L'étude porte sur les écoulements et les transferts en convection naturelle au sein d'empilements de produits alimentaires et plus particulièrement dans des réfrigérateurs ménagers. L'approche est à la fois expérimentale et numérique. Une première étude a été effectuée dans un empilement ordonné de sphères soumis à une convection à faible vitesse d'air (u<0.2 ms-1). L'intensité des échanges convectifs entre les objets et l'air ainsi qu'entre la paroi de la cavité et l'air a été caractérisée en fonction de la position et de la vitesse. Par ailleurs, les résultats ont montré que les autres modes de transfert (conduction et rayonnement entre les objets et avec les parois) sont du même ordre de grandeur que la convection. Des mesures de vitesse d'air (par vélocimétrie par images de particules) et de température ont été effectuées dans une enceinte fermée représentant une maquette d'un réfrigérateur ménager (à l'échelle 1) dans différentes configurations (maquette vide, maquette chargée d'empilement de produits inertes thermiquement et de produits échangeant de la chaleur). Des instationnarités dans le bas de l'enceinte vide ont été mis en évidence (Ra" 3´108). La présence d'empilements d'objets (même inerte thermiquement) modifie de façon importante les écoulements et les transferts. Les simulations numériques effectuées avec un logiciel de mécanique de fluide pour les différentes configurations étudiées sur la maquette et dans le cas d'un réfrigérateur réel ont donné des prédictions satisfaisantes en comparaison avec les mesures expérimentales. Mots clés: milieu macro-poreux, convection naturelle, coefficient de transfert convectif, conduction, rayonnement, réfrigérateur ménager, simulations numérique (CFD).

Le refroidissement par film froid consiste à refroidir une paroi en injectant un écoulement froid pour protéger celle-ci. Un banc d'essai simplifié d'une configuration d'étude de " film cooling " a été réalisé pour caractériser l'influence d'un écoulement chauffé et pulsé provenant d'un trou cylindrique incliné à 30° dans un écoulement principal thermiquement uniforme. Les taux de soufflage (M ) étudiés sont égaux à 0,65, 1 et 1,25. Les pulsations ont été générées par un hautparleur à des fréquences de pulsation adimensionnalisées ( St ) de 0, 0,2, 0,3 et 0,5. Les aspects aérodynamiques et thermiques de l'écoulement ont été étudiés, en utilisant des techniques de mesure avancées incluant la P.I.V. résolue en temps, la thermographie infrarouge, la vélocimétrie filchaud et la thermométrie fil-froid. Pour M = 0,65, sans pulsation, l'écoulement injecté couvre une partie importante de la paroi. Dans le cas pulsé, cette distribution optimale est considérablement dégradée quelque soit la fréquence de pulsation. Pour M =1 et 1,25, sans pulsation, on observe un décollement de l'écoulement qui dégrade la protection film de la paroi. Dans ces cas, la pulsation induit de grandes variations spatiales à la fois de la trajectoire, de l'éclatement du jet et du passage des structures turbulentes dans le sillage de jet à cause de la variation périodique du débit de jet. On a constaté que la pulsation améliore la protection de la paroi pour des excitations de basse fréquence pour des taux de soufflage important (M = 1 et 1,25).

Data from the GARP Atlantic Tropical Experiment (GATE) was analysed in an exploratory manner to discover the characteristics of mesoscale organization of convection and it's relationship to large-scale wind profiles. Automated methods were developed to identify the convective cells and their linear organization. These automated methods use a median high-pass filter to identify enhanced cells and a simple pattern recognition technique to ascertain the linear organization between them. Due to the simplified nature of the algorithm, the whole data set of the 21 day period from the phase 3 of GATE was processed in an economical manner obtaining a large data base which was used in the investigation of clusters and other associated phenomena. The mesoscale organization of convective cells and the widespread areas of lighter precipitation associated with them showed expected characteristics and compared satisfactorily with previous results. A large fraction of the rainfall (64%) fell from the widespread area. The total precipitation had a correlation of 0.94 with the fractional area of the widespread and a correlation of 0.89 with the fractional area of the clusters. The widespread precipitation had a correlation index of 0.97 with it's fractional area and the cluster precipitation had a strong linear relationship to it's area with a correlation of 0.99. These factors argue well for the parameterization of rainfall rate in tropical regions to a high accuracy by the area covered by organized convective cells and widespread areas. It was also seen that there was a good correlation with the number of clusters and number of cores with the total precipitation rate in a given area. These factors create a strong argument for identifying mesoscale systems consisting of convective cells and widespread precipitation as basic units of precipitation in tropical regions, having a characteristic life cycle of their own. The widespread and total precipitation showed very good correlation with upper-level vertical motion. Clusters tended to align parallel with the horizontal low-level wind shear and the degree of alignment appears to depend on the strength of the wind shear. / Graduation date: 1991

The problem of simple mathematical models of laminar natural convective flow within a long vertical parallel-plate channels and circular tubes kept at uniformly heated walls is revisited to seek a clear physical understanding of heat transfer mechanisms. A series solution method to analyze the fully developed flow and an integral solution method to analyze the developing flow are used. Chapters 3, 4, and 5 of this dissertation constitute a series of three-paper manuscripts for submission to archival journals. The channels and circular tubes considered here are assumed to be sufficiently long to yield a fully developed flow thermally as well as hydrodynamically before the exit is encountered. In such fully developed flow situation, the fluid mass flow rate naturally induced into the channel due to buoyancy is found to be a function of the wall heating condition. The predicted average Nusselt number as a function of GrPrD/L not only agrees with the existing literature but also is found to be in a functional form comparable to that proposed by Elenbaas (1942 a and b). Our results show that, in spite of being driven by buoyancy (rather than by a pump or a blower), the flow and heat transfer characteristics in the fully developed regime are essentially the same as those of fully developed laminar forced convection in which the flow is externally driven. This observation is confirmed to be valid also in the study (Chapter 5) of laminar natural convection in the developing (entrance) region within a long vertical parallel-plate channel and circular tube. The mass flow rate, which has to remain invariant with axial location even in the entry region, is determined by the flow in the fully developed region. This is the same mechanism involved in forced convection in which the fluid outside the developing boundary layers (i.e. the core flow) is forced to accelerate in the entrance region. The entrance length of channel natural convection is also discovered to be about the same as that in forced convection. / Graduation date: 2001

