基于光滑粒子元的水-沙两相流冲刷数值仿真

基础河床冲刷逐渐成为水工建筑物等结构破坏的主要原因，传统分析方法基于欧拉网格法模拟，易出现网格畸变、计算不收敛等情况，降低了计算效率与精度.基于拉格朗日坐标系下无网格光滑粒子元法（Smoothed Particle Hydrodynamics，SPH）对河床冲刷过程进行数值模拟；将泥沙相视为非牛顿流体，分为沉积物、推移质、悬移质三种状态；引入Drucker-Prager与Shields应力模型作为泥沙状态转化判断准则，为不同状态泥沙粒子赋予不同流变特性，使计算结果可准确描述水-沙耦合过程. 提出了基于SPH多相流模型的河床冲刷数值仿真改进算法，编写了泥沙冲刷计算模块，采用GPU进行计算加速，最后将所得数值水槽模型与Louvain溃坝试验及其他同类仿真结果进行对照. 结果表明：本文模型能够更为准确地反映冲刷发展整体趋势；各时刻自由液面、水-沙交界面轮廓均方根误差均在合理范围内；数值模型与试验结果吻合度较高.

Numerical Simulation of Scour by Liquid-sediment Two-phase Flow Based on Smoothed Particle Hydrodynamics

Scours at the foundations of hydraulic structures have gradually become the major cause of structural damage. The traditional numerical simulation is based on the Euler mesh method， which is not easier to converge or accompanied by large deformation of grids and eventually results in the loss of solution efficiency and accuracy. This paper numerically simulated the scour process of riverbed by utilizing the Smooth Particle Hydrodynamics（SPH）based on the Lagrange coordinate. It treated the sediment phase as a non-Newtonian phase and divided it into three states： sediment， bed load， and suspended load. To accurately describe the effect of liquid-sediment interaction， the Drucker-Prager and Shields stress model was introduced in the numerical model as the criteria for the transformation judgment between three states of sediment， including assigning different rheological properties for sediment particles in different states. In this study， a modification scouring algorithm based on the two-phase flow was proposed， and the sediment scouring calculation module was built and accelerated by GPU. Finally， a numerical flume model was designed for comparison with the Louvain dam-break experiment as well as a similar numerical model. The conclusions are drawn： the current numerical model proposed could more accurately reflect the overall trend of scour development； the RMSE of the free surface profile of the water and sediment-liquid interface were within a reasonable range at the specified moment； the result of numerical model was in good agreement with the experimental data.