i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3

Available online at www.sciencedirect.com

ScienceDirect

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

Affect of recombiner location on its performance in

closed containment under dry and steam
conditions

Vikram Shukla a,b,*, Sunil Ganju b, Salil Varma a,b, Samiran Sengupta b,
N.K. Maheshwari a,b
a
Homi Bhabha National Institute, Mumbai, 94, India
b
Bhabha Atomic Research Centre, Mumbai, 85, India

highlights

Numerical model is developed & validated for Passive catalytic recombiner device.

Model used to evaluate PCRD performance at different locations & ambient conditions.

It is found that location of PCRD affects its performance inside closed geometry.

The effectiveness of PCRD is maximum when placed near the bottom of closed geometry.

This is in contradiction with assumption of it being best when placed near the top.

article info abstract

Article history: Hydrogen has been the major cause of fire in almost all of the biggest nuclear accidents
Received 30 May 2019 witnessed by world so far. Many approaches have been investigated and developed
Received in revised form worldwide to mitigate the consequences of hydrogen buildup inside the containment of
26 July 2019 Nuclear Power Plants (NPPs) under severe accident scenarios. One such most promising
Accepted 2 August 2019 method is to deploy Passive Catalytic Recombiner Devices (PCRDs). They work on the
Available online 29 August 2019 principle of recombining hydrogen with oxygen from ambient air on catalytic surfaces to
form steam and release of the exothermic heat of reaction. The present work describes the
Keywords: development, validation and application of a CFD based detailed 3D model for hydrogen
Hydrogen recombiners recombination inside PCRD using the governing mass, momentum, energy and species
Computational fluid dynamics conservation equations from first principle. The model has been integrated into the CFD
Heat transfer code FLUIDYN-MP to capture the associated multi-physics phenomena. The integrated tool
has been used to assess the most suitable location within a closed geometry for placing the
PCRD so as to improve its performance and efficiency. Simulations were performed for
different PCRD positions within a closed vessel under dry as well as steam laden condi-
tions. The findings reveal that PCRD location in closed geometry plays important role in its
performance. Moreover lower PCRD position helps in more natural convective mixing
causing better hydrogen transport towards PCRD inlet and hence more hydrogen
consumption.
© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

* Corresponding author. Homi Bhabha National Institute, Mumbai, 94, India.

E-mail address: vikrams@barc.gov.in (V. Shukla).
https://doi.org/10.1016/j.ijhydene.2019.08.015
0360-3199/© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
25958 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3

