Search

We present measurements of the global heat transfer and the velocity field in two Rayleigh-Bénard cells (aspect ratios 1 and 2). We use Fluorinert FC770 as the working fluid, up to a Rayleigh number . The velocity field is inferred from sequences of shadowgraph pattern using a Correlation Image Velocimetry (CIV) algorithm. Indeed the large number of plumes, and their small characteristic scale, make it possible to use the shadowgraph pattern produced by the thermal plumes in the same manner as particles in Particle Image Velocimetry (PIV). The method is validated in water against PIV, and yields identical wind velocity estimates. The joint heat transfer and velocity measurements allow to compute the scaling of the kinetic dissipation rate which features a transition from a laminar scaling to a turbulent Re 3 scaling. We propose that the turbulent transition in Rayleigh-Bénard convection is controlled by a threshold Péclet number rather than a threshold Rayleigh number, which may explain the apparent discrepancy in the literature regarding the “ultimate” regime of convection.

Due to large scale flow inhomogeneities and the effects of temperature, turbulence small-scale structure in thermal convection is still an active field of investigation, especially considering sophisticated Lagrangian statistics. Here we experimentally study Lagrangian pair dispersion (one of the canonical problems of Lagrangian turbulence) in a Rayleigh-B\'enard convection cell. A sufficiently high temperature difference is imposed on a horizontal layer of fluid to observe a turbulent flow. We perform Lagrangian tracking of sub-millimetric particles on a large measurement volume including part of the Large Scale Circulation (LSC) revealing some large inhomogeneities. Our study brings to light several new insights regarding our understanding of turbulent thermal convection: (i) by decomposing particle Lagrangian dynamics into the LSC contribution and the turbulent fluctuations, we highlight the relative impact of both contributions on pair dispersion; (ii) using the same decomposition, we estimate the Eulerian second-order velocity structure functions from pair statistics and show that after removing the LSC contribution, the remaining statistics recover usual homogeneous and isotropic behaviours which are governed by a local energy dissipation rate to be distinguished from the global dissipation rate classically used to characterise turbulence in thermal convection; and (iii) we revisit the super-diffusive Richardson-Obukhov regime of particle dispersion and propose a refined estimate of the Richardson constant., Comment: 9 figures

Turbulent thermal convection is a complex problem: the flow is forced at almost all scales, because of the presence of both large scale circulation and small scale plumes. Moreover the difference of density induced by temperature adds an ingredient in the basic equations of the flow, so that turbulence could be different from the case of pure mechanical forcing. Mixing is in principle different and probably increased because of the structure of the plumes. Moreover the entire flow is not homogeneous (at least at large scales) and the statistical behaviour can be influenced by the large scale flow. One way to probe all these specific effects is to use a Lagrangian point of view, measuring, for example, Lagrangian transport of velocity and temperature, or heat flux. We will show how the inhomogeneity of the mean temperature and velocity fields affect the heat flux. Then the interactions between the mean fields and the Lagrangian fluctuations will be highlighted.

We report joint Lagrangian velocity and temperature measurements in turbulent thermal convection. Measurements are performed using an improved version (extended autonomy) of the neutrally-buoyant instrumented particle that was used by to performed experiments in a parallelepipedic Rayleigh-Benard cell. The temperature signal is obtained from a RFtransmitter. Simultaneously, we determine particle's position and velocity with one camera, which grants access to the Lagrangian heat flux. Due to the extended autonomy of the present particle, we obtain well converged temperature and velocity statistics, as well as pseudo-eulerian maps of velocity and heat flux. Present experimental results have also been compared with the results obtained by a corresponding campaign of Direct Numerical Simulations and Lagrangian Tracking of massless tracers. The comparison between experimental and numerical results show the accuracy and reliability of our experimental measurements. Finally, the analysis of lagrangian velocity and temperature frequency spectra is shown and discussed. In particular, we observe that temperature spectra exhibit an anomalous f^2.5 frequency scaling, likely representing the ubiquitous passive and active scalar behavior of temperature, Comment: 15 pages, submitted

International audience; We present the principle for a micro-sensor aimed at measuring local correlations of turbulent velocity and temperature. The operating principle is versatile and can be adapted for various types of flow. It is based on a micro-machined cantilever, on the tip of which a platinum resistor is patterned. The deflection of the cantilever yields an estimate for the local velocity, and the impedance of the platinum an estimate for the local temperature. The velocity measurement is tested in two turbulent jets: one with air at room temperature which allows to compare with well-known calibrated reference anemometers, and another one in the GReC jet at CERN with cryogenic gaseous helium which allows a much larger range of resolved turbulent scales. The recording of temperature fluctuations is tested in the Barrel of Ilmenau which provides a controlled turbulent thermal flow in air. Measurements in the wake of a heated or cooled cylinder demonstrate the capability of the sensor to display the cross correlation between temperature and velocity correctly.

