays. The air volume of the condensers is connected to four air traps.
Between the air traps and the air volume of the trays there are check valves. The check valves are closed if the pressure in the air traps is
higher than that in the trays. That means that the air will be contained in the air traps with a total volume of about 17 000 m
3
. The total
volume of the containment building is about 49000 m
3
.
2 MODELING OF THE VVER-440/213 CONTAINMENT WITH GASFLOW 2.1
2.1 The GASFLOW 2.1 Code
GASFLOW 2.1 is a best estimate, special purpose computer code developed at Los Alamos National Laboratory (LANL) and
Forschungszentrum Karlsruhe (FzK) to predict the transport, mixing and combustion of hydrogen and other gases, liquid droplets and
aerosols in nuclear reactor containments and other non-nuclear buildings.
Two coordinate systems are available in GASFLOW 2.1: rectangular (Cartesian) and cylindrical. For the present model, the
rectangular coordinate system was used, because the VVER-440 containment basically consists of rectangular compartments. The
computational domain or 3D block is discretized using a rectangular finite difference mesh consisting of computational cells. Selection of
this option means that cylindrical components like the reactor, the steam generators or the hydroaccumulators should also be modeled in a
rectangular mesh. It is possible to define several 3D blocks, which are connected via one-dimensional ducts.
3
2.2 Model of the VVER-440/213 Containment
The developed model of the VVER-440/213 containment consists of two 3D blocks connected with 1D ducts.
- Block 1
Block 1 contains the steam generator room (SG box), the primary system rooms connected to the SG box (pressurizer room,
hydroaccumulator rooms, cable channels, etc.), the pump deck, the connecting corridor, the bubbler tower shaft and the bubbler condenser
trays. The block consists of 27x29x22=17 226 real cells. Cells, which are not part of the hermetic zone, are blocked out by obstacles
(mobs). The total net (free) volume of the block is 31 175 m
3
.
- Block 2
Block 2 contains the four air traps. The block consists of 23x1x22=506 real cells. The total net (free) volume is 17 552 m
3
.
- 1D ducts
The air traps and the twelve bubbler condenser trays are connected via check valves. Twelve 1D ducts, representing the check valves,
connect Block 1 and Block 2. The diameter of one duct is 50 cm, its length is 1 m. Ducts open at 500 Pa pressure difference and they are
closed in the reverse direction. The elevations of the ducts correspond to the level of trays.
The check valve model is essential for the correct description of the operation of the bubbler condenser system.
- Computational Mesh Grid
The applied mesh grid is shown in different planes in Figs. 2 and 3. Fig. 1 shows the plan view of the containment at the 10.5 m level.
The mesh view (Fig. 2) can be compared to the plan view of the containment at the same level (Fig. 1). The outside walls, shown in the
mesh, form the containment boundary. The vertical view of the mesh grid is shown in Fig. 3. Inside walls of the containment are also
modeled in the mesh.
Primary system components, like the reactor pressure vessel, steam, hydroaccumulators, pressurizer and main circulating pumps are
modeled by obstacles (mobs). Obstacles consist of an arbitrary number of cells through which no fluid flow is allowed. No heat conduction
is allowed through the obstacles: these structures are assumed to be insulated.
The primary coolant system components are represented as rectangular objects. Diagonal steam generators are set up of shifted
rectangular mesh cells. A finer mesh grid would certainly lead to a better approximation of curvilinear components. However, computation
time, which largely depends on the number of applied mesh cells, sets another limit to the mesh grid definition. This input involves only the
main primary system components. Smaller components including piping are not currently modeled.
- Heat Structures
Heat conducting structures in the containment are modeled according to GASFLOW 2.1 modeling options. Walls are modeled as thin
surfaces dividing two adjacent layers of fluid that forbid flow across them. Their temperature profile is determined by the adjacent fluid cell
temperatures on both sides, and by their heat capacity and conductivity. Floors and ceilings are modeled as concrete slabs. Slabs are
considered so thick that within the problem time scale, the temperature gradient never penetrates deep enough to affect the temperature
profile near their back side. Construction steel in the containment is modeled as distributed sinks - heat structures, which are assumed to be
distributed within the fluid cell.
- Bubbler Condenser Model
There is no suppression pool model currently available in GASFLOW 2.1. At the same time, the pressure suppression pool containing
1500 m3 of water is a major heat sink influencing the thermal hydraulic behavior of the containment. Modeling of this system is important
for a reasonable description of containment phenomena.
The bubbler condenser pools were modeled in this study as distributed heat sinks. Suppression pool heat transfer is very effective,
therefore the sink thickness is defined as 1mm. Material properties of this sink are the same as those of water, and the sink mass is equal to
the mass of the bubbler condenser water pool. This model can describe the energy transfer from the steam-water mixture to the water.
However, the sink model has its own limitations. Sinks are modeled as a separate matter distributed in the whole volume of the
bubbler condenser chamber. Therefore, dynamic phenomena associated with the pools (e.g., vent clearing, level change) cannot currently
be modeled.