Microfluidic devices could find applications in many areas, such as BioMEMs, miniature fuel cells and microfluidic cooling of electronic circuitry. One of the important considerations of microfluidic device in analytical and bioanalytical chemistry is the dispersion of solute. In this study, we have developed an analytical solution, which considers the axial dispersion of a solute along the flow direction, to simulate convection and diffusion transport in a pressure driven creeping flow for a rectangular shape slit. During flow, the balance of competing effects of diffusion (especially cross-section diffusion) and convective diffusion in the flow direction are investigated. / Singapore-MIT Alliance (SMA)

La thèse porte sur l'étude de la convection capillaire (effet Marangoni) et son couplage avec la convection thermogravitationnelle. Le travail met en oeuvre des mesures de champ de vitesse par vélocimétrie laser, d'une part, et des simulations numériques de ces expériences, d'autre part. Des solutions asymptotiques sont également utilisées, et la transition de la convection stationnaire vers un état d'oscillations spatiales ou spatio-temporelles a été analysée. Le manuscript aborde également des situations impliquant l'effet de thermodiffusion, ainsi que les mouvements dans une bicouche de fluides immiscibles.

<p>QC 20120928</p>

In this thesis, several aspects of the convection, magnetic, and optical auroral dynamics of the high-latitude ionosphere are investigated from multi-instrument observations. The spatial and temporal relationships between nightside radar flow enhancements (NRFEs) and auroral intensifications are studied in Chapter 3. The NRFEs on open field lines usually are associated with very little accompanying auroral and magnetic activity. The NRFEs on closed field lines are often accompanied by optical auroral activity, but there is not a definite one-to-one correspondence. Both the statistical investigation and event study showed that the NRFEs may occur nearly simultaneously with the auroral intensifications. Because existing models associating the tail reconnection process and near-geosynchronous onset of substorms do not explain these correlated radar and optical observations very well, we propose a new model to explain the nearly simultaneous onset of the NRFEs and the auroral intensifications. In Chapter 4 we describe a small postmidnight substorm event on October 9, 2000 during dominantly IMF By+ Bz+ conditions. A sequence of three optical auroral intensifications and Pi2 bursts were found. The first two activations were characteristic of pseudobreakups, while the last and strongest intensification corresponded to a substorm expansive phase (EP). The auroral, magnetic and radar signatures of the event are interpreted as the consequence of three successive drift-Alfven-ballooning (DAB) mode instabilities in the near-geosynchronous orbit plasma sheet (NGOPS). About 10 minutes after the EP onset, there was a second auroral brightening. The convection feature during this second auroral brightening was consistent with the scenario of a Stage-2 EP. We suggest that the first two pseudobreakups, the Stage-1 EP, and the Stage-2 EP are related, respectively, to loading-unloading, directly driven, and internal magnetotail processes. Finally, in Chapter 5, we make some comparisons between the ionospheric plasma convection vortex structure observed by SuperDARN and the associated equivalent current pattern derived from the magnetometer observations. The discrepancies between the equivalent convection (EQC) and the SuperDARN-observed convection (SDC) pattern are explained in terms of the effect of day-night photoionization conductance gradient, and the coupling between field-aligned currents (FACs) and ionospheric conductances. In particular, we found the agreement between the EQC and SDC patterns is rather poor for a counterclockwise convection vortex. We suggest the discrepancies are probably due to a downward FAC-conductance coupling process.

Graduation date: 1992

The accurate prediction of warm-season convective systems, and the heavy rainfall and severe weather associated with them, remains a challenge for numerical weather prediction models. This study looks at one such circumstance in which back-building convection forms perpendicular to, and above the cold pool region behind strong bow echoes. We refer to this phenomenon as a "bow and arrow" because, on radar imagery, the two convective lines resemble an archer's bow and arrow. The "arrow" can extend over hundreds of kilometers and can cause damage from high winds, hail and flooding. Events of this nature pose a particular challenge to forecast because they require an accurate forecast of the earlier convection and the effects of that convection on the environment. In this study, radar and surface observations of four events are presented to identify common environmental conditions prior to the development of the back-building convection. Additionally, simulations of three cases using the Weather Research and Forecasting (WRF) model are analyzed in an attempt to understand the mechanisms responsible for initiating and maintaining the convective line. Due to coarse resolution, observational analyses are only useful for inspection of the synoptic-scale. Model output from numerical simulations is utilized to examine the mesoscale in the vicinity of the convective arrow. Several environmental characteristics are evident in each of the studied cases. Strong southwesterly flow (inducing warm air advection and gradual isentropic lifting), in addition to directional and speed convergence into the convective arrow region possibly contribute to convection initiation. Horizontal wind speed shear and increased wind speed in the area surrounding the arrow may be associated with the linear orientation of the arrow. It seems as though when these ingredients are combined with thermodynamic instability, there is a greater possibility of formation and maintenance of a convective arrow behind a bow echo.