Device (PCRD), which works on the principle of recombining

Nomenclature hydrogen with oxygen from ambient air on catalytic surfaces
to form steam and release of the exothermic heat of reaction.
Р density
These devices are known to recombine hydrogen at much
U velocity vector
lower concentrations and at near-ambient temperatures, thus
p pressure
reducing the risk of gas phase ignition/deflagration/detona-
t the viscous stress tensor
tion of the hydrogeneair-steam mixtures at low hydrogen
g_ the rate-of-strain (or rate-of-deformation)
concentrations.
tensor
Experimental studies together with well validated analyt-
m effective viscosity
ical models based on empirical, semi-empirical and more
k dilatational viscosity (¼0 according to Stokes)
detailed 2D or 3D numerical approaches have gained signifi-
d unit tensor
cance for understanding the behavior and evaluating the
SU source term for momentum equation
performance of catalytic recombiners. However, for the real
Si source term for species i
containment analysis, such models are usable only when they
i any conserved species (scalar)
are built in conjunction with CFD based hydrogen transport
Rj reaction rate of reaction j in terms of leading
codes, in order to resolve the stratification related issues in the
reactant [kmol/m3s]
typical post Loss Of Coolant Accident (LOCA) steam laden
Mi molecular weight of species i [kg/kmol]
containment. Such computational tools can be built to have
ST source term for temperature
the comprehensive capability to predict reasonably well some
Qrj heat of reaction of jth reaction in terms of
P of safety related important information, viz., a) hydrogen
leading reactant [J/kmol] ¼ n111 DHfi ðn2i  n1i Þ
i transport behavior in and around the recombiners, b) effec-
DHf heat of formation
tiveness of homogenization of the gas mixing, c) catalyst
Aj pre-exponential factor
temperature and d) likely progression of the post-accident
zj temperature exponent
scenario, particularly in the context of hydrogen transport
Ej activation energy
and mitigation behavior which has assumed significant
R universal gas constant
importance in the post Fukushima safety era. These models
Xi mole fraction of species i
for the passive recombiner need consideration of various
nji species rate exponents
multi-physics phenomena, such as buoyant gas transport,
T temperature
conjugate heat transfer between catalytic plates and gas
Cp specific heat at constant pressure
mixture, combined convective and radiation heat transfer
q heat flux vector ¼ kVT
mechanisms, species transport by combined convective and
k thermal conductivity
molecular diffusion and the reaction kinetics.
І total radiation intensity
One of the initial works in this field was reported by [1,2]
r position vector
using the CFD tool GASFLOW which was performed for three
s direction vector
dimensional analysis of steam/hydrogen transport with cat-
s' scattering direction vector
alytic recombiners. They applied it to situations involving
s path length
transporting and distributing combustible gas mixtures to
ka absorption coefficient
study gas dynamic behavior (1) in low-speed, buoyancy-
ks scattering coefficient
driven flows, as well as sonic flows or diffusion dominated
n refractive index
flows; and (2) during chemically reacting flows, including
s Stefan-Boltzmann constant ¼ 5.672e-8 W/m2K4
deflagrations [3]. Subsequently applied the code in mecha-
F phase function
nistic 3-D analyzes of steam/hydrogen distributions under
U0 solid angle
severe accident conditions with mitigation involving a large
tpc relaxation time for phase change
number of catalytic recombiners at various locations in two
types of PWR containments of German design. They presented
the developed 3-D containment models, the applied concept
of recombiner positioning, and the calculated results in rela-
Introduction tion to the applied source term [4]. worked on improvements
in the CFD code CFX-4 for analysis of hydrogen behavior
Hydrogen has very wide flammability limits, very low energy within the NPP containments by implementation of models
required for ignition, low radiant heat, very high heat of for several transport and physical phenomena like: steam
combustion, high thermal conductivity and specific heat, and condensation onto walls in presence of non-condensable
high buoyancy cause this difference and add to the difficulty gases, heat conduction, fog and rain formation, material
of controlling hydrogen fires. Owing to these properties, properties and criteria for assessing the hydrogen combustion
release of hydrogen as a consequence of severe accident in a regime expected. Validation of the GASFLOW code with data
Nuclear Power Plant (NPP) can be a potential threat to the from the tests carried out in PANDA facility covering steam/air
integrity of the containment building. One of the most and steam/helium/air mixtures was carried out by [5,6]. used
promising technologies to mitigate the consequences of the TONUS code for hydrogen risk analysis. Several bench-
hydrogen buildup inside the containment of NPP under severe mark exercises based on experimental results obtained on
accident scenario is to deploy Passive Catalytic Recombiner large-scale facilities, such as MISTRA, TOSQAN and RUT, were
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3
performed by them. These studies were used as verification
and assessment procedures for modeling and numerical ap-
proaches of the TONUS code [7]. presented a detailed step by
step procedure for mechanistic analysis of hydrogen behavior
and mitigation during accidents and highlighted the applica-
tion of CFD codes for hydrogen behavior modeling [8]; with an
intention of demonstrating the utility of CFD codes to model
gas movement inside containment during a severe accident,
presented a detailed study on CFD calculations using GAS-
FLOW, FLUENT and CFX for the Paks NPP.
CFD based recombiner modeling and its validation on the
H2-Par and Kali-H2 experiments was published by [9]; where
they presented numerical assessments of the recombiner
models implemented in CFD solvers NEPTUNE_CFD and
Code_Saturne. Simulation of hydrogen mitigation in
hydrogen recombiners was reported from surface chemistry
as well as CFD modeling perspective in [10,11] where they
compared the solution from one a single step reaction model
and multiple reaction model. A new single step rate expres-
sion was also derived from the detailed mechanism [12].
presented simulation of the efficiency of hydrogen
recombiners as safety devices investigating the applicability
of Passive Autocatalytic Rcombiner (PAR) from NPP contain-
ment to typical H2 application using study based on GARAGE
experiment, performed with ANSYS-CFX and REKO-DIREKT
codes. Nicolas [14] performed the numerical simulation of
the impact of the oxygen starvation (i.e. low oxygen/nitrogen
ratio) on the PAR efficiency and on the PAR gas-phase ignition
limit using a dedicated CFD code named SPARK. Antoni [13]
presented modeling of a PAR for hydrogen mitigation by CFD
methods using four turbulence models (k-u, k-e, intermit-
tency, RSM) for simulations of: gas flow, heat and mass
transport, and chemical surface reactions. Michael [15] re-
ported experimental and numerical study on the effect of
carbon monoxide on recombiner performance. Parallel cata-
lytic reactions of H2 and CO on the catalyst surface were
studied in a series of experiments. The authors also used
three different numerical codes REKO-DIREKT (JÜLICH/
RWTH), SPARK (IRSN), and CFX (ANSYS) for the simulation of
parallel CO/H2 recombination to capture the effects observed
in the experiments. Magdalena [16] carried out numerical
modeling of PAR based on the REKO 3 experiments using
ANSYS-FLUENT and tuned the model using their own exper-
imental results. [17]. carried out validation of REKO-DIREKT
code for simulation of PAR operational behavior against
small-scale experiments performed at Forsc-hungszentrum
Juelich in the REKO facilities. J.R. Hoyes et al. (2016) investi-
gated the capabilities of CFD models for predicting hydrogen
stratification in a containment vessel using data from the
NEA/OECD SETH2 MISTRA experiments. They also carried out
simulations to illustrate qualitative effects of hydrogen
stratification on the performance of PAR units [18], developed
a conceptual design of an innovative PAR which uses pool
boiling of water to enhance heat transfer from the catalyst
surface and limit its temperature. Additionally, an in-house
numerical code called HDS that couples CFD techniques
with models for reaction kinetics and water pool boiling heat
transfer was used to carry out numerical analysis of the same
[19]. implemented a CFD PAR model in the ANSYS Fluent code
using a correlation given by the PAR manufacturer. In the
25959
model, PAR unit is considered as a channel where the reaction
takes place without any chemistry details. The CFD model is
validated against two experimental tests (HR2 and HR5),
which were performed in the THAI facility in the frame of the
OECD/NEA THAI project. Recently, transient free convective
flows of a nanofluid with generalized thermal transport be-
tween two vertical parallel plates have been analyzed by [20].
They have used Laplace transform based approach for
obtaining closed form of temperature field and velocity field.
Pressurised Heavy Water Reactor (PHWR) based nuclear
power plants have been the mainstay of the Indian Nuclear
Power Programme. Although characteristically safe reactor
systems, all attempts are in place to further enhance the
safety of Indian PHWRs in the context of hydrogen related
safety issues. As part of the technology development pro-
gramme for hydrogen management in Indian Pressurised
Heavy Water Reactors (IPHWRs), experimental as well as
analytical studies are underway. Accordingly, a CFD based
tool has been developed using FLUIDYN-MP. A 3-D numerical
model of catalytic recombiner was developed and integrated
into the above code which was successfully validated against
data from open literature. The integrated tool has been
further used to predict the behavior of the prototype
recombiners tested in the Hydrogen Recombiner Test Facility
(HRTF).
This paper presents salient features of the recombiner
model, its validation against data from open literature and
results of parametric studies performed to investigate the
influence of various process and design parameters. The
validated and optimized model is then applied for the pre test
calculations on experiments in HRTF.
Recombination process
Recombiner box consists of a rectangular enclosure which is
open at bottom and top. The rectangular PtePd coated cata-
lytic thin stainless steel foils, are stacked in parallel at the
bottom portion of the box with a fixed gap between the foils.
The gas mixture enters the box at bottom opening and leaves
the box from opening at the top of the box as shown in Fig. 1.
Initially, gas mixture (hydrogen-steam-air) enters the
recombiner box due to external convection and molecular
diffusion. As steam hydrogen air mixture comes in contact
with catalytic plates, the exothermic oxidation reaction takes
place on the catalytic surface. The exothermic heat is partly
absorbed by the catalytic plate causing rapid increase in its
temperature.
Part of the exothermic heat is dissipated to the surround-
ing gas mixture by convection and radiation causing rise in its
temperature and initiation of buoyancy induced gas move-
ment in upward direction. With the increase in reaction rate
due to higher catalyst temperature, the buoyancy induced
flow becomes dominant. This causes more and more fresh
steam-hydrogen-air mixture to enter the box from the bottom
and leaner gas mixture to exit from top of the box. Thus there
are three transport phenomena viz. i) movement of gas
mixture by buoyancy through the box, ii) catalytic oxidation
reaction and iii) dissipation of exothermic heat which are to be
modeled.
25960 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3

Fig. 1 e Schematic of the recombiner box.

