Subgrid-scale modeling, which characterizes Large Eddy Simulation (LES), has been used to predict the behavior of a water-fed hydrocyclone operating without an air core. The governing equations were solved by a fractional step method on a staggered grid. The Smagorinsky subgrid-scale model was employed to account for turbulent effects. Numerical results actually capture the main features of the flow pattern and agree reasonably well with experiments, suggesting that LES represents an interesting alternative to classical turbulence models when applied to the numerical solution of fluid flows within hydrocyclones.

Experimental data of a turbulent boundary layer developing over a flat plate, obtained by Digital Particle Image Velocimetry (DPIV) technique, are analyzed making use of proper orthogonal decomposition (POD). Different POD definitions have been used in order to check their ability in educing the various structures dominating the near wall region. Results show a specific sensitivity depending on the POD definition adopted.

The study is based on K-ε turbulence modeling using both structured and unstructured grids. The pressure distributions in the flow field and on the surfaces of the bodies are determined. Different parameters that affect the flow field such as Reynolds number, aspect ratio of the body, and flow direction are considered. Special concern is paid to the separations at the corners of the body and the circulations behind the body. The thermal behavior due to the flow field is also considered. Nusselt number on the heated surfaces of the body is examined. The thermal behavior of a given body is strongly dependent on its aerodynamic characteristics. Comparisons are made between the present results and the available experimental data. Artificial Neural Network (ANN) was used to predict the values of the drag coefficient. Many useful conclusions and suggestions were drawn from the study.

In order to study the aerodynamics of a slot jet impinging a concave wall, flow visualisations, velocity measurements by PIV and pressure measurements are carried out. A statistical data processing allows us to obtain the mean velocity fields and the Reynolds stresses. Among the studied parameters, the effect of the relative curvature of the wall is studied in particular because of a jet beating phenomenon observed for a low relative curvature. Three flow modes are then described.

The ability of turbulence models to predict self-similar mixing layers is investigated. The influence of velocity is well captured but no model reproduces the sensitivity of the mixing layer to density differences. A correction proposed for boundary layer flows hardly affects mixing layer predictions. A correction is proposed but is not satisfactory. At last, compressible turbulence effects are investigated. Without corrections, models cannot predict the spreading rate reduction. Standard corrections predict too weak a reduction. The sonic eddy concept is validated whatever the turbulence model. A form suitable for Navier-Stokes codes is proposed.

This study examines the dynamics of turbulent flow in the wake of a turning ship using the large eddy simulation (LES) technique. LES is applied in conjunction with a random flow generation (RFG) technique originally developed at West Virginia University to provide unsteady inflow boundary conditions. As the ship is turning, the effects of the Coriolis and centrifugal forces on vortical structures are included. The effects of the Coriolis force on the flow-field are assessed and a grid sensitivity study is performed. The predicted turbulence structures are analyzed and compared with the wake of a non-turning ship.

This study deals with the L arge E ddy S imulation of a 3D unsteady shock wave/boundary layer interaction, in sight of turbomachinery applications. The simulation of a compressible flow through a 3D channel (ONERA) has been performed in LES, with a third order WENO scheme (Jiang, Shu, 1996) and the mixed scale model (Tenaud, Ta Phuoc, 1996). The flow has already been studied experimentally at ONERA (Cahen, 1995), and numerically with RANS modeling (Gerolymos, Vallet, 1997). The unsteady results obtained by LES within and downstream of the 3D interaction are presented and analyzed. The mean velocity profiles in the interaction zone, are then compared to both experiments and statistical numerical results.

Results of numerical and experimental investigations of turbulent jet from a round ribbed tube are presented. The round ribbed tube is a coil-inserted tube. The ratio of the coil wire diameter to the tube inside diameter, d/D, is 0.1 and the ratio of the coil pitch spacing to the tube inside diameter, p/D, is 1.2. The coil is placed inside the tube such that it terminates at the tube outlet. The standard k-e turbulence model along with the control volume method is used in the numerical calculations. The numerical results provided more insights into the flow characteristics in the radial planes. Results show that the coil inserts generates a weak swirl at the tube outlet that dissipates at an approximate distance of X/D = 1 where at this location the generated free vortices off the coil become the main mechanism behind the entrainment and mixing process. Comparisons of the numerical and experimental results show over estimation of the results by the numerical method.

