基于GNDS格式热散射律评价库的热中子散射截面计算方法研究

Research on Calculation Methods for Thermal Neutron Scattering Cross Sections Based on GNDS-Format Thermal Scattering Law Evaluated Libraries

摘要: 针对目前广泛采用的ENDF评价核数据库格式扩展困难、协方差数据关联弱、可读性较差等问题，国际上提出了GNDS格式评价核数据库格式，该格式旨在解决传统评价核数据库格式(如ENDF)的历史遗留问题，采用“结构化、现代化、面向应用和长期可持续”的思想，把评价核数据库变成一个完整、自描述、机器可读的核数据知识库。开发能处理GNDS格式的计算程序，对于推动反应堆物理计算的自主化发展具有重要意义。本文首先开发了GNDS格式热散射律评价核数据库处理程序GTES，然后采用自主化先进核截面处理程序AXSP进行了截面计算，最后采用国际临界基准题验证了截面处理精度。针对GNDS格式ENDF/B-VIII.1评价库中一声子模型石墨热散射律数据，采用AXSP计算得到的热散射截面，在0.001 eV以下能区，优于晶体石墨的计算结果且与更吻合实验值，精度提高约26%。针对金属氢化物，采用自适应网格，对比NJOY的固定网格，在高能区可以更好地描述振荡情况。针对同时含有相干弹性散射和非相干弹性散截面核素，创新性提出了散射角分布合并方法，与仅采用干弹性散射截面，最大偏差为17%；与仅采用非相干散射截面相比，最大偏差为32.564%。

Abstract: In response to the challenges of expanding the widely adopted ENDF evaluated nuclear database format, weak correlation of covariance data, and poor readability, the GNDS format for evaluated nuclear databases has been proposed internationally. This format aims to address the legacy issues of traditional evaluated nuclear database formats (such as ENDF). By adopting a “structured, modernized, application-oriented, and long-term sustainable” approach, it transforms evaluated nuclear databases into a comprehensive, self-describing, and machine-readable repository of nuclear data knowledge. Developing computational programs capable of processing GNDS-format data holds significant importance for subsequent reactor physics calculations. This paper first develops GTES, a processing program for GNDS-format thermal scattering law evaluated nuclear databases. Subsequently, the independently developed advanced nuclear cross-section processing program AXSP is employed for cross-section calculations. Finally, the accuracy of the cross-section processing is validated using international critical benchmark problems. For the one-phonon model graphite thermal scattering law data in the GNDS-format ENDF/B-VIII.1 evaluated library, the thermal scattering cross section calculated by AXSP is superior to that of crystalline graphite in the energy region below 0.001 eV, shows better agreement with experimental data, and achieves an accuracy improvement of approximately 26%. For metallic hydrides, the use of an adaptive mesh better captures oscillations in the high-energy region compared to the fixed-grid approach of NJOY. Furthermore, for nuclides containing both coherent and incoherent elastic scattering cross-sections, an innovative method for merging angular scattering distributions has been proposed. Compared to using only one type of scattering cross-section, this method achieves a remarkable improvement of up to 32.564%.