You are on page 1of 23

Nuclear Engineering and Design 240 (2010) 11951217

Contents lists available at ScienceDirect

Nuclear Engineering and Design


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

Some thermal hydraulic challenges in sodium cooled fast reactors


D. Tenchine
CEA/DEN, CEA Grenoble, 17 rue des Martyrs, 38054 Grenoble, France

a r t i c l e

i n f o

Article history:
Received 3 November 2009
Received in revised form 7 January 2010
Accepted 13 January 2010

a b s t r a c t
Sodium cooled fast reactors have been developed in France for nearly 50 years with successively Rapsodie, Phenix and Superphenix plants. Thermal hydraulic challenges have progressively increased with
the power and the size of the reactors. After Superphenix stop in 1997, the thermal hydraulic activity on
sodium cooled fast reactors was drastically reduced for about 10 years. Nowadays, the so-called Astrid
prototype developed in France in the frame of Generation IV deployment requires thermal hydraulic
inputs to support the design and safety studies. This paper aims at summarizing the main thermal
hydraulic challenges in sodium cooled fast reactors, on the basis of the past experience. Numerical
and experimental tools used in the French Atomic Energy Commission (CEA) are briey presented. The
improvement on numerical simulation is emphasized with some examples of recent computations. Of
course, this review is not a comprehensive one as it is mainly based on the author experience. The items
covered in this paper are the subassembly, the core, the upper plenum, the lower plenum, the decay
heat removal, the gas entrainment and the piping. Heat exchanger thermal hydraulics is also briey
mentioned. Several experimental and computed results are presented as simple illustrations without
quantitative information on the data.
2010 Elsevier B.V. All rights reserved.

Contents
1.
2.

3.

4.

5.

6.

7.

8.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1196
Subassembly thermal hydraulics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1196
2.1.
Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1196
2.2.
Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1196
Core thermal hydraulics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1198
3.1.
Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1198
3.2.
Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1198
Upper plenum thermal hydraulics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1199
4.1.
Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1199
4.1.1.
The core outlet region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1200
4.1.2.
The whole upper plenum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1200
4.2.
Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1201
4.2.1.
The core outlet region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1201
4.2.2.
The whole upper plenum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1203
Lower plenum thermal hydraulics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1205
5.1.
Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1205
5.2.
Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1206
Decay heat removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1207
6.1.
Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1207
6.2.
Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1208
Gas entrainment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1209
7.1.
Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1209
7.2.
Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1211
Piping thermal hydraulics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1212
8.1.
Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1212

E-mail address: denis.tenchine@cea.fr.


0029-5493/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.nucengdes.2010.01.006

1196

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

8.2.
Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1212
9.
Others . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1214
10.
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1215
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1216
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1216

1. Introduction

2. Subassembly thermal hydraulics

Sodium cooled fast reactors have been developed in France for


nearly 50 years with successively Rapsodie, Phenix and Superphenix plants as illustrated in Fig. 1. Phenix reactor started in
operation in 1973 and Superphenix design and safety studies were
very intensive since the late 1970s. Superphenix construction was
decided in 1976 and the reactor went critical in 1985. So, many
thermal hydraulic studies were performed to support Superphenix
design and safety analysis. In this period, the strategy was to
develop fast breeder reactors with several Superphenix-2 reactors taking advantage of the Superphenix operating experience
(Gourdon et al., 1990).
Then a second active period started with the launch of the
European Fast Reactor (EFR) project in the late 1980s, in tight collaboration with English and German colleagues (Serpantie et al.,
1989). European colleagues stopped their contribution to the EFR
project in the early 1990s and the project was stopped (Lefevre
et al., 1996). It was the rst signal for a decrease of the fast reactor activity in France too, including of course thermal hydraulic
studies.
In 1997, the decision to stop Superphenix plant was the most signicant event which contributed to reduce drastically the sodium
fast reactor activity in France, except the support to Phenix reactor
(IAEA, 1999). So, we were deeply immersed as shown in Fig. 1
in a long fast reactor desert crossing when a rst wake-up on fast
reactors was initiated by the decision to evaluate the possibility
of gas cooled fast reactors. New challenges for thermal hydraulic
people to deal with gas cooling, gas natural convection, coupling
convection, conduction and heat radiation and other specic problems of gas cooled reactors.
But, the real wake-up for sodium cooled fast reactors in France
was in 2006, with the decision to build a rst Generation IV
prototype by 2020. It was clear that such a short time schedule could be reached only by designing and building a sodium
cooled fast reactor with benet of the past experience and taking into account the Generation IV criteria. The prototype is called
Astrid as mentioned in Fig. 1. New thermal hydraulic challenges
are concerned with the Astrid project to improve safety and
economy of the future Generation IV sodium cooled fast reactors.
The paper presents the thermal hydraulic challenges for sodium
cooled fast reactors based on the past experience in CEA. So,
it is focused on pool type reactors as Phenix, Superphenix and
EFR project. But most topics are common with loop type reactors, even if the design is different. The items covered in this
paper are the subassembly, the core, the upper plenum, the lower
plenum, the decay heat removal, the gas entrainment and the
piping thermal hydraulics. Heat exchangers and steam generators are briey mentioned. For each item, the thermal hydraulic
challenges are recalled. The numerical and experimental tools
used in the CEA past activity are briey presented. The improvement on numerical simulation is emphasized with some recent
examples. The remaining need of validation and qualication is
mentioned.
Several experimental and computed results are presented
as simple illustrations without quantitative information on the
data.

2.1. Needs
Thermal hydraulic analysis of fuel subassemblies is required for
design and safety purposes (IAEA, 2000).
Most of fast reactor subassembly concepts consist in a pin bundle with helical wire spacers. In some cases, grids can be used
instead of helical wires. The pin bundle is surrounded by a hexagonal can and the ow is distributed at the bottom by holes connected
to the cold plenum.
The main parameters to be evaluated are the following ones:
the total subassembly pressure drop including the lower and
upper regions of the subassembly;
the clad temperature and especially the clad maximum temperature;
the hexagonal can temperature for thermo-mechanical analysis.
These previous parameters are required for steady-state conditions
(nominal and partial regimes) and for transient situations (scram
and others). An important safety requirement is the natural circulation case for the passive decay heat removal demonstration.
Other safety situations as pin or bundle deformation, subassembly partial blockage, bubble transport and onset of boiling must
be evaluated.The maximum clad temperature requires an accurate
knowledge of the global and local thermal hydraulics in the pin bundle. One important feature is the transversal ow and the mixing
induced by the wire-wrapped spacer which imposes local thermal
hydraulic coupling between the sub-channels. It is necessary to
take into account three kinds of sub-channels which have different
section areas: triangular in most of the bundle, rectangular along
the hexagonal can and corner ones. The sodium high thermal diffusivity and the local turbulence diffusivity have also a signicant
inuence. The other parameters which inuence the clad temperature are of course the axial and radial power distribution in the pin
bundle.
2.2. Tools
First 3D computations of pin bundle were performed in the
1980s (Basque, 1989). Then, CEA has developed in the 1990s a specic modelling for wire-wrapped pin bundle subassembly based
on sub-channel averaging for pressure, velocity and temperature.
This numerical tool called CADET was developed with the support
of previous analysis in the frame of Phenix and Superphenix studies
(Valentin, 2000). Transport of averaged quantities of mass, momentum and energy between two adjacent sub-channels is based on
analytical and experimental data. The deviated ow rate induced
by the wire is modelled with some assumptions.
For pressure drop evaluation, the comparison of different friction factor models proposed in the literature with the pressure
drop correlation provided by Superphenix subassembly tests has
been done. The best agreement with Superphenix data are obtained
by Rehme (1973) and Cheng and Todreas (1986). The advantage
of Rehme model is its relative simplicity, but the ChengTodreas
model is validated on a larger range of bundle characteristics and
ow regimes. So, CADET is using the ChengTodreas model for

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

1197

Fig. 1. Brief history of sodium cooled fast reactors in France.

the friction factors. A recent survey on friction factors in wirewrapped pin bundle is proposed by Bubelis and Schikorr (2008). The
thermal coupling between two adjacent sub-channels is governed
by three parameters: thermal diffusivity, turbulent diffusivity and
wire effect. An example of CADET computation for a Phenix blanket
subassembly is shown in Fig. 2.
Recently, the sub-channel model has been implemented in the
so-called TRIO U code. TRIO U is the Computational Fluid Dynamic
(CFD) code developed by CEA for 3D thermal hydraulic studies in
components and plena. Besides the sub-channel model used for
design studies, a rened CFD model of the subassembly becomes
possible nowadays to get reference thermal hydraulic behaviour
and detailed data. Such computations provide direct information on
the ow eld, with the effect of helical wires and the inuence of the
hexagonal can. The clad temperature can be calculated everywhere
with a reasonable renement.

A rst rened computation of a Phenix blanket subassembly was performed in 2009 (Bieder et al., 2009). The blanket
subassembly consists in 61 pins with helical wire spacers. The
owrate is 2.85 kg/s, the power is 440 kW and the inlet temperature is 450 C. The mesh is composed of 180.106 nodes
for the whole subassembly with 9 wire spacer rotations. The
computation was performed within a few days on up to 3000 processors of a parallel computer. A global view of the temperature
eld and a local view of horizontal velocity eld are shown in
Fig. 3.
With the unceasingly increase of the computers performance
and the used of massively parallel method in TRIO U code, more
and more rened computations are possible with reasonable time
restitution. So, rened computations can serve as a reference for
the validation of the sub-channel calculations used for parametric
studies. The rened model can be used for studying the thermal

Fig. 2. Outlet temperature in a Phenix blanket subassembly (CADET computation).

1198

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

Fig. 3. Temperature and velocity elds in a Phenix blanket subassembly (TRIO U computation).

hydraulic behaviour in particular situations as partial blockage for


the safety analysis.
Of course, the rened computations need also validation on
experimental data as the mesh size remains limited and the
turbulence models are based on some assumptions. Water and
air test facilities as ESTHAIR air mock-up in CEA (Berthoux and
Carenza, 2008) shown in Fig. 4 are used for the validation of
friction factor and pressure drop calculations. Sodium available
data in forced, mixed and natural convection with local temperature measurements can be used for the qualication of computed
temperature eld and clad temperature. Nevertheless, future subassembly designs which are not included in the range of parameters
already tested should require new experimental data.
3. Core thermal hydraulics

core outlet ow (Betts et al., 1991). Heat transfer with the hexagonal tubes, heat transfer with the outer core bafe and buoyancy
effects must be taken into account in the inter-wrapper ow analysis (Mochizuki, 2007). A thermal stratication should occur and
induce thermo-mechanical stresses on the hexagonal tubes due to
the axial temperature gradient.
A particular situation occurs in the core region when decay heat
exchangers immersed in the hot pool are operating at low ow condition or even in natural circulation. In this case, sodium cooled by
the immersed heat exchangers ows directly in the inter-wrapper
area and contributes to the global cooling of the core, via the heat
exchange with the hexagonal tubes (Kamide, 2001).
So, it appears that the inter-wrapper region plays an important
part in the core design and safety analysis. The physical phenomena
are particularly complicated and require a reliable modelling of the
whole core including the inter-wrapper region.