The present work is concerned with a numerical investigation of the unsteady two-dimensional incompressible turbulent flow around a 45° wedge inclined at an angle of attack of 45°. The incompressible time-dependent Navier-Stokes equations written in a finite volume formulation with an upwind discretization scheme were used in this study. The numerical solutions of the elliptic partial differential equations have been obtained using the standard k-ε turbulence model. The numerical results are compared with available experimental data. The comparison includes the Strouhal number and the distribution of velocities and vorticity in the wake flow. The comparison shows that the Strouhal number, time-averaged streamlines and vorticity pattern agree well with experiments. However, the agreement with the experimental data is less satisfactory for the time-averaged normal velocity profile and root mean square values of fluctuating velocity components.

Numerical and experimental investigations of the flow in a two-dimensional 180°-curved diffuser, with a guide vane interposed in the diffuser, is presented to clarify their characteristic and to improve the curved diffuser performance. Particular attention is focused on the effect of the variation in the interposed position of the guide vane on the flow field and pressure recovery. Measurements for mean longitudinal and transverse velocity profiles as well as the wall static pressure are performed at different downstream stations. Comparisons are made between the present measurements for the diffuser with a guide vane and without a guide vane [10]. The numerical investigation is based on the solution of the governing equations using a finite volume technique employing SIMPLE algorithm on co-located body-fitted grids. The two-equation model of turbulence, standard k-ε model, and the modified k-ε model [15] are employed in this investigation. The study indicated that the emerging velocity distribution is more uniform than that associated with flow in the diffuser without a guide vane. The presence of a guide vane is shown to suppress the formation of the region of flow separation, as a consequence, the performance of the diffuser is improved. The overall pressure recovery is 54% of the inlet dynamic pressure. The comparison between the numerical and experimental results indicates that the modified k-ε model satisfactorily predicts the overall characteristics of the flow in the diffuser. Numerical predictions show that the best position for a guide vane is positioning the vane towards the convex wall of the diffuser. Diffusers with a guide van at B/W 1 = 0.25 to 0.333 give the highest-pressure recovery coefficient.

Environmental flows of extreme importance, such as turbulent atmospheric boundary layer over thick rain forests, may benefit from more realistic mathematical models. Accordingly, flow over layers of dense vegetation can be characterized by some sort of porous structure through which a fluid permeates. For hybrid media, involving both a porous structure and a clear flow region, difficulties arise due to the proper mathematical treatment given at the interface. The literature proposes a jump condition in which shear stresses on both sides of the interface are not of the same value. This paper presents numerical solutions for such hybrid medium, considering here a channel partially filled with a porous layer through which fluid flows in turbulent regime. One unique set of transport equations is applied to both regions. Effects of Reynolds number, porosity, permeability and jump coefficient on mean and turbulence fields are investigated. Results indicate that depending on the value of the stress jump parameters, a substantially different structure for the turbulent field is obtained.

A two-equation turbulence model that can adjust itself from RANS to LES regimes is constructed using the renormalization group. When the grid-spacing gets to coarse for LES, the model tends to a RANS model. In well-resolved regions the model results in a LES subgrid-model. The renormalization group procedure leads to transport-equations that are dependent on the filterwidth and contain no adjustable constants.

A common weak spot of RANS turbulence models is that one turbulence scale is determined from the dissipation rate ε (or some other turbulence quantity). An improved transport equation for ε is presented. It is valid in both high and low Reynolds number flow regions. The new ε-equation is combined with a cubic eddy-viscosity model. The complete model is applied to fully developed channel and pipe flows, the turbulent plane and round jet, a backward-facing step flow and to heat transfer predictions with turbulent impinging jets. The presented model yields accurate results for all these test cases (in particular the plane jet–round jet anomaly is resolved).

Accurate resolution of turbulent wakes is a formidable task. Herein, we challenge this task by testing several compact schemes for approximating the convective derivative of the direct numerical and large-eddy simulation (DNS and LES) methodologies. The stencils house the needed resolution efficiency and numerical stability characteristics to properly resolve turbulent wakes while concurrently maintaining convergent solutions. Turbulent wakes of the circular cylinder are computed for Re = 1260 and 3900 with the results verified by the experimental evidence.

