Search

Flows at tidal-stream energy sites are characterised by high turbulence intensities and by the occurrence of highly energetic large and coherent flow structures. The interaction of the flow with seabed roughness is suspected to play a major role in the generation of such coherent flow structures. The problem is introduced with canonical wall-mounted square obstacles representing abrupt changes of bathymetry, with high Reynolds number flow (Re = 250000). Two methods are used: a numerical model, based on the LBM (Lattice Boltzmann Method) combined with LES (Large Eddy Simulation) and an experimental set-up in a circulating tank. The numerical model is validated by comparison with experimental data. In the case of a wall-mounted square cylinder, large-scale turbulent structures are identified in experiments where boils at the free surface can be observed. LBM simulation allows their three-dimensional characterisation. The dynamic of such large-scale events is investigated by temporal, spatial and spectral numerical analysis. Results show that periodical Kelvin–Helmholtz vortices are emitted in the cylinder wake. Then, they merge to form larger and more coherent structures that rise up to the surface. A wavelet study shows that the emission frequency of the Kelvin–Helmholtz vortices is not constant over time.

International audience; There is a strong tidal energy potential in France, especially in the Alderney Race. In the area of study, where turbines will be installed, bathymetry variations are causing velocity fluctuations with a high turbulence rate in the water column: large coherent turbulent structures can be observed at the sea surface. Such events can have a major impact on the marine tidal turbines behaviour and structural fatigue. To reproduce and analyse these turbulent events, tests are carried out in the wave and current circulating flume tank of IFREMER in Boulogne-sur-Mer. Before trying to reproduce a complex bathymetry, we chose to introduce the topic by studying elementary obstacles representative of real seabed elements (with an aspect ratio of the magnitude of the mean bathymetry variations): a wide square cylinder and an inclined floor. Experiments are carried out with Reynolds number as high as achievable in Froude similitude: Re = 2.5 × 10 5 and Fr = 0.23. The impact of the aspect ratio is studied by comparing results obtained with PIV and LDV measurements on the cube and cylinder cases. The addition of an inclined floor is also investigated. Results show a significant increase of the wake with the aspect ratio. The inclined floor induces a reduction of the shear layer created by the obstacle and modifications on the shedding frequency.

The present study aims at investigating turbulence characteristics in high flow velocity areas like those suitable for marine energy application. The Reynolds number, based on the rugosity height and mean flow velocity, is rather high: R e = 2 . 5 × 1 0 7 . For that purpose, experiments are carried out in a flume tank with R e as high as achievable in Froude similitude (in the tank: R e = 2 . 5 × 1 0 5 and F r = 0 . 23 ). Obstacles are canonical wall-mounted elements chosen to be representative of averaged bathymetric variations: a cube and a cube followed by an inclined floor. First, the wake topology past a canonical wall-mounted cube is illustrated from PIV measurements. Results show a flow behaviour already observed in the literature but for different upstream conditions ( R e and turbulence intensity). Second, the impact of the addition of an inclined floor is studied. It is shown that the inclination causes a squeezing of the cube wake that strongly impacts the shape and intensity of the shear layer (up to 10% more intense with the inclined floor). To fully grasp the turbulence organization in the wake for both test cases, an analysis using both complementary Proper Orthogonal Decomposition and quadrant method is performed. POD acts as a turbulent noise filter and quadrant method decomposes turbulent events. Results show the predominance of ejection (Q2) and sweep (Q4) events in the flow Reynolds shear stress. Q2 events are more energetic although Q4 events prevail. It is observed that the inclined floor causes a persistence of Q2 and Q4 events higher into the water column, more than the impulsion given by the floor altitude variations. The rise of the cube wake due to the inclined floor is thus illustrated using Q4 predominance area.

The turbulent flow and heat transfer in a channel with surface mounted cubical obstacles forms a generic example of a problem that occurs in many engineering applications, for instance in the design of cooling devices. We have performed a numerical simulation of it without using any turbulence models. This approach is the most accurate - but also the most expensive - way of computing complex turbulent flows since all dynamically significant scales of motion are to be solved numerically from the unsteady, incompressible Navier-Stokes equations and the energy equation. In view of the computational complexity, our first concern is to reduce the computational cost as far as we can get. We discretise convective and diffusive operators such that their spectral properties are preserved, i.e. convection ↔ skew-symmetric; diffusion ↔ symmetric, positive definite. Such a symmetry-preserving discretisation is stable on any grid and conserves mass, momentum and kinetic energy if the dissipation is turned off. First, the results of a second-order and a fourth-order, symmetry-preserving discretisation are compared for a fully developed, turbulent flow in a plane channel. The more accurate fourth-order method is applied to perform a numerical simulation of turbulent flow and heat transfer in a channel, where a matrix of cubes is mounted at one wall. Here, the temperature is treated as a passive scalar. The Reynolds number (based on the channel width and the mean bulk velocity) is equal to Re = 13, 000. The results of the numerical simulation agree well with the available experimental data.

