Development of a generalized perturbation theory model for uncertainty propagation in lattice cell codes

这项研究建议的主要目标包括开发一种程序，可用于通过代码DRAGON传播基本核数据中的不确定性。包括这种不确定性分析特征的晶格单元代码将清楚地表示当前分析代码的量子步骤，因为人们将能够评估实验核数据中的不确定性对全球反应堆操作参数的影响。我们建议开发的灵敏度系数评估程序将基于直接方法，但也基于Monte Carlo和更有效的伴随技术。后者的方法是基于广义扰动理论（GPT）的方法已经实现的代码龙。我们还将解决的问题，得出一个GPT的方法来评估与在DRAGON编程的谐振自屏蔽模型的灵敏度系数。最后，一个新的横截面库，包括协方差矩阵将被定义和实施的DRAGON。这项工作将提高用于反应堆物理计算的分析工具的质量和多功能性，使其也能满足下一代反应堆（先进CANDU和第四代）的分析需求。事实上，由于横截面中的误差，在核心功率分布的精度的误差限制，可以使用这些工具进行评估，应大大减少在反应堆安全研究的任意性水平。

The main objective of this research proposal consists of developing a procedure that can be used to propagate through the code DRAGON the uncertainties in basic nuclear data. A lattice cell code that includes such uncertainty analysis features will clearly represent a quantum step over the current analysis codes since one will then be able to evaluate the effect of uncertainties in the experimental nuclear data on global reactor operation parameter. The procedures for sensitivity coefficient evaluation we propose to develop will be based on a direct method but also on a Monte Carlo and a more efficient adjoint technique. This later method is based on the generalized perturbation theory (GPT) method already implemented in the code DRAGON. We will also address the problem of deriving a GPT approach to evaluate the sensitivity coefficients associated with the resonance self-shielding model programmed in DRAGON. Finally, a new cross section library that includes covariance matrices will be defined and implemented in DRAGON. This work will lead to improvements in the quality and versatility of the analysis tools that are used for reactor physics calculations in such a way that they can also satisfy the analysis needs of the next generations of reactors (Advanced CANDU and Generation IV). The fact that the errors limits in the precision of the core power distributions due to errors in the cross sections can be evaluated using these tools should reduce substantially the level of arbitrariness in reactor safety studies.