You are on page 1of 15

This article was downloaded by: [112.215.36.

144]
On: 19 March 2014, At: 17:55
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of Nuclear Science and Technology
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tnst20
Water experiment and numerical simulation on failed
fuel detection and location system of Japan sodium-
cooled fast reactor (JSFR)
Kosuke Aizawa
a
, Jun Ohshima
a
, Hideki Kamide
a
& Naoto Kasahara
a

b
a
Japan Atomic Energy Agency, 4002 Narita, Oarai, Higashi-ibaraki-gun , Ibaraki , Japan
b
Graduate School of Nuclear Engineering and Management , University of Tokyo , Tokyo ,
Japan
Published online: 24 Jan 2012.
To cite this article: Kosuke Aizawa , Jun Ohshima , Hideki Kamide & Naoto Kasahara (2012) Water experiment and
numerical simulation on failed fuel detection and location system of Japan sodium-cooled fast reactor (JSFR), Journal of
Nuclear Science and Technology, 49:1, 47-60, DOI: 10.1080/18811248.2011.636539
To link to this article: http://dx.doi.org/10.1080/18811248.2011.636539
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of
the Content. Any opinions and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied
upon and should be independently verified with primary sources of information. Taylor and Francis shall
not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other
liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or
arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
ARTICLE
Water experiment and numerical simulation on failed fuel detection and location system of Japan
sodium-cooled fast reactor (JSFR)
Kosuke Aizawa
a
*, Jun Ohshima
a{
, Hideki Kamide
a
and Naoto Kasahara
a,b
a
Japan Atomic Energy Agency, 4002 Narita, Oarai, Higashi-ibaraki-gun, Ibaraki, Japan;
b
Graduate School of Nuclear Engineering
and Management, University of Tokyo, Tokyo, Japan
(Received 8 February 2011; accepted nal version for publication 10 August 2011)
A conceptual design study of Japan sodium-cooled fast reactor (JSFR) is in progress as the Fast
Reactor Cycle Technology Development (FaCT) project in Japan. A selector-valve type failed fuel
detection and location (FFDL) system is applied to the JSFR design that has an upper internal structure
(UIS) with a slit above the core and several sampling nozzles for the FFDL are set in the UIS around the
slit to detect the ssion product (FP) from the subassemblies below the slit. Therefore, mixing process in
the UIS of complicated geometry should be known and appropriate arrangement of the sampling nozzles
in the UIS is needed. A water experiment using a 1/5-scale model was carried out to investigate the
mixing process in the UIS and concentration distribution of FP simulant. Experimental results showed
that the sampling nozzles set in the UIS detected the FP simulant concentration within the criteria of
FFDL signal detection, even in case of failed subassemblies under the UIS slit. In addition, identication
of the failed fuel subassembly under the UIS slit was achieved by means of comparing concentration
proles in the UIS. Therefore, the suitable sampling nozzle arrangement was obtained for JSFR. A
numerical simulation using a CFD code was carried out and the simulation method was validated based
on the experimental data of the FP simulant concentrations. The simulation results showed that the
simulation predicts the FP concentration distributions.
Keywords: sodium-cooled fast reactor; selector-valve type FFDL system; UIS; sampling nozzles; ssion
product
1. Introduction
A conceptual design study of Japan sodium-cooled
fast reactor (JSFR) is in progress as the Fast Reactor
Cycle Technology Development (FaCT) project in
Japan [1]. JSFR has 1500 MWe power and 562 core
fuel subassemblies. Thus, failed fuel detection and
location (FFDL) system is one of the signicant issues
of design study. The FFDL system is an instrumenta-
tion device that identies a subassembly which has a
failed fuel pin (failed fuel subassembly). Secured
identication of failed fuel subassembly is of impor-
tance since the plant cannot be resumed into next
operation cycle without take away of the failed fuel
subassembly in a case of fuel failure incident. JSFR
also has failed fuel detection (FFD) system which
detects a fuel failure in the core. The FFD system
monitors a ssion product coming from failed fuel
subassembly at all times. The FFDL system operates
and identies failed fuel subassembly after the
detection of a fuel failure in the core. However, it
becomes dicult to identify the failed fuel subassembly
when the total number of fuel subassemblies increases.
JSFR adopts a selector-valve (SV) FFDL system.
The SV-FFDL system selects and samples sodium from
each fuel subassembly outlet during reactor operation
and indenties the failed fuel subassembly by means of
detection of ssion product gas or delayed neutron
(DN) precursor of ssion product. Figure 1 shows a
conceptual diagram of the SV-FFDL system. The SV-
FFDL system consists of a sampling nozzle, a selector-
valve drum, a drive shaft, an electromagnetic pump, a
neutron detector, ssion and a beta-ray detector. The
sampling nozzle to take sodium is located right above
each fuel subassembly at bottom of an upper internal
structure (UIS) and all sampling lines are connected
individually to the selector-valve drum. The drive shaft
of the selector-valve rotates and moves vertically in
order to select the sampling line. The sampling lines
collect the sodium exiting from each fuel subassembly
*Corresponding author. Email: aizawa.kosuke@jaea.go.jp
{
Present address: Mitsubishi FBR systems Inc., 2-34-17 Jingumae, Shibuya-ku, Tokyo 150-0001, Japan.
Journal of Nuclear Science and Technology
Volume 49, No. 1, January (2012) pp. 4760
http://www.tandfonline.com
ISSN 0022-3131 print/ISSN 1881-1248 online
2012 Atomic Energy Society of Japan. All rights reserved.
http://dx.doi.org/10.1080/18811248.2011.636539
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

