An experimental research was performed to study the effects of surface roughness on the characteristics of separated and reattached turbulent flows in an open channel. A backward facing step was used to induce flow separation. The rough surfaces comprised wire mesh grit-80 and sand grains of average diameter 1.5 mm. In each experiment, the Reynolds number based on the step height and freestream velocity of approach flow was fixed at 3240 and the Reynolds number based on the approach flow depth and freestream velocity was kept constant at 25130. Particle image velocimetry (PIV) technique was used to measure the flow velocity. The results showed that roughness effects on the mean and turbulent quantities are evident only in the recovery region. Moreover, roughness effects on the flow dynamics are dependent on the specific roughness element.

An experimental research was performed to study the effects of surface roughness on the characteristics of separated and reattached turbulent flows in an open channel. A backward facing step was used to induce flow separation. The rough surfaces comprised wire mesh grit-80 and sand grains of average diameter 1.5 mm. In each experiment, the Reynolds number based on the step height and freestream velocity of approach flow was fixed at 3240 and the Reynolds number based on the approach flow depth and freestream velocity was kept constant at 25130. Particle image velocimetry (PIV) technique was used to measure the flow velocity. The results showed that roughness effects on the mean and turbulent quantities are evident only in the recovery region. Moreover, roughness effects on the flow dynamics are dependent on the specific roughness element.

Neste trabalho é apresentado um estudo numérico sobre escoamentos incompressíveis, não isotérmicos, bi e tridimensionais nos regimes laminar e turbulento através da Simulação de Grandes Escalas e da utilização do Método de Elementos Finitos. Para tornar isso possível, é implementada a equação da energia e os termos de forças de campo (empuxo) em um algoritmo numérico desenvolvido em FORTRAN, já existente, que simula escoamentos incompressíveis, isotérmicos, tridimensionais, nos regimes laminar e turbulento. O código desenvolvido abrange escoamentos onde as formas básicas de troca térmica ocorrem por difusão e advecção. No que tange a natureza da convecção térmica é possível analisar escoamentos com convecção forçada, mista ou natural. O método numérico empregado é o de elementos finitos (FEM) e a discretização espacial das equações que governam o fenômeno (continuidade, conservação da quantidade de movimento e conservação da energia) é realizada através do método de Galerkin. Para a análise dos termos temporais nos escoamentos transientes aplica-se o esquema temporal explícito de Taylor-Galerkin. O elemento finito utilizado é o hexaedro isoparamétrico de oito nós. É empregado o método da pseudo-compressibilidade com o objetivo de manter os termos derivados da pressão na equação da continuidade, pois essa ausência gera uma dificuldade adicional na discretização das equações. Para a abordagem da turbulência é empregada a simulação de grandes escalas (LES) com modelagem sub-malha clássica de Smagorinsky para a viscosidade e a difusividade turbulenta. Visando a melhoria no tempo de processamento foi utilizada integração explícita das matrizes dos elementos e a técnica de processamento paralelo OpenMP. São apresentados resultados para escoamentos com vários números de Reynolds, Prandtl e de Grashoff dos campos de velocidade, pressão e temperatura para escoamentos em cavidade bidimensional, nos regimes laminar e turbulento, e para o degrau tri-dimensional no regime laminar. As simulações para escoamentos em cavidades nos regimes laminar e na região de transição são comparados com os resultados de outros autores, se mostrando bastante satisfatórios, tanto no regime transiente como no permanente. Além disso, a inserção das forças de campo no código melhorou os resultados obtidos com o mesmo. As outras simulações são apresentadas como novos casos e tiveram um comportamento qualitativamente satisfatório. / A numerical study about non-isothermal, bi and three-dimensional, incompressible, laminar and turbulent flows is done in this work using Large Eddy Simulation and Finite Element Method. To became this possible, is implemented the energy equation and buoyance forces (in the Navier-Stokes equations) in a numerical algorithm, developed in FORTRAN, already existent, that simulate isothermal, three-dimensional, incompressible, laminar and turbulent flows. The developed code includes flows where the basic forms of heat transfer are diffusion or advection. About the nature of thermal convection it is possible to analyze the forced, mixed or natural convection flows. The numerical method used is the finite element method (FEM) and the spatial discretization of governing equations of phenomena (mass, conservation of momentum and conservation of energy) is done through Galerkin method. To analyze the time-dependent terms in transient flows is employed a time-explicit Taylor-Galerkin scheme. The finite element used is the isoparametric hexahedral with eight nodes. It is used the pseudo-compressibility method to keep the pressure terms in continuity equation, because without these terms there are additional difficulties to obtain the discretizated equations. Regarding the turbulence approach, it is employed the large eddy simulation (LES) and for subgrid-scales is used the classical Smagorinsky model to turbulent viscosity and diffusivity. To minimize the processing time is used explicit integration of element matrix and the multiprocessing technical OpenMP. Results are presented to a wide range of Reynolds, Prandtl and Grashoff numbers for velocity, pressure and temperature fields to laminar and turbulent, bi-dimensional, lid-driven cavity flow and a laminar three-dimensional backward-facing step. Simulations of lid-driven cavity flows in laminar and transitional regimes is compared with others authors results, presenting good agreement, in both transient and permanent regimes. Besides that, the implementation of buoyance forces in the present code improved the results obtained by it. The others simulations are presented like new cases and had qualitatively good behavior.

