A comparative study of rigid-lid and level-set methods for LES of open-channel flows: morphodynamics
A comparative study of rigid-lid and level-set methods for LES of open-channel flows: morphodynamics
复制标题
DOI：
10.1007/s10652-019-09703-y
复制
发表时间：
2019-07
影响因子：
2.2
通讯作者：
A. Khosronejad;M. Arabi;D. Angelidis;E. Bagherizadeh;K. Flora;A. Farhadzadeh
中科院分区：
文献类型：
--
作者：
A. Khosronejad;M. Arabi;D. Angelidis;E. Bagherizadeh;K. Flora;A. Farhadzadeh
明渠水流的计算建模通常使用刚性盖 (RL) 假设来规定水面。这种方法被认为更适合弗劳德数 ≪ 1的亚临界流态。在水工结构周围流动时，自由表面波动会影响引起的流体动力学。事实上，先前对 RL 假设对桥台周围明渠流大涡模拟 (LES) 影响的研究表明，这种假设可能会影响一阶和二阶湍流统计数据。在这项工作中，我们研究了 RL 假设对床形态动力学计算的影响。本文采用完全耦合的流动和床形态动力学虚拟流动模拟器（VFS-地球物理）模型对实验室规模的桥台模型周围的沉积物输送进行数值研究。使用质量守恒水平集（LS）方法和RL假设进行数值模拟，以评估雷诺数为7.9 × 104时自由表面高程对计算的床层形态动力学的影响。我们还进行了一项实验研究，以观察桥台模型周围的冲刷发展情况。将用 LS 和 RL 计算的瞬时冲刷模式和河床剖面与测量值进行比较。对于 LS 和 RL 计算结果，处于或接近床演化准平衡状态的床剖面预测的平均绝对误差分别约为 4.1% 和 4.7%。尽管两种方法的流场计算存在差异，但计算的床形态对两种数值方法似乎相对不敏感。然而，耦合的 LES-LS-形态动力学的计算成本比 LES-RL-形态动力学的计算成本至少大一个数量级。
Computational modeling of open-channel flows is usually carried out using the rigid-lid (RL) assumption to prescribe the water surface. Such an approach deems more appropriate for sub-critical flow regimes with Froude number ≪ 1. When the Froude number is locally high, e.g. flows around hydraulic structures, the free-surface fluctuations affect the induced hydrodynamics. Indeed, prior investigations on the effect of the RL assumption on large-eddy simulation (LES) of open-channel flows around bridge abutments revealed that such an assumption may influence both the first and second order turbulence statistics. In this work, we investigate the effect of the RL assumption on bed-morphodynamics calculations. The fully coupled flow and bed morphodynamics virtual flow simulator (VFS-Geophysics) model is herein employed to numerically investigate the sediment transport around a laboratory-scale model of abutment. Numerical simulations using a mass-conserving level-set (LS) method and a RL assumption are performed to evaluate the influence of free-surface elevation on the computed bed-morphodynamics at a Reynolds number of 7.9 × 104. We also conducted an experimental study to observe the scour development around the abutment model. The instantaneous scour patterns and bed profiles computed with the LS and RL are compared with the measurements. The mean absolute errors of bed-profile predictions at or near quasi-equilibrium of bed evolution, for the LS and RL computed results, are about 4.1% and 4.7%, respectively. Despite the differences in flow field computations of the two methods, the computed bed morphology appeared relatively insensitive to the two numerical approaches. However, the computational cost of the coupled LES-LS-morphodynamics is at least one order of magnitude greater than that of the LES-RL-morphodynamics.