one by one and it is connected to the neutron and beta-
ray detectors via the selector-valve drum. This SV-
FFDL system generally has an advantage that could be
applied to a large-sized reactor by increasing sampling
nozzle number. Several sodium-cooled fast reactors, e.g.
PFR [2], Phenix [3], and Super-Phenix [4], adopted the
SV-FFDL system. However, several fuel subassemblies
in JSFR cannot have the sampling nozzles above their
outlet due to the geometry of UIS.
The UIS of JSFR has a radial slit (UIS slit) for the
fuel handling system. An arm of pantograph-type fuel
handling machine (FHM) goes through the UIS slit and
accesses the fuel subassemblies under the UIS. Thus,
there is no need to move the UIS from the core top and a
single rotation plug systemis adopted. This fuel handling
system using the UIS slit contributes to reduce the
reactor vessel diameter [5]. Figure 2 shows the schematic
view of the UIS design of JSFR and the geometry of the
fuel subassembly outlet and the sampling nozzle. The
UIS is composed of control rod guide tubes (CRGTs)
and six horizontal perforated plates. The horizontal
plates are xed to the CRGTs. The bottom plate is
named as core instrumentation support plate (CIP).
Other ve plates are called as bae plates and are named
as BP1 to BP5 from the bottom to top. As shown in
Figure 2c, each sampling nozzle is set just above a
subassembly outlet and supported by the CIP so as to
take the sodium owing out of the subassembly. How-
ever, the sampling nozzles cannot be set in the UIS slit as
shown in Figure 2d. There are 13 core fuel subassemblies
under the UIS slit. Therefore, several sampling nozzles
are installed in the UIS around the slit so as to sample the
sodium from the failed fuel subassembly under the UIS
slit, even if the sampling nozzles are apart from the fuel
subassembly outlets. Targeted performances of the SV-
FFDL system in JSFR are shown below.
To identify the failed fuel subassembly except for
that under the UIS slit,
To narrow suspected failed fuel subassemblies to
a few subassemblies under the UIS slit.
In order to detect the ssion product gas or DN
precursor of ssion product from a failed fuel
subassembly under the UIS slit, mixing process in the
UIS of complicated geometry should be known and
appropriate arrangement of the sampling nozzles in the
UIS is needed. Thus, a water experiment of uid
mixing in the UIS and development of a numerical
simulation method were carried out.
Objectives of this study are to nd appropriate sam-
pling nozzle arrangement and to reveal its applicability.
2. Approach of this study
2.1. Strategy
2.1.1. Criteria to detect FP concentration
Dilution in the mixing process in the UIS makes
concentrations of ssion product gas and DN precursor
have very low value than those at the outlet of failed
subassembly which is located under the UIS slit. The
criteria of the FP concentration at the sampling nozzle
inlets should be decided by (1) Performance of detectors,
(2) Minimum failure size of the fuel pin to stop
propagation of the fuel failure to the adjacent sub-
assembly, (3) Dilution due to inow in the selector-valve
drum, (4) Decay of DN precursor during transportation
from the sampling nozzle to the detector, (5) Measure-
ment time to scan the entire core. The preliminary design
study [6] of the SV-FFDL system in JSFR showed that
the criteria of FP concentration are 1% which is a
normalized value by the FP concentration at the outlet of
failed fuel subassembly. Taking accounts of some
estimation error of the FP concentration in the UIS,
which has new geometry and complex ow eld, 2% of
normalized FP concentration was set as the target value
for the sampling method of the SV-FFDL system to
detect the failure in the fuel subassemblies under the UIS
slit. As for the subassemblies under the UIS slit, the
mixing area from the subassembly to the sampling
nozzles is so large that transportation phenomena of gas
bubble is quite dicult to predict. Thus, the DN
precursors are main objectives to detect in the SV-
FFDL for the failed subassembly under the UIS slit.
2.1.2. Sampling nozzle arrangement in the UIS
Requirements of nozzle arrangement are as follows,
(1) The FP exiting from any of failed subassemblies
under the UIS slit can be sampled by at least
Figure 1. Conceptual diagram of selector-valve FFDL
system.
48 K. Aizawa et al.
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