Several Rayleigh–Bénard experiments in water are performed with smooth or rough boundaries. We present new thermal transfer measurements obtained with large roughness elements arranged in a square lattice. The data are compared to previous data obtained with smaller elements in the same cell (Tisserand et al., Phys. Fluids, vol. 23, 2011). Experiments in the same apparatus without roughness are presented, as reference results, to allow for comparison. In the rough case, several regimes of heat transfer are identified: one similar to the smooth case, an enhanced heat transfer regime characterized by a modification of the Nusselt versus Rayleigh number relation and a third part where the relation can be similar to a smooth one with a corrected prefactor.

International audience; In this paper we focus on the turbulent transport of a scalar through a channel. The scalar flux, and the corresponding scalar concentration gradient along the channel allow to define a Nusselt number and a Grashof number. While the relation between the large scale velocity field and the input energy rate show a perfect inertial (turbulent) behavior, three different regimes can be distinguished, with different scalings between Nusselt and Grashof numbers.

International audience; We report Particle Image Velocimetry of the Large Scale Circulation and the viscous boundary layer in turbulent thermal convection. We use two parallelepipedic Rayleigh-Bénard cells with a top smooth plate. The first one has a rough bottom plate and the second one has a smooth one so we compare the rough-smooth and the smooth-smooth configurations. The dimensions of the cell allow to consider a bi-dimensional mean flow. Lots of previous heat flux measurements have shown a Nusselt–Rayleigh regime transition corresponding to an increase of the heat flux in presence of roughness which is higher than the surface increase. Our velocity measurements show that if the mean velocity field is not clearly affected by the roughness, the velocity fluctuations rise dramatically. It is accompanied by a change of the longitudinal velocity structure functions scaling. Moreover, we show that the boundary layer becomes turbulent close to roughness, as it was observed recently in the air [Liot et al., JFM, vol. 786, pp. 275-293]. Finally we discuss the link between the change of the boundary layer structure and the ones observed on the Large Scale Circulation.

In this experimental work, the aim is to understand how turbulent thermal flows are enhanced by the destabilization of the boundary layers. Square-stud roughness elements have been added on the bottom plate of a rectangular Rayleigh–Bénard cell in air, to trigger instabilities in the boundary layers. The top plate is kept smooth. The cell proportions are identical to those of the water cell previously operated and described by Salort et al. (Phys. Fluids, vol. 26, 2014, 015112), but six times larger. The very large size of the Barrel of Ilmenau allows detailed velocity fields to be obtained using particle image velocimetry very close to the roughness elements. We found that the flow is quite different at low Rayleigh numbers, where there is no heat-transfer enhancement, and at high Rayleigh numbers where there is a heat-transfer enhancement due to the roughness. Below the transition, the fluid inside the notch, i.e. between the studs, is essentially at rest, though it is slowly recirculating. The velocity profiles on the top of obstacles and in grooves are fairly compatible with those obtained in the smooth case. Above the transition, on the other hand, we observe large incursions of the bulk inside the notch, and the velocity profiles on the top of obstacles are closer to the logarithmic profiles expected in the case of turbulent boundary layers.

International audience; We report preliminary results obtained using new local velocity probes in the Superfluid Helium high REynold number von Kármán flow (SHREK) experiment for different forcing conditions. The presentation will focus on the validation of the signals obtained from a hot-wire and a total head pressure tube in both normal and superfluid phases of liquid helium.

International audience; Natural cross ventilation can be a promising passive solution for summer thermal comfort in buildings. It takes advantage of the night temperature of the air to cool the walls of the building. Although this technique is well-known in hot climate, its use in new buildings requires being able to predict the quantity of heat that can be dissipated. There is indeed a lack of experimental data either to build design rules for engineers or to validate numerical code dedicated to the design. In this study, experimental research is carried out on a full-scale dwelling and a small-scale model, and evidence is given on the relevance of such a comparison. The flow has been estimated with spatially discrete local sensors in both systems, and flow visualizations have been additionally performed in the small-scale controlled model. For windy periods, the indoor airflow is found to be driven by the wind, as expected. For calm periods which are critical for summer comfort, the flows look much more complex with the observation of unsteady flow reversals.

