用两种计算流体力学模型与物理验证模型研究横向坝坎块石缓坡的水力荷载

摘要

横向坝坎块石缓坡（crossbar block ramp）可保持河床稳定并改善河流的生态连通性。确定其设计荷载时，两大不确定性来源是其强烈的紊流流动与下游壅水影响。本文用 OpenFOAM 建立了整条横向坝坎块石缓坡的两个计算流体力学（CFD）模型，通过记录单道坝坎上的荷载（力与力矩）来处理这两个问题。就湍流而言，第一个模型采用雷诺平均 Navier-Stokes（RANS）方法，第二个模型采用大涡模拟（LES）。两个模型的流动均为非定常，并对自由表面加以追踪；网格完全相同，约 3030 万个单元，分布在 480 个处理器核心上，计算的采样时段为 180 s。CFD 模型以物理模型实测的水位与压力加以检验，各模型比尺相同，均为 1∶20/3。每个模型均研究了代表三种典型流态的三级流量。结果表明，各模型都再现了这三种流态的特征。RANS 模型最快可比 LES 快 5.3 倍，但产生了过大的波浪，很可能导致坝坎荷载被高估或低估。LES 模型与物理模型吻合良好，可用于荷载预测；不过若要获得统计上可靠的最大荷载估计，还需更长的采样时段和更多级流量。两个模型都表明，为找出临界荷载，必须考虑尾流干扰流态中的非定常波浪。本文结果可为今后研究横向坝坎块石缓坡水力荷载的方法选择提供依据。

Abstract

Crossbar block ramps provide bed stability and facilitate ecological connectivity in rivers. Two major sources of uncertainty in determining their design loads are their massively turbulent flow and backwater influence. Here, they were addressed using two computational fluid dynamics (CFD) models (created with OpenFOAM) of a complete crossbar block ramp by recording loads (forces and moments) on single crossbars. To model turbulence, the first model used the Reynolds-averaged Navier-Stokes (RANS) approach, and the second model used Large Eddy Simulation (LES). The flow in both models was transient and the free surface was tracked. Their mesh was identical and consisted of about 30.3 million cells distributed on 480 processor cores. The computed sampling interval was 180 s. The CFD models were tested against measurements of water level and pressure from a physical model. All models were in the same scale, 1:20/3. Three discharges representing three typical flow regimes were studied in each model. Characteristics of the flow regimes were reproduced in all models. The RANS model was up to 5.3 times faster, but produced excessive waves, which likely caused over- and underestimation of crossbar loads. The LES model showed good agreement to the physical model and could be used for load predictions. However, a longer sampling interval as well as a larger variety of discharges would be required to obtain stochastically reliable estimations for the maximum loads. Both models indicated that unsteady waves in the wake-interference flow regime must be considered to find critical loads. The results can support decisions on the methods for future investigations of hydraulic loads on crossbar block ramps.