Nuclear Engineering and Design 325 (2017) 57–67

Contents lists available at ScienceDirect

Nuclear Engineering and Design

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

Study of strategies to avoid hydrogen deflagration in venting pipelines MARK

during severe accident scenarios
⁎
Nancy Astrid Solis-Alcantaraa, Armando Miguel Gómez-Torresb, , Javier Ortiz-Villafuerteb,
Carlos Filio-Lópeza, Eduardo Sáinz-Mejíab, José Vicente Xolocostli-Munguíab
a
Escuela Superior de Física y Matemáticas/Instituto Politécnico Nacional, Ciudad de México 07738, Mexico
b
Departamento de Sistemas Nucleares/Instituto Nacional de Investigaciones Nucleares, Ocoyoacac, Estado de México 52750, Mexico

A R T I C L E I N F O A B S T R A C T

Keywords: Venting timing and duration are key issues for the development and assessment of severe accident guidelines and
Mark II mitigation alternatives. In BWRs, venting from wetwell has the advantage of gaining fission product scrubbing.
BWR In this study, two strategies are investigated to avoid hydrogen deflagration in venting pipelines. The starting
GASFLOW point of the vent pipe is a penetration on the wall of wetwell’s suppression chamber of a BWR Mark II con-
Venting strategy
tainment. A three-dimensional pipeline model was developed for the CFD type code GASFLOW, to better de-
Nitrogen inertization
Volume enlargement
termine conditions leading to risk of flame acceleration and hydrogen deflagration. The analysis starts with a
base case, in which venting occurs when pressure reaches 4.5 kgf/cm2 and the vent pipe is full of air. Then, the
first strategy to reduce hydrogen deflagration risk consists of inertization with nitrogen at specific locations
along the vent pipe through rupture disks with three opening pressure setpoints (2.0, 3.0. 4.0 kgf/cm2). Three
different locations are considered in this study. The second strategy is the volume enlargement of the last section
of the vent pipeline. Two different expansions additional to the base case were considered for analysis.
The results show that the inertization with nitrogen at the lower pressure setpoints (2.0 and 3.0 kgf/cm2) does
effectively, for practical applications of safety analysis, highly reduces the risk of flame acceleration anywhere in
the vent pipeline. However, lowering the opening pressure value implies earlier venting. If it is preferable to
keep the disk opening pressure at the higher pressures (4.0 and 4.5 kgf/cm2), the results show that it is necessary
to choose an appropriate location to set the rupture disk, to effectively diminish flame acceleration risk.
Regarding the second venting strategy, the results show that increasing the volume of the last section of the vent
pipe is also an effective way to reduce hydrogen deflagration risk. Thus, although flame acceleration still could
occur, those conditions for that to happen will be restricted to a shorter period. For actual practical applications,
this second strategy seems more plausible to be carried out, because all relevant changes to the vent pipeline
would be focused on the parts already outside reactor building.

1. Introduction additional relevant decision during the evolution of a severe accident

Methodologies for risk assessment studies of severe accidents have
been applied for decades now (US NRC, 1990). Containment analysis is
a fundamental part of the probabilistic safety assessment level 2 for
power reactors, and one key objective of such studies is determining
conditions for venting during severe accident scenarios. After the Fu-
kushima accident, a renewed interest in this topic arose quite re-
markably. Furthermore, venting timing and duration are key issues for
the development and assessment of severe accident guidelines (SAMGs)
and mitigation alternatives (SAMAs). Particularly for BWRs, the option
of the venting action from the wetwell or drywell becomes an
(EPRI, 2012). For those scenarios where steam is being directed to the
pressure suppression pool, via the safety/relief valves and pipelines, the
former alternative has the advantage of gaining fission product scrub-
bing.
Additionally, performance of venting actions directly impacts on
hydrogen control in containment. In this aspect, it is also necessary to
create accurate models of the venting pipelines, since it has been shown
that conditions for flame acceleration inside the vent pipe could be
reached due to hydrogen concentration (Gómez-Torres et al., 2015). In
that study, a model of a vent pipe starting in the drywell of a BWR Mark
II containment was created for the CFD-type code GASFLOW (KIT,

⁎
Corresponding author.
E-mail addresses: astrid_soal@hotmail.com (N.A. Solis-Alcantara), armando.gomez@inin.gob.mx (A.M. Gómez-Torres), javier.ortiz@inin.gob.mx (J. Ortiz-Villafuerte),
cfilio@esfm.ipn.mx (C. Filio-López), eduardo.sainz@inin.gob.mx (E. Sáinz-Mejía), vicente.xolocostli@inin.gob.mx (J.V. Xolocostli-Munguía).