Neste trabalho é apresentado um estudo numérico sobre escoamentos incompressíveis, não isotérmicos, bi e tridimensionais nos regimes laminar e turbulento através da Simulação de Grandes Escalas e da utilização do Método de Elementos Finitos. Para tornar isso possível, é implementada a equação da energia e os termos de forças de campo (empuxo) em um algoritmo numérico desenvolvido em FORTRAN, já existente, que simula escoamentos incompressíveis, isotérmicos, tridimensionais, nos regimes laminar e turbulento. O código desenvolvido abrange escoamentos onde as formas básicas de troca térmica ocorrem por difusão e advecção. No que tange a natureza da convecção térmica é possível analisar escoamentos com convecção forçada, mista ou natural. O método numérico empregado é o de elementos finitos (FEM) e a discretização espacial das equações que governam o fenômeno (continuidade, conservação da quantidade de movimento e conservação da energia) é realizada através do método de Galerkin. Para a análise dos termos temporais nos escoamentos transientes aplica-se o esquema temporal explícito de Taylor-Galerkin. O elemento finito utilizado é o hexaedro isoparamétrico de oito nós. É empregado o método da pseudo-compressibilidade com o objetivo de manter os termos derivados da pressão na equação da continuidade, pois essa ausência gera uma dificuldade adicional na discretização das equações. Para a abordagem da turbulência é empregada a simulação de grandes escalas (LES) com modelagem sub-malha clássica de Smagorinsky para a viscosidade e a difusividade turbulenta. Visando a melhoria no tempo de processamento foi utilizada integração explícita das matrizes dos elementos e a técnica de processamento paralelo OpenMP. São apresentados resultados para escoamentos com vários números de Reynolds, Prandtl e de Grashoff dos campos de velocidade, pressão e temperatura para escoamentos em cavidade bidimensional, nos regimes laminar e turbulento, e para o degrau tri-dimensional no regime laminar. As simulações para escoamentos em cavidades nos regimes laminar e na região de transição são comparados com os resultados de outros autores, se mostrando bastante satisfatórios, tanto no regime transiente como no permanente. Além disso, a inserção das forças de campo no código melhorou os resultados obtidos com o mesmo. As outras simulações são apresentadas como novos casos e tiveram um comportamento qualitativamente satisfatório. / A numerical study about non-isothermal, bi and three-dimensional, incompressible, laminar and turbulent flows is done in this work using Large Eddy Simulation and Finite Element Method. To became this possible, is implemented the energy equation and buoyance forces (in the Navier-Stokes equations) in a numerical algorithm, developed in FORTRAN, already existent, that simulate isothermal, three-dimensional, incompressible, laminar and turbulent flows. The developed code includes flows where the basic forms of heat transfer are diffusion or advection. About the nature of thermal convection it is possible to analyze the forced, mixed or natural convection flows. The numerical method used is the finite element method (FEM) and the spatial discretization of governing equations of phenomena (mass, conservation of momentum and conservation of energy) is done through Galerkin method. To analyze the time-dependent terms in transient flows is employed a time-explicit Taylor-Galerkin scheme. The finite element used is the isoparametric hexahedral with eight nodes. It is used the pseudo-compressibility method to keep the pressure terms in continuity equation, because without these terms there are additional difficulties to obtain the discretizated equations. Regarding the turbulence approach, it is employed the large eddy simulation (LES) and for subgrid-scales is used the classical Smagorinsky model to turbulent viscosity and diffusivity. To minimize the processing time is used explicit integration of element matrix and the multiprocessing technical OpenMP. Results are presented to a wide range of Reynolds, Prandtl and Grashoff numbers for velocity, pressure and temperature fields to laminar and turbulent, bi-dimensional, lid-driven cavity flow and a laminar three-dimensional backward-facing step. Simulations of lid-driven cavity flows in laminar and transitional regimes is compared with others authors results, presenting good agreement, in both transient and permanent regimes. Besides that, the implementation of buoyance forces in the present code improved the results obtained by it. The others simulations are presented like new cases and had qualitatively good behavior.