3.1. Needs
3.2. Tools
The global core thermal hydraulic behaviour must be studied for
design and safety purposes (IAEA, 2000).
From a design point of view, one main target is to obtain as uniform fuel core outlet temperature as possible, although the radial
prole of heat power is largely non-uniform. So, the radial distribution of mass ow rate must be adjusted in each subassembly, taken
into account the pressure drop at the inlet, in the ssile region and
at the outlet of the subassembly. This objective is focused on the
nominal ow condition to optimise the core loading characteristics, but an evaluation of the radial core outlet temperature must
be done for part regimes as well.
Another design requirement is the knowledge of the temperature on the hexagonal tubes for a thermo-mechanical analysis.
This information requires a good evaluation of the inter-wrapper
ow behaviour and the temperature eld in this region. The interwrapper zone may be fed by a by-pass ow adjusted at the bottom
of the subassemblies. Another possibility is that no by-pass ow
is allowed at the bottom of the inter-wrapper area, but a large
ow recirculation should occur from the upper plenum. Due to
the radial pressure prole imposed by the above core structure at
the core outlet, hot sodium will penetrate downwards from the
core outlet region into the inter-wrapper region and interact with
the cold by-pass ow and return upwards to mix with the main

The analysis of the whole core thermal hydraulic behaviour


requires a global core modelling to determine the temperature distribution and the inuence of the inter-wrapper ow. Two levels of
modelling are possible, according to the purpose of the calculation
and according to the accuracy required (Ducros et al., 2008).
The rst level is the system analysis to describe the global
behaviour of the whole plant during transient situations. The core
is modelled with several 1D parallel channels to take into account
the main different core regions, with a possible by-pass channel to
simulate a mass ow rate in the inter-wrapper zone. In France, the
CATHARE system code developed by CEA in collaboration with
AREVA, EDF and IRSN is used for system analysis of all kind of
reactors including sodium cooled fast reactors.
The second level is the 3D global core modelling with a subchannel description of each subassembly and a 3D model of the
inter-wrapper space. The past tool used at CEA was included in
the ELOGE global core modelling, with the CADET sub-channel
model for the subassemblies (described in Section 2.2) and the former TRIO VF code of the inter-wrapper space modelled in a 2D
approach.
The CADET sub-channel model of subassembly is now replaced
by TRIO U to have a unique tool for the whole core thermal

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

1199

Fig. 4. ESTHAIR air facility for measurement of pressure drop and heat transfer correlations.

hydraulic modelling. Another possibility is to couple a global core


model with a rened CFD description of one subassembly. This
coupling process allows for example a detailed description of
the hottest subassembly in tight connection with the global core
behaviour. So, it becomes possible during transient regimes to look
for the local hottest clad temperature at the same time as the whole
core behaviour is calculated.
Fig. 5 shows a TRIO U core calculation using a sub-channel
model for the subassemblies, coupled with the 3D inter-wrapper
space and connected to the upper and lower plenum. Fig. 6 shows
a TRIO U core calculation with one special plate type subassembly modelled with a rened mesh to calculate local velocity and
temperature elds in case of local blockage.
The qualication of whole core calculations requires experimental data taking into account the thermal hydraulic interactions
between the subassemblies, the inter-wrapper ow and the upper
and lower plenum. Some water tests were performed in the past
to study the inter-wrapper ow in connection with the plenum
pressure effect (Serpantie et al., 1989). But, water facilities cannot simulate correctly the inuence of heat transfer through the

hexagonal tubes and the outer core bafe. A few sodium tests are
available on the thermal interaction between subassemblies and
inter-wrapper ow (Kamide, 2001). These tests are performed with
seven subassemblies and they are focused on decay heat removal
situations with cold sodium owing downwards from decay heat
exchangers immersed in the upper plenum. Further experimental programs in sodium are needed to improve the qualication of
global core calculations.
4. Upper plenum thermal hydraulics
4.1. Needs
Thermal hydraulics of the upper plenum is frequently considered as one of the noblest part of the fast reactor thermal
hydraulic activity as many challenges are concentrated in this
region (Francois et al., 1990). We describe hereafter a list of problems which require thermal hydraulic analysis for design or safety
purpose (Azarian et al., 1990). Fig. 7 gives a schematic view of these
problems.

Fig. 5. Core temperature and velocity elds (TRIO U computation).

1200

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

Fig. 6. Core with a partially blocked plate subassembly (TRIO U computation).

4.1.1. The core outlet region


Starting at the core outlet level, an important safety requirement
is the validation of the core outlet temperature measurements.
Thermocouples are located above each fuel subassembly for nominal core monitoring and also to detect any abnormal increase of
temperature. So, they can play a key role in the prevention of reactor accident due to a ow blockage for instance. They also allow
the analysis of the subassembly outlet temperature during transient regimes. As they are the only temperature measurements to
follow the core thermal hydraulic behaviour, it is quite important
to be sure that the thermocouples measure the effective subassembly outlet temperature for all operating conditions. This question
is raised because the thermocouples are generally located a few
centimetres above the core outlet level, as they are mounted on special devices supported by the above core structure. The measured
temperature is sensitive to local characteristics of the subassembly
upper region. The distance from the top of the subassembly to the
thermocouple and the possible radial offset may inuence the measurement as well. In nominal operating condition, the core outlet
ow is deviated by the above core structure and a radial centrifugal
ow occurs which induces a centripetal radial ow just above the
external subassembly heads. So, colder sodium coming from the
external zone can inuence the temperature measurements at the
outlet of the outer fuel subassemblies. Another disturbing effect
may occur at the outlet of subassemblies located at the vicinity
of control rod assemblies were colder sodium is owing and can

Fig. 7. Upper plenum thermal hydraulic challenges.

be entrained towards the thermocouples. Of course, local design


features are very important and some ideas can be proposed to
reduce or to avoid these disturbing effects. Anyway, a rened thermal hydraulic analysis is needed to conrm the reliability of all the
core outlet temperature measurements.
Again at the core outlet level, another question frequently raises
up: is there a risk of thermal fatigue due to temperature uctuations (Kim et al., 1999)? The specic areas described here above
are located in potential mixing zones between ows at different
temperatures. Temperature uctuations of high amplitude (several tenths of Celsius) can be produced by the mixing process and
their range of frequency can be rather large (1100 Hz). So, thermal
fatigue may occur and induce thermal striping at long term on the
concerned structures such as subassembly heads, thermocouples,
supporting grid and so on.
4.1.2. The whole upper plenum
An important component is the above core structure which is
supporting the control rods and the instrumentation at the core
outlet. The subassembly jets at high velocity and the radial ow
below the above core structure may induce vibration. A detailed
velocity and pressure eld with information on the uctuating values is needed to evaluate the risk of ow induced vibration. The
porosity of the above core structure is a parameter which will inuence the thermal hydraulic behaviour in the component itself, in the
above core region and in the whole upper plenum. Moreover, as we
have described in Section 4.1.1, it will have a signicant effect on
the recirculating ow in the inter-wrapper region.
At the sodium free surface, oscillations should occur as a consequence of the hydraulic behaviour in the upper plenum. As designer
try to reduce the vessel dimensions for a given power and a global
ow rate, the free surface oscillations become more signicant.
These oscillations can induce thermal fatigue on the vessel and
other internal structures due to the important temperature gradient at the free surface. So, an objective of the upper plenum thermal
hydraulic studies is to estimate the thermal gradient oscillations at
the free surface and to test various options to reduce the risk of
thermal fatigue.
Another important problem at the free surface is the risk of gas
entrainment. This topic is described in detail in chapter 7 which
concerns the gas behaviour in the primary circuit.
The region of the upper plenum located along the internal wall
which separates the upper plenum and the lower plenum may be
concerned by a signicant temperature gradient and possible thermal stratication. Thermal stratication should occur if a low ow
or a recirculating ow area is arranged, depending on the design
features. If a thermal stratication is present, the thermal interface may be unstable and thermal fatigue can be induced on the
neighbouring structures.

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

The clad failure detection requires a reliable system to avoid a


large quantity of ssion products to be dispersed in the primary
circuit. In Superphenix reactor, the ssion products detectors were
placed outside the vessel but alternative concepts were evaluated
with detectors placed near the intermediate heat exchangers inlets
(Trapp et al., 1990). In both cases, the thermal hydraulic behaviour
of the upper plenum including the above core structure is an
important parameter to determine the transfer function between
the subassembly and the detector. The transit time and the signal attenuation are directly dependent on the velocity eld and
the turbulence characteristics in the upper plenum (Natesan et al.,
2007).
During some transient situations with decrease of ow rate and
temperature evolution at the core outlet, buoyancy forces may
modify the ow pattern in the above core region, in the above core
structure and in the whole upper plenum (Ieda et al., 1990). During a cold shock at the core outlet (scram for example), a thermal
stratication should occur at a level located at the top of the intermediate heat exchanger window (for a pool type reactor) or over
the hot pipe inlet (for a loop type reactor). This thermal stratication can produce thermal stresses on the internal structures which
must be evaluated on the basis of the temperature gradient evolution (Tanaka et al., 1990). Let us have in mind that the thermal
hydraulic behaviour inside the above core structure may be different than the one in the plenum. The porosity and the detailed
design of the above core structure will affect directly the thermal
stresses on its various structures.
Some particular transient regimes concern the decay heat
removal situation. The use of decay heat exchangers immersed
in the upper plenum generates specic ow behaviour. A special
case is the natural convection situation with a tight coupling of
core, plenum and components thermal hydraulics. These topics are
described in detail in chapter 6 which concerns decay heat removal
situation.
4.2. Tools
4.2.1. The core outlet region
The core outlet region is concerned with two main challenges:
the reliability of the subassembly outlet temperature measurements and the risk of thermal fatigue in mixing areas at different
temperatures.
The numerical analysis of such phenomena requires a rened
modelling of the concerned region. As described above the local
thermal hydraulic behaviour may be inuenced by the global complicated ow between the core and the above core structure. So, the
rst step of analysis should cover the whole above core region, with
boundary conditions to be dened at the core outlet and beyond the
above core structure external radius. Such a global evaluation of the
above core region thermal hydraulic behaviour will use a coarser
mesh than the one required for local analysis of temperature mea-

