Professional Documents
Culture Documents
net/publication/263918335
CITATIONS READS
0 386
1 author:
YASSINE Boulaich
National Center for Energy Sciences and Nuclear Techniques
68 PUBLICATIONS 240 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by YASSINE Boulaich on 18 December 2014.
a r t i c l e i n f o a b s t r a c t
Article history: In the present work, we analyze the CREOLE experiment on the reactivity temperature coefficient (RTC)
Received 10 November 2010 by using the three-dimensional continuous energy code (MCNP5) and the last updated nuclear data eval-
Received in revised form 3 June 2011 uations. This experiment performed in the EOLE critical facility located at CEA/Cadarache, was mainly
Accepted 5 June 2011
dedicated to the RTC studies for both UO2 and UO2 –PuO2 PWR type lattices covering the whole temper-
ature range from 20 ◦ C to 300 ◦ C. We have developed an accurate 3D model of the EOLE reactor by using
the MCNP5 Monte Carlo code which guarantees a high level of fidelity in the description of different con-
figurations at various temperatures taking into account their consequence on neutron cross section data
and all thermal expansion effects. In this case, the remaining error between calculation and experiment
will be awarded mainly to uncertainties on nuclear data. Our own cross section library was constructed
by using NJOY99.259 code with point-wise nuclear data based on ENDF-BVII, JEFF3.1 and JENDL3.3 eval-
uation files. The MCNP model was validated through the axial and radial fission rate measurements at
room and hot temperatures. Calculation-experiment discrepancies of the RTC were analyzed and the
results have shown that the JENDL3.3 and JEFF3.1 evaluations give the most consistent values; the dis-
crepancy is less than 0.23 pcm/◦ C. Whereas, when using ENDF-BVII evaluation for UOX configuration,
this discrepancy reaches a value of 0.63 pcm/◦ C. In order to specify the source of this relatively large dis-
crepancy in the case of ENDF-BVII nuclear data evaluation, the keff discrepancy between ENDF-BVII and
JENDL3.3 was decomposed by using sensitivity and uncertainty analysis technique. The thermal energy
range of 238 U(n,␥) absorption cross section was found to contribute to the major part of the observed keff
discrepancy between ENDFB-VII and JENDL3.3 evaluations.
© 2011 Elsevier B.V. All rights reserved.
0029-5493/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.nucengdes.2011.06.009
2928 Y. Boulaich et al. / Nuclear Engineering and Design 241 (2011) 2927–2932
Table 1
Experiment fuel rods characteristics (Erradi et al., 2003).
Pellet diameter 8 mm
Outside diameter of cladding 9.4 mm
Uranium enrichment 3.1% for UO2 fuel rods and natural uranium
for UO2 –PuO2 fuel rods
Plutonium characteristics 80 fuel rods with 3.2% of fissile Pu
120 fuel rods with 2% of fissile Pu
2. CREOLE experiment
Measurements were performed in the following four tempera- where (T) is the density as function of temperature, x(T) is the
ture ranges related to four corresponding driver core loadings to dimension as function of temperature, ˛(T) is the linear coefficient
achieve the initial reactivity excess (Erradi et al., 2003): of thermal expansion whose values are given in Table 2.
Figs. 2 and 3 show the geometry of EOLE reactor as it was pre-
- 1st range: 20–111 ◦ C sented in the MCNP Visual Editor in comparison with real geometry.
- 2nd range: 111–186 ◦ C
- 3rd range: 186–242 ◦ C 4. Results and discussion
- 4th range: 242–289 ◦ C
4.1. Fission rate distributions
3. MCNP model of EOLE reactor
The EOLE MCNP model was validated through the axial and
For the purpose of developing our 3D model of EOLE reactor, we radial fission rate distributions by comparing the measured and
have used the MCNP5 code which allows describing the real geom- calculated values.
etry and materials composition. This leads to minimize the number The axial fission rate is calculated every 2-cm along the experi-
of approximations and to highlight the errors linked to nuclear data mental tube using the track length estimate of flux with multiplier
libraries. The calculations were performed with this Monte Carlo tally (FM4). The measured and calculated axial profiles for UOX
code based on the ENDF/B-VII, JEFF3.1 and JENDL3.3 cross section and MOX configurations at room temperature are displayed in
libraries. These libraries were constructed, for all the isotopes con- Figs. 4 and 5, respectively.