We study the influence of a small tilt angle ( $0 < \theta < 3 \times 10^{-2}$ rd) on the Nusselt number in a 1/2 aspect ratio Rayleigh-Benard cell, at high Rayleigh number (5 x 1011 < Ra < 4 x 1012). The small decrease observed is interpreted as revealing a two rolls structure of the flow. Transitions between different global flows are also observed, on very long times, comparable to the diffusion time on the whole cell. The consequence is that the Nusselt number observed in most high Ra experiments should significantly depend on initial conditions.

International audience; In the environmental context, night cross-ventilation combined with thermal mass of walls is a promising passive solution to reduce the energy consumption of buildings while maintaining comfort, especially during summer. We realize measurements in a full-scale apartment in real conditions and in a reduced scale model of the apartment positioned in a wind tunnel. Concerning the measurement of natural ventilation air flow rate through large openings in a real apartment, an air direction detector coupled with classical anemometers has been developed and calibrated in a controlled test room. To study the flow in controlled conditions, a 1/20th size model that takes into account the details of the apartment geometry was built. Moreover, to understand how heat is removed by the flow and the role of turbulence, we develop a micro-machined sensor able to measure the local heat flux : a 375×35×1 ?m cantilever whose deflection reflects the local fluid velocity, with a temperature-sensitive thin layer deposited on its tip. This combined approach grants access to crossed turbulent statistics, in particular the contribution of the turbulent heat-flux. Data in real buildings are compared to the measurements in the reduced scale model and the validity of the approach based on the similarities is discussed. The results of the measurements in real buildings lead to distinguish between windy and calm periods, the latest being the most critical one regarding summer comfort. During calm period, we observe reversed flows in the measuring section and the duration of their occurrence is analysed.

In this article we deal with the turbulent regimes of Rayleigh–Benard convection, namely the 2/7 regime and beyond. An experiment with He at low temperature allows us to explore a large Rayleigh number (Ra) range up to 2×1014, under Boussinesq conditions, while the Prandtl number (Pr) is equal to and larger than 0.7. Calorimetric measurements evidence a departure from the 2/7 regime above Ra=1011 toward a new regime where the heat transfer is enhanced. Local measurements with two nearby thermometers allows us to relate this change to a laminar–turbulent transition of the velocity boundary layer induced by the large-scale flow near the walls of the cell. The features of the observed new regime match those of the ultimate regime predicted by R. Kraichnan [Phys. Fluids 5, 1374 (1962)] at moderate Pr; in particular, our experimental data show that the thermal boundary layer lies inside the viscous sublayer of the turbulent boundary layer.

In this article we describe an experimental procedure to study Rayleigh–Benard convection with He 4 close to its critical point (tabulated as T c =5.1953 K in the EIT 90, P c =2.275 bar) as the working fluid. The convention takes place in a closed cylindrical cell (20 cm height and 10 cm inner diameter). The experimental challenge lies in accurate temperature and density measurements. To test our system and our thermometry, an independent measurement of the critical temperature is performed and we found 5.1939 K±1.0 mK (EIT 90). Measurements of the Nusselt number versus the Rayleigh number, from the laminar regime (Ra∼10 3 ) up to the highest Ra obtained in a laboratory experiment (2×10 14 ) are reported, with Nu varying from 1 to more than 5000.

We report an experimental study of the swirling flow generated in the gap between two coaxial corotating disks in an enclosed geometry, i.e. when the flow is enclosed in a cylindrical vessel. In this situation the angular momentum generated by the disk rotation tends to concentrate in a strong axial vortex. In the range of high Reynolds numbers explored, we show that two regimes exist, depending on the disk angular velocities. When they rotate at quite different speeds, we observe very intermittent fluctuations of the axial vorticity associated with quasi-periodic vortex bursting, whereas at comparable rotation rates a more stable vorticity structure is observed. We describe these regimes using global measurements; we also show, using local velocimetry, how the structure of the flow affects the small scale turbulence.

We present global heat-transfer and local temperature measurements, in an asymmetric parallelepiped Rayleigh-B ́enard cell, in which controlled square-studs roughnesses have been added. A global heat transfer enhancement arises when the thickness of the boundary layer matches the height of the roughnesses. The enhanced regime exhibits an increase of the heat transfer scaling. Local temperature measurements have been carried out in the range of parameters where the enhancement of the global heat transfer is observed. They show that the boundary layer at the top of the square-stub roughness is thinner than the boundary layer of a smooth plate, which accounts for most of the heat-transfer enhancement. We also report multistability at long time scales between two enhanced heat-transfer regimes. The flow structure of both regimes is imaged with background-oriented synthetic Schlieren and reveals intermittent bursts of coherent plumes.