1201

surements and thermal fatigue. Then, a second step of analysis will


consist in a rened mesh of the concerned area, with boundary
conditions given by the rst calculation of the whole above core
structure region. But this method is limited by the fact that the
instability of the velocity and temperature elds is very important,
particularly for thermal fatigue analysis which requires detailed
knowledge of the temperature uctuations. So, it may be difcult
to dene non-stationary boundary conditions for the rened local
calculation based on the coarser mesh global calculation. Moreover, there may be a feedback from the local thermal hydraulic
behaviour to the global one that cannot be taken into account with
this method. So, another approach consists in a simultaneous calculation with locally rened meshes as a rst attempt is presented
in Fig. 8. Three main options are opened: the xed mesh renement, the adaptative mesh renement and the multi-grid method.
Of course, for a given total number of meshes, the rst approach
may limit the mesh renement. We will not describe here the
advantages and drawbacks of the last two approaches, but the three
methods have the common advantage to ensure a global coherence
for non-stationary calculations at various scales. The use of Large
Eddy Simulation modelling is recommended for the estimation of
the amplitude and frequency range of the temperature uctuations.
In the 1990s for the European Fast Reactor project, the performance of computers limited the numerical approach to about
one hundred thousands meshes. The global thermal hydraulic
behaviour of the core outlet region was calculated with a twodimensional model or a coarse mesh three-dimensional model.
The local analysis was limited to a small area in order to have
rened meshes, with the difculty described previously to dene
coherent boundary conditions. So, the reference approach for these
problems in the 1990s (and before) was the experimental one.
Nowadays, computations with one hundred million of meshes are
performed and the computational time can be acceptable by using
massively parallel techniques. Anyway, the use of experimental
facilities probably remains very useful for the qualication process
and the rapid test of design adjustments.
From an experimental point of view, the thermal hydraulic core
outlet region can be simulated by reduced scale mock-ups using
water. Two main questions arise when designing such a water
model: which similarity condition and which scale? We will not
enter here in a detailed analysis of similarity criteria, but we can
remind some qualitative aspects (Grewal and Gluekler, 1982). As
far as high Reynolds turbulent ows are concerned in the above
core region, water can be used even if the water Prandtl number is much higher than the sodium Prandtl number. The sodium
high thermal diffusivity is not a primordial parameter when looking at the turbulent mixing behaviour in the uid. But the transfer
function of the temperature uctuations from the sodium to the
structure cannot be correctly simulated by water. So, a water model
can be acceptable to verify the reliability of core outlet temperature
measurements. To evaluate the risk of thermal fatigue on specic

Fig. 8. Core coupled to a simplied upper plenum (TRIO U computation).

1202

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

Fig. 9. JESSICA water test facility of EFR upper plenum.

structures, the water model will provide information on the amplitude and frequency range of the temperature uctuations in the
uid. The water model will also be useful to test design evolutions in
order to eliminate the problem. The scale of the water model must
be sufciently high to be able to represent detailed features such
as the subassembly heads, the grids, the thermocouples and their
supporting structures and so on. The water tests must be performed
at a Reynolds number sufciently high to simulate the same ow
regime as the reactor one. A scale of 1/5 or 1/3 may be required, but

a limited sector may be sufcient to have a representative thermal


hydraulic behaviour of the core outlet region for local studies. Such
a compromise can reduce the need of very large ow rate in the
water facility. If nally an accurate knowledge of the temperature
uctuations on the structures is required, sodium experiments may
be needed to take into account the inuence of the sodium thermal
diffusivity.
In the 1980s, we have built in CEA a water facility called PERIGEE
for studying the thermal hydraulic behaviour of the core outlet
region and the whole upper plenum of Superphenix reactor. This
facility was a 90 sector at a reduced scale of 1/5, without any simulation of the inter-wrapper region. In the 1990s, we have built a new
water facility called JESSICA for studying the core outlet region of
the European Fast Reactor. This facility illustrated in Fig. 9 was a 90
sector at a reduced scale of 1/3, with a representation of the interwrapper region to estimate the recirculating ow and its inuence
on the core outlet behaviour.
In parallel to these representative water tests, some analytical
facilities were used to validate the CFD computations and to justify
the similarity criteria when a simulant uid is used. For the mixing jets conguration encountered at the core outlet, two identical
test sections were built, one in air called AIRJECO and the other
one in sodium called NAJECO (Tenchine and Moro, 1997), with a
basic conguration of two co-axial jets at different temperatures
as shown in Fig. 10. The purpose was to compare the elds of temperature with a particular interest in the temperature uctuations
for the thermal fatigue analysis. Some typical results are presented
in Figs. 11 and 12. They illustrate that the air tests can predict
the evolution of the mean temperature and the uctuation intensity in sodium, providing that the Reynolds number is sufciently
high (Tenchine and Moro, 2000). The evaluation of the transfer of
temperature uctuations from the uid to the structure requires
experimental data in sodium, as the high thermal diffusivity of
sodium plays an important role in the transfer function. Recent
sodium data are provided by JAEA on a sodium test facility with
jets mixing near a wall (Kimura et al., 2007). Mean temperature
and temperature uctuations measured in the JAEA experiment

Fig. 10. AIRJECO and NAJECO characteristics.

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

1203

Fig. 11. Comparison of mean temperature.

Fig. 14. Comparison of TRIO U computations and JAEA experimental data.

presented in Fig. 13 are compared with CFD computations using


TRIO U code as presented in Fig. 14. The amplitude and the frequency characteristics of the temperature uctuations are analysed
and they show a rather good agreement between numerical and
experimental results.

4.2.2. The whole upper plenum


The thermal hydraulic behaviour of the above core structure
must be predicted for steady-state and transient situations. The
ow distribution through the above core structure, the transport of
uid between the above core structure and the plenum, the temperature gradients and the possible thermal stratication in the
component during transient regimes must be evaluated.
During Superphenix project and during the European Fast Reactor project, the experimental approach was the unique way to
estimate the thermal hydraulic behaviour of the above core structure. The very complicated internal geometry of this component
including the control rods, the instrumentation, grids and so on
led to a quasi impossibility to use a predictive numerical approach.
So, the same water test facilities as for the above core region presented previously were used, at a reduced scale between 1/8 as
COLCHIX experiment presented in Fig. 15 and 1/3 as JESSICA experiment illustrated in Fig. 9. These facilities were designed to enable
the measurement of the ow distribution in the above core structure in steady-state condition. They also allowed the simulation
of transient regimes with imposed core ow rate and temperature evolutions. So, the temperature evolution in the above core
structure was measured at the same time as the plenum temperature evolution. These simultaneous measurements were used to
estimate the thermal stresses on the above core structure external
shell. One had to pay a particular attention to the similarity criteria used for designing such facilities. As far as the steady-state
ow distribution was concerned, the ratio on Reynolds number

Fig. 12. Comparison of temperature uctuations.

Fig. 13. JAEA mixing jet experiment.

Fig. 15. COLCHIX water model of EFR upper plenum.

1204

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

Fig. 16. TRIO U computation of CORMORAN sodium experiment: velocity (left), mean temperature (left) and temperature uctuations (right).

between the mock-up and the reactor should be high enough to


stay in the same ow regime as in the reactor. For transient regimes
with signicant buoyancy inuence, Richardson number must be
equal in the mock-up and in the reactor. The ratio on Peclet number is directly related to the ratio on Reynolds number, with the
sodium Prandtl number which is about one thousand smaller than
the water Prandtl number. For example, a water model at a scale
of 1/5 based on Richardson number similarity, will have a ratio
on Reynolds number of about 1/100 and a ratio on Peclet number
of about 10. It is well known that Peclet number is the governing
parameter for heat transfer correlations in sodium forced convection regimes. So, the ratio on Peclet number of 10 in a water test may
introduce an over-estimation of the Nusselt number. As far as the
water tests are not used to estimate local heat transfer correlations,
the discrepancy on Peclet number is acceptable. Water facilities
will give velocity and temperature elds in the uid, including the
buoyancy inuence, but the wall temperatures must be estimated
a posteriori by use of an adapted sodium correlation (Costa and
Tenchine, 1995).
The similarity criteria used in water tests for mixed convection
situations was a topic for basic research supported by experimental programs called SUPERCAVNA and CORMORAN, including water
and sodium tests (Vidil et al., 1988). In particular, some generic congurations were studied with a channel ow connected to a cavity
heated (or cooled) at one wall. So, forced convection and mixed
convection regimes have been studied. The occurrence of thermal
stratication was analysed, the interface characteristics (gradient
and uctuation) were studied in detail. Sodium and water results
have been compared to verify the similarity criteria and to validate numerical calculations as shown in Fig. 16 for the CORMORAN
sodium experiment.
One signicant conclusion was that the temperature gradient
at the stratication interface is higher in water than in sodium.
This discrepancy may be due to the Prandtl number inuence at
the interface where the high thermal diffusivity of sodium plays
an important role in the gradient characteristics. This conclusion
was taken into account in the design studies considering water
results for thermal stresses evaluation on the structures. During

the commissioning tests of Superphenix plant, available measurements have been recorded to bring a global qualication of the
predicted results (Francois et al., 1990).
Nowadays, the numerical approach becomes efcient for studying the upper plenum thermal hydraulic behaviour in steady-state
and transient situations (Ducros et al., 2008) as shown in Fig. 17.
The above core structure remains a complicated component which
requires special modelling. As long as we cannot model precisely
all the internals with an acceptable meshing, a porous medium
approach can be used and an experimental validation is needed.
This porous medium model of the above core structure must be
connected to a continuous model of the upper plenum. In the future,
it might be possible to produce a detailed modelling of the whole

Fig. 17. Upper plenum velocity eld with free surface disturbance (TRIO U computation).

