Optimization of inverse-Prandtl of Dissipation in standard k-ε Turbulence Model for Predicting Flow Field of Crossflow Turbine

Abstract

Despite the successful use of the Standard model in simulating turbulent flow for many industrially relevant flows, the model is still less accurate for a range of important problems, such as unconfined flows, curved boundary layers, rotating flows, and recirculating flows. As part of the authors’ effort to extend the model applicability and reliability, this paper aims to study the effects of diffusivity parameter called the turbulent Prandtl number of dissipation rate () on the Standard model performance for predicting recirculating flow in a crossflow water turbine. The value of this parameter was varied from 0.5 to 1.5 in the CFD simulations, and the results were compared to the more sophisticated model, namely the RNG , which has been first qualitatively validated by an experimental result. In addition, the parameter value was also adjusted using the Multi-Linear Regression (MLR) method ranging from 0.42 to 1.5 to complement the CFD simulations. It was observed that reducing the value is effective in minimizing the average deviation of the turbulence properties concerning the RNG model. However, the adjusted model still faces difficulty in accurately predicting the pressure and velocity field. Based on this result, adjusting the constant in the Standard turbulence model has the potential to improve the model performance for modelling recirculating flow in terms of the turbulence properties, but still needs further investigation for the flow properties.

Analysis of Mesh Resolution Effect to Numerical Result of CFD-ROM: Turbulent Flow in Stationary Parallel Plate

“…Richardson extrapolation (RE) method is used to predict the discretization error. RE method has been used to select the optimum mesh in CFD-FOM [39]. The equation are as follows [40]: Calculate mesh size (h) (30) r is mesh refinement factor.…”