http://dx.doi.org/10.1016/j.nucengdes.2017.10.008
Received 30 May 2017; Received in revised form 2 October 2017; Accepted 4 October 2017
0029-5493/ © 2017 Elsevier B.V. All rights reserved.
N.A. Solis-Alcantara et al. Nuclear Engineering and Design 325 (2017) 57–67

2011). The accident scenario considered was a SBO simulated with the been already achieved. In order to define the source term for a CFD
code MELCOR. As one conclusion of that work, the authors suggested to code, as GASFLOW, more accurate data of the wetwell state could be
maintain an inert atmosphere of nitrogen inside the vent pipe to avoid obtained at each time from a safety analysis computer code. This sim-
reaching the hydrogen and oxygen mixture concentration required for plified model is an alternative in case there is no access to such tools
deflagration. (Nancy Astrid Solis Alcantara, 2017). The model starts by setting the
In this work, two strategies to avoid flame acceleration, deflagration following equations for energy and mass conservation:
and/or detonation in venting pipelines are proposed and analyzed. The
U = hsrv Ṁ srv + hH 2 Ṁ H 2 (1)
first one is to inertize sections of the venting pipeline, and the second
one is to modify its geometry at particular sections. The combination of
Ṁ l + Ṁ g = Ṁ srv
these two strategies becomes another alternative. To perform the ana-
lysis, firstly it is proposed a model of a venting pipeline having its Ṁ g = Ṁ H2 (2)
starting point at a penetration on the wall of the suppression chamber
and the condition of a constant total volume
of the wetwell of a BWR Mark II containment. This vent pipe still has a
connection with other vent pipe coming from a penetration to the VTOT = Vg + Vl = Ml vl (p,T ) + Mg vg (psat (T ),T ) (3)
drywell. In this design, thus, the first sections of the pipeline are inside
where U ,h,Ṁ ,V ,v represent the total energy in the system, enthalpy,
the reactor building but not inside the primary containment, and the
mass flow, volume, and specific volume, respectively. The subscripts
last two sections are outside the reactor building. The last section dis-
srv, H2, l, g, TOT, and sat refer to safety/relief valves, hydrogen, liquid,
charges laterally above reactor building. This composed pipeline has
gas, total volume, and saturation conditions, respectively. Eq. (1) can
been modeled with the code GASFLOW. The analysis starts with a base
also be written in terms of an equation of state for water and ideal gas
case, in which venting occurs when pressure reaches 4.5 kgf/cm2 and
equations for the non-condensable gases hydrogen and nitrogen as
the pipeline is full of air and at atmospheric pressure. Then, the option
follows
of using nitrogen as a mean to avoid a direct contact of hydrogen and
oxygen is investigated. Since nitrogen is already in use as inertizing U = Ml ul (p,T ) + Mg ug (psat (T ),T ) + MN 2 CvN 2 T + MH 2 CvH 2 T (4)
agent in BWR containments, it could be used inside some sections of the
In this equation, pressure p and temperature T are independent
vent pipe via the use of rupture disks having different aperture pressure
variables. Also, the total pressure in the gas volume is the sum of the
set points. Three different locations along the pipe vent are considered
partial pressures of the steam, hydrogen and nitrogen, that is
to set up a rupture disk, and three different pressure set points (2.0, 3.0.
4.0 kgf/cm2) for opening the disk are used for the GASFLOW calcula- p = psat (T ) +
MN 2 RN 2 T M R T
+ H2 H2
tions. That is, for the first strategy, nine different venting cases plus the Vg Vg (5)
base case are analyzed and discussed. To study the second strategy, it is
After deriving Eqs. (3)–(5) with respect to time, and using the mass
proposed to enlarge the cross section of the vent pipe at its end section,
conservation (Eq. (2)) plus some algebra, it can be obtained a system of
that is, the part outside the reactor building. Such volume enlargement
three nonlinear differential equations, which can be written in matrix
will decrease the H2 volumetric fraction in that volume, thus effectively
form as
moving away from the flame acceleration region.
In order to calculate the thermodynamic conditions of the con- ̇
⎛ A11 A12 A13 ⎞ ⎛ P ⎞ ⎛ F1 ⎞
tainment and discharge mass flows during an SBO, it was necessary to
⎜⎜ A21 A22 A23 ⎟⎟ ⎜ T ̇ ⎟ = ⎜⎜ F2 ⎟⎟
use a simplified analytical model, as an alternative to obtain those data ⎜ ⎟
⎝ A21 A22 A23 ⎠ ⎝ Ṁ l ⎠ ⎝ F3 ⎠ (6)
from results of system codes. This model is described next.
The coefficients A in that matrix and the right-hand vector F are:
2. Calculation of hydrogen source term for GASFLOW ∂ul
A11 = Ml
∂p
The conjunction of results from system codes as MELCOR, RELAP,
MAAP, etc., with CDF-type codes as GASFLOW is a practical practice for ∂ul ∂ug ∂Psat ∂ug
analysis of containment behavior during severe accident conditions A12 = Ml + Mg + Mg + MN 2 CvN 2 + MH 2 CvH 2
∂T ∂psat ∂T ∂T
(Van Dorsselaere et al., 2015; Paladino et al., 2016). The system codes
provide with boundary and initial conditions for the CFD code. It is not A13 = ul−ug
the purpose of the present analysis to develop and evaluate a coupling
approach, but to solve a practical problem based on data calculated F1 = hsrv−ug (psat , T ) Gsrv + (hH 2−CvH 2 T ) GH 2
with a severe accident code. A simplified model allows to perform a
A21 = 1
sensitivity study to show the different options for finding solutions to
avoid problems in the venting pipe. In this work, a base case of analysis ∂vg ∂psat ∂vg ⎞ ⎤
⎡ dp p−psat p−psat
was carried out from given initial and boundary conditions of the A22 = −⎢ sat + − Mg ⎛⎜ + ⎟

