In this paper an attempt has been made to study the effects of floodplains width and discharges on flow field in prismatic compound channels. A three-dimensional Computational Fluid Dynamic (CFD) model is used to predict the velocity distribution, secondary flow circulation and boundary shear stress in prismatic compound channels with various floodplains widths. The ANSYS-CFX software and three different turbulence models, k-e, k-e Explicit Algebraic Reynolds Stress Models (EARSM) and Eddy Viscosity Transport, are used to solve Reynolds Averaged Navier-Stokes equations. The results of the numerical modeling were then compared with experimental data on prismatic compound channels with 100 mm, 200 mm, 300 mm, and 400 mm floodplain widths. The study shows that all turbulence models are capable to predict the depth-averaged velocity in prismatic compound channels, fairly well. However, to compare with the velocity distribution, discrepancy between experimental data and boundary shear stress calculated by numerical modeling are high. Also only k-e EARSM model is able to predict secondary flow circulations.

The dispersion of hazardous gas in the environment presents dangerous risks for people living close to chemical plants or storages. Since heavy gases tend to stay at lower levels and disperse at a slower pace in the atmosphere, they are potentially more dangerous. In this paper, various mathematical models for turbulence (including k-ε , RNG k-ε , EARSM, LES, DES) and their associated parameters have been assessed, compared and validated against the experimental data in various scenarios to find the most suitable one for atmospheric dispersion of dense-gases. This topic has been investigated and validated by a computational fluid dynamics (CFD) simulation of the Kit-Fox experiment. The precision of the CAD models, practicality, computational resource requirements, and some other factors have been considered and addressed in this paper to achieve a comprehensive solution for atmospheric dispersion. The results here suggest that the proper selection of the turbulence model and the turbulent Schmidt number is crucial. Our results indicate that the most promising combination in the case of atmospheric dense-gas dispersion is the RNG k-ε model with the Schmidt number of 0. 4, considering the demand for accuracy and computational resource.

Today, due to advances in computing power, Reynolds Averaged Navier-Stokes (RANS) solvers are widely preferred for quasi-three-dimensional (Q3D) blade-to-blade analysis. This study investigates the performance of different flux calculation methods and turbulence models with a density-based RANS solver (Numeca®) in blade-to-blade analysis. A block-structured mesh topology is used to create a solution grid around the airfoil. Spatial discretization is performed in the pitchwise direction to represent the quasi three-dimensional flow, while only one computational cell is used in the radial direction to simulate the flow through the Q3D cascade. The computational grid around the airfoil is created with the Autogrid® tool using the block mesh topology. For the convective flow calculations, both the central and upwind methods available in Numeca® are applied separately. The Baldwin Lomax (BL), Spalart Allmaras (SA), Shear Stress Transport (SST), Explicit Algebraic Reynolds Stress Model (EARSM) and k-ε (KEPS) turbulence models are used for the turbulent shear stress calculations. In order to evaluate the aerodynamic performance of the spatial discretization methods and turbulence models, the isentropic Mach distribution on the airfoil surface, the total pressure loss and the exit flow angle behind the blade are compared with the experimental data of six test cases. In the compressor cases, the Spalart-Allmaras turbulence model with the Central scheme gives the best results in terms of average loss prediction, while no turbulence model is superior to the other in terms of exit angle prediction. On the turbine side, EARSM and KEPS give better performance in terms of loss prediction for the low Reynolds case compared to others, while the Spalart-Allmaras turbulence model is better for the high Reynolds cases.