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

1205

upper plenum, including the above core structure. Anyway, numerical tools can already predict the upper plenum global thermal
hydraulic behaviour during transient situations. The occurrence
of thermal stratication and the propagation of the stratication
interface are reasonably well predicted. But, the estimation of the
temperature gradient requires a locally rened mesh, with the
difculty induced by the propagation of the interface. A dynamic
adaptative meshing method may be needed to deal with this problem. A Large Eddy Simulation modelling can be used to access the
temperature uctuations at the interface for the thermal fatigue
analysis.
For the free surface behaviour and the gas entrainment problem,
water test facilities have been used for the European Fast Reactor
project. This topic is described in detail in Section 7.
For decay heat removal situations, the possible use of decay heat
exchangers immersed in the upper plenum requires a modelling of
the interaction between the core and the plenum. Numerical and
experimental methods for that situation are described in Section
6.
5. Lower plenum thermal hydraulics
5.1. Needs
Usually, thermal hydraulics of the lower plenum is not considered as noble as the upper plenum one, but several important
concerns must be analysed to avoid severe thermal stresses on the
structures (IAEA, 1999). For pool type reactors, the lower plenum
generally includes the cold pool receiving the cold sodium from
the intermediate heat exchangers, the primary pumps, the diagrid
located just below the core, the core support structure, the bottom of the vessel with an internal core catcher and a cooling vessel
by-pass ow. The lower plenum is including a low pressure region
upstream to the pumps and a high pressure region downstream to
the pumps (Serpantie et al., 1989).
In steady-state condition, thermal stresses and thermal fatigue
may be induced by the temperature gradient at the outlet of the
intermediate heat exchangers. This vertical gradient is induced
by the ow pattern in the primary side of the heat exchanger.
The sodium which penetrates into the central region of the heat
exchanger is colder than the sodium remaining in the external
region, mainly due to the longer path and consequently the longer
heat exchange surface with the secondary side. The colder sodium
ows out in the lower part of the heat exchanger outlet window, as
the hotter sodium ows out in the upper part of the outlet window.
The temperature gradient may reach several tenths of Celsius and
produce temperature uctuations. So, thermal stresses and thermal
fatigue may be induced on the vessel or the structures concerned
by the heat exchanger outlet jet. Fig. 18 gives a schematic view of
the problem.
The region of the lower plenum located along the internal wall
which separates the upper plenum and the lower plenum
may be concerned by a signicant temperature gradient and possible thermal stratication. Thermal stratication should occur if
a low ow or a recirculating ow area is arranged, depending
on the design features. If thermal stratication occurs, the thermal interface may be unstable and thermal fatigue can be induced
on the neighbouring structures. The problem is similar to the one
described previously for the lower region of the upper plenum. But,
of course, the specic design arrangement and ow pattern in the
lower plenum has to be taken into account to evaluate the thermal stresses on the structures. Fig. 18 gives a schematic view of the
problem.
Apart from the nominal condition, other steady-state situations
may occur and require thermal hydraulic analysis. An important
case is the non-symmetrical situation when one secondary circuit

Fig. 18. Lower plenum thermal hydraulic challenges.

is not operating. For instance, Phenix reactor has been operating


in non-symmetrical condition for the last years. The power was
reduced with only two secondary circuits in operation instead of
three. So, the two intermediate heat exchangers connected to the
non-operating loop were replaced by closed components. The three
primary pumps were still operating during this period. In such a
situation, the thermal hydraulic behaviour of the cold pool is nonsymmetrical and particular thermal stresses may occur on the inner
structures.
As mentioned for the upper plenum, the transient regimes
can induce large change in the lower plenum thermal hydraulic
behaviour. Buoyancy forces can play a signicant role in many transient situations, especially when the primary ow rate is reduced
and the intermediate heat exchanger outlet temperature is largely
modied. In the case of a hot shock (loss of secondary heat sink
for example), the jet at the outlet window of the intermediate
heat exchanger is raising by buoyancy and cold sodium remains
stratied in the bottom of the pool. A thermal gradient occurs at
the stratication interface, with a possible time evolution which
must be estimated for thermal stress analysis. In the case of a cold
shock (which can occur after a hot shock as described previously,
followed by a scram), the jet at the outlet window of the intermediate heat exchanger is oriented downwards and cold sodium
feeds the bottom of the pool. A thermal stratication will occur in
the upper part of the lower plenum, with an interface rising slowly
above the level of the heat exchanger window. Such a transient can
happen for non-symmetrical operating conditions, as it was the
case during the Phenix ultimate tests with a secondary pump trip
simulation.
The lower plenum is including the core support structure as
well. Depending on the reactor design, the core support structure may be cooled by a by-pass ow used for the main vessel
cooling. During transient situations, buoyancy forces can affect
considerably the thermal hydraulic behaviour of the core support
structure where the ow rate is small. As this structure plays an
important role in the core integrity, it is important to estimate
the thermal stresses on it. The evolution of the velocity and temperature elds should be studied for representative hot and cold
shocks.

1206

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

Fig. 19. Phenix cold pool calculation during a secondary pump trip (TRIO U computation).

5.2. Tools
For the European Fast Reactor project in the early1990s, it
appeared that the rened analysis needed for the thermal fatigue
investigation at the intermediate heat exchanger outlet was at the
limit of the computational performance at this time. Effectively,
the radial jet characteristic at the outlet window is dependent
on the ow behaviour in the heat exchanger itself. The global
direction of the jet is inclined downwards, with a maximum
velocity located in the lower part of the window. The vertical
temperature prole is quite heterogeneous, with a much higher
temperature in the upper part than in the lower part. At least,
the mixing process which determines the thermal stresses on the
structures is dependent on the turbulence characteristics at the
heat exchanger outlet. So, the best way to predict the thermal
hydraulics of the heat exchanger outlet region is to couple in the
same model a part of the heat exchanger bundle with the lower
plenum. From a numerical point of view, this kind of computation is now possible. The use of a Large Eddy Simulation model is
recommended to estimate the amplitude and the frequency range
of the temperature uctuations. From an experimental point of
view, the use of water mock-up is limited as it is difcult to simulate with water the tube bundle thermal hydraulics and the heat
exchange with structures strongly dependent on the Prandtl number. Another way for the qualication of numerical results is the
use of sodium data. In Phenix reactor, a pole of thermocouples
has been installed in the cold pool along one intermediate heat
exchanger. This instrumentation has provided information on the

vertical temperature prole for steady-state and transient situations.


The thermal stratication in the upper part of the cold pool in
steady-state conditions can be estimated by a numerical approach.
It requires a modelling of at least a sector of the cold pool, taking into
account the heat transfer through the internal walls separating the
cold pool with other regions at different temperatures. If the oscillations of the stratication interface must be estimated for a thermal
fatigue analysis, the use of a Large Eddy Simulation model is recommended. In the case of a non-symmetrical situation, a larger sector
or the whole plenum should be modelled to evaluate the angular
ow redistribution. Looking at the global pool behaviour, the coupling with the heat exchanger bundle may be not required. But, the
boundary conditions imposed at the heat exchanger outlet window must take into account the specic velocity and temperature
proles provided by a separate heat exchanger computation. Fig. 19
shows a pre-calculation with the TRIO U code of a non-symmetrical
transient situation in Phenix reactor. It corresponds to a secondary
pump trip in one loop of the Phenix reactor. At the beginning of the
transient, the 3D effects appear clearly in front of the heat exchangers connected to the loop concerned with the pump trip. It can be
seen that the thermal stratication induced by the buoyancy forces
leads progressively to a quasi-symmetrical behaviour in the cold
pool.
During the ultimate tests of Phenix reactor in 2009, a secondary
pump trip was performed to verify the numerical computations for
this important non-symmetrical situation. A new pole of thermocouples had been installed previously in the cold pool to measure

Fig. 20. Phenix cold pool measurement during a secondary pump trip.

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

1207

A particular topic concerns the lower plenum in case of severe


accident with core melt down and spreading on the core catcher.
Natural convection heat transfer around the core catcher was
studied in a rst step on water mock-ups (Tenchine, 1978) and
conrmed on sodium facilities (Tenchine and Amblard, 1983).
6. Decay heat removal
6.1. Needs

Fig. 21. COCO water model of EFR lower plenum.

the vertical prole of temperature at the position of one nonoperating intermediate heat exchanger. Fig. 20 shows a recording
of this pole of thermocouples during the secondary pump trip. The
initial thermal stratication in the pool appears clearly. Then, the
hot shock is due to the secondary pump trip and the following cold
shock is due to the scram.
For transient situations, numerical simulation is largely used
to estimate the evolution of the ow and temperature elds in
the cold pool. Of course, buoyancy forces should be modelled as
they play an important role as described previously. As for the previous steady-state global simulation, the coupling with the heat
exchanger bundle may be not required at a rst stage of analysis.
But, during the transient, the evolution of the velocity and temperature proles at the heat exchanger outlet due to buoyancy
effect in the bundle may be signicant. In a rened analysis, the
coupling between the heat exchanger and the cold pool should be
necessary. From an experimental point of view, the use of a water
mock-up is possible for a rst step of qualication (Padmakumar et
al., 2007). The Richardson similarity is required as buoyancy inuence is important. The compromise between the ratio on Reynolds
number and the ratio on Peclet number must be evaluated as
described earlier for the upper plenum test facilities. Fig. 21 shows
a water test facility called COCO used for the European Fast Reactor
project. It was a 1/10 reduced scale model where various transient regimes were simulated. The heat exchanger tube bundle was
partly represented to provide a reasonable velocity prole at the
heat exchanger outlet.
Concerning the lower plenum, numerical and experimental
studies were performed during the Superphenix and the EFR
projects. The numerical approach was difcult as the core support
structure had a complicated design with a partition in connected
volumes. Each volume was fed through a hole at the top of the structure. During transient situations, the buoyancy inuence was very
important as the ow rate in the lower plenum is low. The evaluation of the ow redistribution with buoyancy inuence was a real
challenge. So, water test facilities were used in parallel to verify
the computations. As for the upper plenum and the cold pool, the
water tests were performed at the same Richardson number as in
the reactor to represent correctly the buoyancy effects (Vidil et al.,
1988). The compromise between the ratio on Reynolds number and
the ratio on Peclet number was particularly difcult as the low ow
rate could lead to locally laminar ow regime instead of turbulent
one.