Analytical model  
t ¼ mg_  23m  k
=
ðV,UÞ d (3)
Commercial CFD software FLUIDYN-MP was used to build the
analytical recombiner model. The software is a general pur-
g_ ¼ VU þ ðVUÞT (4)
pose multi-physics CFD code with capability to handle single/
multi phase, multi-component, multi-fluid, transient incom-    
vðrTÞ vðlnrÞ hvp i
pressible/compressible flows with heat transfer, turbulence, Cp þ V,ðrUTÞ ¼ V,q  þ U,Vp þ ST (5)
vt vðlnTÞ p vt
chemical reactions, free surfaces, etc. It uses FVM based fluid
flow solver and FEM based solver for stress and thermal
analysis of structures. The flow and structural solvers are vðrXi Þ
þ V,ðrUXi Þ ¼ V,ðGXi VXi Þ þ Si
coupled by exchanging the forces, heat flux and mesh move- vt
X (6)
ment at each time step. In order to handle two phases, spe- Si ¼ Rj ðn2i  n1i Þ Mi
j
cifically arising due to steam condensation, Homogenous
Equilibrium Model (HEM) available in the code has been used.
1 X
The bulk condensation model is based on the concept of ST ¼ Q Rj (7)
Cp j rj
thermodynamic super saturation of the steam in gaseous
environment with an empirical relaxation constant for
Qrj ¼ heat of reaction of jth reaction in terms of leading
incipient bulk condensation initiation. The wall condensation
reactant [J/kmol]
is based on wall condensation theory in conjunction with the
gas mixture transport towards the wall by diffusive and 1 X
¼ DHf i ðn2i  n1i Þ (8)
convective modes. The physical problems studied in this work n11 i
involve relatively low volume fractions of the liquid phase,
except possibly at the walls, and this liquid phase will nor- Modeling of catalytic recombination
mally be well mixed. Since the interest of the present work
was to determine the effect of condensation on gas distribu- A chemical reaction is generally of the following form:
tion and pressure changes due to turbulence created by
X
i¼j X
i¼k
condensing steam, the homogeneous mixture model is cho- n1i R1i 4 n2i R2i (9)
sen over other models such as interface tracking models and i¼1 i¼1

Eulerian two fluid model.

With n0 s representing the stoechiometric coefficients and
R1 and R2 the reactants and products respectively. In present
Modeling of gas transport
case R11 is H2, R12 is O2, R21 is H2O. Also n11 is 2, n12 is 1 and n 21
is 2.
The reaction rate is usually expressed in terms of the
vr concentrations of the reactants and/or products, and catalyst
þ V,ðrUÞ ¼ 0 (1)
vt temperature. The rate of a reaction, Rj, can be computed in 3
ways in the code: (i) by chemical kinetics, (ii) controlled by the
vðrUÞ mixing of the reactants and (iii) combination of (i) and (ii). In
þ V,ðrUUÞ ¼ V,t  Vp þ SU (2)
vt the Arrhenius model used in the present work, the reaction
rate is completely determined by chemical kinetics. The
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3 25961

generalized Arrhenius form as implemented in the code is as plates to gas phase as well as radiation heat transfer mecha-
follows: nism between the walls have been considered. Heat sources
Y vji which result from the exothermic reaction between hydrogen
Rj ¼  Aj Tzj eEj =R T xi (10) and oxygen are added directly to the catalytic walls. As the
i
catalytic plates are at much higher temperature compared to
The above reaction has been modeled earlier by a detailed the surrounding gas as well as the container, they radiate heat
reaction mechanism as reported by [21,22]. For the present to the gas as well as container walls and thus accelerate the
work, a single step global reaction rate expression as reported natural convection process. The radiative transfer equation
by [23] has been used. This approach assumes a single step (RTE) for an absorbing, emitting, and scattering medium is:
reaction from the stable reactants to the stable products
Z4p
without considering any of the intermediate reactions. So the dІðr; sÞ sT4 ks
þ ðka þ ks ÞІðr; sÞ ¼ ka n2 þ Іðr; s'ÞFðr; sÞdU' (12)
species considered in the reaction are hydrogen, oxygen and ds p 4p
0
water vapor. The rate of the catalytic reaction in the present
case is given by the following expression given by [23]. (kaþks) is the optical thickness or opacity of the medium.
 
00 16:1x106 Homogeneous equilibrium model (HEM) & wall film model
R ¼ 14:exp: $½H2  (11)
Ru T

The recombination reaction on catalyst surface is a multi Flows involving phase change are modeled using homoge-
step process but the CFD model deals with the overall phe- neous equilibrium model (HEM) for phase changes in the bulk
nomena associated with recombination hence the recombi- domain and wall film model for those occurring near or at the
nation reaction has been modeled using single step chemistry walls. The phase change, evaporation and condensation, is
with an aim to reasonably capture the overall phenomena in a modeled by adding the appropriate source or sink term to the
computationally economical manner without losing on the species equations for the liquid phase and the equation for the
accuracy of results. Schefer's equation is a generic single step corresponding vapor. The assumptions used in these models
formulation. The values for activation energy and pre expo- are: (i) both phases are in thermodynamic equilibrium, and
nential factor are catalyst specific. The Schefer's equation has hence have same pressure and temperature, (ii) there is no
been used as it has been widely referred in literature. More- well-defined interface between the two phases and (iii) the
over the experiments on specific catalyst to be used in PCRD fluid, defined as a mixture of gases, follows the equation of
for Indian NPPS are in progress and the data from them will be state of the form p ¼ f (r,T,q). The density r is the mixture
used to fine tune the Schefer equation for specific catalyst to density, i.e., the sum of the macroscopic densities for each
further improve the results. However the results from Sche- individual species including the liquid water. The equation of
fer's data are fare and accurate enough to give the general state for mixture of gases is modified to account for the
trend for behavior of PCRD and its dependence on location for presence of the liquid water. The general equation of state
better performance. thus becomes,
The region covering the plate is modeled as a solid domain
P
ish2ol P
ish2ol
and rest of the domain is modeled as fluid cells. The reaction T Ri ri rT Ri Xi
i i
source and sink terms for the species are applied in the very p¼ ¼ (13)
q q
first fluid cell adjacent to the solid catalyst surface. Since the
reaction is taking place at the surface, the heat release due to The volume fraction of the gaseous mixture is calculated as
the reaction occurs directly on the plate surface. It is assumed rh2ol
q ¼ 1  ql ¼ 1  (14)
that the reaction mechanism is same for all the catalytic rthermo
h2ol
plates and hence, the reaction rate will depend on the pre-
where rthermo
h2ol is 1000 kg/m3.
vailing catalyst temperature and the species concentration in
the adjacent fluid cells.
Bulk condensation and evaporation models
The mass source and sink term, in kg/(m3s), for steam is given
Modeling of heat transfer and radiation
by
 
As the heterogeneous reaction of hydrogen and oxygen is Sh2og ¼ rsat  rh2og;gas t (15)
pc
highly exothermic in nature, the large amount of heat liber-
ated during the reaction heats up the catalyst surface as well
psat
as the gas mixture surrounding it. Thus passive, self initiating rsat ¼ (16)
Rh2og T
and self sustained natural circulation flow of the gas mixture
continuously drives out the recombined gas outside and sucks c2 Tc1

in fresh gas rich in hydrogen. The equation (5) represents psat ¼ 105 e c3 T
(17)
energy conservation which defines the adjoining gas mixture. a value of 0.1 is found to be appropriate for tpc the relaxation
Inside the catalyst plates, heat is transported due to heat time for phase change in large containments. The mass
conduction. Inside the flow channels, the heat flow is due to source and sink term, in kg/(m3s), for liquid water is given by
convection through the gas phase. Conjugate heat transfer
model for representing convective heat dissipation from solid Sh2ol ¼  Sh2og (18)
25962 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3

