Ce travail concerne une étude numérique d’écoulements incompressiblesanisothermes dans une cavité. Dans un premier temps, nous procédons à une modélisation destransferts de chaleur dans une paroi dont l’une de ses faces est recouverte d’une couche dematériau à changement de phase (MCP) de faible épaisseur. Cette modélisation est basée surune condition aux limites de type Signorini. Les équations de transfert sont résolues par uneprocédure itérative spécifique. Cette procédure est ensuite appliquée aux transferts dans unecavité différentiellement chauffée dont l’une des parois est recouverte d’une couche de MCPde faible épaisseur. Les équations qui régissent les transferts d’air sont résolues par uneméthode semi-implicite aux différences finies de second ordre et l’algorithme de projection.Nous validons la procédure en l’appliquant à la cavité entrainée, la marche descendante,l’écoulement autour d’un barreau de section carrée et la convection naturelle dans une cavitédifférentiellement chauffée. Dans un deuxième temps, une étude d’écoulements turbulentsincompressibles dans une cavité ventilée a été effectuée en utilisant un solveur de hauteprécision parallèle développée au LAMPS. Les équations de transfert sont résolues par unschéma compact aux différences finies et l’algorithme de projection. Il est montré notammentque le flux de chaleur appliqué à la paroi inférieure de la cavité influence considérablement lastructure de l’écoulement et les transferts de chaleur ainsi que les champs moyens etfluctuants de la vitesse et de la température. / The aim of this work is about a numerical study of anisothermal incompressible flowsconfined in a cavity. We perform a modeling of heat transfer in a wall where one of its faces iscovered with a thin layer of phase change material (PCM). This modeling is based on aSignorini boundary condition. The transfer equations are solved by a specific iterativeprocedure. This procedure is then applied to a differentially heated cavity, one of the walls ofwhich is covered with a thin layer of PCM. The transfer equations are solved by a semi-implicit method with finite second order differences and the projection algorithm. We validatethe procedure by applying it to the lid-driven cavity, downward motion, flow around a squaresection bar and natural convection in a differentially heated cavity. In a second step, the studyof incompressible turbulent flows in a ventilated cavity was carried out using a parallel highprecision solver developed at LAMPS. The transfer equations are solved by a finite differencecompact scheme and the projection algorithm. It is shown in particular that the heat flowapplied to the lower wall of the cavity greatly influences the structure of the flow and the heattransfers, as well as the mean and fluctuating fields of velocity and temperature.

The problem of laminar and turbulent conjugate mixed convection flow and heat transfer in shallow enclosures with a series of block-like heat generating components is studied numerically for a Reynolds number range of zero (pure natural convection) to typically 106, Grashof number range of zero (pure forced convection) to 1015 and various block-to-fluid thermal conductivity ratios, with air as the working medium. The shallow enclosure has modules consisting of heat generating elements, air admission and exhaust slots. Two problems are considered. In the first problem, the enclosure has free boundaries between the modules and in the second problem, there are partitioning walls between the different modules. The flow and temperature distributions are taken to be two-dimensional. Regions with the same velocity and temperature distributions can be identified assuming repeated placement of the blocks and fluid entry and exit openings at regular distances, neglecting end wall effects. One half of such rectangular region is chosen as the computational domain taking into account the symmetry about the vertical centreline. On the basis of the assumption that mixed convection flow is a superposition of forced convection flow with finite pressure drop and a natural convection flow with negligible pressure drop, the individual flow components are delineated. The Reynolds number is based on forced convection velocity, which can be determined in practice from the fan characteristics. This is believed to be more meaningful unlike the frequently used total velocity based Reynolds number, which does not vanish even in pure natural convection and which makes the fan selection difficult. Present analysis uses three models of turbulence, namely, standard k-ε (referred to as Model-1), low Reynolds number k-ε (referred to as Model-2) and an SGS kinetic energy based one equation model (referred to as Model-3). Results are obtained for aiding and opposing mixed convection, considering also the pure natural and pure forced convection limiting cases. The results show that higher Reynolds numbers tend to create a recirculation region of increasing strength at the core region and that the ranges of Reynolds number beyond which the effect of buoyancy becomes insignificant are identified. For instance, in laminar aiding mixed convection, the buoyancy effects become insignificant beyond a Reynolds number of 500. Results are presented for a number of quantities of interest such as the flow and temperature distributions, local and average Nusselt numbers and the maximum dimensionless temperature in the block. Correlations are constructed from the computed results for the maximum dimensionless temperature, pressure drop across the enclosure and the Nusselt numbers.