Decay heat removal is a major challenge for all types of nuclear


reactors. For sodium cooled fast reactors, passive decay heat
removal based on natural convection is possible. This is one of
the important advantage of these reactors. Practically, active and
passive decay heat removal systems can be used depending on
the situation and the safety requirements. So, many studies have
been performed on this topic and various systems have been
designed. In Superphenix plant, decay heat could be removed by
sodiumair heat exchangers connected to each secondary circuit.
Sodiumsodium heat exchangers immersed in the hot pool and
connected to sodiumair heat exchangers were also available for a
direct cooling of the primary circuit. In the EFR project, the main
strategy for decay heat removal was still based on direct reactor
cooling (DRC) systems (Serpantie et al., 1989), with immersed coolers in the hot pool connected to sodiumair heat exchangers as
shown in Fig. 22. Three DRC systems could operate in forced convection with electromagnetic pumps on the secondary circuit and
three other DRC systems had to operate in natural convection. In all
cases, the air circuit was designed to operate in natural convection
with a sufciently high air stack. Each DRC system was designed
to remove 15 MW, so that the total heat removal capability was
90 MW (2.5% of the nominal power).
On the primary side of DRC system, the interaction between the
immersed coolers and the hot pool is the main thermal hydraulic
problem. Several ow regimes must be considered corresponding
to different operating conditions. For decay heat removal situation
with primary pumps still operating, a mixed convection regime
occurs in the hot pool with cold sodium owing downwards from
the decay heat exchangers. There is a direct interaction on the global
ow behaviour in the hot pool. The cooled sodium can ow inside
the core inter-wrapper space and contribute to a direct cooling of
the core through the inter-wrapper of the subassemblies tubes. The
whole transient procedure including the initial incident and the
scram must be analysed as it can induce an initial thermal stratication in the upper plenum which will inuence the later interaction
between the immersed coolers and the hot pool. For decay heat
removal with primary pumps stopped, a natural convection regime
occurs in the primary circuit. Again, the whole transient procedure
including the initial incident and the scram must be analysed. The
behaviour of the DRC system operating in natural convection is a
key point to demonstrate the reliability of the systems in case of
total plant black out for example. One has to keep in mind that
the immersed coolers do not contribute to the natural circulation
through the core. Immersed coolers will cool the lower part of the
hot pool and contribute to a direct cooling of the core through
the inter-wrapper tubes of the subassemblies. So, it appears that
the design and the position of the immersed coolers are important
parameters to optimise the DRC system efciency.
On the secondary side of DRC system, the ow regime may be
forced convection with electromagnetic pumps or natural convection. In the second case, the main challenge is the onset of natural
convection directly related to the onset of air natural convection
in the tertiary side. It may be necessary to manage a small heat
release through the DRC system even in reactor nominal operating
condition to ensure an initial natural circulation in the secondary
and tertiary sides. There is also a risk of sodium freezing inside

1208

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

Fig. 22. EFR vessel and decay heat removal system.

the sodiumair heat exchangers. So, it is important to study the


behaviour of the whole system when operating in natural convection, for the various possible initial situations.
One can notice that we have focused on DRC system for pool
type reactor as it was operating in Superphenix and it has been the
main system studied for the EFR project. But another option could
be a primary reactor auxiliary cooling system (PRACS) located in a
position which allows natural circulation through the core: in the
upper part of the intermediate heat exchangers for instance.
6.2. Tools
A rst level of analysis of the decay heat removal systems
can be done with system codes. A whole modelling of the reactor is needed including the primary circuit, the secondary circuits
and the decay heat removal circuits. All kind of transient situations must be simulated and these calculations are very helpful to
dene the procedures and to evaluate the efciency of the decay
heat removal systems. During the Superphenix project, the DYN
code was developed and used to analyse the reactor behaviour
in transient situations including decay heat removal situations.
Then, during the Superphenix commissioning tests, some transient
regimes were performed to verify the reliability of the computations. For the present so-called Astrid prototype, the CATHARE code
is used in France (Geffraye et al., 2009), taking advantage of the previous experience and validation of the DYN code. A complementary
verication of CATHARE code for natural convection regime is in
progress on the basis of Phenix ultimate tests performed in 2009.
One test consisted in a scram followed by a trip of the three primary pumps as shown in Fig. 23. The onset of natural convection
in the primary circuit was obtained during several hours with only
heat losses of the secondary loops as heat sink. This test is used
for an international benchmark exercise in the frame of an IAEA
Coordinated Research Project.
The system codes generally using a 1D approach, some
difculties may appear when 3D phenomena occur due to nonsymmetrical situations or local buoyancy effects. Non-symmetrical
situations can be induced by an incident on one loop or the use
of only part of the decay heat removal systems. Buoyancy effects

clearly inuence the thermal hydraulic behaviour in decay heat


removal situations where the ow rate is reduced and the temperature evolution may be important. In each plenum, the expected
phenomena have been described previously. In the circuits, thermal
stratication can occur in horizontal pipes and inuence the global
transfer function of the system. It may also induce severe thermal
stresses on the pipes. So, computational uid dynamics is needed
to estimate theses 3D phenomena. The upper plenum behaviour is
certainly the most complicated one, with the effect of the immersed
coolers and the tight coupling with the core inter-wrapper region
as described previously. Improvement of CFD codes is still needed
to model correctly all these coupled phenomena. The development
of a dynamic coupling between a porous medium approach for the
core region and an open medium approach for the plenum is still in
progress in CEA with the TRIO U code. Another important development is the dynamic coupling of system and CFD codes to take into
account 3D effects on the global system behaviour during transient
situations (Ducros et al., 2008). For instance in CEA, CATHARE and
TRIO U codes can be coupled to perform computations where the
whole plant is modelled with CATHARE and the core and the upper
plenum are modelled with TRIO U.
An experimental data bank for the validation of system and
CFD computations is needed. The experimental approach has been
largely used to study the decay heat removal situations for the EFR
project. German colleagues at FzK have performed a comprehensive experimental study on the RAMONA facility (Hoffmann, 1989).
RAMONA was a 1/20 scale water model of the upper plenum with
boundary conditions imposed at the core outlet and with active
immersed coolers. Transient situations were tested and the interaction between the cold sodium owing downwards from the coolers
and the hot sodium owing upwards from the core was analysed in detail. Several types of immersed coolers were tested and
non-symmetrical situations were simulated. A larger water model
NEPTUN at a scale 1/5 was also used to estimate the scale effect
on the results for a better extrapolation to the reactor (Weinberg
et al., 1995). Another water model called KIWA at a scale 1/10 was
also built at FzK to simulate the whole decay heat removal circuit
including a simplied reactor vessel (2D slab model), the secondary
side and the air cooler.

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

1209

Fig. 23. Phenix natural convection test.

A few sodium experiments are available on the upper plenum


behaviour when decay heat exchangers are operating. JAEA has performed valuable tests on the PLANDTL sodium facility (Nishimura
et al., 2000). This facility could simulate seven subassemblies with
an inter-wrapper ow, the upper plenum with an immersed cooler
and the primary loop with another decay heat exchanger located at
the top of the main heat exchanger. Besides this limited amount of
sodium test facilities, reactor tests simulating decay heat removal
situations were performed in Phenix and Superphenix reactors, as it
was said previously, including natural convection tests for various
scenarios. Of course, reactor tests are very useful for a global code
qualication, but these data are not sufcient for a full validation.
Other sodium experiments were performed to study the loop
behaviour in decay heat removal situation. The JANUS facility
shown in Fig. 24 was built in CEA to study the onset of natural
convection in a loop and the occurrence of thermal stratication in
horizontal pipes at low ow condition (Tenchine and Grand, 1989).
The ILONA large sodium facility was built by Siemens to test the
efciency of sodiumair heat exchangers in realistic conditions.

ties are increased and the free surface is disturbed. As for Phenix
and Superphenix reactors the free surface was smooth. A disturbed
free surface was expected in the European Fast Reactor project with
higher compactness. Up to now, the main source of gas entrainment
at the free surface is the presence of vortices. Such vortices are
induced by concomitant local vorticity and downward ow. This
is an unsteady complicated process with creation, transport and
vanishing of vortices, with or without bubble entrainment in the
ow. The shape of the vortices and the occurrence of small or large

7. Gas entrainment
7.1. Needs
Gas entrainment in the primary circuit of sodium cooled fast
reactors may lead to safety problems in case of accumulation and
transport of gas through the core. A positive reactivity effect should
occur if a large quantity of gas is crossing the core. So, this scenario
is not acceptable and one has to demonstrate that no large quantity
of gas should accumulate and cross the core.
Other problems can be induced by the presence of gas bubbles
in the primary circuit, such as the pump behaviour and the perturbation of possibly ultrasonic measurements. Various potential
sources of gas exist as free surfaces, overows, nucleation, leakage
of a gas seal, as shown in Fig. 25. At the same time, various sinks of
gas also exist as free surfaces and dissolution.
Gas entrainment at the free surface is one important potential
source of gas. At rst, gas can be dissolved in the sodium at the free
surface, but the quantity of gas is generally limited. As the reactor
compactness is an economical objective, the ratio volume/power
is designed as low as possible, so that the upper plenum veloci-

Fig. 24. JANUS sodium facility.

1210

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

Fig. 25. Gas entrainment problems in a pool type fast reactor.

bubbles are depending on the local velocity eld and the turbulence characteristics, as shown in Fig. 26. In parallel to a necessary
physical understanding and modelling of this complicated process,
design adaptations must be studied to avoid or at least to reduce
large bubble entrainment.
Another source of gas entrainment in the primary circuit is the
existence of overows as in Superphenix plant. In such reactor, a
small by-pass ow is arranged at the core inlet to cool the lower
plenum and the main vessel. This cold sodium is owing upwards
along the vessel and it overows to come back in the cold pool. As
the cold pool free surface is lower than the weir level, the sodium
ow penetrates through the free surface with sufcient momentum
to induce gas entrainment.
A third source of bubbles is the nucleation of dissolved gas in
the cold regions, as the sodium solubility decreases when the tem-

perature decreases. So, the main location for nucleation in nominal


condition is the intermediate heat exchanger where the sodium
is cooled. During transient cooling down situation, nucleation can
occur in the hot pool and bubbles may be transported at lower ow
condition than the nominal one. So, one can expect that decantation of bubbles at the free surface of the upper plenum is more
signicant in such transient situation than in the nominal condition.
If the presence of bubbles in the primary circuit cannot be
avoided, one has to estimate the behaviour of the bubbles in the
circuit taking into account the various sizes of bubbles. Different
physical effects should occur during the transport of bubbles as
fragmentation, coalescence, possible accumulation in gas pocket,
decantation, dissolving in hot regions. All these phenomena are
complicated, they can interact and their modelling is a real challenge.