A formulation to include the effects of wall-proximity in a second moment closure model is presented that utilizes a tensor representation for the redistribution term in the Reynolds stress equations. The wall-proximity effects are modeled through an elliptic relaxation process of the tensor expansion coefficients that properly accounts for both correlation length and time scales as the wall is approached. DNS data and Reynolds stress solutions using a full differential approach at channel Reynolds number of 590 are compared to the new model.

Two methods for computational reconstruction of fluctuating flows in the ocean based on field measurement data are suggested. Here full-3D treatment with the spatial scale of O(100m)–O(1000m) is taken into account. In the first method, the fluctuating flow based on measurement data is generated at inflow boundary and introduced into computational domain. In the second method, small wavenumber components of fluctuation in the ocean are forced in computational domain, and large wavenumber components are generated automatically. The results of each method are discussed with comparing with measured data, and effectiveness of each method is shown.

Based on the holographic PIV measurements in the core region of a square duct at Re H ≈ 1.2×10 5 , the tensorial alignment of the deviatoric subgrid-scale (SGS) stress ( τ ij ) is shown to have a bi-modal behavior (Tao et al. , 2001, 2002). To gain further insight into the statistical geometry of τ ij , we employ the Germano decomposition, τ ij = L ij + C ij + R ij (Leonard, 1974; Germano, 1986), and compute the decomposed stress elements directly from the velocity data. Their alignment trends are then investigated using joint probability density functions (pdf), similar to those in Tao et al. (2002). The alignment between the Leonard stress, L ij , and the filtered strain-rate ( S˜ ij ) is found to exhibit a similar bi-modal trend, but the alignment angle between the corresponding eigenvectors is around 40° instead of 32°. The alignment between the SGS Reynolds stress, R ij and S˜ ij , on the other hand, is found to be consistent with an eddy viscosity configuration. In contrast, the alignment between the cross stress, C ij , and S˜ ij is more complicated and does not exhibit any significant trend. Furthermore, previously reported alignment trends of the vorticity vector, ω˜ i , relative to τ ij , are primarily due to the contribution from L ij . The orientation of ω˜ i relative to C ij shows less robust alignment trends, and there is no clear trend in the alignment of ω˜ i relative to R ij .

The wake structure and vortex shedding characteristics of a NACA 0025 airfoil were studied experimentally. Wind tunnel experiments were carried out for three Reynolds numbers and three angles of attack by means of cross-wire measurements, spectral analysis and complementary surface flow visualization. Evidence of wake vortex shedding and flow separation was obtained for most of the cases examined, and dependence of these phenomena on Reynolds number and angle of attack was found. External acoustic excitation at a particular frequency was found to eliminate or reduce the separation region and decrease the airfoil wake. Moreover, the results show that the acoustic excitation has a significant effect on vortex shedding and can improve airfoil performance, i.e. produce an increase of the lift and/or decrease of the drag.

To improve aerodynamic performance of a blade and to maintain its designed performance, many study have been focused on this point. The factors for aerodynamic performances of a blade are its geometry and operating conditions, which are almost fixed during its lifetime. On the other hand, the factor such as surface roughness generated by impure sand grain or oil droplet is variable one. The surface roughness on the blade affects the flow past it by its aerodynamic nature. The blade design can be improved by revealing this change in aerodynamic performance due to surface roughness. In this study, three-dimensional computations are carried out on the blade root with local surface roughness to investigate how the roughness will change the aerodynamic performance. The locality of the surface roughness distribution is simulated with particle tracks in three-dimensional flow. The model replacing the effect of roughness elements with virtual force is used to simulate turbulent flow around blade with local surface roughness.

In this paper, the flow behavior of a reattachment process over a backward facing step flow is reported. The reattachment process is controlled by injection of vortex generator jets. The injection of jets upstream of the step produces the co-rotating longitudinal vortices in a separating shear layer. The experiment of the step response of the injection jet is also conducted in order to investigate the evolution process of the longitudinal vortices. A large scale of primary and counter vortices are observed, when the velocity ratio of the free stream to injected jet is 6. The detailed structure of the longitudinal vortices is clarified. The remarkable effect of the vortices on the separating shear layer downstream of the step is observed.