In a low temperature He gas Rayleigh-B\'enard experiment, Rayleigh numbers from ${10}^{3}$ to more than ${10}^{14}$ are explored. Local velocity is estimated through the time lag between two closeby temperature probes. This allows characterizing of the high Rayleigh regime $(\mathrm{Ra}g{10}^{11})$ as a fully turbulent one, possibly corresponding to the asymptotic regime predicted by R. Kraichnan [Phys. Fluids 5, 1374 (1962)].

We investigate the closed flow between coaxial contra rotating disks, at moderate to high Reynolds numbers. We show that global (i.e. spatially averaged) quantities can be used to characterize the state of the flow and its degree of turbulence. We first report measurements on the driving torque and show how it depends on the manner momentum is imparted to the fluid. We then show that pressure measurements at the flow boundary provide a good estimate of the rms velocity fluctuations in the flow and that it reveals the transition to turbulence in the flow volume. Finally, we show that once the transition has occurred, the knowledge of the same global quantities allows the calculation of fundamental turbulence characteristics such as the rms velocity fluctuations, the effective integral length scale L * , Taylor's microscale λ and Kolmogorov's dissipation length η. That these quantities may be obtained from measuring devices removed from the bulk of the flow is of importance for the study of fluid motion in complex geometries and/or using corrosive fluids

In this paper, we analyze the mean velocity profile and the Reynolds shear stress in a turbulent, inclined, heat pipe. We show that the simplest version of a mixing length model is unable to reproduce the evolution of the velocity profile shape with the inclination angle ψ. An improvement of this model, taking into account some buoyancy effects, gives nice qualitative agreement with the observations. The agreement implies a low value for the gradient Richardson number Ric above which the flow is laminar. While such a low value (Ric ≃ 0.05) is surprising, we found it in agreement with published experimental data, when the information given allowed to calculate the gradient Richardson number Ri.

WOS:000331787700001; International audience; Heat transfer measurements performed by Riedinger et al. (Phys. Fluids, 25, 015117 (2013)) showed that in an inclined channel, heated from below and cooled from above with adiabatic walls, the flow is laminar or intermittent (local bursts can occur in the laminar flow) when the inclination angle is sufficiently high and the applied power sufficiently low. In this case, gravity plays a crucial role in the characteristics of the flow. In this paper, we present velocity measurements, and their derived tensors, obtained with Particle Image Velocimetry inside the channel. We, also, propose a model derived from a jet interpretation of the flow. Comparison between experiment and model shows a fair agreement.

We study Rayleigh-Benard convection in a cell which consists of a cylinder of 20 cm height and 10 cm diameter filled with gaseous helium between 4 and 5 K. The new feature compared to previous studies comes from a very accurate thermocouple measurement of the temperature difference across the cell ΔT (resolution of 20 µK). The Rayleigh number Ra ranges from the conductive regime to 5 × 1012 by changing the gas density from 3 g/m3 to 30 kgjm3 and by scanning ΔT from 1 mK to 800 mK for each density. On the figure the Nusselt number N u versus Ra is presented. The high sensitivity measurement of low ΔT allows to evidence and measure the adiabatic gradient influence on the convective threshold (in our geometry Ra = 30000 ± 4000 without this influence); the adiabatic gradient slows down the convection for points at low ΔT as seen on the figure.

Plusieurs modeles de turbulence relient la cascade d'energie et l'intermittence, caracterisee par l'evolution des densites de probabilite (pdf) des increments longitudinaux de vitesse. Nous appuyant aussi bien sur des modeles recents que sur des resultats experimentaux, nous montrons que la Curtosis de ces pdf permet une estimation simple et directe de la profondeur de la cascade. Cela nous permet de reanalyser les resultats publies d'une simulation numerique et de montrer que la profondeur de la cascade y evolue de la meme facon que pour les experiences de laboratoire en fonction du nombre de Reynolds.

APR'L 1994' ~~~~ ~~' Abstract. For turbulent studies in gaseous 4He at 4 K, a micronic scale anenometer with a broad band (~ l MHz ) has been developed. Typical results are presented, together with possible extensions to cryogenic flowmetry applications. Open turbulent flow experiments are govemed by the Reynolds number : file = UL/v U is the mean flow velocity measured at the same place as L which is some characteristic length for the energy injection (the diameter of a nozzle in a jet experiment for example) and v the kinematic viscosity, that is the ratio between the viscosity and the specific mass. Investigations at high file numbers (a 10~) are rather difficult under clean laboratory experimental conditions. The relatively large v for air (Tab. I) shows that the reaching of these Reynolds numbers needs decametric L sizes, that is windtunnel conditions. Low v values are of clear advantage. The use of N~ at 77 K, with a ten times lower v, has received large attention, despite the technological problems associated with the develoment of metric cryogenic wind tunnels.