Fig. 26. Gas entrainment by vortices at free surface.

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

Fig. 27. Gas entrainment at free surface (TRIO U computation).

7.2. Tools
Various kinds of tools can help for the estimation of the gas
entrainment in the primary circuit. A global system modelling
should be used to describe the whole circuit, providing that adapted
physical modelling should be implemented in the code to take
into account the different physical effects. Computational uid
dynamics may help for the evaluation of the description of the gas
entrainment at the free surface and the bubbles behaviour in the
plenum (Sakai et al., 2008). Of course, experimental facilities are
also needed to support the modelling approach and to provide the
designer with a validation basis when extrapolating to the reactor
condition (Kimura et al., 2008).
For the global system modelling of the gas entrainment, CEA
has used during the European Fast Reactor project a numerical tool
called VIBUL. This code could model the primary circuit divided in
eight regions corresponding to the main components or plenum, for
a population of bubbles with 50 different diameters. Each region
was considered in a 0D approach in which the various physical
effects were modelled as a transfer function for the different classes
of bubbles. Of course, the transfer functions were based on the limited existing knowledge and necessary assumptions. Nowadays,
the CATHARE code (Geffraye et al., 2009) used in France for the
system analysis of all kinds of reactors can be helpful for a global
gas entrainment analysis in sodium cooled fast reactors. CATHARE
code is basically a six equations two-phase system code using a 1D
approach. So, the liquid eld and the gas eld are fully described and
they can interact, but different sizes of bubbles cannot be modelled
in the present version of the code. The present modelling assumes
that small bubbles follow the liquid eld and large bubbles follow
the gas eld, requiring some assumption on the relation between
the bubble size and the computed void fraction. In the future, the
CATHARE code will be able to deal with several sizes of bubbles
with one eld for each class of bubbles. Of course, the various
physical effects as free surface entrainment, transport, decantation, fragmentation, coalescence, nucleation and dissolving should
be modelled in a 0D or 1D approach, on the basis of existing knowledge.
For the 3D numerical approach of gas entrainment, TRIO U code
is improved, validated and used in CEA. The free surface behaviour
can be modelled using a Front-Tracking method as shown in Fig. 27.
The occurrence of vortex and gas entrainment for unsteady situations is predicted by coupling the Front-Tracking method to a
Large Eddy Simulation modelling of the plenum. Direct numerical
simulation is also possible for a rened study of a limited area to
provide coarser approaches with correlations for bubbles entrainment. The transport of bubbles in a plenum could be modelled with
TRIO U code using a Lagrangian approach and assuming no feedback from the bubbles to the liquid eld. Of course, correlations

1211

should be implemented for fragmentation and coalescence modelling. A further rened 3D analysis could be done in the future
using the NEPTUNE code developed by CEA, AREVA and EDF (Guel
et al., 2007). NEPTUNE code is a two-phase CFD code in which liquid
and bubbles elds are modelled separately and can interact, with
bubbles of different sizes. Another aspect is the large range of typical sizes to be modelled, from 1 mm (small bubble size) to several
meters (plenum size). For instance, the occurrence of unsteady vortices and bubble entrainment at the free surface requires a global
modelling of the plenum coupled with a rened modelling of the
vortex area. So, a strategy on the multi-resolution process is needed
based on a xed multi-grids method or an adaptative mono-grid
method.
Experimental data were required to support the numerical
approaches described previously, for the validation process and for
reactor design improvement (Smith, 1990). So, mainly two kinds of
test facilities are needed: basic ones for model improvement and
validation (Moriya, 1998), realistic ones for design improvement
and code verication before extrapolation to the reactor. For Superphenix plant, the upper plenum free surface was smooth and the
main source of gas in the primary circuit was the overow of the
vessel cooling by-pass ow.
During the European Fast Reactor project, the higher compactness has lead to a risk of gas entrainment at the upper plenum
free surface. So, several upper plenum water mock-ups were built
to study this new problem. Two test sections called OREILLETTE
and COLCHIX were used for various thermal hydraulic purposes
including gas entrainment. The reduced scale 1/10 for OREILLETTE
and 1/8 for COLCHIX allowed a Froude number equal to the reactor one for the free surface behaviour with a Reynolds number
reduction of about 1/100. It was seen that the vortex occurs mainly
between the intermediate heat exchanger and the vessel, with a
risk of gas entrainment in the heat exchanger inlet window. The
wake induced by the components and the locally downward ow
due to the plenum recirculation were the main reasons for the
vortex apparition. The phenomenon was very unstable and statistic counting was made to quantify the vortex occurrence. Some
improvements in the design and local devices were tested in these
facilities to reduce the occurrence of vortices and the risk of gas
entrainment. Another large scale upper plenum water test facility operating in AEA-Technology was used to study the scale effect
on vortex occurrence and gas entrainment at the free surface. This
facility called HIPPO was a 90 sector at a scale 0.5 (Serpantie et
al., 1989), allowing tests at the same Froude number as the reactor
with a higher Reynolds number than in OREILLETTE and COLCHIX
facilities. Using a Froude number similarity criteria, it appeared that
non-conservative results could be obtained on small scale models,
with an under estimation of the vortex depth and gas entrainment
occurrence.
In parallel to the realistic upper plenum test facilities, basic studies on vortex and gas entrainment phenomena were made on a
rather simple test section called VEDETTE. Steady and unsteady
vortices were studied as illustrated in Fig. 26. The vortex geometrical characteristics and the local velocity eld were measured to
determine the main local parameters which inuence the vortex
behaviour as shown in Fig. 28. It was seen that the circulation
and the downward velocity were the two important parameters as
described in some analytical models. This has been conrmed and
analysed in recent JAEA studies (Kimura et al., 2008) where rened
vortex and velocity measurements have been done. The work is
still in progress as the extrapolation from a reduce scale model to
the reactor may be non-conservative when using a Froude number
similarity.
For gas entrainment induced by overow in the vessel cooling
circuit, two experimental devices were used at CEA in the 1980s. A
water test facility simulating the Superphenix condition at the scale

1212

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

Fig. 28. VEDETTE test section: PIV measurements of horizontal velocity below the free surface.

1/1 (for a limited width) showed that an optimal overow height


can minimize the gas entrained at the weir. A sodium test facility
was built with the same geometry and the same ow condition
than the water one. Measurements in sodium were rather difcult
but it appeared that the rate of gas entrained in the water facility
was larger than in the sodium facility. So, it could be concluded that
the water tests were conservative.
8. Piping thermal hydraulics
8.1. Needs
The problems described previously were concerning mainly the
reactor vessel. The piping also can be submitted to thermal stresses
induced by thermal mixing or thermal stratication.
In case of mixing between hot and cold sodium, temperature
uctuations occur and can be the origin of thermal fatigue (IAEA,
2002). Thermal striping could occur after a certain time of operation
with the initiation and the propagation of cracks in the pipe wall.
The welding regions are particularly sensitive to thermal striping
and the designer must avoid as far as possible any welding in a mixing region. In Superphenix reactor, several regions in the secondary
circuits were concerned with thermal mixing and a potential risk
of thermal fatigue. One region was the steam generator outlet submitted to the mixing of two different sodium ows in the vertical
portion of pipe connected to the bottom of the steam generator. The
main ow (about 90% of the total ow) is coming from the center
part of the steam generator. The secondary ow (about 10% of the
total ow) is coming from the outer part of the steam generator.
The temperature difference is about 50 K at nominal condition and
it can reach more than 100 K for some reduced ow rate conditions.
The ratio on the mean velocity between the main ow and the secondary ow is about 3. This is a co-axial ow conguration where
the mixing area is located along the pipe wall. Physically speaking,
this is a particular kind of wall ow with an interaction between
the development of the mixing region and the boundary layer.
Another region submitted to thermal mixing in Superphenix
secondary circuit is the mixing tee. A mixing tee is located between
the secondary pump and the intermediate heat exchanger, at the
junction with the pipe coming from the sodiumair heat exchanger
of the secondary decay heat removal system. The mixing tee is connected on a vertical portion of the main pipe with a horizontal
junction for the secondary ow. A particular internal design of the
mixing tee component was previously dened on the basis of air
tests. Several operational situations with various ow conditions
have been analysed. The temperature difference between the two
ows could reach 130 K in the worst conditions. The thermal eld
in the mixing tee and the residual temperature uctuations at the
outlet of the mixing tee were estimated. As the effective temperature which can lead to thermal fatigue is the wall temperature,
an important parameter is the transfer function of the temperature

uctuations from the uid to the wall. In the case of sodium, the
high thermal diffusivity will play an important role which cannot be
simulated by water or air. So, this particular aspect must be studied
to get a better estimation of the real thermal stress on the structure
(Simoneau and Gelineau, 2001). Various ow congurations can be
involved as boundary layer, jet impingement and so on.
Thermal stratication is another important phenomenon which
can occur in the horizontal portions of the piping. Thermal stratication is induced by buoyancy forces, when a transient situation
occurs, with a hot or a cold shock produced at low ow rate condition. Hot uid is owing in the upper part of the horizontal pipe and
cold uid in the lower part. Consequently, the differential dilatation will induce a bending of the pipe with a risk of crack. In the
bends connected to the horizontal portion of pipe, mixing between
hot and cold uid occurs and may propagates in the vertical portions of the pipe. The geometrical conguration of the piping has
an important effect on the stratication behaviour. For instance,
a hot shock propagating at low ow rate in a U shape pipe may
induce a long term thermal stratication with a large temperature
difference between the top and the bottom of the pipe. Effectively
in this case, the cold sodium can be blocked in the lower part of the
pipe as the hot sodium is owing in the upper part. The cold sodium
will be eliminated slowly by thermal diffusion and a severe thermal
stress is induced on the pipe. On the contrary, a cold shock propagating in the same U shape pipe will generate a smaller and shorter
stratication, as the hot sodium can be evacuated progressively in
the vertical portions of the pipe. Thermal stratication may be particularly important in decay heat removal situations, especially for
natural convection regimes. As described previously, it will have a
severe thermo-mechanical effect which could lead to a rupture. It
will also modify the global system behaviour in these critical situations. So, it is important to be able to predict the occurrence and
the level of thermal stratication in the piping during low ow rate
transient situations (Tenchine and Barroil, 1996).
8.2. Tools
For Superphenix plant, the piping thermal hydraulics was investigated mainly by use of water and sodium facilities. Numerical
simulations were also developed for some applications, but the performance of computers at this time was not sufcient to directly
use these simulations for the reactor design. For the European
Fast Reactor project, no signicant activity was performed on piping thermal hydraulics, except for decay heat removal situations
described earlier.
Mixing of hot and cold sodium at the outlet of the steam generator was studied in CEA on a specic sodium facility called GEVEJET
and presented in Fig. 29. This facility simulated the vertical portion
of pipe connected to the steam generator outlet at a scale 1/5. The
main hot sodium ow in the center region and secondary cold ow
in the outer region were represented. Mean temperature and tem-

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

1213

Fig. 31. NAJET sodium facility.

