CFD software applications for transcritical free surface flow

Abstract

Flows in rivers, floodplains and coastal zones are very complex due to uneven bottom topography and irregular boundaries of the flow domain. In particular, when the flow shows strong gradients in water depth and velocity it is very difficult to predict, with accuracy, flow characteristics such as water profiles in all points of the domain. Traditional approaches solve shallow-water flow equations, known as Saint-Venant equations, when one or two dimension solutions can be adequate for obtaining most of the important flow characteristics. However, complex situations can require solving Navier-Stokes equations. In these cases, a two-phase flow problem must be solved and, as water profiles are not known in advance, only a numerical approach can be used to obtain approximate solutions. In addition, flow can be subcritical, supercritical or in a mixed-flow regime. These flow characteristics and complex geometries can make the use of in-house developed software difficult. The arrival of high performance computers and commercial software packages offers new possibilities in the field of numerical hydraulics. However, commercial software packages should be tested on some specific cases; so that these can be used with confidence. In this paper we solve, several cases of free surface flow that consider subcritical, supercritical, critical, oscillatory depth profiles and hydraulic jumps using a commercial package, CFX™. Most of these cases were proposed as benchmark solutions by MacDonald et al. (1997) for non-prismatic cross section, non-uniform bed slope and transition between subcritical and supercritical flow. Hydraulic jump cases consist of experimental data of hydraulics jumps obtained by Gharangik & Chaudhry (1991) for incident flow with Froude numbers of 2.3 and 4.23. In all simulated cases flow was described using a homogeneous model for each phase of the flow. Turbulence was modeled by using the well-known k-ε model. In addition, sensitivity to turbulence level in the entrance of flow domain was done to assure independence of results with this variable. Experimental facilities were properly represented in order to assure exact correspondence between boundary conditions of the model and the actual facility. Results obtained with CFX™ show excellent agreement with analytical solutions, for subcritical, supercritical, transitional and hydraulic jump cases. Special care with grid selection and entrance boundary condition is crucial to simulate with accuracy these types of flows. In particular, when a proper structured mesh is used, quality results are highly improved. Finally, results show to be insensitive to entrance turbulence conditions