Neste trabalho é apresentado um estudo numérico sobre escoamentos incompressíveis, não isotérmicos, bi e tridimensionais nos regimes laminar e turbulento através da Simulação de Grandes Escalas e da utilização do Método de Elementos Finitos. Para tornar isso possível, é implementada a equação da energia e os termos de forças de campo (empuxo) em um algoritmo numérico desenvolvido em FORTRAN, já existente, que simula escoamentos incompressíveis, isotérmicos, tridimensionais, nos regimes laminar e turbulento. O código desenvolvido abrange escoamentos onde as formas básicas de troca térmica ocorrem por difusão e advecção. No que tange a natureza da convecção térmica é possível analisar escoamentos com convecção forçada, mista ou natural. O método numérico empregado é o de elementos finitos (FEM) e a discretização espacial das equações que governam o fenômeno (continuidade, conservação da quantidade de movimento e conservação da energia) é realizada através do método de Galerkin. Para a análise dos termos temporais nos escoamentos transientes aplica-se o esquema temporal explícito de Taylor-Galerkin. O elemento finito utilizado é o hexaedro isoparamétrico de oito nós. É empregado o método da pseudo-compressibilidade com o objetivo de manter os termos derivados da pressão na equação da continuidade, pois essa ausência gera uma dificuldade adicional na discretização das equações. Para a abordagem da turbulência é empregada a simulação de grandes escalas (LES) com modelagem sub-malha clássica de Smagorinsky para a viscosidade e a difusividade turbulenta. Visando a melhoria no tempo de processamento foi utilizada integração explícita das matrizes dos elementos e a técnica de processamento paralelo OpenMP. São apresentados resultados para escoamentos com vários números de Reynolds, Prandtl e de Grashoff dos campos de velocidade, pressão e temperatura para escoamentos em cavidade bidimensional, nos regimes laminar e turbulento, e para o degrau tri-dimensional no regime laminar. As simulações para escoamentos em cavidades nos regimes laminar e na região de transição são comparados com os resultados de outros autores, se mostrando bastante satisfatórios, tanto no regime transiente como no permanente. Além disso, a inserção das forças de campo no código melhorou os resultados obtidos com o mesmo. As outras simulações são apresentadas como novos casos e tiveram um comportamento qualitativamente satisfatório. / A numerical study about non-isothermal, bi and three-dimensional, incompressible, laminar and turbulent flows is done in this work using Large Eddy Simulation and Finite Element Method. To became this possible, is implemented the energy equation and buoyance forces (in the Navier-Stokes equations) in a numerical algorithm, developed in FORTRAN, already existent, that simulate isothermal, three-dimensional, incompressible, laminar and turbulent flows. The developed code includes flows where the basic forms of heat transfer are diffusion or advection. About the nature of thermal convection it is possible to analyze the forced, mixed or natural convection flows. The numerical method used is the finite element method (FEM) and the spatial discretization of governing equations of phenomena (mass, conservation of momentum and conservation of energy) is done through Galerkin method. To analyze the time-dependent terms in transient flows is employed a time-explicit Taylor-Galerkin scheme. The finite element used is the isoparametric hexahedral with eight nodes. It is used the pseudo-compressibility method to keep the pressure terms in continuity equation, because without these terms there are additional difficulties to obtain the discretizated equations. Regarding the turbulence approach, it is employed the large eddy simulation (LES) and for subgrid-scales is used the classical Smagorinsky model to turbulent viscosity and diffusivity. To minimize the processing time is used explicit integration of element matrix and the multiprocessing technical OpenMP. Results are presented to a wide range of Reynolds, Prandtl and Grashoff numbers for velocity, pressure and temperature fields to laminar and turbulent, bi-dimensional, lid-driven cavity flow and a laminar three-dimensional backward-facing step. Simulations of lid-driven cavity flows in laminar and transitional regimes is compared with others authors results, presenting good agreement, in both transient and permanent regimes. Besides that, the implementation of buoyance forces in the present code improved the results obtained by it. The others simulations are presented like new cases and had qualitatively good behavior.