Fig. 29. GEVEJET sodium facility.

perature uctuations were measured downstream in the uid and


at the wall with thermocouples of 0.25 mm and 0.5 mm in diameter. Special devices designed and built by AEA-Technology were
mounted on GEVEJET test section to measure the wall temperature at a distance of 0.25 mm from the surface. These devices were
used to estimate the transfer function of the temperature uctuations from the uid to the wall for the thermal fatigue analysis. The
attenuation of the temperature uctuations from the uid to the
wall was clearly identied for various ow conditions, due to the
signicant inuence of sodium thermal diffusivity in the boundary layer (Tenchine, 1994). As expected, the attenuation effect was
increasing with the frequency as shown in Fig. 30. Sodium thermal diffusivity acts as a signicant low pass lter as it has been
conrmed later (Wakamatsu et al., 1995).

Fig. 30. Boundary layer attenuation of temperature uctuations.

Another mixing ow conguration encountered in Superphenix


reactor is the mixing tee case described previously. A special mixing tee component was designed to reduce the risk of thermal
fatigue on the pipe wall. A sodium test facility shown in Fig. 31
and called NAJET was built to test the efciency of the mixing tee
for various ow conditions. The mixing tee component and the
upstream and downstream pipes were represented at a scale 0.3.
Mean temperature and temperature uctuations were measured
in the component and downstream with thermocouples of 0.5 mm
in diameter. The attenuation of the temperature uctuations at
the wall was measured in one section and conrmed the results
obtained on the GEVEJET test section. More recently, mixing tee
congurations have been studied in water models in CEA and in
other laboratories for Light Water Reactors and Liquid Metal Fast
Reactors (Kamide et al., 2009). Detailed measurements of velocity
and temperature are available, including frequency analysis of the
uctuations in the uid and at the wall. These experimental results
are used for the validation of numerical simulations using various
turbulence modellings.
In CEA, Large Eddy Simulation is preferentially used in TRIO U
code for mixing tee problems. Such modelling is able to estimate the
characteristics of the temperature uctuations including the frequency distribution. An example of TRIO U computation of a water
mixing tee experiment performed at JAEA is presented in Fig. 32. It
shows the rather good prediction obtained with Large Eddy Simulation when using rened meshes (Coste et al., 2006). Code validation
based on water models is useful but not sufcient to ensure a right
estimation of the temperature uctuations in sodium, especially at
the wall (Muramatsu, 1999). The sodium high thermal diffusivity
plays a leading part in the boundary layer and the attenuation of the
temperature uctuations is highly dependent on this physical property. So, sodium experiments are probably still needed to reach a
higher condence in the numerical predictions of the temperature
uctuations at the wall for new ow conditions.
Thermal stratication in piping has been investigated for Superphenix secondary circuits. As a rst step, water facilities as shown
in Fig. 33 were used to study the onset and the propagation of
thermal stratication in horizontal portion of pipes during transient situations at low ow rate (Tenchine and Martin, 1980). Then,
the inuence of the geometry was analysed to point out the cases
where thermal stratication can stay for a long time with a large
temperature gradient between the upper part and the lower part
of the horizontal pipe. The U shape in case of hot shock and the
inversed U shape in case of cold shock were seen as the most severe
geometrical conditions. A specic water facility representing the
Superphenix secondary circuit was used to simulate various tran-

1214

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

Fig. 32. TRIO U computation of a JAEA mixing tee experiment.

sient situations. The scale 1/10 allowed a Richardson similarity to


represent correctly the buoyancy inuence which is predominant.
Later, sodium tests were performed to get some criteria on the
occurrence of thermal stratication. The main sodium facility used
in CEA for that purpose was the JANUS experiment presented in
Fig. 24 and also used for decay heat removal studies (Tenchine and
Grand, 1989). The test results were plotted as a function of two
parameters, respectively the Richardson number and the normalized slope of the thermal shock. The last parameter appeared to be
very important, as it governs the local buoyancy effect at the propa-

gating thermal front. New sodium tests are needed to complete and
improve the criteria on thermal stratication in piping. Up to now,
the numerical simulation was not intensively used in CEA to predict thermal stratication, but it should be developed in the future
for the Astrid project. One challenge might be the simulation of the
mixing process which occurs in the bends as shown in Fig. 34 and
in the vertical portions of the piping.
9. Others
Of course, other thermal hydraulic problems should appear during the design of sodium cooled fast reactors. We can point out

Fig. 33. Water facility for thermal stratication.

Fig. 34. Stratication and mixing in a bend.

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

1215

Fig. 36. Water facility for a sodiumair heat exchanger.

Fig. 35. Pool immersed heat exchanger.

some of them concerning important components (Carbonnier et


al., 1990).
Sodiumsodium heat exchangers are used to transfer the heat
between the primary circuit and the secondary circuit, or between
the primary circuit and the decay heat removal circuit. In French
reactors, they usually are designed as tube and shell heat exchangers (Serpantie et al., 1989), with a vertical bundle of straight tubes
inserted between two thick plates. Some supporting grids are generally located along the bundle to avoid any ow induced vibration.
Primary sodium is owing around the tubes, as secondary sodium
is owing inside the tubes. On the primary side, one has to check
the uniform distribution of the ow at the inlet window connected
to the upper plenum. The ow eld and the temperature eld in
the bundle play an important role in the efciency of the heat
exchanger (Gajapathy et al., 2008). Moreover, the thermal stresses
induced by the possible differential dilatation between the inner
and outer parts of the bundle must be evaluated as the whole bundle is connected to the two thick plates. By-pass ows along the
inner and outer shells must be estimated as well. As the ow pattern
is complicated with transverse ows at the inlet and outlet of the
bundle, it is important to use adapted pressure drop and heat transfer correlations. As various ow situations may occur, including low
ow rate condition with buoyancy inuence, the correlations must
also cover these situations. On the secondary side, the inlet plenum
must be designed to get a ow distribution in the bundle as uniform
as possible. Several water and sodium heat exchanger test facilities
were used for Superphenix reactor and EFR project as shown in
Fig. 35. Numerical simulation was also developed to simulate the
bundle with a sub-channel modelling. Rened computations were
performed to study the inlet and the outlet plenum of the secondary
side.
Sodiumair heat exchangers are commonly used in decay heat
removal systems to transfer heat between the DHR secondary circuit and the outside. Various types of sodiumair heat exchangers
have been used or designed. In Superphenix reactor, horizontal
parallel V-shape tubes were used for the decay heat removal systems connected to the primary and secondary circuits. Sodium
is owing inside the tubes and air is owing around the tubes

in a transverse direction. On the sodium side, one had to check


the sodium ow distribution in the tubes to optimise the efciency. As the tubes were connected in parallel to an inlet collector
and an outlet collector, the design needed some adaptation to
ensure a uniform distribution of the sodium. Hydraulic water
models as the one shown in Fig. 36 were used to measure the
ow distribution for various ow conditions. On the air side, the
characteristics of the nned tubes and the ow distribution in natural circulation across the tubes play an important role on the
heat exchanger efciency (Mochizuki and Takano, 2009). Sodium
tests in reactor conditions were needed to check the global thermal hydraulic behaviour. A nal verication was done during
the commissioning tests of Superphenix plant, for various situations including a natural convection regime of the whole DHR
system. For the EFR project, a new design of sodiumair heat
exchanger was studied, so-called drum type. The drum type design
consisted in helical tubes assembled in a cylindrical shape and connected to inlet and outlet ring collectors. Air is owing in natural
convection across the cylindrical bundle in a transverse direction. Hydraulic water models were used to optimise the sodium
ow distribution for various ow conditions. Sodium tests at a
scale nearly 1:1 were performed by Siemens on the ILONA loop
representing the whole secondary circuit of the EFR decay heat
removal system (Stehle et al., 1994). Steady-state and transient
situations were simulated to conrm the acceptable behaviour of
the sodiumair heat exchanger for stand-by and operating conditions.
Steam generator is a key component of sodium cooled fast reactors. Phenix reactor had modular steam generators with horizontal
W-shape tubes. In Superphenix plant, one through steam generators with helical tubes were built and tested. One possible design
for EFR project was a straight tubes bundle. Of course, many design
and safety studies on these various kinds of steam generators have
been done. Thermal hydraulic analysis using numerical and experimental approaches have been performed (Carbonnier et al., 1990).
As the author has not been involved in this activity, no summary is
given on this topic.
10. Conclusion
Sodium cooled fast reactors have been developed in France for
nearly 50 years. After Rapsodie, Phenix, Superphenix, the European Fast Reactor project, the new target for 2020 is Astrid.
Astrid aims at being a prototype to support Generation IV deployment in the middle of the century. The past experience shows
that thermal hydraulic challenges induce an intensive activity
to support the design and the safety requirements. We have
summarized some important thermal hydraulic problems encountered in the past reactors and projects. The items covered in
this paper are the subassembly, the core, the upper plenum, the
lower plenum, the decay heat removal, the gas entrainment and
the piping. Heat exchangers have been briey mentioned. As
economical and safety requirements for Astrid are higher than