Y. Boulaich et al. / Nuclear Engineering and Design 241 (2011) 2927–2932 2929
Table 2
Linear thermal expansion coefficients (Erradi and Santamarina, 2008).
˛ (◦ C−1 ) 8379 × 10−6 + 5792 × 10−9 T 5770 × 10−6 + 6154 × 10−9 T 18 × 10−6
For both of these configurations, there is an overall satisfactory and lower part of fuel elements, while in the MOX configuration
agreement between calculation and experiment especially around neutrons are reflected by graphite and aluminum from upper part
the center region with differences essentially within the uncertain- and by aluminum in the lower part.
ties. The discrepancies between calculation and experiment toward
The observed asymmetry in the axial profiles of fission rates for the edge for the MOX configuration, at ∼30-cm from the core
the two configurations is due to the presence of the graphite in the mid-plane, are mainly due to the lower and upper aluminum
upper part of the fuel element as reflector. The UOX configuration reflectors (Klann et al., 2004). In fact, we can notice that the
is characterized also by the presence of stainless steel in the upper difference in the lower part is greater because of the large quantity
ENDF-BVII
JEFF3.1 1.00 ENDF-BVII
1.0 JENDL3.3 JEFF3.1
EXP JENDL3.3
0.98 EXP
0.9
Normalized fission rate
0.6
0.90
0.5
0.88
measured and calculated radial profiles for the UOX and MOX con-
1.00
figurations, respectively, at room temperature. One can find that
the calculated values are reasonably in good agreement with the
experiment except near the boundaries because of the proximity 0.95
of the void region.
The axial fission rate profiles can also be used to determine the
axial buckling by fitting the results with a cosine function (Baeten 0.85
et al., 2004). Practically the cosine function can be used in a region
where the influence of the upper and lower water reflector is neg- 0.80
1 2 3 4 5 6 7 8 9 10
ENDF-BVII H(cm)
JEFF3.1
JENDL3.3
1.05 Fig. 7. Calculated radial fission rate distribution for MOX configuration with fitting
EXP
curves in comparison with the experiment.
1.00
0.95
ligible. Results of the axial buckling compared to the experimental
Normalized fission rate
0.90 values are summarized in Table 3. As it can be seen, the very large
0.85 extrapolated distance (active fuel 70 cm) due to the streaming effect
near the boundaries is well predicted by the calculation.
0.80
It must be noted that all of these calculations were performed by
0.75 using the last updated nuclear data evaluations (ENDF-BVII, JEFF3.1
0.70 and JENDL3.3) with 432 millions of neutron histories (7200 active
cycles with 60,000 neutrons each, and 100 discarded cycles for con-
0.65
0.60 Table 3
0.55 Extrapolated length in the central test loop at room and hot temperatures for the
UOX configuration by using ENDF-BVII nuclear data evaluation in comparison with
-40 -30 -20 -10 0 10 20 30 40 experimental values.
4.4. Reactivity temperature coefficient In the case of MOX configuration, the discrepancy is within the
experimental uncertainties for each evaluation (<0.23 pcm/◦ C). It is
In a given temperature range [T1 , T2 ], the calculation of the reac- probable that the relatively large discrepancy produced when using
tivity temperature coefficient was performed by calculating the ENDF-BVII is concealed by compensation with plutonium isotopes
effective multiplication factors keff (T1 ) and keff (T2 ) at temperatures contribution.
