三维中子动力学程序cosKIND针对核电厂仿真应用的典型基准题验证

Typical Benchmarks Verification of Neutron Kinetic Code cosKIND for the Simulation Application of Nuclear Power Plant

重大专项"CAP1400全范围模拟机自主化核心技术研究"课题需要自主开发应用于全范围实时仿真计算的物理程序,该程序需具有较高的精度和效率指标要求。为获取COSINE软件包中子动力学程序cosKIND仿真改造的定量依据,对基于节块展开法和全隐向后差分的两群六组时空中子动力学求解器进行二维TWIGL和三维LMW国际基准题测试。在测试过程中,针对基准题特点设计专门的程序测试接口,采用三维轴向全反射边界模型等效模拟二维堆芯问题,并采用逐网格线性截面变化近似模拟控制棒动作。测试结果显示,该求解器的计算结果与参考值相比整体偏小,求解器的计算精度基本满足仿真计算的需求,且计算精度随时间尺度和空间尺度的变化趋势基本合理;但是通过对黑盒测试结果的深入分析表明,该求解器和参考文献中的程序由于截面引入方式的问题,在时间步长较粗时可能导致较大的计算误差。该测试结果为后序基于cosKIND来进行开发优化的求解器提供了定量依据,也为后期cosKIND程序应用于全范围模拟机提出了改进建议。

The self-developed physical calculation code for CAP1400 FSS needs high accuracy and efficiency. The solver of cosKIND, which based on the NEM and fully implicit finite difference method, is tested by the TWIGL and LMW benchmarks. Special interfaces are set up and the test uses the three-dimensional axial reflection boundary model to simulate the two-dimensional core problem and uses the linear section of each mesh to simulate the movement of control rods. It shows that the overall calculation results are smaller than the reference, the accuracy satisfies the demand of the simulation, and the accuracy basically meets the trends over the time step and the meshing size. However, the method of cross-section introduction should be improved to reduce the errors when the solver uses longer time steps, and the same results are founded in the reference. This test provides quantitative analysis for the development of the solver and improvement suggestions for the application of cosKIND to FSS.