| 摘要: |
| 轴向函数展开中子输运计算方法作为一种新的全堆高保真输运计算途径,具有稳定迭代收敛的优势,能避免传统2D/1D方法在强非均匀条件下可能出现的数值不稳定性。然而,现有研究大多仅采用一阶轴向试探函数,为达到足够精度需要细化轴向网格,轴向层数和内存占用显著增加。针对这一问题,将轴向试探函数扩展至二阶,推导了对应3×3系数矩阵的解析矩阵逆和矩阵指数求解方法,并在3D C5G7国际基准算例上实现和验证了二阶求解器的性能。结果表明:在较粗轴向网格条件下,二阶轴向展开方法能够显著提高燃料栅元功率分布计算精度(将功率计算相对误差降低至一阶方法的1/2~1/5),并使达到同等精度所需的轴向分层数减少为一阶方法的约1/3,从而将内存需求降低约50%。同时,二阶近似保持了稳定的数值迭代特性。然而,由于粗轴向网格条件下传统粗网扩散差分(coarse mesh finite difference,CMFD)迭代加速方法效率下降,二阶方法收敛所需迭代步数较一阶方法有所增加,总计算时间未出现明显缩减。总体而言,本文提出的二阶轴向函数展开方法在保持全堆输运计算高精度的同时,大幅降低了对轴向网格划分和内存资源的要求,对于开发高效高保真反应堆物理计算工具具有重要意义。 |
| 关键词: 轴向函数展开方法 间断有限元方法 二阶试探函数 C5G7基准题 |
| DOI:10.11918/202504076 |
| 分类号:TL323 |
| 基金项目:国家自然科学基金(12375167) |
|
| Second-order discontinuous finite element characteristic-line coupled method for neutron transport calculations |
|
Zhang Guangchun,Peng Jinghan
|
|
(Institute of Nuclear Science and Technology, School of Energy Science and Engineering, Harbin Institute of Technology, Harbin 150001, China)
|
| Abstract: |
| The axial function expansion method for neutron transport offers a novel full-core, high-fidelity transport calculation approach with the advantage of stable iterative convergence, avoiding the numerical instabilities that traditional 2D/1D methods may encounter under strongly heterogeneous conditions. However, most existing axial expansion implementations use only first-order (linear) axial trial functions; achieving sufficient accuracy therefore requires very fine axial meshes, leading to a substantial increase in the number of axial layers and high memory usage. To address this issue, this paper extends the axial expansion to second-order (quadratic) trial functions. Analytical solutions for the inverse and matrix exponential of the resulting 3×3 coefficient matrix are derived, and the performance of the second-order solver is implemented and validated on the 3D C5G7 international benchmark. The results demonstrate that with relatively coarse axial meshes, the quadratic approximation significantly improves the pin-wise power calculation accuracy (reducing the relative power error to about 1/2-1/5 of that of the first-order method), and decreases the number of axial layers required to achieve equivalent accuracy to roughly one-third of that needed by the linear method, thereby reducing the memory requirement by roughly 50%. Concurrently, the quadratic method also retains stable iterative convergence behavior. However, due to the diminished effectiveness of the conventional coarse-mesh finite-difference (CMFD) acceleration scheme on large axial mesh cells, the quadratic approximation requires more iterations to converge than the first-order method, so no significant reduction in total computing time was observed. In summary, the proposed second-order axial expansion method maintains high-fidelity in whole-core transport calculation accuracy while substantially lowering the mesh density and memory demand, which is of great significance for the development of efficient high-fidelity reactor-physics simulation tools. |
| Key words: axial functional expansion method discontinuous finite element method quadratic trial functions C5G7 benchmark problem |