Numerical modeling of a 90° openchannel confluence flow using openfoam CFD

• Ano de defesa: 2014
• Autor(a) principal: Victor Manoel Pavanelli Lira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: meio ambiente e recursos hídricos; Saneamento
• Link de acesso: http://hdl.handle.net/1843/BUBD-9ZKGDZ

Numerical modeling of flows is an important tool used before building proper physical models of hydraulic structures. In some projects, when the construction of physical models is not economically feasible, numerical models are used to directly evaluate the performance of these structures and to determine final details in their design. A large variety of numerical methods and turbulence models has been developed in the last three decades seeking to represent the enormous range of types of flows existing in nature or in industry. Although, there is still a gap between the foretold potential of this models and the assurance of their accuracy for different cases. Accordingly, the aims of this study are to set up a three-dimensional numerical model and to evaluate the accuracy of different turbulence models of reproducing the flow characteristics of a 90º open channel confluence, which, though consisting of a common and simple geometry, produces a roundly three-dimensional flow, not easy to be reproduced in numerical models. OpenFOAM, a free and open source CFD software, was used in this research. The experimental data used for validating the numerical models was taken from the experiments made by Weber et al. (2001). The most three dimensional flow scenario was analyzed, when q* = main channel inflow / total outflow = 0,25. Velocity fields were compared to evaluate the accuracy of simulation results from three different turbulence models [the Re- Normalization Group (RNG) k- model, the k- model and a Large Eddy Simulation (LES) model]. The water-surface was treated by a rigid lid approach. The general flow behavior generated by the numerical models was in good agreement with the experimental results, regardless of the rigid lid approachs natural inability of modeling the free-surface. However none of the turbulence models could reproduce properly the secondary current or the helicoidal current downstream the junction. Though the LES model was meant to be more powerful, its accuracy was worse than the two-equation turbulence models. No big difference was found in the performance of the RNG k- and the k- models. However, since the RNG k- better predicted the separation zone, it can be said that it was the most efficient turbulence model for the case analyzed in this research.