International audience; In a Rayleigh-Bénard cell at high Rayleigh number, the bulk temperature is nearly uniform. The mean temperature gradient differs from zero only in the thin boundary layers close to the plates. Measuring this bulk temperature allows to separately determine the thermal impedance of each plate. In this work, the bottom plate is rough and the top plate is smooth; both interact with the same bulk flow. We compare them and address in particular the question whether the influence of roughness goes through a modification of the bulk flow.

We report experimental evidence that the Shraiman and Siggia conjecture (Phys. Rev. A, 42 (1990) 3650) on temperature boundary layer shape, in turbulent thermal convection, is correct. Besides, we also show that the spatial distributions of temperature fluctuations present exponential tails, which is compatible with a logarithmic dependence on z of the temperature fluctuation amplitudes.

We report experimental evidence of a new property of turbulent thermal convection. We show that the spectra of temperature fluctuations obtained from local and averaged measurements coincide. This new result is theoretically demonstrated by using only the hypothesis of isotropy of the turbulence field and it could be applied to the transport of passive scalars in any turbulent flow.

International audience; The flow generated by heat convection in a long, vertical channel is studied by means of particle imagery velocimetry techniques, with the help of the thermal measurements from a previous paper (Gibert et al 2009 Phys. Fluids 21 035109). We analyse the mean velocity profiles and the Reynolds stresses, and compare the present results with the previous ones obtained in a larger cell and at a larger Reynolds number.We calculate the horizontal temperature profile and the related horizontal heat flux. The pertinence of effective turbulent diffusivity and viscosity is confirmed by the low value of the associated mixing length. We study the one-point and two-point statistics of both velocity components. We show how the concept of turbulent viscosity explains the relations between the local probability density functions (pdf) of fluctuations for temperature, vertical and horizontal velocity components. Despite the low Reynolds number values explored, some conclusions can be drawn about the small scale velocity differences and the related energy cascade.

11 pages; International audience; Following a previous study [Gibert , Phys. Rev. Lett. 96, 084501 (2006)], convective heat transfer in a vertical channel of moderate dimensions follows purely inertial laws. It would be therefore a good model for convective flows of stars and ocean. Here we report new measurements on this system. We use an intrinsic length in the definition of the characteristic Rayleigh and Reynolds numbers. We explicit the relation between this intrinsic length and the thermal correlation length. Using particle imaging velocimetry, we show that the flow undergoes irregular reversals. We measure the average velocity profiles and the Reynolds stress tensor components. The momentum flux toward the vertical walls seems negligible compared to the shear turbulent stress. A mixing length theory seems adequate to describe the horizontal turbulent heat and momentum fluxes, but fails for the vertical ones. We propose a naive model for vertical heat transport inspired by the Knudsen regime in gases.

We study inhomogeneous turbulence in the presence of a strong large scale vortex. In order to understand its influence on the observed properties of the turbulence, we determine the large scale structure and dynamics of the flow, using LDV and visualisations of the motion of solid particles. We then describe the smale scales fluctuations of velocity.We show that the features of turbulence vary with the distance to the vortex,and are also influenced by its dynamics.

We have developed a small, neutrally buoyant, wireless temperature sensor. Using a camera for optical tracking, we obtain simultaneous measurements of position and temperature of the sensor as it is carried along by the flow in Rayleigh-B\'enard convection, at $Ra \sim 10^{10}$. We report on statistics of temperature, velocity, and heat transport in turbulent thermal convection. The motion of the sensor particle exhibits dynamics close to that of Lagrangian tracers in hydrodynamic turbulence. We also quantify heat transport in plumes, revealing self-similarity and extreme variations from plume to plume., Comment: 4 pages

See Also * Phys. Rev. Focus 17, story 9; International audience; We measure the relation between convective heat flux and temperature gradient in a vertical channel filled with water, the average vertical mass flux being zero. Compared to the classical Rayleigh-Bénard case, this situation has the advantage of avoiding plates and, thus, their neighborhood, in which is usually concentrated most of the temperature gradient. Consequently, inertial processes should control the convection, with poor influence of the viscosity. This idea gives a good account of our observations, if we consider that a natural vertical length, different from the channel width, appears. Our results also suggest that heat fluxes can be deduced from velocity measurements in free convective flows. This confers to our results a wide range of applications.

Experimental data of a fluid turbulence is analysed with respect to the statistical dependency of eddies of different length scales. The joint probability distributions of velocity fluctuations of two different length scales L 1, L 2 are evaluated. We quantify statistical dependency by calculating the correlation function as well as a suitably defined Kullback information.