The turbulent flow and heat transfer in a channel with surface mounted cubical obstacles forms a generic example of a problem that occurs in many engineering applications, for instance in the design of cooling devices. We have performed a numerical simulation of it without using any turbulence models. This approach is the most accurate - but also the most expensive - way of computing complex turbulent flows since all dynamically significant scales of motion are to be solved numerically from the unsteady, incompressible Navier-Stokes equations and the energy equation. In view of the computational complexity, our first concern is to reduce the computational cost as far as we can get. We discretise convective and diffusive operators such that their spectral properties are preserved, i.e. convection ↔ skew-symmetric; diffusion ↔ symmetric, positive definite. Such a symmetry-preserving discretisation is stable on any grid and conserves mass, momentum and kinetic energy if the dissipation is turned off. First, the results of a second-order and a fourth-order, symmetry-preserving discretisation are compared for a fully developed, turbulent flow in a plane channel. The more accurate fourth-order method is applied to perform a numerical simulation of turbulent flow and heat transfer in a channel, where a matrix of cubes is mounted at one wall. Here, the temperature is treated as a passive scalar. The Reynolds number (based on the channel width and the mean bulk velocity) is equal to Re = 13, 000. The results of the numerical simulation agree well with the available experimental data.

The turbulent flow and heat transfer in a channel with surface mounted cubical obstacles forms a generic example of a problem that occurs in many engineering applications, for instance in the design of cooling devices. We have performed a numerical simulation of it without using any turbulence models. This approach is the most accurate - but also the most expensive - way of computing complex turbulent flows since all dynamically significant scales of motion are to be solved numerically from the unsteady, incompressible Navier-Stokes equations and the energy equation. In view of the computational complexity, our first concern is to reduce the computational cost as far as we can get. We discretise convective and diffusive operators such that their spectral properties are preserved, i.e. convection ↔ skew-symmetric; diffusion ↔ symmetric, positive definite. Such a symmetry-preserving discretisation is stable on any grid and conserves mass, momentum and kinetic energy if the dissipation is turned off. First, the results of a second-order and a fourth-order, symmetry-preserving discretisation are compared for a fully developed, turbulent flow in a plane channel. The more accurate fourth-order method is applied to perform a numerical simulation of turbulent flow and heat transfer in a channel, where a matrix of cubes is mounted at one wall. Here, the temperature is treated as a passive scalar. The Reynolds number (based on the channel width and the mean bulk velocity) is equal to Re = 13, 000. The results of the numerical simulation agree well with the available experimental data.

The turbulent flow and heat transfer in a channel with surface mounted cubical obstacles forms a generic example of a problem that occurs in many engineering applications, for instance in the design of cooling devices. We have performed a numerical simulation of it without using any turbulence models. This approach is the most accurate - but also the most expensive - way of computing complex turbulent flows since all dynamically significant scales of motion are to be solved numerically from the unsteady, incompressible Navier-Stokes equations and the energy equation. In view of the computational complexity, our first concern is to reduce the computational cost as far as we can get. We discretise convective and diffusive operators such that their spectral properties are preserved, i.e. convection ↔ skew-symmetric; diffusion ↔ symmetric, positive definite. Such a symmetry-preserving discretisation is stable on any grid and conserves mass, momentum and kinetic energy if the dissipation is turned off. First, the results of a second-order and a fourth-order, symmetry-preserving discretisation are compared for a fully developed, turbulent flow in a plane channel. The more accurate fourth-order method is applied to perform a numerical simulation of turbulent flow and heat transfer in a channel, where a matrix of cubes is mounted at one wall. Here, the temperature is treated as a passive scalar. The Reynolds number (based on the channel width and the mean bulk velocity) is equal to Re = 13, 000. The results of the numerical simulation agree well with the available experimental data.