Numerical analysis of air-water flows in hydraulic structures using computational fluid dynamics (CFD)

Abstract

The new legal regulations derived from climate change dictate that hydraulic structures must be designed to handle flood events associated with return periods up to 10,000 years. This obviously involves adapting the existing infrastructure to meet such requirements. In order to avoid risks in the restitution of the flow discharged to rivers, such as bank overflows or streambed erosion and scour processes, hydraulic design must be supported by reliable tools capable of reproducing the behavior of hydraulic structures. In the work presented herein, a fully three-dimensional CFD model to reproduce the behavior of different types of air-water flow in hydraulic structures is presented. The flow is assumed to be turbulent, isotropic and incompressible. Several RANS turbulence models are tested and structured rectangular meshes are employed to discretize the analyzed domain. The presence of two fluids is modeled using different VOF approaches and simulations are run using the PIMPLE algorithm. The model is implemented using the open-source platform OpenFOAM and its performance is compared to the commercial code FLOW-3D. The analysis is conducted separately on two different parts of hydraulic structures, namely: the spillway and the stilling basin. Additionally, a case of practical application, where the model reproduces the flow of a real-life case, is also presented in order to prove the suitability of the model to actual design cases. Mesh independence and model validation using experimental data are checked in the results of all the case studies. The sensitivity of the presented model to certain parameters is extensively discussed using different indicator variables. Among these parameters are turbulence closure, discretization scheme, surface tracking approach, CFD code or boundary conditions. Pros and contras of each of them are addressed. The analyzed turbulence models are the Standard k ? ?, the Realizable k ? ?, the RNG k ? ?, and the SST k ? ?. The discretization schemes under study are: a first-order upwind method, the second-order limited Van Leer method, and a second-order limited central difference method. The VOF approaches analyzed are the Partial VOF, as implemented in OpenFOAM, and the TruVOF, as implemented in FLOW-3D. In most cases, the Standard k ? ? model provides the most accurate estimations of water free surface profiles, although the rest of variables, with few exceptions, are better predicted by the RNG k ? ?. The latter model generally requires slightly longer computation times. The SST k ? ? reproduces correctly the phenomena under study, although it generally turned out to be less accurate than its k ? ? counterparts. As regards the comparison among VOF approaches and codes, it is impossible to determine which one performs best. E.g. OpenFOAM, using the Partial VOF, managed to reproduce the in- ternal hydraulic jump structure and all derived variables better than FLOW-3D, using the TruVOF, although the latter seems to capture better the momentum transfer and so all derived variables. In the case of flow in stepped spillways, OpenFOAM captures better the velocity profiles, although FLOW-3D is more accurate when estimating the water free surface profile. It is worth remark- ing that not even their response to certain model parameters is comparable. E.g. FLOW-3D is significantly less sensitive to mesh refinement than OpenFOAM. Given the result accuracy achieved in all cases, the proposed model is fully applicable to more complex design cases, where stilling basins, stepped spillways and hydraulic structures in general must be investigated.