Geometrical modeling of the bubbler condenser is shown in Fig. 3. The 12 bubbler condenser chambers located in the bubbler tower
shaft are modeled as rectangular boxes with distributed sinks inside. Flow area from the shaft to the bubbler condenser is defined with the
fractional area option of the code. On the other side, connection to the air traps through check valves is modeled with the 1D duct option of
the code.
4
- Initial and Boundary Conditions
Initially the containment pressure is 0.993 bar, the temperature is 323 K, relative humidity is 60%. Wall, slab and sink temperatures
are equal to the atmosphere temperature. At boundaries of the computational domains, a rigid free-slip boundary condition is defined. This
means that the entire computational volume is enclosed within rigid, impenetrable walls at which there is free slip or the gradient of the
tangential velocity component is zero. This is a good approximation, because the computational boundaries are solid surfaces and the mesh
resolution is not fine enough to represent the near-wall velocity gradients.
Hydrogen burns were suppressed to analyze the hydrogen mixing behaviour during the whole transient.
3 ACCIDENT SEQUENCE AND HYDROGEN SOURCE
The calculated unmitigated sequence is a medium break LOCA with failure of safety injection, and containment heat removal. As an
initiating event a 100 mm break in the hot leg was considered. It was assumed that the engineering safety systems (low and high pressure
emergency core cooling, and containment spray systems) were not available.
In-vessel calculations were performed by the MAAP4/VVER code. This code provided the blowdown source of water, steam and
hydrogen to the containment, which was used as input to the GASFLOW 2.1 code as function of time. The total amount of hydrogen
generated during the in-vessel period is 304 kg. The bulk of hydrogen is generated before support plate failure (274 kg). The maximum
hydrogen generation rate is 1.11 kg/s, but the max. release rate from the primary system is about 0.4 kg/s.
The blowdown source is shown in Fig. 4. Until 236 s the source contains only water. Between 236 s and 753 s a mixture of steam and
water is discharged. From this time on, only steam and H
2
is discharged. The H
2
source starts at 1240 s and it practically finishes at 3470 s.
During this time 265 kg of H
2
is released. An additional 8 kg of H
2
is released between 3470 s and 4775 s, the last entry in the source table.
The location of the blowdown source is situated in the left part of the SG box in the centerline in y-direction, at elevation of 10 m,
facing to the reactor (to east, cf. Fig. 6). The isenthalpic expansion option of GASFLOW 2.1 was used to calculate the expansion process
from the primary system pressure to the containment pressure.
4 RESULTS OF GASFLOW 2.1 CALCULATION
The calculation was performed on a SUN SPARC 20 workstation at ANL. The problem time of the calculation was 3600 s. Total
computational time took almost six weeks.
4.1 Thermal Hydraulic Transient Results
After the start of the accident, the pressure rises rapidly in the containment. The pressure distribution is quite homogenous in the
whole containment, except the air traps, where the pressure buildup is slower, and it reaches its maximum value of 2 bar at 200 s (Fig 5.).
After 200 s the mass flow rate of the source drops significantly, therefore the pressure also decreases. Between 550 and 750 s a mixture of
steam and water is released with a mass flow rate one order of magnitude less than before, and the pressure drops again. After 750 s, only
steam is released with continuously decreasing mass flow rate. The heat removal from the containment atmosphere due to condensation on
heat structures is higher than the energy input, which results in pressure decrease. The average pressure in the containment at 3600 s is
about 1.3 bar. The pressure in the localization tower drops below that of the air traps at about 1100 s and the check valves close. This
occurs before the hydrogen release starts.
The average temperature of the containment atmosphere rises rapidly and reaches its maximum value of about 372 K at 500 s then it
decreases to 360 K at about 3600 s.
The spatial distribution of the temperature is influenced by the flow patterns within the containment, and therefore it is less
homogeneous than the pressure distribution. In the first seconds, when the primary coolant is discharged at high pressure, there is a strong
flow through the corridor to the bubbler tower. The velocity of the flow is about 2 to 4 m/s. Temperature contour plots at 10 m level at 200
s are shown in Fig. 6. The plots show how the temperature decreases with the distance form the break location. There are high
temperatures, about 170 to 180 C near the break. The temperature values are between 110 and 130 C in the bulk of the containment.
Temperatures in the upper part of the bubbler tower do not exceed 100 C. Vertical temperature distributions show that the hottest part is
located at the entrance to the localization shaft. The hot gases flow upward in the localization tower and then return cooled down, as
reflected very well by the flow pattern. The temperature decreases with the elevation in the hydroaccumulator and pressurizer rooms.
After 500 s, when the discharge flow decreases drastically, and the source contains mainly steam, the flow pattern changes in the
corridor. Until 500 s the gases move towards the localization tower trough the whole cross section of the corridor. Starting from 500 s the
hot gases move towards to the tower at the upper part of the corridor and cooler gases return to the steam generator room in the lower
region. This difference between the two sides of the corridor disappears after 500 s.
