Numerical discretization and subgrid‐scale model effects on large‐eddy simulations of a stable boundary layer
Numerical discretization and subgrid‐scale model effects on large‐eddy simulations of a stable boundary layer
复制标题
DOI：
10.1002/qj.2888
复制
发表时间：
2016-06
影响因子：
8.9
通讯作者：
G. Matheou
中科院分区：
文献类型：
--
作者：
G. Matheou
在中等稳定边界层的模拟中，研究了大涡模拟（LES）模型性能的各个方面。LES采用常系数Smagorinsky-Lilly子网格尺度（SGS）闭合。考虑了三个模型参数：网格间距、SGS模型常数和平流离散化精度（分辨力）阶数。研究了二阶、四阶和六阶全保守非耗散平流格式。考虑的所有三个模型参数都显著影响LES结果。根据模型常数的值，确定了两种主要的误差产生机制。在模型常数较高的情况下，无论是在模型自旋上升的短时间内，还是在整个模拟过程中，都会观察到虚假的湍流坍缩。尽管这种虚假的模型特征之前已经被记录，并且可能在低分辨率模拟中被期望，但它取决于平流离散化的顺序，这意味着显著的离散化和SGS闭合相互作用。当模式常数值较低时，数值离散误差占主导地位，导致小尺度能量的积累和对地表热通量大小的过度预测。势温分布的差异与地表热通量密切相关。总体而言，四阶和六阶方案的性能明显优于二阶方案。四阶和六阶格式之间的差异相对较小，并且六阶格式增加的计算费用在大多数应用中可能无效，至少对于本研究中考虑的低阶统计量而言是如此。尽管Smagorinsky-Lilly闭包的结果显示了对所检查的所有模型参数的持续依赖，但对于几个参数组合，相对于参考模拟的差异很小。因此，与先前研究的结论相反，封闭可以准确地捕获适度稳定的流量。
Aspects of a large‐eddy simulation (LES) model performance are investigated in simulations of a moderately stable boundary layer. The LES utilizes the constant‐coefficient Smagorinsky–Lilly subgrid‐scale (SGS) closure. Three model parameters are considered: grid spacing, SGS model constant and order of accuracy (resolving power) of the advection discretization. Second‐, fourth‐ and sixth‐order fully conservative non‐dissipative advection schemes are examined. All three model parameters considered significantly affect the LES results. Depending on the value of the model constant, two main error‐producing mechanisms are identified. For high values of the model constant, spurious turbulence collapse, either during the short period of model spin‐up, or for the entire simulation duration, is observed. Even though this spurious model characteristic was previously documented, and perhaps expected for low‐resolution simulations, it depends on the order of the advection discretization, implying a significant discretization and SGS closure interaction. For low values of the model constant, numerical discretization errors dominate, leading to accumulation of energy at small scales and over‐prediction of the magnitude of the surface heat flux. Differences in potential temperature profiles are well correlated with the surface heat flux. Overall, the fourth‐ and sixth‐order schemes perform significantly better than the second‐order scheme. The differences between the fourth‐ and sixth‐order schemes are relatively small and the increased computational expense of the sixth‐order scheme may not be effective in most applications, at least for the low‐order statistics considered in this study. Even though the results of the Smagorinsky–Lilly closure show persistent dependence on all model parameters examined, for several parameter combinations the differences with respect to a reference simulation are small. Thus, in contrast to the conclusions of previous studies, the closure can accurately capture moderately stable flows.