1216

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217

in the last century, thermal hydraulic challenges increase. The


numerical and experimental tools used in the past activity in CEA
have been presented succinctly for each item. The improvement
on numerical simulation is emphasized with some examples of
recent computations. The considerable improvement in the computing performance allows rened thermal hydraulic calculations
which will contribute to reach the required level of condence.
Nevertheless, experimental data are still needed to qualify the
numerical simulations. The review of past experimental activity
aims at providing some support to the new generation of thermal hydraulic people in charge of designing experimental facilities.
Fortunately, in France the relay between the old generation of
the 20th century and the new generation of the 21st century can
be done with a several years of common activity on the Astrid
project.
Acknowledgments
I would like to acknowledge all my past and present CEA colleagues who have been involved in thermal hydraulic activity for
sodium cooled fast reactors. My brief review is widely based on
their analytical, numerical and experimental work. I have a particular thought for Jean-Alain Toly who died several years ago. He
was a colleague and a friend involved in several topics mentioned
in the paper as mixing tees, thermal stratication and decay heat
removal.
References
Azarian, G., et al., 1990. Sodium thermal-hydraulics in the pool LMFBR primary
vessel. Nuclear Engineering and Design 124 (3), 417430.
Basque, G., 1989. 3D computation of thermal-hydraulics in a wire-wrapped rod bundle in the natural and mixed convection regimes using the TRIO VF code. In:
NURETH-4, Karlsruhe, Germany, October 1013.
Berthoux, M., Carenza, A., 2008. Pressure loss and heat exchange in a rod bundle
representative of ETDR start-up core: ESTHAIR experiment in hot air similarity.
In: NUTHOS-7, Seoul, Korea, October 59.
Betts, C., et al., 1991. European studies on fast reactor core interwrapper ows. In:
Int. Conf. Fast Reactor and Related Fuel Cycles (FR91), Kyoto, Japan, October
28November 1.
Bieder, U., et al., 2009. CFD investigations of a full scale helical wire-wrapped 61pin fuel bundle by using the code TRIO U. In: Annual Meeting on NUCLEAR
TECHNOLOGY 2009, Dresden, Germany, May 1214.
Bubelis, E., Schikorr, M., 2008. Review and proposal for best t of wire-wrapped
fuel bundle friction factor and pressure drop predictions using various existing
correlations. Nuclear Engineering and Design 238 (12), 32993320.
Carbonnier, J.L., et al., 1990. Thermo-hydraulics of LMFBRs components. Nuclear
Engineering and Design 124 (3), 379389.
Cheng, S., Todreas, N.E., 1986. Hydrodynamic models and correlations for bare and
wire-wrapped hexagonal rod bundles: bundle friction factors, subchannel friction factors and mixing parameters. Nuclear Engineering and Design 92 (2),
227251.
Costa, J., Tenchine, D., 1995. Liquid metals. In: Int. Encyclopedia of Heat and Mass
Transfer. CRC Press, pp. 675677.
Coste, P., et al., 2006. Large Eddy Simulation of a mixing tee experiment. In: ICAPP06,
Reno, USA, June 48.
Ducros, F., et al., 2008. A proposal for a global multi-scale/multi resolution thermalhydraulic approach for Gen IV concepts. In: International Workshop on
Thermal-Hydraulics of Innovative Reactor and Transmutation System (THIRS),
Karlsruhe, Germany, April 1416.
Francois, G., et al., 1990. Assessment of thermal-hydraulic characteristics of the
primary circuit. Nuclear Science and Engineering 106 (1), 5563.
Gajapathy, R., et al., 2008. Thermal hydraulic investigations of intermediate heat
exchanger in a pool-type fast breeder reactor. Nuclear Engineering and Design
238 (7), 15771591.
Geffraye, G., et al., 2009. CATHARE 2 V2.5 2: a single version for various applications.
In: NURETH-13, Kanazawa, Japan, September 27October 2.
Gourdon, J., et al., 1990. An overview of Superphenix commissioning tests. Nuclear
Science and Engineering 106 (1), 110.
Grewal, S., Gluekler, E., 1982. Water simulation of sodium reactors. Chemical Engineering Communications 17 (1), 343360.
Guel, A., et al., 2007. NEPTUNE: a new software platform for advanced nuclear
thermal hydraulics. Nuclear Science and Engineering 156 (3), 281324.
Hoffmann, H., 1989. Thermohydraulic investigations of decay heat removal systems
by natural convection for liquidmetal fast breeder reactors. Nuclear Technology
88 (1), 7586.

IAEA, 1999. Status of liquid metal cooled fast reactor technology. TECDOC-1083,
Vienna, Austria.
IAEA, 2000. LMFR core thermohydraulics: status and prospects. TECDOC-1157,
Vienna Austria.
IAEA, 2002. Validation of fast reactor thermomechanical and thermohydraulic codes.
TECDOC-1318, Vienna, Austria.
Ieda, Y., et al., 1990. Experimental and analytical studies of the thermal stratication
phenomenon in the outlet plenum of fast breeder reactors. Nuclear Engineering
and Design 120 (23), 403414.
Kamide, H., 2001. Investigation of core thermal hydraulics in fast reactors:
inter-wrapper ow during natural circulation. Nuclear Technology 133,
7791.
Kamide, H., et al., 2009. Study on mixing tee behaviour in a tee piping and numerical
analysis for evaluation of thermal striping. Nuclear Engineering and Design 239
(1), 5867.
Kim, J.B., et al., 1999. Protection of Kalimer upper internal structure against thermal
striping loads. In: SMIRT-15, Seoul, Korea, August 1520.
Kimura, N., et al., 2007. Experimental investigation on transfer characteristics of
temperature uctuation from liquid sodium to wall in parallel triple-jet. International Journal of Heat and Mass Transfer 50, 20242036.
Kimura, N., et al., 2008. Experimental study on gas entrainment at free surface
in reactor vessel of a compact sodium-cooled fast reactor. Journal of Nuclear
Science and Technology 45 (10), 10531062.
Lefevre, J.C., et al., 1996. European fast reactor design. Nuclear Engineering and
Design 162 (23), 133143.
Mochizuki, H., 2007. Inter-assembly heat transfer of sodium cooled fast reactors:
validation of the NETFLOW code. Nuclear Engineering and Design 237 (19),
20402053.
Mochizuki, H., Takano, M., 2009. Heat transfer in heat exchangers of sodium cooled
fast reactor systems. Nuclear Engineering and Design 239 (2), 295307.
Moriya, S., 1998. Estimation of hydraulic characteristics of free surface vortices
extension vortex theory and ne model test measurements. Central Research
Institute of Electric Power Industry Report U93004, Tokyo, Japan.
Muramatsu, T., 1999. Evaluation of thermal striping phenomena at a tee junction
of LMFBR piping systems with numerical methods. In: SMIRT-15, Seoul, Korea,
August 1520.
Natesan, K., et al., 2007. Thermal hydraulic study on detection of random failure of
fuel by delayed neutron detection system. Nuclear Engineering and Design 237
(23), 22192231.
Nishimura, M., et al., 2000. Transient experiments on fast reactor core thermalhydraulics and its numerical analysis. Inter-subassembly heat transfer and
inter-wrapper ow under natural circulation conditions. Nuclear Engineering
and Design 200, 157175.
Padmakumar, G., et al., 2007. Comprehensive scale model for LMFBR reactor assembly thermal hydraulics. International Journal of Nuclear Energy Science and
Technology 3 (4), 325344.
Rehme, K., 1973. Simple method of predicting friction factors of turbulent ow
in non-circular channels. International Journal of Heat and Mass Transfer 16,
933950.
Sakai, T., et al., 2008. Proposal of design criteria for gas entrainment from vortex dimples based on a computational uid dynamics method. Heat Transfer
Engineering 29 (8), 731739.
Serpantie, J.P., et al., 1989. Progress in LMFBR design: the european fast reactor
compact primary system. In: SMIRT-10, Anaheim, USA, August 1418.
Simoneau, J.P., Gelineau, O., 2001. Simulation of attenuation of thermal uctuations
near a plate impinged by jets. In: ICONE-9, Nice, France, April 812.
Smith, M.R., 1990. Techniques for the investigation of scaling criteria for gas entrainment mechanisms in liquid metal cooled fast reactors. General Electric Company
Journal of Research 8 (1), 4956.
Stehle, H., et al., 1994. Large scale experiments with a 5 MW sodium/air heat
exchanger for decay heat removal. Nuclear Engineering and Design 146 (13),
383390.
Tanaka, N., et al., 1990. Prediction method for thermal stratication in a reactor
vessel. Nuclear Engineering and Design 120 (23), 395402.
Tenchine, D., 1978. Study of Transient and Permanent Flow in the Event of Natural
Convection in a Conned Environment. PhD Report. Lyon University, March 15.
Tenchine, D., 1994. Boundary layer attenuation in turbulent sodium ows. In:
IAEA Specialists Meeting on Correlation Between Material Properties and
Thermal Hydraulic Conditions in LMFBR, Aix en Provence, France, November
2224.
Tenchine, D., Amblard, M., 1983. Heat transfer by natural convection in sodium from
downward facing surfaces. In: Second International Topical meeting on Nuclear
Reactor Thermal Hydraulics, Santa Barbara, USA, January 1114.
Tenchine, D., Barroil, J., 1996. Transfer of temperature uctuations at the interface
of a stratication. In: Second European Thermal-Sciences, Rome, Italia, May 29
31.
Tenchine, D., Grand, D., 1989. Onset of natural circulation in a sodium loop. In:
NURETH-4, Karlsruhe, Germany, October 1013.
Tenchine, D., Martin, R., 1980. Thermal stratication in a horizontal sodium pipe. In:
International Seminar of ICHMT, Dubrovnik, Yougoslavia, September 15.
Tenchine, D., Moro, J.P., 1997. Experimental and numerical study of coaxial jets. In:
NURETH-8, Kyoto, Japan, September 30October 4.
Tenchine, D., Moro, J.P., 2000. Experimental and computational study of turbulent
mixing jets for nuclear reactors applications. In: ISTP-12, Istanbul, July 1620.
Trapp, J.P., et al., 1990. Comparison of failed fuel detection mock-up with Superphenix measurements. Nuclear Science Engineering 106, 8893.

D. Tenchine / Nuclear Engineering and Design 240 (2010) 11951217


Valentin, B., 2000. The thermal hydraulics of a pin bundle with an helical wire
wrap spacer. Modeling and qualication for a new sub-assembly concept. IAEATECDOC-1157, Vienna, Austria.
Vidil, R., et al., 1988. Interaction of recirculation and stable stratication in a rectangular cavity lled with sodium. Nuclear Engineering and Design 125, 321
332.

1217

Wakamatsu, M., et al., 1995. Attenuation of temperature uctuations in


thermal striping. Journal of Nuclear Science and Technology 32 (8),
752762.
Weinberg, D., et al., 1995. Summary report of NEPTUN investigations into the transient thermal hydraulics of the passive decay heat removal. Report FZKA 5666,
Forshungszentrum Karlsruhe, Germany.

You might also like