We study the statistics of turbulent velocity fluctuations in the neighbourhood of a strong large scale vortex at very large Reynolds number. At each distance from the vortex core, we observe that the velocity spectrum has a power law ``inertial range'' of scales and that intermittency -- defined as the variation of the probability density function (PDF) of velocity increments as the length of the increment is varied -- is also present. We show that the spectrum scaling exponents and intermittency characteristics vary with the distance to the vortex. They are also influenced by the large scale dynamics of the vortex., submitted to europhys lett, 6 pages, 5 figures

We discuss the new experimental approach of turbulence offered by the use of gaseous 4He around 4 K. Both the intrinsic properties of the working fluid and the possibilities of cryogenic instrumentation give access to a wide range of control parameters. Examples of new transitions are described in open jet flow and in Rayleigh-Benard experiments.

A Rayleigh-Benard convection experiment is performed with Helium at low temperature, as the working one-phase fluid [1]. Helium is closed in a cylindrical cell 20 cm high and 10 cm in diameter. Three experimental parameters, independently adjustable, and accurately measured for each fixed condition, allow to study the Rayleigh Benard problem from the conductive regime up to the turbulent convective one. These parameters are: i) the average fluid density in the cell, measured with a 0.7% precision (from 8 g/m3 up to 135 kg/m3 as well as close to the critical point: 69.6 kg/m3, Tc = 5.1953 K, Pc = 2.275 bars); ii) the mean absolute temperature (from 2 K up to 8 K) controlled within a 0.1 mK stability and measured with a 2 mK absolute accuracy; iii) the temperature gradient along the cell height, ΔT (from less than 1 mK up to 5 K), adjusted via the heating power applied on the bottom cell plate, and measured with a 20 mK resolution, 1% uncertainty thermocouple. Two thermometers of 200.tm size and 2 mm vertically apart of each other are located in the cell, at half height and half way between the centre and the wall.

International audience; A summary of experimental results on structure functions obtained using extended self-similarity in various flow configurations (jet, grid, mixing layer, duct flow, cylinder) at Reynolds numbers ranging between 30 and 5000 is presented.

In this paper we report measurements of the heat flux in a slightly tilted channel (angle less than 45°), filled with water, that connects two chambers: the hot in the lower part and the cold on the top. We show that different regimes develop depending on the angle and the applied power. We put in evidence a hard turbulent regime, a soft turbulent regime, a laminar regime, and an intermittent one. In the last regime, the flow oscillates between laminar and turbulent, which locks the temperature gradient to a constant value. We characterize those regimes thanks to the measurement of the axial gradient of temperature and to the measurement of the power. We model them giving descriptions in term of Nusselt and Rayleigh numbers. The soft turbulence to hard turbulence transition is interpreted as the birth of the inertial range of developed turbulence. This transition, which appears in several systems, is particularly clear here, thanks to its consequences on heat transport properties.

We describe an experiment where we have measured the general features of the thermal boundary layer of turbulent thermal convection in a fluid layer heated from below. The influence of the boundary layer on the heat transport properties and on the statistics of temperature fluctuations in the bulk is also discussed.

Using an axisymmetric jet with helium at low temperatures we have studied the velocity intermittency on a large range of Taylor scale based Reynolds number Rλ. The results show for the first time evidence for a transition between two types of turbulence in a jet. Above the transition (Rλ>180) a clear separation occurs between small and large scales: in the intermediate range, the existence of a Reynolds dependent scaling law is confirmed. It illustrates in which way the viscosity can have some influence up to close to the integral scale in developed turbulent flows.

To investigate the influence of a rough-wall boundary layer on turbulent heat transport, an experiment of high-Rayleigh convection in water is carried out in a Rayleigh-Benard cell with a rough lower plate and a smooth upper plate. A transition in the heat transport is observed when the thermal boundary layer thickness becomes comparable to or smaller than the roughness height. Besides, at larger Rayleigh numbers than the threshold value, heat transport is found to be increased up to 60 %. This enhancement cannot be explained simply by an increase in the contact area of the rough surface since the contact area is increased only by a factor of 40 %. Finally, a simple model is proposed to explain the enhanced heat transport.

Recent experimental, numerical and theoretical advances in turbulent Rayleigh-Benard convection are presented. Particular emphasis is given to the physics and structure of the thermal and velocity boundary layers which play a key role for the better understanding of the turbulent transport of heat and momentum in convection at high and very high Rayleigh numbers. We also discuss important extensions of Rayleigh-Benard convection such as non-Oberbeck-Boussinesq effects and convection with phase changes.