one of the sampling nozzles in the UIS with at
least 2% of normalized concentration.
(2) The number of suspicious subassemblies esti-
mated by the SV-FFDL system should be lower
than four when one of the fuel subassemblies
under the UIS slit has a failed fuel pin.
A water experiment was carried out to investigate
the mixing process in the UIS and concentration
distribution of simulated FP material (FP simulant).
The nozzle arrangement was decided by a prediction
analysis and experimental data of the concentrations at
the sampling nozzles.
2.1.3. Applicability of SV-FFDL system under reactor
conditions
The FP concentration under the reactor conditions
will be predicted by a three-dimensional numerical
simulation. Thus, the numerical simulation method is
validated based on the experimental data of the FP
stimulant concentrations.
2.2. Water experiment of FP concentration
The water experiments were carried out to measure
the FP simulant concentration around the UIS slit.
Figure 3 shows a conceptual diagram of the water
experimental apparatus. The experimental apparatus
consists of the test section, FP simulant mixing tank,
FP simulant injection units, sampling pipes which have
the sampling nozzle at tips and concentration meter.
Figure 4 shows schematic of the square shaped test
section and its modeled domain in the reactor upper
plenum. The model scale of the test section is 1/5 and
water at room temperature is used as a working uid.
Domain bounded by the dotted lines in Figure 4b
shows the modeled region. The test section models the
Figure 2. Schematic views of UIS and geometry of fuel subassembly and sampling nozzle.
Journal of Nuclear Science and Technology, Volume 49, No. 1, January 2012 49
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

UIS around the slit in the rectangular geometry. Salt
water is used as the FP simulant. It is noted that
inuence of molecular diusion on the concentration
distribution is ignorable in a turbulent jet [7]. The
molecular diusions of FP material are dierent from
that of salt water. However, such dierence has no
inuence in this setup where multiple jets of high
Reynolds (Re) number 1 6 10
5
govern the ow
eld. Thus, the salt water can be a good stimulant of
FP in sodium ow. Salt water is provided into a
selected fuel subassembly, and water is provided into
the others. Salt concentration of water which is
sampled at the sampling nozzle is measured.
Figure 4. Schematic and modeled domain of test section.
Figure 3. Schematic diagram of the water experimental apparatus.
2.3. Numerical simulation method
In this study, prediction analyses of concentration
distribution in the 1/5-scale water experimental were
carried out and further validation of numerical
simulation method was carried out by using the
experimental results of concentration distribution.
The commercial computational uid dynamics code,
STAR-CD was employed as the simulation code. In
the design calculation of the JSFR, STAR-CD has
been used to see ow elds in the reactor vessel and the
main components. A numerical simulation using
STAR-CD showed a good agreement of velocity
proles in the UIS including the slit with those
50 K. Aizawa et al.
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

obtained by a 1/10-scale water experiment of JSFR
reactor upper plenum [8]. Applied numerical options
in this study were the same as in the simulation above
and listed in Table 1. Renormalization-group (RNG)
k e model, Monotone advection and reconstruction
scheme (MARS) and Standard wall function were used
for turbulence model, space discretization scheme and
the wall function, respectively.
In a preliminary analysis, comparison between
the RNG ke model, normal ke model and CHEN
ke model was carried out for the 1/10-scale water
experiment. This comparison showed that the velocity
proles in the upper plenum were in good agreement
between the analyses used the RNG ke model and
the experiments. Thus, the RNG ke was used in this
study.
3. Experimental setup
3.1. Test section
As previously explained, the model scale of the test
section is 1/5. The test section models the reactor upper
plenum from the core top to the dipped plate in vertical
direction (see Figure 4). The core top is modeled using
a perforated block which has holes corresponding to
the subassembly outlets. The outlets of the radial
neutron shielding were not modeled because of low
ow velocity. The diameter of each hole is 19 mm and
1/5 of the channel diameter of the hollow type upper
neutron shielding in a subassembly. In order to
simulate the ow distribution across the core in
JSFR, the orices were set at the inlets of subassembly
channels. The geometry of UIS, e.g. the UIS slit
distances between the UIS bae plates and ow holes
at the bae plates is modeled precisely. Each ow hole
at the bae plates is 27 mm in diameter. Dimensions
of the UIS slit in Y-direction (see Figure 4) and X-
direction are 96 mm and 435 mm, respectively. An
FHM plug is set instead of the FHM in full power
operation in order to optimize ow eld in the upper
plenum [9]. The FHM plug is also modeled precisely.
The diameter of FHM plug is 176 mm. The FHM plug
is located outside the UIS slit above the outer core fuel
subassemblies. The lower end of FHM plug is set at the
height of the BP1.
3.2. Measurement of FP simulant concentration
Sampling nozzle positions were decided by a
preliminary analysis (see section 5.1) and ow visua-
lization (see section 4.2). The sampling nozzle positions
in the experiment were located at the slit edge of BP2,
the side and bottom walls of FHM plug as shown in
Figure 5. The pitch of the sampling nozzles along the
slit edge of BP2 is the same as that of subassemblies.
The slit edge of the CIP hangs over the outlets of the
subassemblies on the next line to the center line in the
UIS slit. Thus, it was expected that the sampling
nozzles at the CIP edge could detect the FP from the
subassemblies on the side lines below the UIS slit. In
Table 1. Applicable numerical options for 1/5-scale test
analysis.
Items Methods
Numerical method Finite volume method with
unstructured mesh
Turbulence model RNG k e
Dierencing scheme
for convection terms
Monotone advection and
reconstruction scheme (MARS)
Mesh scheme Combination of hexahedron and
tetrahedron mesh elements,
Discontinuous mesh for locally
detailed mesh arrangement
Figure 5. Sampling nozzle positions in the experiment.
Journal of Nuclear Science and Technology, Volume 49, No. 1, January 2012 51
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