⎣
dT T Vg ∂p
⎝ sat ∂T ∂T ⎠ ⎥⎦
containment thermodynamic conditions during the discharge of steam
through the safety/relief (SRV) pipelines to the pressure suppression p−psat
pool (PSP) from an SBO simulated with MELCOR. Then, when the A23 = − vg
Vg
venting action starts, the evolution of containment conditions and the
steam and hydrogen source term for the GASFLOW calculation are p−psat
F2 = − vg Gsrv
computed from a simplified model of the evolution of the containment Vg
thermodynamic conditions. The steam flow through the SRV pipes
provided by MELCOR is the same one used by Gómez-Torres et al. ∂vl
A31 = Ml
(2015). Then, it is computed the pressure and temperature fields, hy- ∂p
drogen concentration and mass flows inside the vent pipe, from the
wetwell to the discharge point above the reactor building. ∂vl ∂vg dpsat ∂vg ⎞
A32 = Ml + Mg ⎛⎜ + ⎟
The proposed model for calculation of the evolution of the ther- ∂T ∂p
⎝ sat dT ∂T ⎠
modynamic state of the PSP and suppression chamber assumes that
thermodynamic equilibrium between the liquid and gas volumes has A33 = vl−vg

58
N.A. Solis-Alcantara et al.
Fig. 1. Cross section view of the venting pipe.
F3 = −Gsrv vg
This system of equations is solved at each time step using an Euler
explicit integration scheme, by some programming in the language
PHYTON, following a procedure similar to that shown by Gómez-Torres
et al. (2015). The thermodynamic properties of water were taken from
the formulation IFC-97 (IAPWS, 1996).
3. Vent pipe geometric model and test cases
Fig. 1 shows the geometric model used to represent the circular
cross section of the vent pipe and Fig. 2 shows the pipe trajectory from
the suppression chamber to outside the reactor building. As shown in
Fig. 1, 32 radial nodes were used to achieve a sufficiently accurate
model of the pipe circular area. In GASFLOW, it is possible to define
rooms for performing some specific calculations described later and
that are related to deflagration calculations. In this model, the sections
separated by vertices are divided in rooms. Thus, as shown in Fig. 2,
there are nine rooms in this vent pipe model, being room 9 the end
section of the pipeline.
For the venting scenario using nitrogen to avoid initially the direct
contact of hydrogen and oxygen in the vent pipe, the rupture disk is
located at three different locations: start of rooms 3, 4, and 5, which are
still at heights corresponding to the wetwell level (see Fig. 2). Further,
three pressure set points for disk opening were considered: 2.0, 3.0, and
4.0 kgf/cm2. Thus, a total of nine test cases were used in the
59
Nuclear Engineering and Design 325 (2017) 57–67
calculations, and compared against the result of the base case, in which
there is no rupture disk and the venting valve opens when pressure
reaches 4.5 kgf/cm2. Three different pressure setpoints are used in this
analysis, besides the base case, because the reactor operator has the
option of manually opening the venting valves of the drywell and
wetwell, and not waiting until design pressures are reached. This is
particularly relevant to this study because early venting might be pre-
ferable for certain scenarios than having an increasing risk of hydrogen
accumulation at higher pressures.
For the case of enlarging the last section of the vent pipe (room 9),
two additional cross sections areas were used for the venting perfor-
mance analysis. Fig. 3 shows these new geometries. In this figure, in
Expansion 1 the pipe radius has been increased one single node of the
GASFLOW mesh, leading to an increase of 567 cm2, for the whole room
9 length. In Expansion 2, one more radial node is added to those in
Expansion 1, but only after the second half of room 9, leading to a total
area of 1944 cm2 in that section, that is, the cross-section area is
1215 cm2 in the first half of the pipe but 1944 cm2 in the second half.
Table 1 shows a resume of all the different cases under study.
4. Results
Given that venting pipes discharge to the surrounding environment
of the reactor building, it is clear that, the pipeline will be at atmo-
spheric conditions. Since the interest of this study is mainly on the
amounts of nitrogen and oxygen available, it has been considered that
the air in the pipe line consists only of 80% nitrogen and 20% oxygen.
Also, due to computational power limitations, all studied cases are
adiabatic, however, some discussion is presented later regarding the
effect of heat transfer in the calculation.
4.1. Base case
The analysis starts with a base case, in which venting occurs when
pressure reaches 4.5 kgf/cm2. In this case, there is no rupture disk be-
cause there is no inertized region inside the pipe. As soon as the venting
pressure is reached, the venting valve opens and the atmosphere of
wetwell start flowing inside the vent pipe (start of room 1). For this
case, after approximately one second, all the O2 has already left the
venting path, as shown by Gómez-Torres et al. (2016). Thus, this ana-
lysis is focused on the first second of venting, because after that time
Fig. 2. Isometric view of the venting pipe (left), detailed
room definition and location of rupture disks (right).
N.A. Solis-Alcantara et al. Nuclear Engineering and Design 325 (2017) 57–67

