Criticality Benchmark Test of CENDL-3.2 Based on ENDITS-2.1

The Chinese Evaluated Nuclear Data Library CENDL-3.2 was officially released in June 2020, and the criticality benchmark test results based on the Evaluated Nuclear Data Integral Testing System ENDITS-1.0 were introduced on the ND2019 meeting. At that time, the criticality benchmark test results showed that the CENDL-3.2 library gave the best results compared with ENDF/B-Ⅷ.0, JENDL-4.0 and JEFF-3.3 libraries. In order to further evaluate the quality of the CENDL-3.2 library and provide feedback for future evaluation data improvement, the ENDITS-2.1 was developed with 2237 criticality benchmark experiments adopted from the ICSBEP2006. The CENDL-3.2 library was tested with the extended benchmark test suit and the results were compared with those based on the latest international evaluation nuclear data libraries, including ENDF/B-Ⅷ.0, JEFF-3.3 and JENDL-5.0 libraries, and also compared with the results calculated with the CENDL-3.1 and CENDL-NP-1.3 libraries. During the validation, the ACE libraries for fast neutron were generated with the NJOY 2016 code with the same processing parameters, and the ACE library for thermal neutrons all shared the ENDF/B-Ⅷ.0 library. Firstly, all the benchmark test results were analyzed using the statistical chi-squares and the normalized chi-square values based on ENDITS-1.0 and 2.1 were compared. It was confirmed again by the validation results based on the ENDITS-2.1 that the accuracy of the criticality calculation of the CENDL-3.2 was significantly improved compared with the old version. The chi-square decreases from 14.81 to 8.39 which is a little better than that of ENDF/B-Ⅷ.0 library, which means the accuracy of the criticality calculations for the CENDL-3.2 library reaches the international leading level. At the meantime, one may notice that the critical calculation accuracy ranking of each evaluation library based on the ENDITS-1.0 is consistent with the ranking based on the ENDITS-2.1, which indicates that the selection of benchmark experiments in ENDITS is well representative. Secondly, according to the types of critical systems, the improvement of the critical calculation accuracy of the CENDL library was analyzed. Compared with the CENDL-3.1 library, the critical calculation accuracy of a total of 13 critical systems for the new library is improved, as the normalized chi-square reduction of the critical k_eff by more than 40%. Applying trend analysis against E_ALF of the benchmarks, combined with the sensitive nuclides of the critical benchmark, a preliminary analysis of the nuclear data sources of the k_eff improvement for each critical system was made, and shows that the improvement of the criticality performance is mainly due to the evaluation data for ²³⁵U, ²³⁹Pu, ²³³U, ²³²Th, ⁵⁶Fe, ¹H and W isotopes. Finally, based on the cumulative chi-square, normalized chi-square and Pearson chi-square tests related to sensitive nuclides, the source nuclides of the critical calculation bias of the CENDL-3.2 library were analyzed, the list of high-priority nuclides to be improved for the CENDL-3.2 library was given, and the evaluation data of a total of 16 nuclides were identified to be improved, such as ^235,238U, ^239,240Pu, ²³²Th, etc.