Numerous practical applications exist where dispersed solid particles are transported within a turbulent accelerating or decelerating gaseous flow. The large density variation between phases creates the potential for significant differences in velocity known as velocity slip. Flow over a backward facing step provides a well characterized, turbulent, decelerating flow useful for measuring the relative velocities of the solid and gaseous phases in order to determine velocity slip and particle drag. Numerous investigations have been conducted to determine the gas phase velocity in a backward facing step for both laminar and turbulent flows and therefore the gas phase flow is well know and documented. Furthermore, some studies have also been conducted to determine the velocity of various sizes of spherical particles in a backward facing step and compared with their corresponding gas phase velocities. Few if any velocity measurements have been made for non-spherical particles in a backward facing step. In this work, a Phase Doppler Particle Analyzer (PDA) was used to measure gas and particle phase velocities in a backward facing step. The step produced a 2:1 increase in cross sectional area with a Reynolds number of 22,000 (based on step height) upstream of the step. Spherical particles of 1 – 10 μm with an average diameter of 4μm were used to measure the gas phase velocity. At least three sizes in the range of (38 – 212 μm) for four different particles shapes were studied. The shapes included: spheres, flakes, gravel, and cylinders. Since the PDPA is not able to measure the size of the non-spherical particles, the particles were first separated into size bins and a technique was developed using the PMT (photo multiplier tubes) gain to isolate the particle size of interest for each size measured. The same technique was also used to measure terminal velocities of the particles in quiescent air. The measured gas phase velocity and spherical solid phase particles were in good agreement with previous measurements in the literature. The results showed relative velocities between the particles and gas phase to be in the range of 0 – 3 m/s which is in transition between stokes flow and fully developed turbulent flow. Drag coefficients were an order magnitude higher for non-spherical particles in turbulent flows in comparison to stokes flow which agreed reasonably well with quiescent terminal velocity drag. This information is valuable for modeling turbulent two-phase flows since most assumptions of the drag are currently based on correlations from empirical data with particles moving through a still fluid.

Micro particle image velocimetry (microPIV) is a non-intrusive tool for visualizing flow in micron-scale conduits. Using this investigative instrument, two experimental studies were performed to understand flow behaviors in microfluidic channels - a sudden expansion step flow and laminar velocity profile variation in diffusion driven mixing. First, flow in a backward facing step feature (1:5 expansion ratio) in a microchannel was taken as the subject of microPIV flow visualization. The onset and development of a recirculation flow was studied as a function of flow rate. This flow pattern was further used to investigate two major parameters affecting microPIV measurements; the depth-of-focus and recording time-intervals between images in a microPIV image pair. The onset of recirculation was initiated at flow rates that correspond to Reynolds numbers, Re > 95, which is well beyond the typical working range of microfluidic devices (Re=0.01-10). The recirculation flow has a 3D structure due to the dimensions of the microchannel and the effect of no slip condition on the walls. Ensemble cross-correlation was found not to be sensitive to variations of depth-of-focus and the output flow fields were similar as long as the overall optical focus remained within the upper and lower bounds of the microchannel. However, variations of time intervals between images in a microPIV pair, resulted in quantitatively and qualitatively different flow patterns for a given constant flow rate and depth-of-focus. In the second experiment, the effect of the laminar velocity profile and its variation on mixing phenomena at the reduced scale is studied. It is shown that the diffusive mass flux between two miscible streams, flowing in a laminar regime in a microchannel, is enhanced if the velocity at their diffusion interface is increased. Based on this idea, an in-plane passive micromixing concept is proposed and implemented in a working device (sigma micromixer). This mixer shows considerable mixing performance by periodically varying the flow velocity profile, such that the maximum of the profile coincides with the transversely progressing diffusion fronts repeatedly throughout the mixing channel. microPIV has been used to visualize the behavior of laminar flow inside the micromixer device and to confirm the periodic variation of the velocity profile through the mixing channel.

Helicopter operations in the vicinity of small naval surface vessels often require excessive pilot workload. Because of the unsteady flow field and large mean velocity gradients, the envelope for flight operations is limited. This experimental investigation uses a 1:144 scale model of the U.S. Navy destroyer DDG-81 to explore the problem. Both active and passive flow control techniques were used to improve the flow field in the helicopter's final decent onto the flight deck. Wind tunnel data was collected at a set of grid points over the ship's flight deck using a single component hotwire. Results show that the use of porous surfaces decreases the unsteadiness of the flow field. Further improvements are found by injecting air through these porous surfaces, causing a reduction in unsteadiness in the landing region of 6.6% at 0 degrees wind-over-deck (WOD) and 8.3% at 20 degrees WOD. Other passive configurations tested include fences placed around the hangar deck edges which move the unsteady shear layer away from the flight deck. Although these devices cause an increase in unsteadiness downstream of the edge of the fence when compared to the baseline, the reticulated foam fence caused an overall decrease in unsteadiness in the landing region of 12.1% at 20 degrees WOD. / Master of Science

No description available.

No description available.