Fig. 3. Cross sectional area enlargement of the room 9.

(a) Base Case (b) Expansion 1 (c) Expansion 2

Table 1
Description of cases under analysis.

Case Description

Base Venting setpoint: 4.5 kgf/cm2. No N2 for inertizing. No expansions

1 Rupture disk at start of room 3. Rooms 1 and 2 inertized with N2. Venting setpoint: 2.0 kgf/cm2
2 Rupture disk at start of room 3. Rooms 1 and 2 inertized with N2. Venting setpoint: 3.0 kgf/cm2
3 Rupture disk at start of room 3. Rooms 1 and 2 inertized with N2. Venting setpoint: 4.0 kgf/cm2
4 Rupture disk at start of room 4. Rooms 1, 2, and 3 inertized with N2. Venting setpoint: 2.0 kgf/cm2
5 Rupture disk at start of room 4. Rooms 1, 2, and 3 inertized with N2. Venting setpoint: 3.0 kgf/cm2
6 Rupture disk at start of room 4. Rooms 1, 2, and 3 inertized with N2. Venting setpoint: 4.0 kgf/cm2
7 Rupture disk at start of room 5. Rooms 1, 2, 3, and 4 inertized with N2. Venting setpoint: 2.0 kgf/cm2
8 Rupture disk at start of room 5. Rooms 1, 2, 3, and 4 inertized with N2. Venting setpoint: 3.0 kgf/cm2
9 Rupture disk at start of room 5. Rooms 1, 2, 3, and 4 inertized with N2. Venting setpoint: 4.0 kgf/cm2
Expansion – 1 Pipe cross section area increased on 567 cm2, for the whole room 9 length
Expansion – 2 Pipe cross section area increased on 729 cm2 to the additional first expansion, on the second half of room 9

there is no more safety concern. The sigma criterion is a parameter to
judge the possibility of a laminar flame becoming turbulent and ac-
celerating in a specific volume (room). In GASFLOW, a direct evalua-
tion of the sigma criterion is available, which then is divided by a cri-
tical sigma value coming from experiments (KIT, 2011). This
dimensionless parameter is referred as sigma index.
σindex = (h2,h2 o,o2,T )/ σcritical (h2,h2 o,T )
The sigma index provides data of hydrogen concentration in a
cloud, where a cloud is defined as all computational volumes that are
combustible in a specified room. A sigma index value greater than 1
implies that the risk of flame acceleration cannot be excluded and must
be considered in safety assessments. Fig. 4 shows the sigma index value
during the first second of the venting action in all nine rooms. Note that
all rooms did reach sigma index values greater than the limit one.
However, it is in room 9 where the sigma index remains closer to the
limit for more than 500 ms.
To refine more the calculation, seventeen axial volumes were
defined inside the room 9 and, additionally to the sigma index value,
the 7λ criterion or DDT characteristics were evaluated in each axial
volume of room 9. In GASFLOW, a function (originally developed at the
Kurchatov Institute) relates the detonation cell size to the concentra-
tions of dry hydrogen and steam in air at a given temperature, T, that is:
λ = λ ([h2],[h2 o],T ) (8)
(7)
Fig. 5 (upper) shows the sigma index during the first second of the
venting action in all axial volumes inside room 9, and Fig. 5 (lower)
presents the D/7λ ratio, which indeed is far from 1.0 but it is not zero as
in the other places.
Since there are several volumes in which both sigma and D/7λ ratio
are not zero, the combustion model of GASFLOW was activated. The
Model of Combustion in GASFLOW models the combustion of hydrogen
in air with the following single (global) reaction:
ω̇
a (H2) + b (O2) + c (N2) + d (H2 O) ⇔ e (H2 O) + f (H2) + g (O2) + c (N2)
(9)

Fig. 4. Sigma index values in all sections of the vent pipe-

line for the base case.

