Use of integral experiments for the assessment of a new 235U IRSN-CEA evaluation

The Working Party on International Nuclear Data Evaluation Co-operation (WPEC) subgroup 29 (SG 29) was established to investigate an issue with the 235 U capture cross-section in the energy range from 0.1 to 2.25 keV, due to a possible overestimation of 10% or more. To improve the 235 U capture crosssection, a new 235 U evaluation has been proposed by the Institut de Radioprotection et de Surete Nucleaire (IRSN) and the CEA, mainly based on new time-of-flight 235 U capture cross-section measurements and recent fission cross-section measurements performed at the n_TOF facility from CERN. IRSN and CEA Cadarache were in charge of the thermal to 2.25 keV energy range, whereas the CEA DIF was responsible of the high energy region. Integral experiments showing a strong 235 U sensitivity are used to assess the new evaluation, using Monte-Carlo methods. The k eff calculations were performed with the 5.D.1 beta version of the MORET 5 code, using the JEFF-3.2 library and the new 235 U evaluation, as well as the JEFF-3.3T1 library in which the new 235 U has been included. The benchmark selection allowed highlighting a significant improvement on k eff due to the new 235 U evaluation. The results of this data testing are presented here.


Introduction
The Collaborative International Evaluated Library Organization Pilot Project, known as the CIELO project, is an initiative aimed at improving data evaluation by combining efforts from several institutions around the world. Over the last couple of years, evaluation work has been done to improve data evaluation for six isotopes, namely 1 H, 16 O, 56 Fe, 235−238 U, and 239 Pu. The Working Party on International Nuclear Data Evaluation Co-operation SG 29 was established to investigate an issue with the 235 U capture cross-section in the energy range from 0.1 keV to 2.25 keV [1]. The WPEC criticality calculation results indicated an overestimation of the 235 U capture crosssection of 10% or more, starting around 100 eV. To understand and solve the problem, recommendation was made to perform new capture cross-section measurements followed by a resonance evaluation. Hence, time-offlight capture cross-section measurements were done at the Rensselaer Polytechnic Institute (RPI) [2] and at the Los Alamos National Laboratory (LANL) [3]. Furthermore, recent fission cross-section measurements have been performed at the n TOF facility from CERN [4], which provided strong support to the fission standard values. Indeed, the normalization of these experimental data in the [7.8-11] eV energy range supported the standard value in this energy region and also reinforced the standard averaged fission cross-section values in the resonance region. These new measurements were used together with the SAMMY [5] computer code, developed by ORNL, to re-evaluate the capture and fission 235 U resonance parameters in the energy range from thermal to a e-mail: raphaelle.ichou@irsn.fr 2.25 keV. Integral quantities, such as fission and capture resonance integral and Westcott factors were also included in the evaluation. The CEA DIF was in charge of the evaluation at high energy and recommendations done by the CIELO organization [6,7] concerning the 235 U PFNS (prompt fission neutron spectrum) among others, were followed. Finally, a common IRSN-CEA evaluation called "u235 irsncea 1" was released. IRSN has carried out benchmark calculations, using few-selected benchmarks extracted from the International Criticality Safety Benchmark Evaluation Project (ICSBEP) Handbook [8] for testing the new 235 U evaluation over the whole energy range. The impact of the inclusion of the new evaluation was studied for selected critical benchmarks sensitive to the 235 U capture cross-section in the thermal, epithermal and fast energy ranges. The benchmark calculations were performed using Monte-Carlo codes.

Benchmark selection
The DICE database, associated with the ICSBEP Handbook, was used to select the experiments sensitive to 235 U cross-sections. The available integral sensitivities, mainly calculated with the TSUNAMI-3D module of SCALE, were extracted from the database in the thermal (E < 0.625 eV), epithermal (0.625 eV < E < 100 keV), and fast (E > 100 keV) energy ranges, for both the 235 U capture and fission cross-sections. Experiments showing too high measurement uncertainty were discarded. When possible, benchmarks showing the highest integral sensitivity to 235