order to conrm the sampling ability at the slit edge of
CIP, the sampling nozzles were located at the CIP
above L8 of the side-line subassembly (see Figure 8)
and also G1 of the normal subassembly apart from the
UIS slit as the reference.
Water was sampled at the sampling nozzle so as to
equal the sampling velocity into the nozzle with the
ow velocity near the nozzle. This method can
minimize the inuence of water sampling on the ow
eld in the UIS. The sampled water was stored 10 ml in
a beaker outside the test section. It took 1.5 to 5 s
according to the sampling ow velocity. The salt
concentration in the beaker was measured by a
refractive index method at a constant temperature.
The measurement accuracy was 0.01% (weight %)
based on a calibration. In the experiments, the salt
water of 10% concentration was provided into a failed
fuel subassembly channel. The target concentration of
2% is relative value normalized by the concentration at
the subassembly outlet. Thus, the absolute concentra-
tion was 0.2%. The measurement accuracy of the
concentration is enough to detect this target
concentration.
3.3. Applicability of partial model
Since the test section of experimental apparatus is a
partial model of the reactor upper plenum, its
similarity of the ow eld is signicant. In order to
conrm such applicability of the partial model, a
numerical simulation of the 1/5-scale experimental
apparatus was conducted, and the comparison was
done between ow elds of the 1/5-scale experimental
apparatus and JSFR.
Figure 6 shows the analysis geometry for the 1/5-
scale partial model. In this comparison, the lower end
of the FHM plug was located at the height of the BP4.
The numerical simulation method was shown in the
section 2.3. The perforated plates of the UIS were
modeled as walls and each of ow holes was modeled
in the mesh geometry. The 1/5-scale partial model is
symmetric on the center line of the UIS slit. Thus, a
half region of the test section was modeled in the mesh
scheme. Total number of the mesh elements was
approximately 12,000,000. As the inlet boundary
condition, the ow velocities at the subassembly outlets
and distribution across the core of JSFR were applied.
As the outlet boundary condition, constant pressure
was given. A slip condition and a non-slip condition
were applied at the top wall (free surface in the
experiment) and at the structural wall surface, res-
pectively. Numerical simulation of JSFR [10] was
already done using the same methods shown in the
section 2.3.
Figure 7 shows calculated velocity elds in the UIS
slit cross-section of the 1/5-scale test and JSFR,
respectively. The magnitude of velocity is shown by
the vector length and color. Figure 8 shows calculated
horizontal velocity eld at the height of just below the
BP2. As previously explained, the sampling nozzle
positions were located at the slit edge of BP2. The
concentration distribution at the sampling nozzles of
BP2 depends on the ow pattern in the UIS slit. The
vertical velocity distribution in the UIS slit from core
outlet to the BP2 and also horizontal one at the slit
edge of BP2 are signicant to evaluate the FP
concentration distribution. As shown in Figure 7,
ow patterns of inclined ow toward reactor vessel
from subassembly outlets in the UIS slit were in good
agreement between 1/5-scale test and JSFR. As shown
in Figure 8, the ow patterns across the BP2 slit edge
into the UIS slit were in good agreement between 1/5-
scale test and JSFR.
Furthermore, following ndings were obtained
from the 1/10-scale water experiment [8]. The 1/10-
scale water experiment which modeled the whole
reactor upper plenum of JSFR showed that the ow
eld around the UIS slit was mainly determined by the
jet through the slit. Further, cross ows were limited
through vertical cross-sections which correspond to the
boundary walls in parallel to the UIS slit in the 1/5-
scale partial model.
These results showed the ow elds around the UIS
slit were in good agreement between the 1/5-scale
partial model and JSFR. Therefore, the applicability of
the 1/5-scale partial model was conrmed for the
estimation of FP concentration in the UIS.
4. Experimental results
4.1. Experimental condition
The experimental condition and Re number at the
subassembly outlet are shown in Table 2. Since the SV-
FFDL system is operated mainly during full power
operation of JSFR [6], the data of reactor in Table 2
correspond to the value under the rated condition. The
diameter of the ow hole at the UIS bae plate is Figure 6. Analysis geometry and modeling region.
52 K. Aizawa et al.
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