60
N.A. Solis-Alcantara et al.
where the coefficients a, b, c, d, e, f, and g are generally referred to as
the stoichiometric coefficients. The reaction rate ω is determined from a
modified Arrhenius law. Fig. 6 (left) shows the pressure for detectors
located at the middle of each room. On the right of Fig. 6, the energy
delivered to the system due to the combustion of H2 is presented.
From the results presented in Figs. 5 and 6, it is clear that some
options must be explored in order to avoid a direct contact of hydrogen
and oxygen and/or move beyond the flame acceleration region. Those
strategies are described with detail in the next sections.
4.2. Use of rupture disk for nitrogen inertization
When the rupture disk opens at the start of room 3, at each of the
different pressure setpoints, a nitrogen plug initially inside the rooms 1
and 2 flows together with the atmosphere of wetwell entering the vent
pipe. It is then necessary to quantify the effectiveness of this action to
prevent hydrogen concentration reaching values leading to deflagra-
tion. Moreover, some previous studies show that it is at the end section
where flame acceleration could occur (Gómez-Torres et al., 2015).
Therefore, only the results corresponding to room 9 are shown. Fig. 7
shows the sigma index value reached in room 9 for each different
opening pressure setpoint. Note that the only scenario leading to a
sigma index value greater than 1 is the base case, which indicates that
setting a rupture disk at the start of room 3 effectively reduced flame
acceleration risk at the end of the pipeline. Nevertheless, all other cases
did almost reach the limit value, so it was necessary to perform a more
61
Nuclear Engineering and Design 325 (2017) 57–67
Fig. 5. Sigma criterion and D/7λ ratio in all axial volumes
of room 9.
Fig. 6. Pressure detectors at the middle of each room (left)
and combustion energy liberated in the system due to
combustion (right).
detailed analysis, and thus the combustion model of GASFLOW was
activated to better predict the pressure and temperature fields in the
rooms of the vent pipe.
Fig. 8 shows the pressure evolution profile of a detector located at
the middle of room 9, with the combustion model already activated,
once the venting action started. In this figure, it can be noted that the
base case leads to the earliest and highest pressure peak at the end of
the vent pipe. This figure further corroborates that the venting action
by itself still possesses a flame acceleration risk in the vent pipeline. The
peak pressure magnitude reached from the other cases is similar among
them. Note, also that the pressure peaks all occur within a quite short
time range, of about 0.04–0.05 s, from the beginning of the venting
action, having all of them a width of about 25 ms.
By analyzing the temperature evolution profiles corresponding to
the cases shown in Fig. 8, it was confirmed that there exists a high risk
of flame acceleration and possible deflagration at that part of the vent
pipe if no nitrogen is used for inertization, since the temperature peak
in room 9 reached 1400 K. However, even when rupture disk opens at
4.0 kgf/cm2, temperature reached almost 1000 K and a pressure of al-
most 15 kgf/cm2, so flame acceleration and possible deflagration is also
probable in this case too. The cases for rupture disk pressures of 2.0 and
3.0 kgf/cm2 showed a similar temperature rise, but the peak values
reached only up to 750 K and 780 K, respectively. In all cases, the
temperature rise lasts about 0.1 s, and they coincide with the timing of
the pressure sharp rise.
Proceeding similarly to the presentation of results for room 3, the
N.A. Solis-Alcantara et al.
results for the case when the rupture disk is located at the start of room
4 are presented next. Fig. 9 shows the sigma index value reached in
room 9 for each different opening pressure setpoint. The sigma values
for all other rooms were 0.0. In comparison to the previous case, only
for the base case the possibility of flame acceleration clearly exist in
which the sigma index is consistently close to one. In the other cases,
due to some oscillations in the sigma index it was decided to perform a
more detailed comparison analysis and the combustion model was also
62
Nuclear Engineering and Design 325 (2017) 57–67
Fig. 7. Sigma values corresponding to the different opening
pressure setpoints for the case of rupture disk located at the
start of room 3.
Fig. 8. Pressure evolution during venting action for the case
of rupture disk located at the start of room 3, with the
GASFLOW combustion model activated.
Fig. 9. Sigma index values corresponding to the different
opening pressure setpoints for the case of rupture disk lo-
cated at the start of room 4.
activated.
The resulting pressure evolution in all rooms is shown in Fig. 10,
which confirms that only for the base case there exists the risk of flame
acceleration in room 9. Temperature profiles did not show any sharp
increase for the three cases using nitrogen. For the case of the rupture
disk at the start of room 5, in none of the rooms existed risk of flame
acceleration, as shown in Fig. 11. The cases of setpoints at 2.0 and
3.0 kgf/cm2 cannot be seen on this figure simply because the resulting
N.A. Solis-Alcantara et al.
sigma value is zero during the whole 1.0 s simulation. Note that for the
case of 4.0 kgf/cm2, the conditions of flame acceleration appear and
disappear practically instantaneously.
4.3. Room 9 vol increase
Figs. 12 and 13 show the sigma values obtained in room 1 and room
9, respectively, when the volume of room 9 is increased, as described
63
Nuclear Engineering and Design 325 (2017) 57–67
Fig. 10. Pressure evolution during venting action for the
case of rupture disk located at the start of room 4, with the
GASFLOW combustion model activated.
Fig. 11. Sigma values corresponding to the different
opening pressure setpoints for the case of rupture disk lo-
cated at the start of room 5.
Fig. 12. Sigma index values in room 1 for the three different
values of volume of room 9: All three cases yielded the same
results.
above. Note that in room 1 the possibility of flame acceleration is al-
most instantaneous with the venting action, while in room 9 it would
occur a few tenths of second later. Further, when there is no volume
increase, in room 9 the duration of risk of flame acceleration and
possible deflagration is almost two orders of magnitude higher than
that in room 1. Additionally, if room 9’s volume increases, the duration
of flame acceleration risk is reduced in one order of magnitude. That is,
flame acceleration still could occur, but the conditions for that to
N.A. Solis-Alcantara et al. Nuclear Engineering and Design 325 (2017) 57–67