The enthalpy of water vapor is taken as the sum of At the interface of the fluid and structure the surface tem-
enthalpy of liquid water and enthalpy of latent heat of evap- perature of the structure is used as a boundary condition for
oration. So, internal energy source term is derived using the the fluid flow equations and the heat flux from the fluid (in
liquid water and water vapor's enthalpy values. terms of the local film coefficient and temperature) is used as
a boundary condition for the structure. The condensation at
Wall film model the walls is modeled by varying the thickness of the static
The convective heat transfer coefficient is obtained by using film using the rate of condensation, evaporation and falling.
Reynolds analogy formulation and the mass transfer coeffi- The effect of film on heat transfer is taken into account by
cient is calculated based on Chilton-Colburn empirical anal- varying the local film coefficients and also by adding the heat
ogy between heat and mass transfer. of evaporation.
The wall condensation/vaporization rate 2s , is given by
( )
dns As rthermo
h2 ol   Validation of recombiner model
2s ¼ max  ; hd As rh2 og;gas  rs;sat (19)
2dt
The model for catalytic recombiners was validated by
The mass source or sink term for steam, in kg/(m3s), due to
comparing the results of the analysis against the experi-
phase change involving structural surfaces, is given by (for all
mental data obtained in REKO-3 facility. This facility has three
the surfaces)
modules (Module1, Module2, and Module3). Module1 works as
1X inlet and feeds gas mixture to Module2, which contains four
Sstr;h2og ¼  2s (20)
V s catalytic plates where hydrogen recombines with oxygen to
form water. Module3 works as outlet. The whole unit is ther-
The internal energy source or sink term, in J/(m3s) is given
mally insulated on the sides to prevent heat losses. Four
by (for all the surfaces)
catalyst sheets (stainless steel sheets coated with wash coat/
8 9 platinum catalyst material) having dimensions of
1X < 2 I ðT Þ =
s h2og 
Sstr;ie ¼  max
s
(21) 143 mm  143 mm x 1.5 mm are used in the experiments. The
V s : hd As rh2 og;gas  rs;sat Ih2 og ðTÞ ;
facility has been used to investigate the behavior of a
recombiner under controlled conditions.
The work done by the gas mixture due to wall condensa-
The geometry for numerical computations is shown in
tion, in J/(m3s), is given by (for all the surfaces)
Fig. 2. For the model development and validation, various
1X simulations were performed with increasing complexity of
Swork ¼  2s Rh2 og T (22)
V s simulation in which a dry gas mixture of air and hydrogen was
fed to the recombiner. These simulations differed in number
When the wall film thickness exceeds the critical film
of channels, gas transport properties and heat losses to the
thickness (critical film thickness, dcritical , is the maximum wall
ambient.
film thickness, which is allowed to stick with the wall face)
Since the geometry of recombiner with parallel plates is
then the excess water is drained out from the wall. The
close to regular orthogonal shape, a structured grid approach
following equations give the mass and energy sink term for
has been used here as this allows better control of the mesh
liquid water to drain out the excess wall film thickness. The
size near and away from the plate walls by the use of appro-
mass sink term for liquid water, in kg/(m3s) is given by (for all
priate power laws. Unstructured grid can also be used but
the surfaces)
since the dimensions of interest here range from few milli-
1  n  meters near and in between the recombiner plates to the order
Sh2ol;drain ¼ As rthermo
h2 ol ds  dcritical (23) of meters in the open domain of vessel (containment), its
Vdt
quality cannot be perfectly controlled and hence is avoided
The internal energy sink term, in J/(m3s) is given by (for all
here.
the surfaces)
In order to simulate REKO-3 prototype results using the
Sie;drain ¼ Sh2ol;drain Ih2 ol ðTÞ (24) present analytical model, 4 catalyst plates along with walls of
the facility were considered. For simulation purpose, only
Wall film thickness,
plates P1, P2 and the left box wall were chosen, since the
2s dt system has symmetry with plane S2 as shown in Fig. 2. The
dnþ1 ¼ dns þ (25)
s
As rthermo temperature profiles were calculated at the surfaces of P1 and
h2 ol
P2, while hydrogen concentrations were calculated along
planes S1 and S2. Grid sensitivity studies were performed
Method of solution initially and on this basis, a grid size of 0.5 mm (flow direction)
and 0.25 mm (direction normal to flow) was found to be the
Along with the above equations for the fluid, three- optimal size. As part of the studies for three channel geome-
dimensional heat conduction equation is solved in the try, several simulations were performed by changing the gap
structure surrounding the fluid domain. At each time step between the catalyst plate and the box wall as this dimension
the equations in the fluid and structure are solved sequen- is not reported in the respective references. The outer wall of
tially. The equations can be discretized using three- the box was modeled by an insulated boundary. Typical re-
dimensional, body fitting, unstructured or structured mesh. sults for the temperature and concentration profiles are given
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3 25963

Fig. 2 e Geometry for 3 channel and single channel cases for simulation of REKO-3 experiments.

Fig. 3 e Profiles of catalytic plate temperature and hydrogen concentration.

in Fig. 3. As can be seen from these results, only the lower

portion of the catalytic plate is seen to be participating in the
catalytic reaction. This can be attributed to lower inlet H2
concentrations (2%) used in the reported experiment.
Several parametric studies were also performed for a single
channel to study the effect of (i) inlet concentration and (ii)
inlet flow velocity of hydrogen. These were also compared
against the experimental data and typical results are pre-
sented in Fig. 4.
It can be seen that the steady state concentration profiles
are matching well with experimental results for all the values
of concentrations at inlet. The temperature profiles for the
cases with 1% (v/v) and 2% (v/v) are closer to experiments but
this is not the case for 4% (v/v). The model predicts higher
temperatures at the leading edge of the plate for the higher
concentration case. This could be attributed to the depen-
dence of gas diffusivity on temperatures which was not
accounted for in the simulation. Since the availability of
hydrogen is less in the upper part of the plate (Fig. 3), plate
temperatures are less here. As there is no significant change in
diffusivity, the results in this region of the plate match well
with the experimental results.
Application of the model for pre test calculations
of HRTF experiments
After validating the model developed in the CFD code
FLUIDYN-MP, the experiments to be conducted in the
Hydrogen Recombiner Test Facility were simulated. HRTF is a
facility to assess the performance of the PCRD developed at
BARC. These experiments are aimed at optimizing the
recombiner design and its performance and generate an
extensive database for validation and fine-tuning of numeri-
cal models of nuclear reactor containment problems.
HRTF is a cylindrical carbon steel vessel of 3 m diameter
and around 9 m height with toro-spherical top and bottom
heads. The vessel has a volume of 60 m3. The facility contains
a recombiner device which comprises of a rectangular box
made of stainless steel sheets open at top and bottom. The
approximate dimensions of the box are 400 mm  700 mm x
1000 mm. At the lower portion of this box, catalytic plates are
stacked at a pitch of 10 mm between them. Plates are coated
with platinum/palladium catalyst. The gas is free to enter and
exit at bottom and top ends of the box respectively. Fig. 5
25964 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3

Fig. 4 e Catalytic plate temperature and hydrogen concentrations with different concentrations of H2 at the inlet.