27 mm in the experiment and larger by a factor of 1.4
than that of the subassembly outlet. The ow velocity
at core outlet was same with that in JSFR. Re numbers
are over 1.00 6 10
5
in the experiments and JSFR at
the core fuel subassembly outlet, thus ow regimes
in both ows are fully turbulent. The dierence of
Re number between the experiment and JSFR may
inuence the pressure loss characteristic in the bae
plate. However, as previously explained, the full
reactor 1/10-scale water experiment showed that the
ow eld around the UIS slit was mainly determined
by the jet through the slit. Therefore, the dierence
of the pressure loss in the UIS bae plates has less
impact on the ow around the UIS slit. Thus, it is
believed that the ow through the UIS slit in the
experiments is similar to that in JSFR.
Figure 9 shows experimental cases in the 1/5-scale
experiment. Thirteen cases have been conducted
for all subassemblies located along the center line
(C2C12) of UIS slit and below the FHM plug (L12
and R12). The case name is referred as Case-C2
for the failed fuel subassembly of C2. In addition,
two cases have been conducted for the subassembly
(L8) on the next line to the center line in the UIS
slit, which is just below under the slit boundary and
for a normal subassembly (G1) apart from the UIS
slit.
Figure 8. Velocity eld at the height of the BP2.
Table 2. Experimental condition.
Item JSFR Water test
Diameter of subassembly
outlet hole
a
0.095 m 0.019 m
Subassembly outlet velocity
b
5.3 m/s 5.3 m/s
Re number 1.86 6 10
6
1.00 6 10
5
Diameter of ow holes
at UIS bae plates
0.135 m 0.027 m
Note:
a
Channel diameter of hollow-type upper axial reector, the
narrowest part;
b
Average of fuel subassemblies in the modelling
region of the water test.
Figure 7. Velocity eld in the UIS slit cross-section.
Journal of Nuclear Science and Technology, Volume 49, No. 1, January 2012 53
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

4.2. Flow visualization
First of all, ow visualization experiments were
performed in order to conrm the ow pattern of the
jets exiting from the subassemblies under the UIS slit
by means of injection of dye or small diameter babbles
into each of subassemblies under the UIS slit (see
Figure 9). The ow visualization showed that the jets
exiting from subassemblies on the center line in the
UIS slit mixed with surrounding uid and steadily
spread. The outer region of the jets touched the edge of
BP2. As for the subassemblies below the FHM plug,
the jets impinged on the bottom of FHM plug.
Figure 10 shows the visualized image of ow exiting
from the subassembly C12 which is the outermost
subassembly on the center line (see Figure 9). As shown
in Figure 10, the ow from C12 impinged on the
bottom of the FHM plug and some part of ow rose
along the side wall of the plug. However, the ow from
C12 did not reach any of the bae plates. Therefore, it
is necessary to set the sampling nozzles at the FHM
plug in order to sample the ow from C12 of the core
fuel subassembly.
4.3. Concentration distribution
Figure 11 shows time trends of salt concentration at
several sampling nozzles in Case-C2. Each of data was
average value for 5 s and normalized by the concen-
tration at the outlet of C2 subassembly. This result
showed that it took 120 s to reach stable salt
concentration. Therefore, the concentration measure-
ment was started after 120 s from the start of the salt
water injection.
Figure 12 shows the concentration distribution
along the slit edge at BP2 in Case-C2 to -C6.
Horizontal axis shows the sampling nozzle number at
BP2 shown in Figure 5. The subassemblies of C2 to C6
are on the center line in the UIS slit. The measured
concentrations were in good agreement with the left
and right side sampling nozzles, e.g. 2L3 and 2R3 (see
Figure 9. Experimental cases.
Figure 10. Stream regime injecting the dye ink into C12.
Figure 11. Time trends of salt concentration in Case-C2.
Figure 12. Concentration distribution at BP2 exiting from
C2 to 6.
54 K. Aizawa et al.
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