Fig. 13. Sigma index values in room 9 for the three different
values of volume of room 9.

happen will be restricted to a shorter period. All other cases fall in borders in the Shapiro diagram depend on pressure and temperature,
between the results for rooms 1 and 9. and Fig. 20 corresponds to the case of room temperature and 1 bar of
Given the data in Figs. 12 and 13, the GASFLOW combustion model pressure, which are close to the initial conditions of the atmosphere in
was activated. Figs. 14 and 15 show the pressure evolution profile in the vent pipe. Once the venting action starts, the pressure and tem-
rooms 1 and 9, respectively, once the venting action started. Note that perature in the pipe will be mostly those of the wetwell during the
increasing the volume does lead to a delay in the starting of the pressure accident scenario under study, so the cases shown in Fig. 20 are only
peak in the rooms. It should be also noted that the pressure peaks in indicative of how the venting strategies analyzed actually change the
room 1 are neither due to flame acceleration nor a deflagration, but conditions towards a safer zone.
they are due to the reflected pressure wave resulting from the H2
combustion and possible deflagration in room 9 (they came several
5. Discussion
milliseconds after the peaks in room 9). In room 9 the magnitude of the
pressure peak is almost half of that of the base case with the original
The present paper has shown two options to effectively minimize
volume of that room. Thus, the risk of deflagration has been effectively
the risk of hydrogen flame acceleration and possible deflagration.
reduced. This is further confirmed in Figs. 16 and 17, where the tem-
Naturally, there exists other different alternatives. For example, it could
perature profile evolution in all rooms are presented for a detector lo-
be argued that a rupture disk at the end of an inertized vent pipe would
cated in the middle of each room.
simplify a flame acceleration or deflagration analysis, mainly because
Figs. 18 and 19 show the volumetric fractions of the gas mixture in
the nitrogen plug will push effectively air surrounding the pipe end,
room 9 during the venting action. It can be noted that by increasing the
thus avoiding mixing of hydrogen and oxygen. In real life, however, this
volume in room 9, the concentration of hydrogen has clearly decreased
option would require monitoring the whole vent pipeline, to ensure it is
after the first 0.5 s, which further confirms that the risk of flame ac-
properly inertized, and during an accident scenario some reactor
celeration and deflagration has been also lowered.
building zones could be inaccessible. Additionally, in BWRs the hottest
Throughout the previous analysis, it has been shown that the stra-
areas for radiation exposure are in the drywell, so from the radiological
tegies proposed are practical means to face the hydrogen deflagration
protection perspective it is recommendable to work in wetwell zone. It
issue. A summary of the results of the different cases analyzed can be
should be recalled that in the geometric model proposed venting pi-
also concentrated in a Shapiro diagram, as shown in Fig. 20. In this
peline, rooms 1 through 4 are still below wetwell height but outside
figure, Rupture disk 1 refers to the case of a disk set at start of room 3,
primary containment, providing more shielding to workers. Therefore,
and so on. In the three cases the venting setpoint was 4.0 kgf/cm2. The
it was considered a more favorable option to determine the shortest

Fig. 14. Pressure evolution in room 1 during venting action

for the three different values of volume of room 9.