Fig. 5 e Locations of diffuser and PCRD in HRTF (left) and schematic of recombiner box (right).

shows the location of diffuser and recombiner in HRTF as well (adiabatic). The conditions used for dry as well as wet cases
as the schematic of recombiner box. are given in Table 1.
The recombiner can be hung in the vessel at three heights
(near top, middle & near bottom). The facility also has a steam Geometry and mesh
supply unit comprising of an electrical boiler for supplying
saturated steam to vessel. Steam condensation in the vessel For premix cases, as diffuser is not present and the central
due to loss of heat from the walls is minimized by insulating axis of recombiner box passes through the central axis of
the vessel from outside. The injection mechanism for containment, hence quarter symmetry was considered. Hex-
hydrogen inside the vessel consists of a perforated pipe. The ahedral elements were considered for modeling fluid as well
pipe is located at a vertical distance of 875 mm from the bot- as structural domain. The mesh for recombiner geometry
tom and is kept with its axis in the horizontal plane and having thin catalyst plates was selected based grid sensitivity
coinciding with the vessel diameter. The pipe length is 3 m study performed during validation exercise of the recombiner
and is supported on both ends with vessel walls. The diffuser model. Similarly the mesh for modeling the vessel geometry
has 450 holes of 5 mm diameter on one half side of its was also selected based on a separate mesh sensitivity study
circumference (0 to 180 ). for hydrogen distribution inside the closed vessel. Fine mesh
was used in the regions where gradients were sharp especially
Cases simulated in present work between the plates and at inlet of the recombiner box. Other
than these regions, relatively coarse mesh was used to reduce
The present simulations aim to evaluate the performance of computational time. As fine mesh is not required at all parts of
the recombiner at different locations in dry and wet condi-
tions. Since the gas mixture is present uniformly all over the
geometry for premixed cases, simulation domain has been Table 1 e Input for dry & wet condition cases.
considered as quarter symmetry. Pressure Temperature Concentration (% v/v)
In dry condition, the insulated vessel walls are assumed to (bar) (oC)
H2 Steam Air
be at the vessel gas temperature. Whereas wet condition
include non-condensing hydrogeneairesteam gas mixture in Dry case 1.165 30 8 e 92
Wet Case 2.168 105 8 50 42
the containment as the vessel walls are kept insulated
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3 25965

Table 2 e Mesh details.

Scenario Recombiner Position Fluid Structure
Elements Nodes Elements Nodes
Premix Cases Bottom 97119 127302 6700 12042
Middle 97218 127525 6700 12042
Top 97119 127302 6700 12042

the domain, a non-conformal mesh was considered sur- fluid domain, the interface of the catalyst plates and sur-
rounding the recombiner box to maintain sufficient fine mesh rounded fluid domain has been applied with surface reaction
inside the box and coarse mesh outside it. Thus number of and the total energy released due to exothermic reaction is
elements in the domain could be reduced without compro- applied on to the catalyst plate.
mising the ability to capture gradients between the catalyst Symmetric boundary condition has been applied on the
plates. Table 2 explains the details about the mesh used for faces representing symmetry. Radiation is modeled between
premix cases. catalyst plates and box cover with an emissivity of 0.8.
For recombiner box, to represent the symmetrical condi-
Flow configuration tion at symmetrical faces, adiabatic boundary condition (i.e.
heat flux ¼ 0) has been applied.
The flow has been considered compressible with mixture of
gases containing hydrogen, nitrogen, oxygen, steam and Method of solution
liquid water. Buoyancy Gravity model has been considered for
correct diffusion of hydrogen as it is a low density gas. The In order to solve the set of non dimensional governing equa-
appropriate reference density used for both dry as well as wet tions, pressure based 1st Order Euler Implicit solver (i.e. MP-
condition cases are provided below. NSNT) has been used for transient simulation. The methods
and schemes used in the solver are listed in Table 4:
Ref. density for dry condition ¼ 1.2341
Ref. density for wet condition ¼ 1.4678 Pressure-velocity coupling e Semi-Implicit Method for
Pressure Linked Equations-Consistent (SIMPLEC)
keε Chen-Kim model is considered for predicting turbulence Convection scheme e Upwind Differencing Scheme (UDS)
parameters inside the domain however k-u model was also
tried but k-ε model was found better during validation studies.
More complex models are difficult to use because of their high Results and discussion
computational requirements adding to the complexity and
sluggishness of solution. Homogeneous Equilibrium Model For dry as well as steam environment conditions, three
(HEM) is used for bulk evaporation and condensation. recombiner locations have been considered for simulation i.e.
top, middle and bottom recombiner position. Hence total six
Initial conditions cases have been considered for simulation. For comparison
between these cases, the contours and vectors have been
The initial conditions used for simulations are presented in plotted on symmetrical planes. Trace point locations have
Table 3. been considered in fluid as well as in structural domain to get
the time history of variables like concentration of hydrogen,
Boundary conditions steam and temperature etc. The selected locations of the trace
points are different for each scenario depending on the loca-
The outer walls of the domain are applied with adiabatic no- tion of the recombiner box. For example, in case where
slip wall condition. Since the surface reaction is modeled in recombiner is placed near top of the vessel, Fig. 6 shows the
location of trace points placed in fluid as well as on catalyst
plates. Various trace points in fluid domain were considered at
Table 3 e Initial conditions. top and bottom of the vessel, below and above the catalyst
Fluid Domain plates, and above the recombiner box like as shown in front
view of vessel. Trace point locations in top view explain about
Scenario Variable Dry Condition Wet Condition
the distribution of trace points at each elevation. Similarly the
Premix Cases Pressure (Pa) 116524.0 216837.0 trace points on catalyst plates were considered at all four
Temperature (K) 303.0 378.0
corners and at center as well.
Vol. fr. of H2 8% 8%
The trace points on catalyst plate were considered on two
Vol. fr. of N2 72.68% 33.18%
Vol. fr. of O2 19.32% 8.82% plates in such a way that one is near to symmetry plane (Plate
Vol. fr. of Steam e 50% e 36) and the other one near to recombiner box wall (Plate e 1).
All other cases for middle and bottom recombiner positions in
Structural Domain
dry as well as wet conditions were also analyzed in same
Scenario Variable Dry Condition Wet Condition
manner with respect to the location of trace points and other
Premix Temperature (K) 303.0 378.0 parameters to be analyzed in the results.
25966 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3

Table 4 e Solver parameters for different flow variables.

Variable Number of inner Convergence criterion Relaxation Convection
iterations for inner iterations factor scheme
Pressure 2000 0.01 0.3 UDS
U (axial velocity) 20 0.1 0.3 UDS
V (transverse velocity) 20 0.1 0.3 UDS
W (transverse velocity) 20 0.1 0.3 UDS
Temperature 20 0.1 0.3 UDS
H2 20 0.1 0.3 UDS
O2 20 0.1 0.3 UDS
H2O 20 0.1 0.3 UDS

Fig. 6 e Location of sensor points in fluid and on catalyst plates for top recombiner case.