Figure 5) due to the symmetric ow pattern. Thus, the
averaged concentrations at the left and right side
sampling nozzles were used to show the distributions in
Figure 12. High concentrations above the target value
of 2% were obtained at three sampling nozzles in each
case. Concentration peaks were detected at the outer
sampling nozzles (larger nozzle No. in Figure 9) from
the failed subassembly outlet in each case. This is due
to the inclined ow in the UIS slit from the
subassembly outlet to the BP2 and toward the reactor
vessel as shown in Figure 7. Concentration peaks were
measured at the same sampling nozzles in two cases of
Case-C2 and Case-C3 as well as Case-C5 and Case-C6.
Thus, it is dicult to identify the failed subassembly by
means of the peak concentration position.
Figure 13 shows the concentration distribution
along the slit edge at BP2 and also along the side
wall of FHM plug in Case-C7 to -C11. As shown in
Figure 13, high concentrations over the target value were
detected at several sampling nozzles set at BP2 in each
case. The concentration peaks at BP2 were measured by
the same sampling nozzles in Case-C9, -C10, and -C11.
As shown in Figure 13b, high concentrations at the side
wall of FHMplug were detected in case of outer positions
of the failed subassembly. As for the subassembly C12,
the measured concentrations at all nozzles set at BP2
were lower than the target value. The subassemblies L12
and R12 also showed much lower concentrations than
the target value.
Figure 14 shows the concentration distribution at
the bottom of FHM plug in Case-C12, -L12, and -R12.
The high concentrations over the target value were
measured at the bottom of FHM plug in three cases.
The sampling nozzles on the slit center line detected the
high concentrations over the target value in three cases.
When the left hand side nozzles on the line of DL
(see Figure 5) were used, high concentrations were
detected for the failed subassembly of L12 at left side
position.
Most of the sampling nozzles are set at the CIP for
normal fuel subassemblies. Then, further two cases
have been conducted for the subassembly (L8) under
the slit edge of CIP and the normal subassembly apart
from the UIS slit (G1). As a result, very high
concentrations over 90% were measured at the
sampling nozzles set at the CIP. It was conrmed
that the sampling nozzles at the CIP could detect the
higher concentrations in case of the fuel subassembly
under the slit edge and also the normal subassemblies
apart from the UIS slit.
Several experiments were done in order to conrm
the experimental accuracy and reproducibility of the
concentration measurement. Figure 15 shows the
concentration distributions along the slit edge at BP
in Case-C2 obtained by three times of independent
measurements. The distributions were in good agree-
ment and this result showed that the measured salt
concentration was stable and reproducible even though
the ow pattern around the UIS slit was complicate.
5. Numerical simulation
5.1. Flow pattern
Sampling nozzle arrangement used in the experi-
ment had been decided based on a prediction analysis.
Simulation code, analytical options and boundary
conditions were the same as in the previous analysis
(see section 3.3). Analysis geometry and mesh model
were nearly same as that of the previous analysis. The
lower end of the FHM plug was located at the height of
the BP1 as well as the experiments. Figure 16 shows
calculated velocity eld in the UIS slit cross-section of
the 1/5-scale test. The magnitude of velocity is shown
by the length of vector and color.
The concentration distribution is obtained from the
calculated ow velocity distribution and the turbulent
diusion coecient distribution. The concentration
distribution analyses were conducted for the C2
subassembly at the innermost subassembly, C6 sub-
assembly around middle subassembly and C12 sub-
assembly at the outermost subassembly on the slit center
line. Figure 17 shows calculated concentration distribu-
tion where the salt water was released from the
subassembly C2. The concentration is a relative one
and normalized by the concentration at the subassembly
of C2. As shown in Figure 17a, the high concentration
ow rose in the UIS slit slantwise toward the reactor
vessel. High concentration region spread steadily due to
the mixing as the ow rises from the outlet of
subassembly. As shown in Figure 17b and c, the high
concentration region reached the bae plates at BP2.
Therefore, the sampling nozzles were set at BP2 in the
Figure 13. Concentration distribution (C711).
Journal of Nuclear Science and Technology, Volume 49, No. 1, January 2012 55
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

experiment. Figure 18 shows calculated concentration
distribution in Case-C12. The high concentration region
impinged on the bottom of the FHM plug and spread
along the side wall of the plug. However, the high
concentration region did not reach BP1, BP2, BP3, and
BP4 as seen in Figure 18. Therefore, the sampling
nozzles were set at the bottom and side wall of FHM
plug in order to monitor the C12 subassembly.
5.2. Validation of concentration analysis
In order to validate the simulation model, compar-
ison of the FP simulant concentrations between the
experiments and analyses was conducted. The mea-
sured concentration was only 26% (normalized by the
value at the failed subassembly outlet) at the maximum
in case of failed subassemblies along the slit center line
except C12. Thus, it is so dicult to predict these small
concentration valves precisely. As shown later, the nal
goal of the numerical analysis is to obtain the FP
concentration distribution corresponding to the failed
subassembly position, not only the maximum point of
the sampling nozzle, but the second and third largest
points, which are useful information to identify the
failed subassembly. Thus, the points of comparison
between experiment and analysis are the concentration
prole (especially the rst to third largest points of the
sampling nozzle) and that the discrepancy of the
concentration between experiment and analysis should
be within factor of 2 at the maximum point. For
example, when the experimental result is 2% at the
maximum point, the analysis targets should be in the
range from 1 to 4% at the same point.
Figure 15. Reproducibility of the concentration
measurements in Case-C2.
Figure 16. Velocity eld in the UIS slit cross-section.
Figure 14. Concentration distribution at the bottom of
FHM plug exiting from C12, L12 and R12.
56 K. Aizawa et al.
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

The velocity eld in the UIS is signicant to predict
the FP concentration distribution. However, the
velocity prole was not measured due to some
limitations in the experimental setup. On the other
hand, detailed velocity proles in the UIS slit were
measured in the full reactor 1/10 water model of JSFR
reactor upper plenum. Numerical simulation of velo-
city eld using the similar method and mesh scheme
was also carried out. Comparison of velocity distribu-
tions between experiments and analyses showed good
agreements [8]. Thus, it is believed that this numerical
analysis also provided well predicted velocity proles in
this 1/5-rectangular test model.
Figure 19 shows the comparison of the con-
centration distribution in Case-C2. Horizontal axis
shows the X distance (see in Figure 9). Numbers in
gure show the sampling nozzle number at BP2. As
shown in Figure 19, the calculated prole was in good
agreement with the experimental data including the
concentration peak position. The maximum dierence
between the analysis and experimental results was
2.5% or less and the discrepancy is within factor of 2 at
the maximum point.
The comparison of the concentration distributions
at BP2 in Case-C6 is shown in Figure 20. The
calculated prole was also in quite good agreement
with the measured data. The maximum dierence was
2% or less, and the discrepancy is within factor of 2 at
the maximum point.
Figure 19. Comparison of concentration distributions at
BP2 in Case-C2.
Figure 17. Calculated concentration distribution in Case-C2.
Figure 18. Calculated concentration distribution in Case-C12.
Journal of Nuclear Science and Technology, Volume 49, No. 1, January 2012 57
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