64
N.A. Solis-Alcantara et al.
length leading to safe hydrogen venting, starting from the wetwell.
However, it was also clear that inertizing only rooms 1 and 2 would not
be safe because the amount of nitrogen could be not enough to avoid
contact of oxygen and hydrogen trapped by vortices at various pipe
corners. Therefore, analyses started by setting a rupture disk at start of
room 3, and from it on. As shown by results, if the rupture disk is set at
start of room 5, flame acceleration and deflagration risks practically
disappeared.
65
Nuclear Engineering and Design 325 (2017) 57–67
Fig. 15. Pressure evolution in room 9 during venting action
for the three different values of volume of room 9.
Fig. 16. Temperature evolution in all rooms during venting
action for the first increase of volume of room 9.
Fig. 17. Temperature evolution in all rooms during venting
action for the second increase of volume of room 9.
One limitation of the nitrogen plug option is that in practice it may
be a onetime solution, because once the pipeline is open, steam con-
densation will pull into more air. However, once the venting action
stops, it exists the alternative of trying purging and injection of nitrogen
again, if the accident conditions allow. One other issue not taken into
account in this study is the large amount of hydrogen generated during
the molten core concrete interaction (MCCI), because it only reaches to
the time when primary containment would automatically start venting.
N.A. Solis-Alcantara et al.
Hydrogen generation during MCCI phase has been recognized as a key
issue in deflagration and detonation analysis (NEA/CSNI, 2017). For
hydrogen thus generated, the combination of nitrogen plug and volume
enlargement may become a good venting strategy.
Fig. 20. Location of some of the venting strategies analyzed in a Shapiro diagram.
66
Nuclear Engineering and Design 325 (2017) 57–67
Fig. 18. Evolution of volumetric concentration of the gas
mixture in room 9 for the first increase of volume of room 9.
Fig. 19. Evolution of volumetric concentration of the gas
mixture in room 9 for the second increase of volume of
room 9.
In relation to the uncertainty of the calculations, and particularly to
the values of the sigma index criterion, as consequence of the results of
the HYCOM project (Breitung et al., 2005), it was recommended to use
the sigma criterion in combination with CFD results, as in this work, to
lower uncertainty. Moreover, in containment venting calculations, it is
expected to obtain a more conservative value of the sigma criterion.
This is so, because the sigma criterion was originally developed based
on data from test in tubes with constant flow areas, but one of the re-
sults of the HYCOM project was that the critical conditions for fast
combustion may change with flow geometry change. Therefore, al-
though the sigma index values computed by GASFLOW have some
uncertainty related to the geometric model developed in this work, the
tendency of this uncertainty is towards the conservative side, because
there exists an effective decrease of the expansion ratio due to volume
enlargement. On the other side, however, flame acceleration change
falls in the accuracy range of the sigma criterion, and thus this criterion
should not be considered over-conservative.
Other source of uncertainty is the mesh size used in the geometric
model. For this type of applications, giving the length of a pipe line
required for venting from the wetwell in primary containment to dis-
charge outside and above the reactor building, it makes very expensive
to test convergence for several mesh sizes. The geometric model used in
this study was created by following the directions in the code doc-
umentation, with the aim of satisfying the minimum accuracy needed in
hydrogen distribution calculations for real containment dimensions.
N.A. Solis-Alcantara et al.
6. Conclusions
A 3D model of a vent pipeline has been developed for the code
GASFLOW version 3.2, to the end of testing two different strategies for
hard venting during a severe accident in a BWR Mark II containment.
The venting pipeline starts at a penetration on the suppression chamber
wall of the wetwell. The analysis starts with a base case, in which
venting occurs when pressure reaches 4.5 kgf/cm2. Then, for the first
strategy, a rupture disk was located at the start of three different rooms
of the vent pipe filled with a nitrogen plug. This disk is set to open at
three different pressures. Thus, nine cases were analyzed for this
venting strategy and compared against the base case. The results show
that if the rupture disk is located at the start of room 3, it is still highly
probable that a hydrogen flame acceleration and possible deflagration
will occur at the end section of the vent pipe for the rupture disk