JEFF-3.2 library
The impact of the new preliminary 235 U resonance evaluation in benchmark calculations was investigated using the selected benchmarks. The new set of 235 U resonance parameters was converted into the ENDF format and included in the JEFF-3.   to about +900 pcm (HECTOR) for all the selected benchmarks.
The C/E results, obtained with MORET 5, using full JEFF-3.2 library together with the new 235 U evaluation are shown in Fig. 2. The JEFF-3.2 C/E results (in red) for solutions sensitive to the thermal energy range are rather low but in a good agreement, within the 3σ uncertainty. But for cases sensitive to higher energies, the results are mainly overestimated. The C/E results, except for a few cases, are located at a upper limit of the uncertainty range or even higher. The two exceptions are BIGTEN and GODIVA results which perfectly match the experimental data with the JEFF-3.2 library. The C/E results obtained with the new 235 U in a JEFF-3.2 environment tend to be on average improved, compared to the full JEFF-3.2 results, especially in the intermediate energy range where a bias reduction is observed. For the benchmarks sensitive to the thermal energy range, the k eff results, which were a bit underestimated with JEFF-3.2, are increased and the C/E results are improved. Concerning the benchmarks sensitive to the intermediate and fast energy ranges, the averaged C/E tends to be decreased, so that the agreement becomes better. The k eff values of the ZEUS cases, for which the new 235 U resonance parameters have a strong impact in the intermediate energy range, become in good agreement and show a clear bias reduction. However, there remain some cases for which the C/E is improved but still stand outside or at the limit of the experimental uncertainty range: the HMF-070-003 and HMI-001 benchmarks. The HMI-001 benchmark, which is a 235 U (93%)/Fe cylinder reflected by stainless steel, sensitive to iron and aluminium, still shows a C/E result well above the uncertainty range. This could indicate that there could be a problem for iron and aluminium evaluations. The two exceptions, where the C/E gets worse, are the HECTOR and BIGTEN cases. The HECTOR experiment, which is an homogeneous fuel mixture of uranium oxide (92% enriched 235 U) and graphite, is very sensitive to graphite. This could be an indication that a more detailed study is necessary to understand the HECTOR benchmark results. For BIGTEN, which is an intermediate-enriched uranium benchmark with fast spectrum containing a large amount of 238 U, the new file leads to a strong decrease of 700 pcm, whereas JEFF-3.2 showed an excellent C/E result. In this case, including the new 238 U evaluation, proposed by CEA and IRMM, together with the new 235 U, is relevant. The C/E results obtained using JEFF3.2 library, together with the new 235 U and 238 U evaluations are shown in Fig. 3 and compared to the previous results. The new BIGTEN result, using both 235 U and 238 U resonance parameters, perfectly matches the experimental data. This points out the importance of evaluating the isotopes all together as they are strongly affected by each other.
Finally, the C/E results obtained with the new modified 235 U in a JEFF-3.2 environment, are overall in good agreement, and clearly tend to improve the results. The results tend to be even more improved when using the new 238 U evaluation. These results indicated that the new evaluations seemed to be going in the right direction. Consequently, both the new 235 U and 238 U evaluations were included in the JEFF-3.3T1 library.

JEFF-3.3T1 library
In order to evaluate the impact of these new 235 U and 238 U resonance parameters inclusion in JEFF-3.3T1 library, IRSN has performed data testing using JEFF3.3T1 library. The full JEFF-3.3T1 library was processed with GAIA 1.1 and the k eff calculations were performed for the same benchmark selection with the 5.D.1 beta version of the MORET 5 [11] code. The C/E results obtained using full JEFF-3.3T1 library are shown in Fig. 4. The results clearly indicate a discrepancy compared to full JEFF-3.2 results with a strong improvement in the intermediate energy range. Regarding the considerable amount of modified isotopes between JEFF-3.2 and JEFF-3.3T1 libraries, it is not clear how to achieve a final conclusion. However, these modified files tend to either unchange or decrease the k eff compared to the results using the new 235 U added in JEFF-3.2. The most notable results are the ZEUS cases whose k eff values are clearly decreased and fall nearby the 3σ uncertainty. The ZEUS benchmark is a graphite moderated and Cu reflected assembly, very sensitive to Cu that has been modified between JEFF-3.2 and JEFF3.3T1, which can explain the new ZEUS results. It could also be explained by the new carbon graphite in JEFF-3.3T1 coming from JENDL4 library. The HECTOR benchmark is neither improved. Nevertheless, JEFF-3.3T1 results are showing an averaged improvement for the selected benchmarks compared to JEFF-3.2 and work is going on to understand and improve the results.

k eff contributions
In order to explain the impact of the new evaluations on the benchmark k eff results, especially the impact of with: P = Production, A = Absorption, L = Leakage And if the leakage term is neglected: with: And the macroscopic absorption cross section is: Absor ption = f ission + (n,2n) + (n,3n) + (n,γ ) (5) Table 2 shows the integrated reaction rates over all energy bins for the FLATTOP benchmark, obtained using MCNP6 [12] and both full JEFF-3.2, JEFF-3.2 with new 235 U and with both new 235 U and 238 U. It emphasizes the fact that a small variation of a reaction rate has a strong impact on k eff . Using the new 235 U in JEFF-3.2 increases the total capture rate of around 6% and decreases the fission rate by 0.5%; this affects the k eff by a decrease of around 250 pcm. When adding the new 238 U file, production and absorption rates are almost unchanged but k eff value is increased by around 150 pcm. The indicated changes may be related to the leakage term in connection with the 238 U (elastic and inelastic) scattering cross-section.

Conclusions
IRSN and CEA have produced a new evaluation for 235 U in the thermal to fast energy range. IRSN, which is involved in the data testing, has carried out key critical benchmarks selection, regarding the sensitivities to the 235 U crosssection. MORET calculations have been performed for the selected benchmarks, using new evaluation in JEFF-3.2 library. The benchmark selection allowed highlighting a significant effect on k eff and on C/E results, with an observed improvement when using the new 235 U resonance parameters compared to full JEFF-3.2. An even better improvement with the new resonance parameters 238 U has been noted. The new 235 U evaluation was made available for inclusion in the JEFF cross-section library and has been included in the JEFF3.3T1 version. A collaboration work between IRSN, CEA Cadarache and CEA DIF is ongoing to keep improving the results via benchmark testing.