For the transient analysis of various cases, 1200 s have
been fixed as the simulation time. This is because as the
concentration decreases, the temperature comes down and
hence the rate of recombination reaction decreases due to
temperature dependence of reaction kinetics. This leads to
progressive sluggishness of the transient. One of the initial
cases was run for a longer duration but it was observed that
the trends are mostly captured during the initial time when
concentration is more. Hence it was decided to run all the
cases for 20 min of simulation to capture the transient for
comparison purpose. Moreover the simulation for 1200 s itself
takes around 6 months for each case to complete. Hence it is
not practical to carry out simulation of longer duration.
Cases for dry condition with premixed hydrogen
Fig. 7 shows the comparative contour plots for hydrogen
concentration after 1200 s for three dry cases having different
recombiner positions. This figure shows that case having
bottom positioned recombiner in the facility consumed the
maximum hydrogen during 1200s for dry cases.
Initially as the 8% hydrogen in air mixture is present uni-
formly in the vessel, the reaction starts immediately at the
catalyst plates. This exothermic reaction leads to increase in
temperature. This change in temperature drives the flow due
to buoyancy causing hot product gas moving out of the
recombiner box from the top and fresh hydrogen rich gas
entering the recombiner from bottom. The product gases
travel upward due to natural convection, till they reach the
top of vessel. During this travel they displace the hydrogen
already present in this region. Thus a natural convection loop
is developed wherein hot gases move upward near the axis of
the vessel and the colder gases (rich in premixed hydrogen)
move downwards near the vessel walls. So at initial time the
velocities in the box accelerate and attain a maximum and
then decreases to stabilize at a lower value. The time duration
for the natural convection loop to establish and flow stabili-
zation is different for three cases having different recombiner
positions.
For the top recombiner case, the velocities in the box attain
a maximum of 1.6 m/s and then decreases to low value of
around 0.5 m/s, as against a maximum of 3.0 m/s and de-
creases up to around 1.0 m/s for the bottom recombiner case.
The flow stabilization takes place during 100se700s for the top
recombiner case where as for the bottom recombiner case,
due to the large amount of hydrogen present above the
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3
Fig. 7 e H2 concentration (v/v) Contours at 1200s on ZX-Symmetry plane for top (left), middle (centre) and bottom (right)
recombiner positions in dry cases.
recombiner box, the process of hot product gases replacing
and pushing hydrogen downward continues for the entire
simulation time.
The maximum concentration of hydrogen in the facility by
the end of 1200s is present at the bottom of the vessel. In top
and middle positioned recombiner cases, it is similar to the
initial concentration of hydrogen (8% v/v), whereas in bottom
positioned recombiner, its maximum concentration in the
domain reduced to around 5% v/v for dry cases. This is pri-
marily due to the hot gases coming out of recombiner box
moving upwards and promoting better buoyancy driven flows
for lower recombiner positions. The effect is not prominent
for the case having recombiner box at top location as the hot
gases coming out from the box are already near the top of
vessel and hence are not able to promote buoyancy driven
flows and mixing.
Fig. 8 shows the comparative temperature contour plots for
the three different cases at 1200 s. Since the bottom placed
recombiner case recombines maximum amount of hydrogen
in the domain, the temperatures are higher in most parts of
the domain due to more exothermic heat liberated during the
recombination reaction.
As seen in the contour plots of hydrogen concentration,
there is no significant planar distribution of concentration at
any horizontal plane inside the domain. Hence the trace point
data for five trace points on any particular XeY plane show
almost the same value. Therefore, H2 concentration profiles
are plotted on single trace point in each XeY plane. Data on
25967
trace point 1 which is on the axis of the vessel has been plotted
at elevations such as bottom of the catalyst plates, top of the
plates, top of the recombiner box, top of the vessel and bottom
of the vessel. Such plots are made for all three cases having
top, middle and bottom recombiner positions. Figs. 9 and 10
show hydrogen and steam concentrations respectively at
various elevations for top, middle and bottom recombiner
positions. Concentration of hydrogen inside the entire domain
is 8% at initial time and starts decreasing over time due to
surface reaction. The drop in concentration at elevations
above catalyst plates and above recombiner box is more rapid
for all three cases. This is because hot product gases of
recombination which are lean in hydrogen, rise upwards
causing sharp decrease in hydrogen concentration at these
locations. The hydrogen concentration at trace point near the
top of the vessel has a gradual drop as compared to trace
points near plate top and box top. This is because the upward
moving product gases of recombination have more distance to
cover and hence have more hydrogen to displace during their
travel up to the vessel top. At trace point near the bottom of
the vessel, the drop in hydrogen concentration is slowest. This
is because the hot product gases leaving the recombiner box
rise upwards and accumulate at the top by displacing the
hydrogen present there towards the lower elevations. This
effect takes maximum time to reach the bottom of the vessel.
For the bottom recombiner case, since there is very little gas
volume present below the box, the hydrogen present there is
effected by this process more rapidly as compared to the
25968 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3

Fig. 8 e Temperature Contours at 1200s on ZX-Symmetry plane for top (left), middle (centre) and bottom (right) recombiner
positions dry cases.

Fig. 9 e Time history plots for hydrogen concentration (v/v) on trace point 1 at various elevations for top (left), middle (centre)
and bottom (right) recombiner positions in dry cases.