Natural convection appears widely in nature, and remains a subject of major interest for the understanding of geophysical phenomena. The flows that are linked to it are of great magnitude and remain difficult to reproduce in the laboratory. In order to reach similar regimes in the laboratory, we propose to use fluids whose viscosity and other properties are very different from those of natural phenomena, such as cryogenic helium and fluorocarbon. The use of cryogenic helium in convection cells has made it possible to reach the highest Rayleigh numbers in the laboratories, up to 10^16, and to highlight some particular regimes. Two non-intrusive optical techniques, shadowgraphy and schlieren, have been developed and validated. In particular, shadowgraphy allows, thanks to an image correlation algorithm, to obtain the velocity field in the Rayleigh-Bénard cell. It has been possible to perform a study, in fluorocarbon, at high Rayleigh number and to propose a discussion on the transition to the "ultimate" regime of Rayleigh-Bénard convection by comparing with the literature data. In parallel, the cryogenic Rayleigh-Bénard experiment has been developed to perform the same type of measurement in liquid helium. First, a cell allowing both to generate a convection flow in liquid helium, and to allow optical measurements, has been designed and set up. One of the major motivations of the thesis is that our approach, based on the combined study of heat transfer and velocity by global resolution of the velocity field, can provide elements of answer relating to the "ultimate" regime of convection. Indeed, several recurrent hypotheses on the change of fluid behavior or plume emission statistics at the transition could be proven by optical measurements of the velocity field of this type.; La convection naturelle apparaît largement dans la nature, et demeure un sujet d’intérêt majeur pour la compréhension des phénomènes géophysiques. Les écoulements qui lui sont liés sont de grande ampleur et restent difficiles à reproduire en laboratoire. Afin d’atteindre en laboratoire des régimes semblables, on se propose d’utiliser des fluides dont la viscosité et les autres propriétés sont très différentes de celle des phénomènes naturels, comme l’hélium cryogénique et le Fluorocarbone. L’utilisation de l’hélium cryogénique dans les cellules de convection a permis d’atteindre les plus grands nombres de Rayleigh des laboratoires, jusqu’à 10^16, et de mettre en évidence certains régimes particuliers. Deux techniques optiques non intrusives, l’ombroscopie et la strioscopie, ont été mises au point et validées. En particulier, l’ombroscopie permet, grâce à un algorithme de corrélation d’image, d’obtenir le champ de vitesse dans la cellule de Rayleigh-Bénard. Il a été possible de réaliser une étude, dans le fluorocarbone, à grand nombre de Rayleigh et de proposer une discussion sur la transition au régime « ultime » de convection de Rayleigh-Bénard en comparant avec les données de la littérature. Parallèlement, l’expérience de Rayleigh-Bénard cryogénique a été développée afin de réaliser le même type de mesure dans l’hélium liquide. En premier lieu une cellule permettant à la fois de générer un écoulement de convection dans l’hélium liquide, et de permettre des mesures optiques, a été conçue et mise en place. Une des motivations majeures de la thèse est que notre approche, basée sur l’étude combinée du transfert thermique et de la vitesse par résolution globale du champ de vitesse, puisse permettre d’apporter des éléments de réponse relatifs au régime « ultime » de convection. En effet, plusieurs hypothèses récurrentes sur le changement de comportement du fluide où des statistiques d’émission de panache à la transition pourraient être prouvées par des mesures optiques du champ de vitesse de ce type.