Figure 21 shows the comparison of the concentra-
tion distributions at the bottom of FHM plug in Case-
C12. The labels of sampling nozzles were shown in the
gure. DLR# means the average value of DL#i and
DR#i. The measured concentrations on the slit center
line (see Figure 21a) were nearly 20% and much higher
than the target value of 2%. One of the reason is that
the exiting from the C12 subassembly impinged the
FHM plug directly. The calculated concentrations were
larger than the experimental results, especially front
side (DC1). However, the calculated data on the side
lines (see Figure 21b) were smaller than the measured
concentrations, especially front side (DLR1). Further,
the calculated data on the second side lines (see
Figure 21c) were also smaller than the measured data.
In addition, analyses results in Figures. 19 and 20 were
also slightly lower than the concentrations measured at
the slit edge of BP2 where outer rim of the jets exiting
from the slit center subassemblies reached as shown in
Figure 17c. These results indicated that concentration
diusion in the analyses results was smaller than that in
the experiments. It is assumed that the causes of the
dierence between experimental and numerical results
are the numerical model, the mesh divisions around the
UIS structure, and the inlet conditions of velocity and
turbulence energy distribution at the subassemblies. In
these numerical simulations, at velocity and at
turbulence energy distributions at each of the fuel
subassembly outlets were provided as the inlet condi-
tions. The inuence of inlet conditions was conrmed by
a concentration analysis which modeled only one fuel
subassembly. That analysis showed that the concentra-
tion diusion becomes large at BP1 height by a precise
model of the handling head and upper neutron shielding
which cause turbulence in the subassembly [11].
6. Discussion
6.1. Estimation method of failed fuel location
Failed fuel subassembly is basically identied by
the location of sampling nozzle which has the
concentration peak. However, the peak positions of
the concentration were the same sampling nozzle in
Cases-C2 and C3. The Cases-C5 and C6 also had
the common peak position, and further the Cases-C9,
-C10, and -C11. In these groups, identication of the
failed fuel subassembly is dicult when only the peak
position is considered. Thus, it is needed to take into
account the concentration distribution along the BP2
slit edge. The concentration peaks were detected by
sampling nozzle No. 4 in Cases-C2 and C3 as shown in
Figure 12. In Case-C2, the second-highest concentra-
tion was detected by nozzle No. 3, and the third-
highest concentration by No. 5. On the other hand, in
Case-C3, the second-highest concentration was de-
tected by No. 5, and the third-highest concentration by
No. 6. Thus, the identication of the failed fuel sub-
assembly of C2 or C3 is possible by means of compar-
ing the concentration distributions along the BP2 slit
edge. Identication of the failed fuel subassembly C5
or C6 is also possible as shown in Figure 12. However,
the concentration proles in Cases-C9, -C10, and -C11
along the BP2 slit edge are similar and it is dicult to
identify the failed fuel subassembly only by comparing
the concentration proles along the BP2 slit edge as
shown in Figure 13a. Thus, the concentration data
along the side wall of FHM plug is used in order to
conduct identication of the failed fuel subassembly C9
or C10 or C11. The concentration prole in Case-C11
is dierent from that in Case-C9 and -C10 as shown in
Figure 13b. The concentration proles in Case-C9 and
-C10 along the side wall of FHM plug are similar.
Although the highest concentration was detected by
nozzle No. 11 at BP2 in Case-C9, the highest
concentration was detected by sampling nozzle SC4
at the side wall of FHM plug in Case-C10. Therefore,
identication of failed fuel subassembly C9 or C10 or
C11 is possible by means of comparing the concentra-
tion proles along the BP2 slit edge and those along the
side wall of FHM plug.
As previously explained, the FP concentration under
the reactor conditions will be predicted by the three-
dimensional numerical simulation. Failed fuel subassem-
bly will be identied by comparing between the predicted
and the measured concentration distributions along the
sampling nozzles when a fuel subassembly under the UIS
slit has a pin failure. Therefore, numerical simulation is
required to predict the concentration prole with higher
accuracy. The validation results based on the experi-
mental data of the concentrations showed that the
numerical analysis method can predict the concentration
prole, i.e. the rst and second highest concentration
positions of the sampling nozzle. Thus, applicability of
the numerical simulation was conrmed to predict the FP
concentration under the reactor conditions.
6.2. Sampling nozzle arrangement around UIS slit
The experimental results showed that the sampling
nozzles set in the experiment could detect the FP
Figure 20. Comparison of concentration distributions at
BP2 in Case-C6.
58 K. Aizawa et al.
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

simulant concentration at least 2% of normalized
concentration, even in case of the failed subassemblies
under the UIS slit. In addition, identication of the
failed fuel subassembly under the UIS slit was achieved
by means of comparing concentration proles in the
UIS. These results satised the requirement of nozzle
Figure 21. Comparison of concentration distributions at the bottom of FHM plug in C12 case.
Journal of Nuclear Science and Technology, Volume 49, No. 1, January 2012 59
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