Fig. 10 e Time history plots for steam concentration (v/v) on trace point 1 at various elevations for top (left), middle (centre)
and bottom (right) recombiner positions in dry cases.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3
middle and top recombiner positions. The steam concentra-
tion at all trace points rise proportionally in accordance
with fall in hydrogen concentration. This is evident from Figs.
9 and 10.
Fig. 11 shows the time history plot of catalyst plate tem-
perature at the centre of the plate for plate e 1 and plate e 36
for top, middle and bottom recombiner positions. The
maximum temperature observed at centre of both plates is
around 850 K with the trace points placed at same positions on
both plates behaving similarly. It is worth noting that the
maximum plate temperature is observed at the bottom of the
plate as it sees maximum hydrogen concentration. The
maximum temperature value at plate bottom for all the cases
is around 950 K.
The temperatures of both the plates fall most rapidly for
the top recombiner position while the fall is gradual for mid-
dle and bottom recombiner positions due to the constant flow
of hydrogen into the recombiner box due to buoyancy induced
flows driven by hot gases coming out of the recombiner box.
The trace points on plate e 36 had at higher temperatures,
since the availability of hydrogen is more at center plates than
at end plates of the box due to recirculation of the flow while
entering into the recombiner box.
Cases for wet condition with premixed hydrogen
Cases with steam laden conditions were analyzed in same
manner as the dry cases with respect to the recombiner po-
sition, the location of trace points and other parameters to be
analyzed in the results. Figs. 12 and 13 show the comparative
contour plots for concentration of hydrogen and steam
respectively after 1200 s for these three cases having different
recombiner positions. These figures show that case having
bottom positioned recombiner in the facility consumed the
maximum hydrogen during 1200s for wet cases as was in the
case of dry cases.
The maximum concentration of hydrogen in the facility in
top and middle positioned recombiner by the end of 1200s is
similar to the initial concentration of hydrogen (8% v/v), where
as in bottom positioned recombiner, its maximum concen-
tration in the domain reduced to around 6% v/v for these wet
cases. This is primarily due to the hot gases coming out of
recombiner box moving upwards and promoting better
buoyancy driven flows for lower recombiner positions. The
effect is not prominent for the case having recombiner box at
top location as the hot gases coming out from the box are
Fig. 11 e Time history plots of catalyst plate temperature at plate centre for plate 1 and plate 36 for top (left), middle (centre)
and bottom (right) recombiner positions in dry cases.
25969
already near the top of vessel and hence are not able to pro-
mote buoyancy driven flows and mixing.
Fig. 14 shows the comparative temperature contour plots
for the three different cases at 1200 s. Since the bottom placed
recombiner case recombines maximum amount of hydrogen
in the domain, the temperatures in this case are higher in
most part of the domain due to more exothermic heat liber-
ated during the recombination reaction.
Since there was no significant difference between trace
point data for five trace points on any particular plane as seen
in contour plots, H2 concentration profiles were plotted on
trace point 1 i.e. on the axis of the vessel at various elevations
such as bottom of the catalyst plates, top of the plates, top of
the recombiner box, top of the vessel and bottom of the vessel.
Such plots were made for all three different cases having top,
middle and bottom recombiner positions. Fig. 15 shows
hydrogen concentration at various elevations for top, middle
and bottom recombiner positions. All these plots reveal that
hydrogen concentration falls sharply above the catalyst plates
and near the top of recombiner box for all the cases. This is
because of the exothermic recombination reaction taking
place at the plates resulting in hot gases rising upwards and
replacing the hydrogen present there. For all these cases, the
hydrogen concentration near the bottom of the catalyst plates
falls gradually. This is because of the natural convection flow
developed around the recombiner box resulting in fresh sup-
ply of hydrogen at the box inlet (bottom) from other locations
in the vessel.
The hydrogen concentration at the top of vessel falls fast-
est for the top recombiner position to reach zero value at
around 500 s. The H2 concentration for the middle and bottom
recombiner positions also fall to almost zero value near the
end of simulation. However, the hydrogen present near the
bottom of the vessel takes maximum time to decrease and
remains almost constant for top and middle recombiner po-
sitions where as for the bottom recombiner position it starts to
fall at around 300 s and reaches to around 7% (v/v) at the end
of simulation.
Fig. 16 shows the profiles of steam concentration on the
central trace points at various elevations such as bottom of
the catalyst plates, top of the plates, top of the recombiner
box, top of the vessel and bottom of the vessel for top, middle
and bottom recombiner positions. Steam concentration rises
with respect to time in accordance with the fall of H2 con-
centration as seen from these plots. There is a rapid rise in
steam concentration at elevations above catalyst plates and
25970 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3

Fig. 12 e H2 concentration (v/v) Contours at 1200s on ZX-Symmetry plane for top (left), middle (centre) and bottom (right)
recombiner positions in wet cases.

Fig. 13 e Steam concentration (v/v) Contours at 1200s on ZX-Symmetry plane for top (left), middle (centre) and bottom (right)
recombiner positions in wet cases.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3 25971

Fig. 14 e Temperature Contours at 1200s on ZX-Symmetry plane for top (left), middle (centre) and bottom (right) recombiner
positions wet cases.

Fig. 15 e Time history plots for hydrogen concentration (v/v) on trace point 1 at various elevations for top (left), middle
(centre) and bottom (right) recombiner positions in wet cases.

above recombiner box for all three cases. This is because hot
product gases of recombination containing steam, rise up-
wards causing sharp rise in steam concentration at these lo-
cations. The steam concentration at trace point near the top of
the vessel has a gradual rise as compared to trace points near
plate top and box top. This is because the upward moving
product gases of recombination have more distance to cover
and hence have more volume to disperse during their travel
up to the vessel top. At trace point near the bottom of the
vessel, the rise in steam concentration is slowest. This is
because hot product gases leaving the recombiner box rise
upwards and accumulate at the top. This effect takes
maximum time to reach the bottom of the vessel. For top
recombiner position case, the steam concentration rises
steeply at all elevations except near the bottom of the vessel.
The rise in steam concentration is relatively gradual for the
bottom recombiner position due most prominent natural
convection effect when recombiner is placed near the bottom
of the vessel. This is because of maximum travel height for hot
gases coming out from recombiner. The same causes mixing
and homogenization in the vessel hence the steam concen-
trations rise gradually in this case.
25972 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3
Fig. 16 e Time history plots for steam concentration (v/v) on trace point 1 at various elevations for top (left), middle (centre)
and bottom (right) recombiner positions in wet cases.
Fig. 17 e Time history plots of catalyst plate temperature at plate centre for plate 1 and plate 36 for top (left), middle (centre)
and bottom (right) recombiner positions in wet cases.
Fig. 17 shows the time history variation of catalyst plate
temperature at the centre of the plate for plate e 1 and plate e
36 for top, middle and bottom recombiner positions. The
maximum plate temperature is observed at the bottom of the
plate as it sees maximum hydrogen concentration which de-
creases along the plate height. The maximum temperature
value at plate bottom for all the cases is around 1050 K. As the
central trace point is located away from the plate bottom, the
maximum value observed at plate centre as seen in Fig. 17 is
around 870 K.
The temperatures of both the plates fall rapidly for the top
recombiner position while the fall is gradual for middle and
bottom recombiner positions. This is due to constant
hydrogen flow into box due to buoyancy induced flows driven
by hot gases coming out of the recombiner box.
Conclusion
Numerical model has been developed and validated for PCRD
using commercial CFD software FLUIDYN. The validated
model has been implemented for pre test calculations of ex-
periments in Hydrogen Recombiner Test Facility (HRTF). The
performance of the recombiner has been investigated by
placing the recombiner box in different locations in the vessel,
in different conditions (i.e. dry and steam) at premixed
hydrogen environment. The conclusion of all these studies is
as listed below.

Parametric study on PCRD location demonstrates that
exothermic recombination causes natural convection &
promotes mixing.

For Dry cases, the temperature observed on catalyst plates
in all the cases is similar and it is around 950 K.

The behavior of wet condition cases is almost similar to the
dry cases. In spite of the higher initial temperature in wet
cases, the overall temperature rise and performance of
catalyst plates does not increase much as compared to the
dry cases. This is because of the high heat capacity of
steam present in the domain.

The bottom positioned recombiner in the vessel consumed
the maximum hydrogen during simulation time for dry as
well as wet cases.

The maximum concentration of hydrogen in the facility in
top and middle positioned recombiner during the analysis
is close to the initial concentration of hydrogen (8% v/v),
where as in bottom positioned recombiner, its maximum
concentration in the domain is around 5% v/v for dry and
6% v/v for wet cases.

Models for condensation in bulk as well as at the walls
have been used in the simulation owing to the steam
formed during recombination reaction and also the steam
already present in wet cases. However due to the
exothermic reaction, the water vapor formed during
recombination gets heated up and does not condense in
any of the cases presented.

The findings reveal that PCRD location plays important role
in its effectiveness to curtail hydrogen stratification inside
closed geometry.