During this thesis, we have studied turbulent thermal convection. This manuscript can be divided in three parts. In the first part, we have studied the influence of roughness on the efficiency of thermal transport in a cylindrical and in a rectangular Rayleigh Benard cell. Thanks to thermal measurements, we have observed a transition when the thickness of the thermal boundary layer is equal to the characteristic height of the roughness. After the transition, a new convective regime appears. We purpose here a model which accounts for these observations. In a second part, we have studied turbulent thermal convection in a vertical long channel. In this configuration, PIV measurements unveil that the mean flow can be divided in two columns : either the hot water goes up at the right side et the cold water goes down at the left side or the opposite. Furthermore, we have analyzed the influence of the power on the mean flow at once for the long times (time reversals of the columns) and for the short times (mean velocity profiles, fluctuations measurements and thermal plumes velocity) In a third part, a second free parameter is included : the tilted angle. PIV experiments and thermal measurements allowed us to study the competition between turbulence and stratification in the tilted long channel. A model which explains turbulent measurements is developed.; Dans ce manuscrit, nous nous sommes intéressés au phénomène de convection thermique turbulente. La première partie du manuscrit analyse l'influence de la rugosité sur l'efficacité du transport thermique dans des cellules de convection de Rayleigh-Bénard cylindrique et rectangulaire. Au vu des mesures thermiques, une transition claire apparaît lorsque l'épaisseur de la couche limite thermique est égale à la hauteur caractéristique de la rugosité. Au delà de la transition, un nouveau régime apparaît et il ne peut être expliqué par une simple augmentation de la surface d'échange de la plaque rugueuse avec le fluide. Nous proposons ici un modèle qui rend compte de ces observations. Dans la seconde partie, la convection thermique turbulente a été étudiée dans un canal vertical long. Dans cette configuration, des mesures de vélocimétrie par imagerie de particules (PIV) ont montré que l'écoulement était décomposable en deux colonnes de fluide : le fluide chaud montant sur la partie droite du canal et le froid descendant sur la partie gauche ou réciproquement. Nous avons étudié de façon systématique l'influence de la puissance injectée sur l'écoulement aux temps longs (retournement global des colonnes de fluide) et courts (obtention des profils de vitesse et de leur fluctuation et mesure de la vitesse des panaches thermiques). Dans une troisième et dernière partie, en plus de la puissance, nous ajoutons un second degré de liberté : l'angle d'inclinaison du canal. Des mesures PIV et des mesures thermiques nous ont permis d'étudier la compétition entre la turbulence et la stratification dans ce canal incliné. Un modèle rendant compte des mesures turbulentes est également proposé.

Strong vortex flows with a controlled stability are easily produced in the laboratory. We used the von Karman geometry, where a confined flow is generated inside a cylinder, in the gap between two coaxial discs. We study the interaction between vorticity and turbulence, at all scales of the energy cascade. At large scale, visualisations using air bubbles or tracer particles, and laser Doppler or hot-wire velocimetry show the existence of two flows regimes. When the discs rotate at the same speed, the flow is essentially a global rotation, with a strong axial vortex, stable in time. When angular velocities of the discs are different, the axial vortex is unstable. The turbulent cascade in these two regimes is analysed using the usual tools of turbulence, which appear to be still relevant : the cascades can be described as multiplicative, infinitely divisible, with an almost log-normal structure. Qualitatively, everything behaves as if the turbulence was homogeneous. However, the large scale inhomogeneities have quantitative effects on the evolutions of power spectra and structure functions along the scales, and they alter the cascade depth. The global rotation slows down the energy transfer rate from large to small scales, and even reverses it locally, whereas the differential rotation increases the energy transfer rate. We have also analysed experimental data using a recently proposed description of the turbulent cascade based on a Langevin modelling, both in the classical case of a turbulent jet, and then in the case of the von Karman flows. We establish links with others turbulence models and we show some limitations of the Langevin approach. In particular, the diffusion coefficient is too small to account for intermittency, as is clearly seen on the distributions of velocity increments.; Les écoulements fermés entre deux disques coaxiaux corotatifs permettent de créer facilement en laboratoire des tourbillons à grande échelle, avec une stabilité contrôlée : nous pouvons donc examiner l’interaction entre la vorticité et la turbulence, sur toute la gamme d’échelles de la cascade d'énergie. À grande échelle, des visualisations à l’aide de bulles, de billes, et des mesures de vitesse par vélocimétrie laser Doppler, et à film chaud, permettent la distinction de deux écoulements moyens. Si les deux disques tournent à la même vitesse, l'écoulement est en rotation globale, avec un tourbillon central, stable temporellement. Si les disques tournent à des vitesses différentes, le disque le plus rapide pompe le fluide au niveau de l’axe de rotation et le tourbillon est instable. Les outils habituels d’analyse de la turbulence ont été utilisés, et se révèlent toujours pertinents : les cascades sont multiplicatives, infiniment divisibles, et les noyaux des propagateurs sont pratiquement log-normaux. Qualitativement, tout se passe comme si la turbulence était homogène. Néanmoins, les inhomogénéités à grande échelle ont des effets quantitatifs sur les évolutions en fonction des échelles des spectres de puissance et des moments des incréments de vitesse, et modulent la profondeur de la cascade. La rotation globale freine le transfert d'énergie des grandes vers les petites échelles, voire l'inverse localement, tandis que la rotation différentielle l’accélère. Une description très récente de la cascade turbulente à l’aide d’une équation de Langevin est mise en œuvre sur ces signaux de vitesse et sur celui d’un jet turbulent comme typique de la turbulence homogène et isotrope. De nombreux liens avec les autres modèles de turbulence sont établis, qui mettent en évidence les limites du modèle : le coefficient de diffusion est insuffisant pour rendre compte de l’intermittence des signaux. Ceci est clairement démontré sur les distributions des incréments de vitesse.