opening setpoint at 4.0 kgf/cm2. For the lower setpoints (2.0 and
3.0 kgf/cm2), setting the rupture disk in that section is still an option,
because temperature rise is lower than that at 4.0 kgf/cm2, although a
slightly lower sharp pressure rise still occurs. However, lowering the
opening pressure value implies earlier venting. For this first strategy,
the results show that setting the disk at the start of rooms 4 or 5 highly
reduces the probability of flame acceleration anywhere in the vent pipe,
being this latter location that practically nullifies the possibilities of
flame acceleration and thus deflagration. The decision on the best lo-
cation for the rupture disk will depend on architectural issues at each
particular containment.
Regarding the second venting strategy, the results show that in-
creasing the volume of the last section of the vent pipe is also an ef-
fective way to reduce hydrogen deflagration risk. Thus, although the
sigma index still reached values greater than 1, those conditions for that
to happen will be restricted to a shorter period. Further, by increasing
the volume in room 9, the volume fraction of hydrogen will decrease in
that same room, lowering the risk of hydrogen deflagration. For actual
practical applications, this second strategy seems more plausible to be
carried out, because all relevant changes to the vent pipeline would be
focused on the parts already outside reactor building.
Acknowledgements
The authors greatly acknowledge the support of SENER-CONACYT
67
Nuclear Engineering and Design 325 (2017) 57–67
to the project FSE-2013-04-213519 through their Energy Sustainability
Fund, and the Comisión Nacional de Seguridad Nuclear y Salvaguardias
for providing the MELCOR data of SRVs discharge flow during a SBO
severe accident.
References
Breitung, W., Dorofeev, S., Kotchourko, A., Redlinger, R., Scholtyssek, W., Bentaib, A.,
L'Heriteau, J.-P., Pailhories, P., Eyink, J., Movahed, M., Petzold, K.-G., Heitsch, M.,
Alekseev, V., Denkevits, A., Kuznetsov, M., Efimenko, A., Okun, M.V., Huld, T.,
Baraldi, D., 2005. Integral large scale experiments on hydrogen combustion for se-
vere accident code validation-HYCOM. Nucl. Eng. Des. 235 (2–4), 253–270.
Electric Power Research Institute, 2012. Investigation of Strategies for Mitigating
Radiological Releases in Severe Accidents. BWR Mark I and Mark II Studies. Final
Report 1026539.
Gómez-Torres, A.M., Sáinz-Mejía, E., Xolocostli-Munguía, J.V., López-Morones, R., Filio-
López, C., Royl, P., 2015. CFD analysis of hydrogen volumetric concentrations in a
hard venting containment system of a Mark II BWR. Ann. Nucl. Energy 85, 552–565.
Armando Gómez-Torres, Vicente Xolocostli-Munguía, Eduardo Sáinz-Mejía, 2016. A
parametric study using CFD code GASFLOW on venting strategy for a hard venting
containment system of a Mark II BWR Reactor. In: NUTHOS-11: The 11th
International Topical Meeting on Nuclear Reactor Thermal Hydraulics, Operation and
Safety Gyeongju, Korea.
Karlsruher Institut für Technologie, 2011. GASFLOW 3.2: A Computational Fluid
Dynamics for Gases Aerosols, and Combustion. Volume 2 User’s Manual. Karlsruhe.
NEA/CNSI, 2017. Safety research Opportunities Post-Fukushima. Initial Report of the
Senior Expert Group. NEA/CNSI/R(2016) 19.
Paladino, D., Andreani, M., Guentany, S., Mignot, G., Kapulla, R., Paranjape, S., Medhat,
S., Kisselev, A., Yudina, T., Filippov, A., Kamnev, M., Khizbullin, A., Tyurikov, O.,
Liang, Z., Abdo, D., Brinster, J., Dabbene, F., Kelm, S., Klauck, M., Götz, L., Gehr, R.,
Malet, J., Bentaib, A., Bleyer, A., Lamaitre, P., Porcheron, E., Benz, S., Jordan, T., Xu,
Z., Boyd, C., Siccama, A., Visser, D., 2016. Outcomes from the EURATOM–ROSATOM
ERCOSAM SAMARA projects on containment thermal-hydraulics for severe accident
management. Nucl. Eng. Des. 308, 103–114.
The International Association for the Properties of Water and Steam, 1996. Release of the
IAPWS Formulation 1995 for the Thermodynamic Properties of Ordinary Water
Substance for General and Scientific Use.
United States Nuclear Regulatory Commission, 1990. Severe Accident Risks: An
Assessment for Five U.S. Nuclear Power Plants. NUREG-1150, vol. 1. Division of
Systems Research, Office of Nuclear Regulatory Research. U.S. Nuclear Regulatory
Commission, Washington, DC.
Van Dorsselaere, J.-P., Auvinen, A., Beraha, D., Chatelard, P., Herranz, L.E., Journeau, C.,
Klein-Hessling, W., Kljenak, I., Miassoedov, A., Paci, S., Zeyen, R., 2015. CFD recent
severe accident research synthesis of the major outcomes from the SARNET network.
Nucl. Eng. Des. 291, 19–34.
Nancy Astrid Solis Alcantara, 2017. Analisis de condiciones de venteo de un BWR con
contención tipo MARK II, Master in Science Thesis, Instituto Politécnico Nacional,
Escuela Superior de Física y Matemáticas, México.