The investigation establishes the fact that effectiveness of
recombiner box is maximum when it is placed near the
bottom of the vessel. This is in contradiction with the
earlier assumption that recombiner performance can be
increased by placing the same at higher altitudes due to the
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 2 5 9 5 7 e2 5 9 7 3
possibility of hydrogen rich pockets in the upper regions
because of its low density.

This investigation is vital for the hydrogen mitigation
programme for severe accident scenarios as position of the
recombiner box plays vital role in the mitigation scheme.
Acknowledgement
The technical support from Transoft International Bangalore
-India, the developers of FLUIDYN MP is sincerely
acknowledged.
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.ijhydene.2019.08.015.
references
[1] Travis JR, Spore JW, Royl P, Lam KL, Wilson TL, Necker GA,
Nichols BD, Redlinger R, Hughes ED, Wilkening H,
Baumann W, Niederauer GF. “GASFLOW: a computational
fluid dynamics code for gases, aerosols, and combustion; vol.
1, theory and computational model; vol. 2, user's manual;
and vol. 3, assessment manual,” Los Alamos national
laboratory report LA-13357-MS (1998), and
forschungszentrum karlsruhe report FZKA-5994. 1998.
[2] Royl P, Rochholz H, Travis JR, Necker G, Breitung W.
“Three dimensional analysis of steam/hydrogen transport
with catalytic recombiners in nuclear reactor containment
using computer code GASFLOW”, 15th international
conference on structural mechanics in reactor technology,
post-conference seminar on containment of nuclear
reactors. Seoul, Korea: Hoam Convention Center; 1999.
August 23-24.
[3] Royl P, Rochholz H, Breitung W, Travis JR, Necker G. Analysis
of steam and hydrogen distributions with PAR mitigation in
NPP containments. Nucl Eng Des 2002;202:231e48.
[4] Martın-Valdepenas JM, Jimenez MA, Martın-Fuertes F,
Fernandez JA. Improvements in a CFD code for analysis of
hydrogen behaviour within containments. Nucl Eng Des
2007;237:627e47.
[5] Royl Peter, Travis John R, Breitung Wolfgang, Kim Jongtae,
Kim Sang Baik. “GASFLOW validation with Panda tests from
the OECD SETH Benchmark covering steam/air and steam/
helium/air mixtures”, XCFD4NRS Workshop. 10-12
September 2008. Grenoble.
[6] Kudriakov S, Dabbene F, Studer E, Beccantini A, Magnaud JP,
Paill‘ere H, Bentaib A, Bleyer A, Malet J, Porcheron E, Caroli C.
The TONUS CFD code for hydrogen risk analysis: physical
models, numerical schemes and validation matrix. Nucl Eng
Des 2008;238:551e65.
[7] Breitung W. Analysis methodology for hydrogen behavior in
accident scenarios. In: International conference on hydrogen
safety; 2005. 120009, Pisa, 8-10 September.
25973
[8] Heitsch Matthias, Risto Huhtanen, Zsolt Te chy, Chris Fry,
Kostkac Pa  l, Jarto Niemi, Berthold Schramm. CFD evaluation
of hydrogen risk mitigation measures in a VVER 440/213
containment. Nucl Eng Des 2010;240(Issue 2):385e96.
[9] Mimouni Stephane, Mechitoua Namane, Ouraou Mehdi. CFD
recombiner modelling and validation on the H2-Par and Kali-
H2 experiments. Science and Technology of Nuclear
Installations July 2011;2011(3). 10.1155/2011/574514.
[10] Prabhudharwadkara Deoras M, Aghalayamb Preeti A,
Iyer Kannan N. Simulation of hydrogen mitigation in
catalytic recombiner: Part-I: surface chemistry modeling.
Nucl Eng Des 2011;241:1746e57.
[11] Prabhudharwadkar Deoras M, Iyer Kannan N. Simulation of
hydrogen mitigation in catalytic recombiner. Part-II:
formulation of a CFD model. Nucl Eng Des 2011;241:1758e67.
[12] Reinecke Ernst-Arndt, Kelm Stephan, Jahn Wilfried,
€ kel Christian, Allelein Hans-Josef. “Simulation of the
Ja
efficiency of hydrogen recombiners as safety devices”,
International Conference on Hydrogen Safety. September 12-
14, 2011 [San Francisco, CA].
[13] Roze  Antoni. Modeling of a passive catalytic recombiner for
_ n
hydrogen mitigation by CFD methods. Bled Slovenia: 22nd
International Conference Nuclear Energy for Europe;
September 9-12, 2013.
[14] Meynet Nicolas, Ahmed Bentaı̈b, Giovangigli Vincent. Impact
of oxygen starvation on operation and potential gas-
phaseignition of passive auto-catalytic recombiners”.
Combust Flame 2014;161:2192e202.
[15] Klauck Michael, Reinecke Ernst-Arndt, Kelm Stephan,
Meynet Nicolas, Ahmed Bentaı̈, Allelein Hans-Josef. Passive
auto-catalytic recombiners operation in the presence of
hydrogen and carbon monoxide: experimental study and
model development. Nucl Eng Des 2014;266:137e47.
[16] Orszulik Magdalena, Adam Fic, Bury Tomasz. CFD modeling
of passive autocatalytic recombiners. Nukleonika
2015;60(2):347e53.
[17] Reinecke Ernst-Arndt, Kelm Stephan, Steffen Paul-Martin,
Klauck Michael, Allelein Hans-Josef. Validation and
application of the REKO-DIREKT code for the simulation of
passive autocatalytic recombiner operational behavior. Nucl
Technol November 2016;196:355e66.
[18] Agrawal Nilesh, Seik Mansoor Ali V. Balasubramaniyan,
“Innovative hydrogen recombiner concept for severe
accident management in nuclear power plants”. Nucl Eng
Des 2017;323:359e66.
[19] Halouane Y, Dehbi A. CFD simulation of hydrogen mitigation
by a passive autocatalytic recombiner. Nucl Eng Des
2018;330:488e96.
[20] Ahmed 1 Najma, Shah 1,2 Nehad Ali, Ahmad 3 Bakhtiar,
Shah 3 Syed Inayat Ali, Ulhaq 3 Sami, Rahimi-
Gorji Mohamad. Transient MHD convective flow of fractional
nanofluid between vertical plates. Journal of Applied and
Computational Mechanics 2019;5(4):592e602.
[21] Heitsch Matthias. Fluid dynamic analysis of a catalytic
recombiner to remove hydrogen. Nucl Eng Des
2000;201:1e10.
[22] Appel C, Mantzaras J, Schaeren R, Bombach R, Inauen A,
Kaeppeli B, Hemmerling B, Stampanoni A. An experimental
and numerical investigation of homogeneous ignition in
catalytically stabilized combustion of H2/air mixtures over
platinum. Combust Flame 2002;128(Issue 4):340e68.
[23] Schefer RW. Catalyzed combustion of H2/air mixtures in a
flat plate boundary layer: II, numerical model. Combust
Flame 1982;45:171e90.