T1 and T2 , respectively, and using the following expression:
4.5. Sensitivity and uncertainty analysis
1 keff (T1 ) − keff (T2 )
˛=
T keff (T1 )keff (T2 ) As it is shown above, integral parameters are poorly reproduced
by ENDF-BVII. Sensitivity analysis is a useful tool to identify the
It must be noted that the reactivity temperature coefficient calcu- nuclear data contribution to the discrepancy between calculated
lations will include take into account both Doppler coefficient and integral quantities in fission reactors. Hence, in order to under-
moderator temperature coefficient based on changes in moderator stand the relatively large discrepancy between results calculated
density and the moderator and reflector temperatures. by JENDL3.3 and ENDF-BVII evaluations for UOX configuration, a
Results of the RTC calculation are compared to the experimental preliminary sensitivity and uncertainty study on the multiplication
values in Figs. 8 and 9. factor was carried out to evaluate the impact of neutron cross-
For the UOX configuration, one can found that the calculation- section at room temperature (Takeda and Van Rooijen, 2009).
experiment discrepancy is within the experimental uncertainties These sensitivity calculations were processed by nuclide and
(<0.19 pcm/◦ C) when using JENDL3.3 or JEFF3.1 nuclear data evalu- reaction type using the PERT card of MCNP code for fifteen energy
ations; whereas this discrepancy reaches an absolute value of 0.63 groups (McKinney and Iverson, 1995; Hendricks et al., 1999). Once
when we use ENDF-BVII. This relatively large discrepancy can be the sensitivity coefficient matrix Sr for each integral parameter and
awarded mainly to uncertainties on nuclear data. the covariance matrix D are available, the uncertainty of the inte-
Table 4
Reactivity excess for UOX and MOX configurations calculated by using different nuclear data evaluations at room temperature in comparison with experimental values.
UO2 clean lattice 325 +259 pcm +185 pcm +86 pcm
UO2 –PuO2 lattice 295 +350 pcm +302 pcm +229 pcm
Table 5
Uncertainty (in pcm), by isotope, due to cross-sections for the UOX configuration.
5. Conclusion
References
Fig. 9. Calculation-experiment discrepancy on the RTC as function of temperature
Baeten, P., et al., 2004. Critical experiment with low-moderated MOX rods in VENUS.
for MOX configuration with fitting curves.
In: Workshop Proceedings , Paris, France, 14–15 April.
Chakir, E., 2005. Analyse des benchmarks expérimentaux sur le coefficient de tem-
pérature des réseaux des réacteurs à eau par la méthode de Monte Carlo et
gral parameter can be evaluated as follows (Salvatores and Jacqmin, implication sur les données nucléaire de base. PhD Thesis, University Mohamed
2008; Salvo, 2002): V, Rabat.
Erradi, L., 1982. Etude des effets de températures dans les réseaux caractéristiques
R2 = SR+ DSR des réacteurs nucléaires de la filière à eau ordinaire. PhD Thesis, Université de
Paris Sud Orsay.
In our case, each covariance matrix will be Erradi, L., 2007. Creole PWR reactivity temperature coefficient experiment. Cea
Report, CEA/Cadarche/DER/SPRC.
replaced by a diagonal matrix representing the
Erradi, L., Santamarina, A., 2008. Interprétation de l’expérience CREOLE sur le coeffi-
discrepancy between ENDF-BVII and JENDL3.3 cient de température des REP UOX et MOX par les codes APOLLO2.8 et TRIPOLI4.5
nuclear data evaluations in the fifteen energy groups (Rochman avec la bibliothèque JEFF3.1.1. Note technique de cea, Novembre.
et al., 2007; NEA/NSC/DOC, 2006). Erradi, L., Santamarina, A., Litaize, O., 2003. The reactivity temperature coefficient
analysis in light water moderated UO2 and UO2 –PuO2 lattices. Journal of Nuclear
Table 5 presents the contribution to the multiplication factor Science and Engineering 144, 47–73.
change of nuclide and reaction type. The energy dependence of the Hendricks, J.S., et al., 1999. MCNP perturbation capability for Monte Carlo criticality
contribution is shown in Table 6 for 238 U capture reaction. calculations. In: Sixth International Conference on Nuclear Criticality Safety ,
Versailles, France, September 20–24.