arrangement shown in section 2.1.2. Therefore, the
sampling nozzle arrangement (see Figure 5) used in the
experiment is suitable for JSFR.
7. Conclusions
A selector-valve type FFDL system is applied to the
JSFR design that has UIS with a slit above the core
and several sampling nozzles for the FFDL are set in
the UIS around the slit to detect the FP from the
subassemblies below the slit. Therefore, mixing process
in the UIS of complicated geometry and appropriate
arrangement of the sampling nozzles in the UIS were
investigated.
A water experiment was carried out to measure the
concentration distribution of FP simulant around the
slit in the UIS. A numerical simulation was also carried
and the simulation method was validated based on the
experimental data of the FP simulant concentrations.
The following results were obtained.
(1) Experimental results showed that the sampling
nozzles set in the experiment could detect the FP
simulant concentration at least 2% of normal-
ized concentration, even in case of the failed
subassemblies under the UIS slit. In addition,
identication of the failed fuel subassembly
under the UIS slit was achieved by means of
comparing concentration proles in the UIS.
(2) A numerical simulation using a commercial CFD
code of STAR-CD was applied to the experi-
mental analysis, where the second-order dieren-
cing scheme of MARS, k e RNG turbulence
model, and total mesh number of 12,000,000 were
used. The simulation results showed that the
simulation predicts the concentration distribu-
tions. Thus, it was shown that the numerical
simulation can predict the FP concentration
distributions under the reactor conditions.
(3) The suitable sampling nozzle arrangement in
the UIS was obtained for JSFR. The numerical
simulation can provide the FP concentration
prole corresponding to the failed subassembly
position. The comparison between predicted
prole and measured proles along this sam-
pling nozzle array can identify the failed fuel
subassembly.
The present study includes the results of Devel-
opment of elevated temperature structural design
method for fast reactor vessels and failed fuel detection
and location system entrusted to JAEA by the
Ministry of Education, Culture, Sports, Science and
Technology of Japan (MEXT).
Acknowledgments
The authors are grateful to Dr K. Ikeda of NDC for the
support in the numerical simulation. The authors are also
grateful to Dr Shiraishi and Mrs Sago of Mitsubishi Heavy
Industry for the support in the experiment.
References
[1] S. Kotake, Y. Sakamoto, S. Kubo, N. Uto, Y.
Kamishima, K. Aoto, and M. Toda, Development of
advanced loop-type reactor in Japan, Nucl. Technol. 170
(2010), pp. 133147.
[2] R.W. Scott, PFR components, Nucl. Eng. Int. 16 (1971),
pp. 646650.
[3] P. Coulon and L. Martin, Phenix Operation Experi-
ence; Fuel Management and Fuel Cycle, Proceedings
volume 1 5.6 International conference on fast reactors
and related fuel (FR91), Kyoto, Japan, Oct 28Nov 1
(1991), pp. 5.6.110.
[4] M. Banal, et al. Construction of the worlds rst full-
scale fast breeder reactor, Nucl. Eng. Int. 23 (1978), pp.
4360.
[5] A. Katoh, Y. Chikazawa, H. Obata, and S. Kotake,
Development of advanced fuel handling machine for
JSFR, J. Nucl. Sci. Technol. 47 (2010), pp. 642651.
[6] K. Aizawa, K. Fujita, H. Kamide, and N. Kasahara,
Detection Capability and Operation Patterns of a
Selector-Valve Failed-Fuel Detection and Location Sys-
tem for Large Sodium-cooled Reactor, Proceedings of
2011 International Congress on Advances in Nuclear
Power Plants (ICAPP 2011), Nice, France, May 25,
(2011), pp. 605613.
[7] G.N Abramovich, The Theory of Turbulent Jets, MIT
Press, Cambridge, USA, 1963.
[8] H. Kamide, K. Aizawa, J. Ohshima, O. Nakayama, and
N. Kasahara, Investigation on slit jet through upper
internal structure (UIS) in highly compact vessel of
sodium-cooled fast reactor, J. Nucl. Sci. Technol. 47
(2010), pp. 810819.
[9] N. Kimura, K. Hayashi, H. Kamide, M. Itoh, and
T. Sekine, Experimental study on ow optimization in
upper plenum of reactor vessel for a compact sodium-
cooled reactor, Nucl. Technol. 152 (2005), pp. 210222.
[10] Ministry of Education, Culture, Sports, Science and
Technology of Japan (MEXT), Development of elevated
temperature structural design method for fast reactor
vessels and failed fuel detection and location system,
JP21456071 (2007) in Japanese.
[11] Ministry of Education, Culture, Sports, Science and
Technology of Japan (MEXT), Development of elevated
temperature structural design method for fast reactor
vessels and failed fuel detection and location system,
JP21806618 (2009) in Japanese.
60 K. Aizawa et al.
D
o
w
n
l
o
a
d
e
d

b
y

[
1
1
2
.
2
1
5
.
3
6
.
1
4
4
]

a
t

1
7
:
5
5

1
9

M
a
r
c
h

2
0
1
4

You might also like