Performance of Various Turbulence Models for Simulating Sub-critical High-Reynolds Number Flows over a Smooth Cylinder
Performance of Various Turbulence Models for Simulating Sub-critical High-Reynolds Number Flows over a Smooth Cylinder
复制标题
DOI：
10.2514/6.2021-2762
复制
发表时间：
2021-08
期刊：
影响因子：
--
通讯作者：
Michael D. Mays;S. Laizet;S. Lardeau
中科院分区：
文献类型：
--
作者：
Michael D. Mays;S. Laizet;S. Lardeau
模拟雷诺数为140，000的具有低水平入口湍流的光滑圆柱体上的流动仍然是评估湍流闭合模型和数值方法性能的稳健测试案例。本研究考虑了各种封闭水平，雷诺平均Navier-Stokes方程（RANS），分离和大涡模拟（DES/LES）和混合RANS/LES，以确定其适用于这种情况下，考虑到其敏感性的空间分辨率和所使用的数值方案。在这项研究中选择的RANS和DES封闭都不能捕获正确的物理行为的流动，主要是由于在模型配方，防止在自由剪切层中形成不稳定性的弱点。LES壁自适应局部涡粘性（WALE）模型表现良好，具有足够好的细化网格，但它仍然是一个计算要求很高的方法。一种新的尺度分辨混合（SRH）模型，正式来自时间滤波的Navier-Stokes方程，表现出良好的协议与实验的数量感兴趣。与其他RANS和混合RANS/LES模型相比，SRH模型在粗网格上的表现要好得多，并且可以产生与LES WALE模型相似的结果。这项工作的主要结论是，强大的行为的SRH模型，再加上其潜在的大幅减少计算需求，使其成为一个很好的候选人，研究高度分离的外部流在高雷诺数。
Simulating the flow over a smooth cylinder with low-level inlet turbulence for a Reynolds number equal to 140,000 remains a robust test case to evaluate the performance of turbulence closure models and numerical methods. This study considers a variety of closure levels, Reynolds-Averaged Navier-Stokes (RANS), Detached and Large Eddy Simulations (DES/LES) and hybrid RANS/LES, to determine their applicability to this case, with consideration given to their sensitivities to the spatial resolution and to the numerical schemes used. Neither the RANS nor DES closures selected in this study are able to capture the correct physical behaviour of the flow, largely due to weaknesses in the model formulations that prevent the formation of instabilities in the free shear layer. The LES Wall-Adapting Local Eddy-viscosity (WALE) model performs well with a sufficiently well refined mesh but it remains a computationally demanding method. A novel Scale Resolving Hybrid (SRH) model, formally derived from temporal filtering of the Navier-Stokes equations, shows an excellent agreement with experiment for the quantities of interest. The SRH model performs far better on a coarse mesh by comparison to other RANS and hybrid RANS/LES models and can produce results similar to the LES WALE model. The main conclusion of this work is that the robust behaviour of the SRH model, coupled with its potentially substantial reduction in computational demand, makes it an excellent candidate to study highly-separated external flows at high Reynolds numbers.