As it can be seen from these tables, the multiplication factor Klann, R.T. et al., 2004. Final Report of the International Nuclear Energy Research
was found to be most sensitive to the thermal energy range of 238 U Initiative OSMOSE Project (FY01-FY04). Nuclear Engineering Argonne National
capture cross section. In fact, for LWR calculations, the 238 U(n,␥) Laboratory, December 17.
Leal, L., et al., 2007. ORNL Methodology for Covariance Generation for Sensitiv-
reaction is considered as the most important reaction in the reso- ity/Uncertainty Analysis. In: Proceedings of the 8th International Conference on
nance (Litaize et al., 2002). Nuclear Criticality Safety (ICNC 2007), Vol. II , May 28–June 1, St. Petersburg,
A total uncertainty of ∼227 pcm is obtained by squaring each Russia, pp. 25–29.
Litaize, O. et al., 2002. Analysis of the Mistral Experiment with APOLLO2 – Qualifi-
component of Table 5 and adding positive values or subtracting cation of Neutronic Parameters of UOX and MOX. PHYSOR 2002, Seoul, Korea,
negative ones (Leal et al., 2007). October.
MacFarlane, R.E., 2000. Understanding NJOY, Trieste, 13 March–14 April.
MacFarlane, R.E., 2002. NJOY-99 Nuclear Data Processing System. Los Alamos
Table 6 National Laboratory, Los Alamos, USA.
238
Uncertainty (in pcm), by energy, due to U capture cross-sections for the UOX McKinney, G.W., Iverson, J.L., 1995. MCNP perturbation technique for criticality
configuration. analysis. In: ICNC Meeting , Albuquerque, NM, September 17–21.
NEA/NSC/DOC, 2006. Benchmark on the KRITZ-2 LEU and MOX critical experiments.
Group E (MeV) cap Nuclear Science, NEA/NSC/DOC (2005) 24, Final Report.
Rochman, D. et al., 2007. Preliminary Cross Section and n-Bar Covariances for WPEC
1 1.10E−07 136.04
Subgroup26. Brookhaven National Laboratory, BNL-77407-2007-IR, January.
2 5.40E−07 37.25
Rose, P.F., 1991. ENDF-201, ENDF/B-VI Summary Documentation, 4th edition. BNL-
3 4.00E−06 23.71 NCS-17541.
4 2.26E−05 54.09 Salvatores, M., Jacqmin, R., 2008. Uncertainty and target accuracy assessment for
5 4.54E−04 13.95 innovative systems using recent covariance data evaluations. NEA/WPEC-26, a
6 2.04E−03 12.53 Report by the Working Party on International Evaluation Co-Operation of the
7 9.12E−03 −7.55 NEA Nuclear Science Committee.
8 2.48E−02 −25.79 Salvo, J.D., 2002. Contribution à l’Etude des Incertitudes des Paramètres Neu-
9 6.74E−02 −5.02 troniques d’un Cœur Compact et Hétérogène: Le Réacteur d’irradiation Jules
10 1.83E−01 38.52 Horowitz. Report CEA-R-6015, Décembre.
11 4.98E−01 5.63 Sunny, C.S., et al., 2008. KAMINI reactor benchmark analysis. Annals of Nuclear
12 1.35 3.34 Energy 35.
13 2.23 9.86 Takeda, T., Van Rooijen, W.F.G., 2009. Sensitivity Coefficients for Fast Reactor Core
Analysis. Research Institute for Nuclear Engineering, University of Fukui.
14 6.07 −8.17
X-5 Monte Carlo Team, 2003. MCNP – A General Monte Carlo N-Particle Transport
15 19.60 −0.10
Code, Version 5. Los Alamos National Laboratory, April 24.
Total 156.66